A GRAMMAR OF WAMPIS by JAIME GERMÁN PEÑA A DISSERTATION Presented to the Department of Linguistics and the Graduate School of the University of Oregon in partial ful"llment of the requirements for the degree of Doctor of Philosophy December 2015 DISSERTATION APPROVAL PAGE Student: Jaime Germań Peña Title: A Grammar of Wampis This dissertation has been accepted and approved in partial ful"llment of the requirements for the Doctor of Philosophy degree in the Department of Linguistics by: Doris L. Payne Chairperson Scott DeLancey Core Member Spike Gildea Core Member Claudia Holguín Institutional Representative and Scott L. Pratt Dean of the Graduate School Original approval signatures are on "le with the University of Oregon Graduate School. Degree awarded December 2015 ii © 2015 Jaime Germán Peña This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs (United States) License. iii DISSERTATION ABSTRACT Jaime Germań Peña Doctor of Philosophy Department of Linguistics December 2015 Title: A Grammar of Wampis This dissertation constitutes the "rst attempt at describing the grammar of Wampis (Spanish: Huambisa), a language spoken in the Peruvian Amazon. Wampis belongs to the so-called Jivaroan family of languages and is closely related to sister languages Awajun, Shuar, Shiwiar and Achuar. The grammar introduces the Wampis people and some aspects of their culture and history before analyzing the major aspects of the language from a grammatical perspective. Wampis possesses a complex prosodic system that mixes features of tone and stress. Vowel elision processes pervade most morphophonological processes. Nasalization is also present and spreads rightward and leftward through continuants and vowels. Every word in Wampis needs at least one high tone, but more can occur in a word. Morphologically, Wampis is a very rich language. Nouns and especially verbs iv have very robust morphology. A#xes and enclitics contribute di$erent meanings to words. Some morphemes codify semantic categories that are not grammatically codi"ed in many other languages, such as sudden realization, apprehensive and mirative modalities. An outstanding feature of Wampis is the pattern of argument indexation on the verb, which follows an uncommon pattern in which the verb agrees with the object (and not with the subject) if the object is a Plural Speech Act participant. Parallel to this pattern of argument indexation is the typologically uncommon pattern of object marking in Wampis, whereby a third person object noun phrase is not marked as an object if the subject is a "rst plural, second singular or second plural person. Wampis exhibits a nominative-accusative alignment. All notional objects (direct, indirect, object of applicative) are treated identically in the syntax. The preferred order is A P V. Wampis also possesses a sophisticated system of participant tracking, which is instantiated in the grammar via switch-reference markers. Another typologically uncommon feature of Wampis is the presence of a sub-system of switch-reference markers that track a participant that is not a subject. Throughout the twenty-one chapters of this grammar, other issues of Wampis related to di$erent areas of phonology, morphology and syntax are also addressed and described from a functional and a typological perspective. v CURRICULUM VITAE NAME OF AUTHOR: Jaime Germań Peña GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED: University of Oregon, Eugene Ponti"cia Universidad Católica del Perú, Lima, Perú DEGREES AWARDED: Doctor of Philosophy, Linguistics, 2015, University of Oregon Master of Arts, Linguistics, 2009, University of Oregon Bachelor of Arts, Linguistics, 2002, Ponti"cia Universidad Católica del Perú AREAS OF SPECIAL INTEREST: Morphology, Typology, Syntax, Historical Linguistics, Amazonian Languages, Languages of South America, Spanish, Documentary Linguistics, Language Contact PROFESSIONAL EXPERIENCE: Instructor, Department of Linsguistics, University of Oregon, 2014, 2010, Summer 2007, Summer 2009 Teaching Assistant, Department of Linguistics, University of Oregon 2010-2012, 2014-2015 vi Instructor, Romance Languages, University of Oregon, 2006-2010 Teaching Assistant, College of Arts-General Studies, Ponti"cia Universidad Católica del Perú, 2000-2005 GRANTS, AWARDS, AND HONORS: Hendricks Scholarship, University of Oregon, 2014–2015 Doctoral Dissertation Improvement Grant, A Grammar of Huambisa, National Science Foundation, Award #1226222, 2012-2015 Helen Martin Scholarship, University of Oregon, 2011 Clarice Krieg Scholarship, University of Oregon, 2011 Hendricks Scholarship, University of Oregon, 2010 General Scholarship, University of Oregon, 2009 Clarice Krieg Scholarship, University of Oregon, 2008 Luvaas Fellowship, College of Arts and Sciences, University of Oregon, 2008 Summer Research Award, Documentation of Huambisa and Aguaruna, Center for Latino/Latina and Latin American Studies-University of Oregon, 2008 Everett Del Monte Scholarship, College of Arts and Sciences, University of Oregon, 2007 General Development Scholarship, University of Oregon, 2007 Graduate Teaching Fellowship, 2005-2012, 2014-2015 vii PUBLICATIONS: Peña, Jaime. To appear. Constructions with has(a) in Wampis. In S. Overall, R. Vallejos, & S. Gildea (Eds.), Non-verbal predication in Amazonian languages. John Benjamins. Peña, Jaime. 2011. Review of The Motivated Syntax of Arbitrary Signs. Cognitive constraints on Spanish clitic clustering, by Erica Garcı́a. Studies in Language 34(3): 717-25 Peña, Jaime. 2011. Review of Uchumataqu: The Lost Languages of the Urus of Bolivia, by Katja Hannß. Studies in Language 33(4): 1010-16. Peña, Jaime. 2009. A historical reconstruction of Peba-Yagua linguistic family. MA Thesis, University of Oregon. Peña, Jaime. 2008. Complex clauses in Chipaya: main strategies for complementation, relativization and adverbial subordination. Proceedings of the Conference on Indigenous Languages of Latin America III, University of Texas at Austin. Payne, Doris L. & Jaime Peña (Eds.). 2007. Selected Proceedings of the 37th Annual Conference on African Linguistics. Sommerville, MA: Cascadilla Peña, Jaime. 2007. Perdere il cielo. Breve revisione dei concetti di contatto, colpa e contagio nei popoli indegeni dei Peru. Trickster, rivista del Master in Studi Interculturali, 4, Universita ́di Padova Peña, Jaime. 2006. La primera lexicografía andina: paralelismos, préstamos y reajustes semánticos en Quechua y Aimara (1532-1640). In Mar Campos, Félix Córdoba and José Ignacio Pérez Pascual (Eds.), América y el Diccionario. A Coruña, Spain: Universidade da Coruña. viii ACKNOWLEDGMENTS I wish to express my sincere appreciation to the Wampis people who taught me so much more than their language, specially to Juan Luis Nuningo Puwai and his family, Shapiom Noningo Sesén, Atilio Nuningo Noningo, Andrés Nuningo Graña, Dina Ananco Ahuananchi, Lidia Wajarai, Clara Navarro, Josué Yacum, Eder Yacum, and Óscar Jimpikit and his family. I am grateful to the Wampis leaders Juan Luis Nuningo, Julio Hinojosa and Alfonso Graña for allowing me to carry out my research and for helping me make contacts in di$erent communities of Kanus. In Jaén, my appreciation to the people of Centro Waimaku and to the Carranza family for welcoming me at their house at di$erent times while traveling for "eld work. I would like to express my gratitude to my advisor, Professor Doris L. Payne, for supporting me all these years. I really do not know where I would be without Doris’ guidance; certainly not "nishing a dissertation. She has very patiently read and re-read through every piece of this work and has given my chaotic thoughts a legible form. Our conversations have greatly enriched my understanding not only of language, but of how to do research in general. Scott DeLancey and Spike Gildea are great professors and scholars, and great people. As their student or assistant, I have learned a lot from them. I am honored to have them in my committee and grateful for their comments and criticism. I am grateful to Claudia Holguín for supporting my project since we met for the "rst time and for her comments on my work. ix All of my friends at the Department of Linguistics of the University of Oregon have helped me keep my sanity at acceptable levels, especially Rosa Vallejos (and Óscar and Álvaro), Michael Ahland, Yolanda Valdez. Shahar Shirtz, Manuel Otero, Danielle Barth, Ellen Kress and Matt Stave provided insightful comments on the Wampis data or helped with my questions about English. My thanks also to all of my professors and sta$ at the Department of Linguistics and to the people at Romance Languages, with whom I worked when I started my graduate career in Oregon. My research was supported by the Department of Linguistics, a 2008 Summer Research Grant by the Center for Latino/ Latina and Latin American Studies, and a Doctoral Dissertation Improvement Grant awarded by the National Science Foundation (Award 1226222). I would be nowhere without the love and support of my parents, Isaac Jaime and Zadith. My sister Ketty has been my best friend since before I was born, and I am always happy to see her, her husband Yuri and my nephew Iván. My in-laws are always supportive of our endeavors, I am especially grateful to Teresa Arellano for everything she has done for us. I am indebted to Adriana Da Silva, who has helped me greatly with the maps. Adriana not only found time to help me, but also to be a great mother to our sons Joaquín Alejandro, Nicolás Jaime and Octavio Carlo. All of you are the greatest thing to ever happened upon me. Language fails me to express how much I love you all. x To the Wampis People. To the memory of the dear people who left this physical world while I was writing this dissertation, especially my grandmother Migue and my Oregonian brother James D. Wood. To Adriana, to my parents, to Joaquín, Nicolás and Octavio. xi TABLE OF CONTENTS Chapter Page I. INTRODUCTION 1................................................................................................... 1.1. Introduction 1...................................................................................................... 1.2. The Jivaroan Linguistic Family 3......................................................................... 1.3. Other Proposed Jivaroan Connections and Language Contact 8.......................... 1.4. Previous Studies of Jivaroan Languages 12......................................................... 1.5. Previous Studies about Wampis 14...................................................................... 1.6. Socio-Linguistic Situation 16............................................................................... 1.6.1. Varieties of Wampis 17............................................................................... 1.6.2. Auto-Denomination 21................................................................................ 1.7. Non-Verbal Communication 25........................................................................... 1.8. This Dissertation 28............................................................................................. 1.9. Theoretical Framework 29................................................................................... 1.10. Methodological Considerations: Fieldwork and Data 34.................................. 1.10.1. Fieldwork 34.............................................................................................. 1.10.2. Data, Data Gathering and Processing 45................................................... 1.11. Typological Pro"le: The Wampis Language at a Glance 52............................... II. THE WAMPIS PEOPLE 56...................................................................................... 2.1. Introduction 56.................................................................................................... 2.2. Brief Ethnohistory of the Wampis 57.................................................................. 2.3. Current Situation: The Wampis Today 69............................................................ 2.4. Material Culture 71.............................................................................................. 2.5. Tsantsa ‘Shrunken-Head’ 77................................................................................. 2.6. Economy and Subsistence 78............................................................................... 2.7. Traditional Beliefs and Religion 79.................................................................... 2.7.1. Evangelization 84........................................................................................ 2.8. Oral Tradition and Folklore 85............................................................................ 2.9. Cultural Transmission, Social Life and Organization 86...................................... xii Chapter Page III. SEGMENTAL PHONOLOGY AND PHONETICS 94............................................... 3.1. Introduction 94.................................................................................................... 3.2. Consonant Phonemes 95...................................................................................... 3.2.1. Oppositions 97............................................................................................. 3.2.2. Stops 100.................................................................................................... 3.2.2.1. Voiceless Bilabial Stop /p/ 100........................................................... 3.2.2.2. Voiceless Dental Stop /t/̪ 101............................................................. 3.2.2.3. Voiceless Velar Stop /k/ 102.............................................................. 3.2.2.4. Glottal Stop /ʔ/ 102............................................................................ 3.2.3. Nasal Consonants 103.................................................................................. 3.2.3.1. Bilabial Nasal /m/ 104........................................................................ 3.2.3.2. Alveolar Nasal /n/ 105....................................................................... 3.2.3.3. Marginal Consonantal Nasal Phoneme /ɲ/ 105................................ 3.2.3.4. Nasal Consonants in Coda and Word-Final Positions 106................... 3.2.4. Fricatives 111.............................................................................................. 3.2.4.1. Alveolar Fricative /s/ 111................................................................... 3.2.4.2. Palato-Alveolar Fricative /ʃ/ 112........................................................ 3.2.4.3. Glottal Fricative /h/ 112..................................................................... 3.2.5. A$ricates 114.............................................................................................. 3.2.5.1. Voiceless Alveolar A$ricate /t͡s/ 115.................................................. 3.2.5.2. Voiceless Palato-Alveolar /t͡ʃ/ 116...................................................... 3.2.6. The Rhotic /r/ 116...................................................................................... 3.2.7. Marginal Palatal /j/ 117.............................................................................. 3.3. Consonant Allophonic Variation 119.................................................................. 3.3.1. Bilabial Labialization 120............................................................................ 3.3.2. Consonant Palatalization 120...................................................................... 3.3.3. A$ricativization of /ʃ/ 128.......................................................................... 3.3.4. Dissimilation of A$ricates 128.................................................................... 3.4. Optional Variation of Consonants 129................................................................. 3.4.1. Fricativization of /t͡ʃ/ 129........................................................................... 3.4.2. /h/ → [f] 130.............................................................................................. xiii Chapter Page 3.5. Distribution of Consonants 131........................................................................... 3.6. Vowel Phonemes 135........................................................................................... 3.6.1. Vowel Quality 136....................................................................................... 3.6.2. Vowel Oppositions 138................................................................................ 3.6.3. Oral Versus Nasal Contrasts 139.................................................................. 3.6.4. Distribution of Vowel Clusters 139.............................................................. 3.7. Vowel Allophony 140.......................................................................................... 3.7.1. Assimilation of /a/ 140.............................................................................. 3.7.2. Lower Allophones of High Vowels 142........................................................ 3.7.3. Glides 142.................................................................................................... 3.7.3.1. Fricativization of [w] 143................................................................... 3.8. Other Hypo-Articulatory (Casual Speech) Phenomena 144................................. 3.8.1. De-Palatalization 144.................................................................................. 3.8.2. Palatalization of [w] and /h/ 145............................................................... 3.8.3. Alternation [mʷɨ] ~ [mu] 145.................................................................... 3.8.4. Lenition of /ʃ/ 145...................................................................................... 3.8.5. On the Voicing of Stops 146........................................................................ 3.8.6. On Devoicing of Vowels 148....................................................................... 3.9. Loanwords 150.................................................................................................... 3.9.1. Spanish Loanwords 150............................................................................... 3.9.2. Quechua Loanwords 153............................................................................. 3.9.3. Loanwords from Other Languages 155........................................................ 3.10. Notes on Transcription 157............................................................................... IV. ALLOPHONIC GLIDE DERIVATION, EPENTHETIC GLIDE INSERTION AND PHONOTACTICS 159........................................................................................... 4.1. Introduction 159.................................................................................................. 4.2. Derivation and Insertion of Glides 160................................................................ 4.2.1. Importance of the Ordering of Rules 164.................................................... 4.2.2. Properties of Glides 168.............................................................................. 4.2.3. Epenthetic Versus Allophonic Glides 169.................................................... xiv Chapter Page 4.2.4. On the Approximant [ɰ] 173...................................................................... 4.3. Phonotactics 174.................................................................................................. 4.3.1. Syllable Structure in Wampis 175............................................................... 4.3.2. Phonotactic Restrictions 189....................................................................... V. MORPHOPHONOLOGY 192.................................................................................. 5.1. Introduction 192................................................................................................. 5.2. Morpheme Boundary 193.................................................................................... 5.3. Reduction of the Last Vowel of a Stem 194........................................................ 5.3.1. Set I Nominalizer -inu 195........................................................................... 5.3.2. Set II Nominalizer -u 198............................................................................. 5.3.3. High A$ectedness -a(́u) 201........................................................................ 5.3.4. Low A$ectedness -i 205............................................................................... 5.3.5. Imperfective -a 208...................................................................................... 5.3.6. Third Person Perfective -i~̃-ɨ ̃ 210................................................................ 5.4. Vowel (Mora) Reduction of the Su#x\Clitic at Morpheme Boundary 210.......... 5.4.1. Plural Imperfective -ina 212........................................................................ 5.4.2. Non-Imperfective Plural -ara 212................................................................ 5.4.3. Declarative -i 214........................................................................................ 5.4.4. Apprehensive -ai 215................................................................................... 5.4.5. Sudden Realization and Tag Question =api 215......................................... 5.5. Vowel Harmony 216............................................................................................ 5.5.1. Restrictive =kɨ 216..................................................................................... 5.5.2. Do While Moving -ki(ni) 218...................................................................... 5.6. Di$erent Forms of Su#xes or Clitics 219............................................................ 5.6.1. Locative =nVma 220................................................................................... 5.6.2. High A$ectedness -a(́u) 221........................................................................ 5.6.3. Do While Moving -ki(ni) 221....................................................................... 5.6.4. Locative =(n)i ̃and Di$erent Subject -(n)i ̃ 222........................................... 5.6.5. Copula Clitics 223........................................................................................ 5.7. Other Phenomena 228......................................................................................... xv Chapter Page 5.7.1. Degemination of Stops 228.......................................................................... 5.7.2. Optional Gemination of Consonants 229..................................................... 5.7.3. Optional Metathesis 229.............................................................................. 5.8. Vowel-Switching for Marking Possession 231.................................................... VI. PROSODY 232..................................................................................................... 6.1. Introduction 232.................................................................................................. 6.2. Previous Works and Preliminary Notions 232..................................................... 6.3. The Mora and Mora Requirement 237................................................................. 6.4. Stress in Wampis 240........................................................................................... 6.4.1. Stress and Tone in Bimoraic Words in Wampis 241.................................... 6.4.2. Stress in Words with More than Two Moras 242......................................... 6.4.3. Acoustic Correlates of Stress 248................................................................. 6.4.4. Vowel Elision 251........................................................................................ 6.4.5. Word-Final and Word-Internal Vowel Elision 252....................................... 6.4.5.1. Elision of CV in N.CV Sequence 257.................................................. 6.4.5.2. Special Vowel Elision in CVV Sequences 259..................................... 6.4.5.3. Immunity to Elision 261..................................................................... 6.5. Tone in Wampis 265............................................................................................ 6.6. Induced Metrical High Tone Shift in Non-Predicates 269.................................... 6.6.1. Two and Three Mora Stems 270.................................................................. 6.6.2. Special Tone E$ect with -t͡ʃi ‘Diminutive’ and =nVma ‘Locative’ 271......... 6.6.3. Four Mora Stems 275.................................................................................. 6.6.4. Nominal Morphemes that Have Lexical Tone 275....................................... 6.6.5. Grammatical High Tone 278...................................................................... 6.6.6. Distribution of Tone in Predicates 280........................................................ 6.7. Summary of Stress and Tone 283........................................................................ 6.8. The Prosodic Word 284....................................................................................... 6.9. Notes on Utterance-Level Prosody 286................................................................ 6.9.1. Declarative Intonation Contour 286............................................................ 6.9.2. Non-declarative Intonational Contour 287.................................................. xvi Chapter Page 6.9.3. Prosody of the anemamu, the Protocolar Salutation Ceremony of the Wampis 289....................................................................................... 6.10. Nasal Prosody in Wampis 291........................................................................... 6.10.1. Alternation between Ṽ and VN and between Ṽ and V 292........................ 6.10.2. Nasality as High Pitch 294........................................................................ VII. INTRODUCTION TO MORPHOLOGY 296............................................................ 7.1. Introduction 296.................................................................................................. 7.2. Wampis Morphological Typology 297................................................................. 7.3. Root, Stem and Word 298.................................................................................... 7.4. Grammatical Categories and Roots 302............................................................... 7.5. Fuzzy Boundaries and Clear-Cut Boundaries 308................................................ 7.6. Bound and Free Morphemes 313......................................................................... 7.6.1. A#xes: Derivation and In/ection 313......................................................... 7.6.2. Clitics 319.................................................................................................... 7.7. Reduplication 321................................................................................................ 7.8. Compounds 325................................................................................................... 7.9. Particles 328........................................................................................................ VIII. CLOSED WORD CLASSES 329............................................................................ 8.1. Introduction 329.................................................................................................. 8.2. Pronouns 330....................................................................................................... 8.2.1. Personal Pronouns 330................................................................................ 8.2.1.1. Combining Forms of 1sg and 2 Personal Pronouns 334...................... 8.2.1.2. Genitive Forms of Personal Pronouns 335.......................................... 8.2.1.3. Use of 3 Person Pronouns as De"nite Articles 336............................. 8.2.2. Cataphoric Hesitation Pronoun naa 337...................................................... 8.3. Demonstratives 339............................................................................................. 8.3.1. Demonstratives and Agreement 343............................................................ 8.3.2. The Similative Constructions with -mamtin 346......................................... 8.3.3. Other Functions of Demonstratives 349...................................................... xvii Chapter Page 8.3.4. Resumptive Verbs from Demonstratives 353............................................... 8.4. Tikit ͡ʃi ‘Other’ 353................................................................................................. 8.5. Interrogative Words 354...................................................................................... 8.5.1. Morphosyntactic Properties of Interrogative Words 358............................. 8.5.2. Interrogative Words in Non-Interrogative Constructions 373...................... 8.6. Adverbs 375......................................................................................................... 8.6.1. Manner Adverbs 376................................................................................... 8.6.2. Time Adverbs 385........................................................................................ 8.6.2.1. Su#x -tin ‘Time’ 388........................................................................... 8.6.3. Location adverbs 389................................................................................... 8.6.4. aía and aíatɨkɨ ‘Only’ 391............................................................................ 8.7. Non-Numeral Quanti"ers 393.............................................................................. 8.8. Numerals 397....................................................................................................... 8.8.1. Morphological Properties of Numerals: Distributive Reduplication and -a ́‘Iterative’ 403.................................................................................... 8.9. Noun Modi"er uɨántu ‘Group’ 405....................................................................... 8.10. Conjunctions 407............................................................................................... 8.11. Interjections 412................................................................................................ 8.12. Ideophones 414.................................................................................................. 8.13. Grammatical Particles 420................................................................................. IX. ADJECTIVES 422.................................................................................................. 9.1. Introduction 422.................................................................................................. 9.2. De"ning the Adjective in Wampis 423................................................................ 9.3. Adjectives vs Verbs 426....................................................................................... 9.4. Adjectives vs Nouns 426...................................................................................... 9.5. Morphology Associated with Adjectives 431....................................................... 9.5.1. Diminutive -t͡ʃi with Adjectives 432............................................................. 9.5.2. The Negative Nominalizer -t͡ʃau and Adjectives 432.................................... 9.5.3. Adjectives and Case and Post-Positional Clitics 433.................................... 9.6. The Adjective Phrase (AdjP) 436........................................................................ xviii Chapter Page 9.7. Syntactic Operations of Adjectives 438............................................................... 9.7.1. Adjectives as Modi"ers of Head of NP 439.................................................. 9.7.2. Adjectives in Copular and Non-Verbal Constructions 439........................... 9.7.3. Gradability 440............................................................................................ 9.7.3.1. Adjectives in Comparative Constructions 441..................................... 9.7.3.2. Comparison of Equality 443................................................................ 9.7.3.3. “Less” Comparison 444....................................................................... 9.7.3.4. Adjectives in Superlative Constructions 445....................................... 9.8. Derivation of Adjectives 446................................................................................ 9.8.1. Zero Derivation 446..................................................................................... 9.8.2. Derivation with -taku 447........................................................................... 9.8.3. Derivation with -rama 449........................................................................... 9.8.4. Unproductive Adjectivalizer -patinu 450..................................................... 9.8.5. Augmentative =hakin 450.......................................................................... 9.9. Semantic Types of Adjectives 451....................................................................... 9.10. Discoursive Functions of Adjectives 452............................................................ X. THE NOUN AND NOUN MORPHOLOGY 455........................................................ 10.1. Introduction 455................................................................................................ 10.2. De"ning the Noun in Wampis 455.................................................................... 10.3. Noun Classes 457............................................................................................... 10.3.1. Proper Nouns 457...................................................................................... 10.3.2. Kinship Terms 458..................................................................................... 10.3.3. Location Nouns 460................................................................................... 10.3.4. Compound Nouns 461............................................................................... 10.4. Morphology 462................................................................................................ 10.4.1. Marking on the Possessed Noun 464......................................................... 10.4.1.1. Type I Nouns 464.............................................................................. 10.4.1.2. Type II Nouns 465............................................................................. 10.4.1.3. Possession Marking of Derived Diminutive Nouns 468.................... 10.4.1.4. Possession Marking of Idiosyncratic Nouns 468.............................. xix Chapter Page 10.4.1.5. Arbitrariness of the Type I/Type II Distinction 469.......................... 10.4.2. Diminutive -t͡ʃi 472..................................................................................... 10.4.3. Attributive -tinu 475.................................................................................. 10.4.4. Benefactive -nau 477................................................................................. 10.4.5. Negative Nominalizer -t͡ʃau 480................................................................. 10.4.6. Case 481................................................................................................... 10.4.6.1. Nominative 482................................................................................ 10.4.6.2. Accusative =na 483......................................................................... 10.4.6.2.1. Accusative =na in Ditransitive Constructions 486............... 10.4.6.3. Genitive 486...................................................................................... 10.4.6.3.1. Possible Historical Development of the Genitive 488........... 10.4.6.4. Vocative 491..................................................................................... 10.4.6.4.1. Vocative Forms of Kinship Terms 492.................................. 10.4.6.4.2. Vocative Form of Foreign Names 493.................................. 10.4.6.4.3. Other Cultural/Familiar Vocatives and Terms to Address People 494............................................................... XI. THE NOUN PHRASE AND POSTPOSITIONS 496.................................................. 11.1. Introduction 496................................................................................................ 11.2. The Noun Phrase (NP) in Wampis 496.............................................................. 11.3. NP Constructions 498........................................................................................ 11.3.1. The Determined NP 498............................................................................ 11.3.2. The Possessive Construction 498............................................................... 11.3.3. The Attributive Construction 499.............................................................. 11.3.4. The Plural Construction 501...................................................................... 11.3.5. Construction with uɨantu ‘Group’ 503........................................................ 11.4. Case 505............................................................................................................ 11.5. Postpositional Clitics and the Marking of Oblique Relations 505...................... 11.5.1. Locatives 506............................................................................................. 11.5.1.1. Locative =nVma 508........................................................................ 11.5.1.2. Locative =(n)i ̃ 512........................................................................... xx Chapter Page 11.5.1.3. Locative V́ 513.................................................................................. 11.5.2. Ablative =ia 516....................................................................................... 11.5.3. Allative =ni 520....................................................................................... 11.5.4. Comitative =hai ̃ 521................................................................................. 11.5.5. Comitative =tuk 525................................................................................. 11.5.6. Plural Speech Act Participant =tí 526...................................................... 11.5.7. First =á 529.............................................................................................. 11.5.8. A$ective =a ́ 529....................................................................................... XII. INTRODUCTION TO THE VERB: CLASSES, VERB DERIVATION, STRUCTURE AND CONJUGATION PATTERNS 531............................................ 12.1. Introduction 531................................................................................................ 12.2. De"ning the Verb in Wampis 532...................................................................... 12.3. Valency and Transitivity: Verb Classes 533....................................................... 12.3.1. Labile Verbs 536........................................................................................ 12.4. Auxiliary Verbs 538........................................................................................... 12.5. Other Lexical Types of Verbs 538...................................................................... 12.5.1. Lexically Negative Verbs 538.................................................................... 12.5.2. Verbs that Lexicalize a Speci"c Type of Patient or Location 539.............. 12.6. Verb Derivation 540.......................................................................................... 12.6.1. Zero Derivation 540................................................................................... 12.6.2. Denominal Verbalizer -ma 541.................................................................. 12.6.3. Other Unproductive Verbalizers 542......................................................... 12.6.4. Derivation with -tika and -ni: Resumptive Verbs 543................................ 12.7. Basic Structure of the Wampis Verb 547........................................................... 12.8. Verb Conjugation Patterns 553.......................................................................... 12.8.1. Imperfective Conjugations 554.................................................................. 12.8.1.1. Imperfective Conjugation I 554........................................................ 12.8.1.2. Imperfective Conjugation II 556....................................................... 12.8.1.3. Imperfective Conjugation III 558...................................................... 12.8.1.4. Derived Verb Conjugations 559........................................................ xxi Chapter Page 12.8.2. Perfective Conjugations 560...................................................................... 12.8.2.1. Perfective Conjugation I 560............................................................. 12.8.2.2. Perfective Conjugation II 562........................................................... 12.8.2.3. Perfective Conjugation III 563.......................................................... XIII. THE VERB I: “DERIVATIONAL” LEVEL 566..................................................... 13.1. Introduction 566................................................................................................ 13.2. Valence Operators 566....................................................................................... 13.2.1. Valence-Decreasing Morphology 567........................................................ 13.2.1.1. Detransitivizer -na 568...................................................................... 13.2.1.2. “Placeholder” -na 569....................................................................... 13.2.1.3. Detransitivizer -pa 571...................................................................... 13.2.1.4. Re/exive -ma(ma) 571...................................................................... 13.2.1.5. Reciprocal -nai 576........................................................................... 13.2.2. Valence-Increasing Operators 577............................................................. 13.2.2.1. Causatives 577.................................................................................. 13.2.2.1.1. Causative V- 579................................................................... 13.2.2.1.2. Causative -mitika 582......................................................... 13.2.2.1.3. Causative -ka 583................................................................. 13.2.2.2. Transitivizer -ki 584.......................................................................... 13.2.3. Applicative -ru ~ -tu 584........................................................................... 13.2.3.1. Homophony between the Applicative Forms and the 1sg Object Markers 592............................................................. 13.2.3.2. Double Marking of Applicative? 593................................................ 13.2.4. Object Marking on the Verb 594............................................................... 13.2.4.1. First Person Object 595..................................................................... 13.2.4.2. Second Person Object: 3>2 600...................................................... 13.2.4.3. Second Person Object: 1>2 602....................................................... 13.2.4.4. A Paradigm of Object Marking 604................................................... 13.2.4.5. Applicative and Objects 606............................................................. 13.2.5. The Su#x -nin ‘Almost’ 608...................................................................... xxii Chapter Page 13.3. Verbal Slot 3: Aktionsart, Imperfective, Durative, Present Habitual and Potential Su#xes 609................................................................................. 13.3.1. Overview 609............................................................................................ 13.3.2. Aktionsart 611........................................................................................... 13.3.2.1. High A$ectedness Aktionsart -a(́u) 619............................................ 13.3.2.2. Low A$ectedness Aktionsart -i 620................................................... 13.3.2.3. Intensive Aktionsart -ka 623............................................................. 13.3.2.4. Distributed Action Aktionsart -ra 624............................................... 13.3.2.5. Attenuative Aktionsart Su#x -sa 626............................................... 13.3.2.6. Do Action While Moving -ki(ni) 629................................................. 13.3.2.7. Do in Proximity -ri 630..................................................................... 13.3.2.8. Do in Another Location -u 631.......................................................... 13.3.3. Imperfective -a and Plural Imperfective -ina 631...................................... 13.3.4. The Durative -ma 633............................................................................... 13.3.5. The Present Habitual -na 634.................................................................... 13.3.6. Potential -mai 635..................................................................................... 13.3.7. Verb Stems: “Aspect” 636.......................................................................... 13.4. Negation 637..................................................................................................... 13.5. Non-Imperfective Plural -ara 639...................................................................... XIV. THE VERB II: “INFLECTIONAL” LEVEL 641...................................................... 14.1. Introduction 641................................................................................................ 14.2. Tense 643........................................................................................................... 14.2.1. Present Tense 644..................................................................................... 14.2.2. Past Tense 644........................................................................................... 14.2.2.1. Just-Done Actions 646...................................................................... 14.2.2.2. Recent Past 647................................................................................. 14.2.2.3. Intermediate Past 648....................................................................... 14.2.2.4. Distant Past 650................................................................................ 14.2.2.5. Remote Past 651............................................................................... 14.2.2.6. Habitual Past 652............................................................................. xxiii Chapter Page 14.2.3. Future Tense 654....................................................................................... 14.2.3.1. Immediate Future 654....................................................................... 14.2.3.2. De"nite Future 655........................................................................... 14.3. Person 656......................................................................................................... 14.3.1. 1 and 2 Person S/A 657............................................................................. 14.3.2. Third Person S/A Marking 659.................................................................. 14.3.3. The Typologically Uncommon Argument Indexation Pattern of Wampis 662........................................................................................... 14.3.3.1. Indexation of S Arguments 666......................................................... 14.3.3.2. Indexations of Arguments in Transitive Clauses 667........................ 14.4. Mood 670........................................................................................................... 14.4.1. Declarative 670......................................................................................... 14.4.2. Exclamative 671........................................................................................ 14.4.3. Commands and Manipulative Mood: Imperative, Jussive, Hortative and Prohibitive 672.................................................................... 14.4.3.1. Imperative 672.................................................................................. 14.4.3.2. Familiar Imperative 674.................................................................. 14.4.3.3. Jussive 676........................................................................................ 14.4.3.4. Hortative 677.................................................................................... 14.4.3.5. Prohibitive -pa 678............................................................................ 14.4.4. Apprehensive -ai 679................................................................................. 14.4.5. Inferential -tai 681..................................................................................... 14.4.6. Narrative Modality timaji 683................................................................... XV. NOMINALIZATION 686...................................................................................... 15.1. Introduction 686................................................................................................ 15.2. The Nominalized Verb 687................................................................................ 15.3. Nominalizers in Wampis 692............................................................................. 15.4. Derivational Nominalization 695....................................................................... 15.4.1. Set I -inu, ‘Agentive Nominalizer’ 696....................................................... 15.4.2. Set I -tinu, ‘Future Nominalizer’ 698.......................................................... xxiv Chapter Page 15.4.3. Set I -ta, ‘Action Nominalizer’ 700............................................................. 15.4.4. Set I Nominalizer -tai,̃ ‘Non-Agentive Nominalizer’ 701............................ 15.4.5. Set II Nominalizer -u, ‘Subject Nominalizer’ 704....................................... 15.4.6. Set II Nominalizer -mau, ‘Non-Subject Nominalizer’ 706.......................... 15.4.7. Negative Nominalizer -t͡ʃau 708................................................................. 15.4.8. Nominalizer -na 708.................................................................................. XVI. SINGLE VERB CONSTRUCTIONS WITH LEXICAL VERBS, AND NOTES ON WORD ORDER AND ALIGNMENT 710................................... 16.1. Introduction 710................................................................................................ 16.2. Notes on Constituent Order 710........................................................................ 16.3. Alignment and Grammatical Relations 711....................................................... 16.4. Intransitive Constructions 720........................................................................... 16.4.1. Existential Construction 721...................................................................... 16.4.2. Weather Constructions 722....................................................................... 16.5. Copular Constructions 723................................................................................ 16.6. The Transitive Construction 723........................................................................ 16.7. The Ditransitive Construction 725..................................................................... 16.8. Quotative Construction 726............................................................................... 16.9. The Possession Predication Construction 727.................................................... XVII. POSSESSIVE, EXISTENTIAL, LOCATIONAL, ATTRIBUTIVE AND EQUATIONAL CLAUSES 728..................................................................... 17.1. Introduction 728................................................................................................ 17.2. Overview of Functions of Non-Verbal Predication 728.................................... 17.3. Overview of the Structure of Non-Verb Lexical Predications in Wampis 729.... 17.4. Non-Verbal Clauses 731..................................................................................... 17.4.1. Juxtaposition Constructions: Equative, Proper Inclusion, Attributive 731........................................................................................... 17.4.2. Negation in Juxtaposition Constructions 733............................................ 17.4.3. The Construction NP + NP-tinu: Possession 735....................................... xxv Chapter Page 17.4.4. Tense Reference and Juxtaposition 736.................................................... 17.5. Copular and Existential Predications 737.......................................................... 17.5.1. Brief Overview of the Copulas and the Existential 737............................. 17.5.2. Morphosyntactic Distinctions 739............................................................. 17.5.2.1. The Expression of TAM Categories 739............................................. 17.5.2.2. The Marking of Plurality in Non-Verbal Predicates 744................... 17.5.2.3. The Marking of Negation in Verbal Predicates 747.......................... 17.5.2.4. Copular and Existential Predicates in Subordinate Clauses 749....... 17.5.3. Functions of the Copula a, Copula Clitics and the Existential a 751......... 17.5.3.1. The Copula a 751.............................................................................. 17.5.3.2. The Copula Clitics 753...................................................................... 17.5.3.3. Existential Clauses 757...................................................................... 17.5.3.3.1. Possession with Existential a + Applicative 758.................. 17.5.3.3.2. Predicate Locatives 760....................................................... 17.6. Other Copulative Verbs 767............................................................................... XVIII. DISCOURSE-ORIENTED AND MOOD PHRASE-LEVEL CLITICS 769............... 18.1. Introduction 769................................................................................................ 18.2. Overview 769.................................................................................................... 18.3. Restrictive =kɨ 770........................................................................................... 18.4. Focus =ka 776.................................................................................................. 18.5. Additive =ʃa 781............................................................................................... 18.6. Speculative =ʃa 782.......................................................................................... 18.7. Inferential =t͡su 782.......................................................................................... 18.8. Interrogative =ka 784....................................................................................... 18.9. Sudden Realization =api and Tag Question =api 785...................................... 18.10. Mirative 787................................................................................................... XIX. SWITCH-REFERENCE AND SUBORDINATION 792........................................... 19.1. Introduction 792................................................................................................ 19.2. The Subordinate Verb 794................................................................................. xxvi Chapter Page 19.2.1. Morphological Structure of the Subordinate Verb 798.............................. 19.3. Subordinators 800............................................................................................ 19.4. Person Marking 804........................................................................................... 19.4.1. A Note on the Nasal Marking of Person in Subordinate Verbs 806........... 19.4.2. Di$erent Subject -(n)i ̃ 807......................................................................... 19.5. Non-Temporal Subordination 809..................................................................... 19.5.1. Same Subject -sa 809................................................................................. 19.5.2. 1/3 Di$erent Subject -tai ̃ 812.................................................................... 19.6. Simultaneous Subordination -ku 814................................................................. 19.6.1. Same-Subject Simultaneous Subordinate Clauses 815............................... 19.6.2. Di$erent Subject in Simultaneous Clauses 817........................................ 19.7. Sequential Subordinate Clauses ∅ 818............................................................. 19.7.1. Same-Subject Sequential Clauses 820....................................................... 19.7.2. Di$erent Subject Sequential Clauses 823.................................................. 19.8. Imperfective Di$erent Subject Subordination ∅ 824........................................ 19.9. Same-Subject Only Subordinators 825............................................................... 19.9.1. Frustrative -tahkama ́ 826........................................................................... 19.9.2. Terminative -hkama ́ 829............................................................................ 19.9.3. Purpose -tasa 832....................................................................................... 19.9.4. Reduplicative -kaua 834............................................................................ 19.10. Non-Canonical Switch-Reference 836.............................................................. 19.10.1. Non-Subject to Subject -ma 838............................................................... 19.10.2. Subject to Object -tatamana 840.............................................................. 19.11. Conditional Clauses with -ka 842.................................................................... 19.12. Concessive Clauses with -ʃa 846...................................................................... 19.13. A Brief Note on “Clause-Chaining” 848........................................................... XX. RELATIVIZATION, COMPLEMENTATION AND OTHER COMPLEX CONSTRUCTIONS 851......................................................................................... 20.1. Introduction 851................................................................................................ 20.2. Relativization 851.............................................................................................. xxvii Chapter Page 20.2.1. Relative Clauses with Encliticized Demonstrative 852.............................. 20.2.2. Relatives Clauses with Set II Nominalizers -u and -mau 853..................... 20.2.2.1. Externally and Internally-Headed Relative Clauses 854................... 20.2.2.2. Headless Relative Clauses 858.......................................................... 20.3. Complementation 860....................................................................................... 20.3.1. Complementation Using Nominalization and Subordination 861............. 20.3.2. Desiderative Constructions 863................................................................. 20.3.3. Quotative Construction 865....................................................................... 20.4. Other Adverbial Constructions Involving Nominalization 865.......................... 20.5. “Tautological In"nitive” Construction with -tai ̃ 866.......................................... XXI. THIS GRAMMAR IN BROADER CONTEXT 870................................................. APPENDICES A. ABBREVIATIONS 882...................................................................................... B. SELECTED TEXTS 885..................................................................................... REFERENCES CITED 907............................................................................................ xxviii LIST OF FIGURES Figure Page 1.1. Map of the Jivaroan Languages 4......................................................................... 1.2. Subgrouping of the Jivaroan Family Based on Stark (1985) 7............................. 1.3. Map of Fieldwork Sites 36.................................................................................... 3.1. Non-Local Palatalization of an /n/ after an /in/ Sequence of a Preceding Syllable 124......................................................................................... 3.2. Spectrogram Showing a Surface Palatal Nasal [ɲ] in Word Boundary 126......... 3.3. Plot-Chart of Wampis Vowel Acoustic Space 138................................................ 3.4. Spectrogram of [wiśuta̪] Showing a Very Reduced Vowel at the End 150.......... 4.1. Syllable Structure at an Intermediate Stage (Before Vowel Elision) 176............. 4.2. Maximal Syllable Structure in Wampis 182......................................................... 6.1. Spectrogram Showing Pitch Contour of [wa ̃á] ‘hole’ 238.................................... 6.2. Spectrogram Showing Pitch Contour of [waãńam] ‘in the hole’ 239................... 6.3. Parsing Directionality in Wampis 243.................................................................. 6.4. Spectrogram of Framed Token of [tikiʃ́kiʃa] ‘one more’ 251............................... 6.5. Spectrogram Showing the Pitch Track for [túkɨ] ‘whole’ 267.............................. xxix Figure Page 6.6. Spectrogram Showing the Pitch Track for [tukɨ]́ ‘always’ 268............................. 6.7. Spectrogram with Pitch Track for [maḿa] ‘manioc’ 272..................................... 6.8. Spectrogram with Pitch Track for [mamaat́͡ʃ] ‘little manioc’ 273........................ 6.9. Spectrogram with Pitch Track for [wampiśnúmka] ‘in Wampis’ 274.................. 6.10. Spectrogram of the Word [ut͡ʃitiʃ́a] ‘we the young yoo’ 277.............................. 6.11. Spectrogram with Pitch Track for [hitinkartintija] ‘we the teachers "rst’ 278........................................................ 6.12. Spectrogram with Pitch Track for [jɛinta]́ ‘Help her/him!’ 283......................... 6.13. Prosodic Contour of ‘This story is called "Auhu"’ 287........................................ 6.14. Prosodic Contour of ‘How can we live?’ 288..................................................... 6.15. Prosodic Contour of ‘I tell you thanks’ 290........................................................ 8.1. Distance-Oriented Scheme of Wampis Demonstratives 340................................. 11.1. Structure of Noun Phrase 497............................................................................ 16.1. Wampis Simple Lexical Verb Word Order 711................................................... xxx LIST OF TABLES Table Page 1.1. Previous Studies on Other Jivaroan Languages 13............................................... 1.2. Dialectal Situation along the Santiago River According to Garciá-Rendueles (1996-1999) 19................................................................... 1.3. Possible Dialectal Variation Proposed for Wampis 21......................................... 1.4. Field Trips to Wampis Communities 38............................................................... 1.5. Primary Wampis Teachers and Expert Collaborators 42...................................... 3.1. Wampis Consonant Inventory 96......................................................................... 3.2. Distribution of Nasals in Front of Obstruents 109................................................ 3.3. Re/exes of Proto-Jivaroan *r and *h 114............................................................ 3.4. Distribution of Consonants in Wampis 133.......................................................... 3.5. Wampis Vowel Inventory 136.............................................................................. 3.6. Means for F1 and F2 Measurements of Wampis Vowels 137............................... 3.7. Distribution of Vowel Clusters 139...................................................................... 3.8. Assimilation of /a/ 141........................................................................................ 3.9. Sample of Spanish Loanwords in Wampis 152..................................................... xxxi Table Page 3.10. Sample of Quechuan Loanwords in Wampis 155............................................... 3.11. Some Wampis Words from Other Amazonian Languages 156........................... 4.1. Provisional Distribution of Syllable Types in Wampis 177................................. 4.2. Distribution of Syllables Types in Wampis 183.................................................... 5.1. Morphophonological Processes in Wampis 193................................................... 5.2. Morphemes that Trigger Reduction of Last Vowel (Mora) of the Stem 195......... 5.3. Verbs that Never Reduce their Final Mora when Receiving the High A$ectedness Aktionsart Su#x -a(́u) 204..................................................... 5.4. Morphemes that Reduce Their First Vowel (Mora) in Morpheme Boundary 211.................................................................................... 5.5. Morphemes that Exhibit Di$erent Forms 220...................................................... 5.6. Forms of Copula Clitics in Wampis 223............................................................... 6.1. Measurement of Vowel and Consonant Length in Pre-Tonic, Tonic and Post-Tonic Positions 249.................................................... 6.2. Bound Morphemes that Allow Internal Vowel Elision 260.................................. 6.3. Bound Morphemes Immune to Vowel Elision in Wampis 263............................. 6.4. Morphemes that Induce High Tone Shift in Two and Three Mora Words 270..... xxxii Table Page 6.5. Autosegmental Tone Morphemes in Wampis 279................................................ 6.6. Verbal Su#xes that Bear a Lexical High-Tone 280.............................................. 6.7. Verbal Su#xes that Sometimes Surface with Unpredictable High Tone 281....... 7.1. Examples of Roots and Propositional Act-Types in Wampis 305......................... 7.2. Examples of Roots Used in Di$erent Propositional Act Functions 309................ 7.3. Morphosyntactic Properties of Major Word Categories in Wampis 312............... 7.4. Aktionsart Su#xes in Wampis 314...................................................................... 7.5. Canonical In/ection and Derivational Properties 317.......................................... 7.6. List of Clitics Found in Wampis 320.................................................................... 8.1. Personal Pronouns in Wampis 331....................................................................... 8.2. Combining Forms of Personal Pronouns 334....................................................... 8.3. Genitive Forms of Personal Pronouns 335........................................................... 8.4. Morphemes Attested as Marking Agreement between the Pronoun naa and Its Noun Co-Referent 339.............................................................................. 8.5. Demonstratives with Case, Post-Positional and Discourse-Related Clitics 341.... 8.6. Morphemes that Display Agreement in the Presence of a Demonstrative 343..... xxxiii Table Page 8.7. Interrogative Words in Wampis 355.................................................................... 8.8. Person Markers that Occur on Question Words uruka and itura 366.................... 8.9. Attested Subordinating and Switch-Reference Morphology Associated with Interrogative Words 368.............................................................................. 8.10. Semantic Organization of Manner Adverbs in Wampis 377............................... 8.11. Person-Marking Adverbs 382............................................................................. 8.12. Su#xes Used by Person-Marking Manner Adverbs 383..................................... 8.13. Time Adverbs in Wampis 386............................................................................ 8.14. Location Adverbs in Wampis 390....................................................................... 8.15. Nouns Grammaticalizing as Location Adverbs 390............................................ 8.16. Non-Numeral Quanti"ers in Wampis 394.......................................................... 8.17. Wampis Numerals 399....................................................................................... 8.18. Reduplication of Numerals ‘one’, ‘two’, ‘three’ 405........................................... 8.19. Words that Function as Conjunctions in Wampis 408........................................ 8.20. Common Interjections in Wampis 413............................................................... 8.21. Sample of Wampis Ideophones 418................................................................... xxxiv Table Page 8.22. Probable Grammaticalization Path of has and hak 421...................................... 9.1. Semantic Properties of Prototypical Parts of Speech (from Croft 2001) 424....... 9.2. Morphosyntactic Properties of Nouns and Adjectives 427................................... 9.3. Nominal Su#xes that Can Be Received by Adjectives 431.................................. 9.4. Case and Adpositional Clitics that Can Be Received by Adjectives 434............... 9.5. Syntactic Positions in an AdjP 437....................................................................... 9.6. Sample of Adjectives Derived with -rama 449..................................................... 9.7. Types of Adjectives in Wampis According to Their Semantics 452...................... 10.1. Kinship Terms for Siblings 459.......................................................................... 10.2. Kinship Terms for Siblings-in-Law and Cousin 459............................................ 10.3. Other Gender-Based Kinship Terms 460............................................................ 10.4. Compound Word Combinations 462.................................................................. 10.5. Structure of the Wampis Noun 463.................................................................... 10.6. Type I Possessed Noun Marking 465.................................................................. 10.7. Type II Possessed Noun Marking 466................................................................. xxxv Table Page 10.8. Criteria for Vowel-Grade in Type II Possessed Noun Marking 467................... 10.9. Paradigms of Type I and Type II Person/Possessive Marking 467..................... 10.10. Possessive Marking with Nouns Derived with the Diminutive -t͡ʃi 468............. 10.11. Type I and Type II Kinship and A#ne Terms 470.......................................... 10.12. Type I and Type II Body Part Terms 471........................................................ 10.13. Type I and Type II General Nouns 471........................................................... 10.14. Nouns that Can Be either Type I or Type II 472............................................... 10.15. Nominative and Genitive Forms of Personal Pronouns in Wampis. 489.......... 10.16. Nominative, Acusative and Genitive forms of numi 'tree' and naiapi ‘swallow-tailed kite’ 490................................................................. 10.17. Sample of Nominative, Accusative and Genitive Forms of Nouns 491............. 10.18. Vocative Forms 492......................................................................................... 10.19. Exceptional Vocative Kinship Terms. 493....................................................... 11.1. Functions of Locative Markers 506.................................................................... 11.2. Sample of Words that Can Receive the Locative V́ 514...................................... 11.3. Words that Occur with the Locative when Carrying the Ablative 517............... xxxvi Table Page 12.1. Common Intransitive and Transitive Verbs 534................................................. 12.2. Auxiliary Verbs in Wampis 538......................................................................... 12.3. Unproductive Verbalizers in Wampis 542.......................................................... 12.4. Verbs Derived with -tika and -ni 544.................................................................. 12.5. Overall Structure of the Wampis Verb 548........................................................ 12.6. Su#xes that Occupy Position 3 in the Wampis Verb Structure 554.................. 12.7. Su#xes that Insert a Vowel ɨ in the Imperfective Stem 560.............................. 13.1. Valence-Decreasing Su#xes in Wampis 567...................................................... 13.2. Wampis Derivations with Detransitivizer -na 568.............................................. 13.3. Verbs where “Placeholder” -na Occurs 570........................................................ 13.4. Some Verbs Pre"xed with the Causative V- 580................................................ 13.5. Sample of Verbs that Select Forms -ru and -tu of Applicative 586.................... 13.6. Combination of Applicative and 1sg Object su#xes 593................................... 13.7. Verb Structure (Simpli"ed) 594......................................................................... 13.8. First Person Singular and Plural Object Su#xes 596......................................... xxxvii Table Page 13.9. Second Person Singular and Plural Objects Indexes 601.................................... 13.10. Forms for Marking 1A Participants Acting upon 2P Participants 602............. 13.11. Subject Markers, 1 and 2 Persons 603............................................................. 13.12. Paradigm of Object Marking on the Verb 605................................................. 13.13. Combination of Applicative and 1 Person Object Forms 606........................... 13.14. Combinations of the Applicative with 2 Person Object Markers 608............... 13.15. Su#xes that Fill Morphological Position 3 in the Verb Structure 610............. 13.16. Sample of Verb Roots with Their Preferred Aktionsart Su#xes 618................ 13.17. Aspectual Verb Stems in Wampis 637............................................................. 14.1. Wampis Tense Distinctions 643......................................................................... 14.2. Temporal Interpretation of Past Tenses 645...................................................... 14.3. Combinatorial Properties of Past Tense Su#xes 646......................................... 14.4. Paradigm of Conjugation with the Remote Past Tense 651............................... 14.5. Marking of SAP S and SAP A Acting on 3 Person 657...................................... 14.6. Marking of 1A Acting on 2P 657........................................................................ xxxviii Table Page 14.7. 3 Person Subject Su#xes 659............................................................................ 14.8. Indexation of Subjects and Objects on the Verb 665.......................................... 14.9. Imperative Constructions with 2 Singular and Plural Persons 672.................... 14.10. Paradigm of Person Marking with Apprehensive Modality 680....................... 15.1. Su#xes that Occupy Morphological Position 3 in the Verb Piece 689.............. 15.2. Nominalizing Su#xes in Wampis 693............................................................... 15.3. Morphosyntactic Properties of Underived Nouns, Nominalizations and Adjectives 694............................................................................................ 15.4. Distributional Features of Wampis Nominalizers 695........................................ 16.1. Summary of Object NP Marking 716................................................................. 17.1. Distinctions between Copula a, Copula Clitics and Existential a 738................. 17.2. Wampis Copula Clitics. 753............................................................................... 17.3. Set I Lexical Verbs Used to Code Spatial Notions. 762....................................... 17.4. Sample Posture Verbs in Wampis. 764............................................................... 17.5. Sample Wampis Positional Verbs 766................................................................ 18.1. Mood and Discourse-Oriented Vlitics 770.......................................................... xxxix Table Page 19.1. TAM Properties of Independent, Subordinate and Nominalized Verb 796......... 19.2. Types of Verbs and Associated Clause Types 797.............................................. 19.3. Morphological Structure of the Canonical Switch-Reference Subordinate Verb 798....................................................................................... 19.4. Morphological Structure of the Non-Canonical Switch Reference Verb 798...... 19.5. Wampis Subordinators 801................................................................................ 19.6. Aspectual Stems Associated with Speci"c Subordinators 802............................ 19.7. Person Markers Used in Canonical Switch-Reference Subordinate Clauses 805................................................................................... 19.8. Paradigm of -(n)i ̃ 808......................................................................................... 19.9. Paradigm of Verb puhu ‘live’ in SS and DS Sequential Clauses 820.................. 19.10. Paradigm of puhu ‘live’ in Imperfective DS clauses 825................................... 19.11. Frustrative Constructions in Jivaroan languages 828...................................... 19.12. Non-Canonical Switch-Reference Markers in Wampis 837.............................. 19.13. Person Marking in Conditional Clauses 843..................................................... xl CHAPTER I INTRODUCTION 1.1. Introduction This dissertation o$ers a comprehensive grammar of the Wampis (Huambisa) language, as it is currently spoken in the Santiago river1 area of Peru. The language name is written (the underlying form is /uampisa/, which is pronounced [wampiś]) by the speakers); therefore, assuming there is no accent mark in the English alphabet, throughout this dissertation I will use the term Wampis to refer to the language. The present work is based mainly on the variety spoken in the Middle Santiago, in the communities of Puerto Galilea, Huabal and Boca Chinganaza; and on the Upper Santiago variety as spoken in the community of Candungos. Though some di$erences between the Middle and Upper Santiago varieties occur and will be referred to in some parts of this work, no major statements about Wampis internal variation will be made (however, some basic notes on this topic are given in §1.6.1). The claims made in this work are based on the analysis of primary data gathered via extensive "eldwork in Wampis communities. Wampis is spoken by around 10,000 people in the 1. The Santiago river is known by the name of Kanus (/kanusa/) by the Wampis. 1 departments of Amazonas and Loreto, in Northeast Peru, principally along the Santiago and the Morona River basins. Near the border with Ecuador and crossing it toward the Yaupi and Coangos rivers, the Wampis and Shuar (both belonging to the Jivaroan family) languages come in contact, making this an area of transition, rather than a strict linguistic border. Group relations between the Shuar and the Wampis were cut in the 1940s because of the war between Ecuador and Peru, and remained interrupted for most of the 20th century. A consequence of these wars was the imposition of an international borderline that segregated what for centuries had been a continuous territory. This turn of events also divided a few Wampis and Shuar families that lived in both sides of the border. Nowadays, as peace between Ecuador and Peru is stable, relations across the border have been slowly re-establishing since the beginning of the 21st century. In this chapter, I provide a general introduction to the grammar. In §1.2–§1.3, I introduce the Jivaroan linguistic family and some possible connections to other languages. Sections §1.4–§1.5 discuss previous works on Jivaroan languages. The sociolinguistic situation of Wampis is described in §1.6. In §1.7, I provide a few notes on non-verbal communication among Wampis people. A summary of the dissertation 2 organization is provided in §1.8. Theoretical and methodological considerations are discussed in §1.9–§1.10. Finally, I present a summary of some typological features of Wampis in §1.11. 1.2. The Jivaroan linguistic family The Wampis language belongs to the so-called Jivaroan linguistic family.2 Other established members of the Jivaroan family are Awajun (Aguaruna), Shuar, Achuar and Shiwiar.3 The Jivaroan family is composed by a group of languages spoken in the eastern slopes of the Andes and the lowland Amazon forest of Peru and Ecuador. Figure 1.1 shows the approximate location of Jivaroan languages. 2. The term (/t͡ʃit͡ʃama/) ‘word, language’ has been proposed as the name of the family instead of Jivaroan (Katan Jua 2011). The term or has negative connotations in some varieties of Spanish where it became almost synonym of ‘savage’. The term might sound insulting to the Shuar of Ecuador but, as far as I can tell, it does not have the same strong negative sense for the Wampis. 3. Throughout this dissertation I use the o#cial names of Jivaroan languages as used by the native speakers for their languages and as they appear in o#cial documents of Peru and Ecuador. The names in parenthesis correspond to the imposed Spanish spelling of their names, which has been followed by other scholars who write in other languages, such as English or French—those same spellings have been used more frequently in the literature about the Jivaroan peoples and languages. For a list of o#cial names of native languages of Peru, one can consult http://bdpi.cultura.gob.pe/lista-de-pueblos-indigenas; and for Ecuador, see http:/ /www.conaie.org/index.php/en/. 3 Figure 1.1. Map of the Jivaroan Languages All Jivaroan languages are closely related and have some degree of mutual intelligibility. In this regard, one can consider them as part of a dialectal continuum. Achuar and Shiwiar, especially, are occasionally considered to be one linguistic entity because of their tight resemblance, and thus are sometimes appear under the name 4 Achuar-Shiwiar (see for instance Fast et. al. (1996)). In contrast, modern speakers of Wampis are well aware of the linguistic di$erences between Wampis and their other Jivaroan relatives. They readily identify certain features as belonging to Wampis or not. A good way to see this situation is to consider Jivaroan as a complex of “ethnolinguistic dialects” (Gnerre 2010: 29) that have developed as a result of ethnic identities driven by forces “external” to the language(s) in the last centuries; the external forces include intertribal wars, the rubber boom exploitation between the mid-1800s and beginning of the 1900s, national and international political pressures, and so on. Following his own terminology, Kaufman (1990; 2007) considers Jivaroan as a “language area” with two emergent languages, which he calls Jívaro (Shuar, Achuar, Shiwiar, Wampis (Huambisa in Kaufman’s works)) and Awajun (Aguaruna in Kaufman’s works). His proposal seems based on the traditional internal subgrouping of the Jivaroan family (see discussion below). However, Kaufmann’s schema omits certain important linguistic characteristics that would need to be taken into account to clearly de"ne relationships within the area.4 It also runs the risk of imposing a methodologically-derived construct that does 4. In fact, there is little information about some Jivaroan varieties. Point in case is Wampis, which lacks prior grammatical description. Important potentially de"ning features have not been well studied for all Jivaroan varieties. 5 not correspond to how speakers conceive their own languages in their reality. While their awareness of linguistic di$erences reveals a relatively close relationship with their linguistic relatives, it is worth noticing that the speakers of Wampis consider Wampis to be a language. The same can be said, as far as I can tell, for speakers of the other Jivaroan varieties. Since there is no universal consensus on a scienti"cally rigorous de"nition of the terms language and dialect (Good & Cysouw 2013)—linguistic basis for delimiting dialects from languages clearly is very weak; “dialect” and “language” are rather socio-political constructs—, I will consider Wampis as the native language spoken by the ethnic Wampis.5 Traditionally, the internal subgrouping of the Jivaroan family considered there to be two branches: on one hand, Awajun; on the other, the Shuar subgroup—composed of Shuar, Achuar, Shiwiar and Wampis (Stark 1985; Corbera Mori 1994; Wise 1999; Gnerre 2010). This traditional subgrouping has been followed in most broader classi"catory works (Kaufman 1990; Campbell 1997; Fabre 2005 [modi"ed 07-22-2007]; Kaufman 2007).6 Awajun was seen as more conservative, thought to have 5. In this case, this de"nition correlates with the fact that the population of Wampis speakers almost completely equals the ethnic Wampis population. 6. Note that Campbell (2012) does not assume any subgrouping. 6 kept a hypothesized velar nasal *ŋ proto-phoneme, whereas the other members would have innovated a rhotic from the velar: *ŋ > r (Payne 1981; Turner 1992).7 The traditional subgrouping of Jivaroan languages is shown in Figure 1.2. Figure 1.2. Subgrouping of the Jivaroan family based on Stark (1985) The internal subgrouping shown in Figure 1.2 has been questioned by Overall (2007: 5; 2008), who claims that there is no basis for subgrouping. According to Overall, the velar nasal does not have phonemic status in Jivaroan languages, and therefore is not reconstructable. Further, he submits that a rhotic proto-phoneme *r needs to be reconstructed, rather than *ŋ. In Awajun, there is a merger of *r with a velar *h, but the *r is a shared retention and not an innovation in the hypothesized Proto-Jivaroan Shuar Sub-Group Awajun Wampis Shuar Achuar-Shiwiar 7. Turner actually speaks of a velar vibrant (“vibrante velar”), rather than nasal. 7 Shuar subgroup. Thus, Overall states that the internal classi"cation of the Jivaroan family remains unresolved. His claims are supported by the present work (see §3.2.3.4). 1.3. Other proposed Jivaroan connections and language contact The genetic relationship of the Jivaroan family with other languages or language families has not been proven. There is insu#cient or contrary evidence for all of the proposals, except perhaps the relationship between Jivaroan and Palta, en extinct language formerly spoken in Ecuador: there are at least reasonable arguments for this connection (see discussion below). Greenberg (1987) proposed a Jivaro-Kandozi (Candoshi, Kandoshi) grouping within the Equatorial branch of his hypothesized Equatorial-Tucanoan stock. This proposal has been criticized by Gnerre (2010: 131–138) and Kaufmann (1990), who have shown Greenberg’s proposal shortcomings. The connection between Jivaroan languages and Kandozi was further pursued by David Payne (1981) with a more robust comparative work. However, the author himself later backed away from his claim (Payne 1990b: 84), arguing that most supposed cognates were borrowings, leaving only very few basic vocabulary items to compare. Kaufmann (1990: 42) "nds some lexical basis in support of a hypothesized Jivaro-Cahuapana stock (Swadesh 1959; Suárez 8 1974), but this proposal has not been pursued further. On the basis of lexical and morphological comparisons, Gnerre suggests that there may be a connection with the Arawak family: “The proto-history of the Jivaroan languages is that of a proto-language with a Maipure-Arawak component, spoken in the Andean area” (2010: 158),8 but his analysis is rather inconclusive.9 Perhaps the most interesting proposal regarding genetic relationships with other languages is the connection between the extinct Palta language and Jivaroan. Palta was spoken in today’s Ecuadorian territory and around the colonial Jaen area in what is today North Peru,10 where it was referred to by the name Xoroca. Rivet (1934) and Loukotka (Loukotka 1968), based on only 4 lexical items documented, included Palta within the Jivaroan family. Gnerre (1975), Taylor & Descola (1981) and Taylor (1991), based on ethno-historical colonial sources and toponymy, have provided more reasonable arguments in favor of the Palta-Jivaroan connection (see also Torero (2002: 8. Original in Spanish: “La protohistoria de las lenguas jíbaras es el de una protolengua con un componente Maipure-Arawak, hablada en el área andina”. 9. Karsten (1935: 539) had stated earlier that the Jivaro did not belong to the Arawak stock and that its a#nities were undetermined. 10. The colonial Jaen area was located, roughly, on the eastern slopes of the Northern Andes of what is today Peru. Jaen was an area of contact: it connected the Coast and the Andean regions with the Lowland Amazon forest. 9 273–298), who proposed a Palta-Jíbaro family, and Adelaar (2004: 396–397)). Unfortunately, the list of actual lexical materials for Palta (only 4 words) is too small to demonstrate convincingly the link between Jivaroan and Palta from a comparative perspective. Other extinct languages of the colonial Jaen area likely had contact with Jivaroan languages too (Torero 2002: 295; Adelaar 2004: 405–407). Because of its geographical location near the Marañon River and its low altitude, the colonial Jaen area enjoyed a strategic position for contact and trade between the Andes and the lowland Amazon. Jaen attracted people of apparently di$erent ethnic backgrounds, and the area conceivably was a multilingual complex before the arrival of the Spanish. Unfortunately, such languages have scarce or no documentation. In Ecuador, Jivaroan languages also had close historical ties with the area of Canelos, another important center of trade and cultural exchange between the Andes and the Upper Amazon in colonial times, and probably before European contact (Whitten, Jr. 2013). The Canelos today speak Quichua (the variety of Quechua spoken in Ecuador). The history of contact between Jivaroan languages and other languages has yet 10 to be studied in depth. Over the last two centuries, as Spanish has expanded and become dominant in Peruvian Amazonia, it may have in/uenced Jivaroan languages beyond the lexical level. Quechua is another language that has been in contact with Jivaroan languages for at least the last four centuries, and it has been suggested that there may be a Jivaroan substrate in Ecuadorian Quechua (Muysken 2010; Muysken 2013). Contact with Kandozi has occurred apparently since pre-colonial times, to such an extent that some scholars consider the modern Kandozi to be culturally (Renard- Casevitz, Saignes, & Taylor 1986) or even potentially ethnic Jivaroan (Siverts 1978), though linguistically they speak an unrelated language. Contact with Cariban and Tupi- related languages has also been explored (Gnerre 2010). It is also very possible that the Jivaroan have /uid contact with other languages of the Eastern Amazon; for instance the relation between Jivaroan, Zaparoan and other neighboring languages of the Amazon in colonial and modern times still awaits more profound study. Altogether, in the past Jivaroan languages likely were a$ected by contact with several other languages, most of which have scarce or no documentation and are already extinct. A "nal point to take into account is the contact among the Jivaroan languages themselves. A necessary topic of future research for better understanding the transitions 11 within the Jivaroan family is the study of those varieties near the Jivaroan territorial borders. In this regard, the study of Wampis "lls a gap as Wampis shares its borders with Shuar, Achuar and Awajun; actually, Wampis is located at the heart of the greater modern Jivaroan territory. 1.4. Previous studies of Jivaroan languages In general, Jivaroan languages have received some attention from linguists. Though much work remains to be done within the family, there are already a number of valuable studies of Jivaroan languages. Shuar and Awajun have received more attention judging by the quantity of grammars, sketch grammars or vocabularies related to these languages. Not coincidentally, these are the groups geographically closer to main roads and provincial big towns/cities, and thus they have had more contact with Western culture. Some of the most substantial modern published works about Jivaroan languages that I had access to prior to the writing of this dissertation are listed in Table 1.1. These works were consulted and, in some cases, will be discussed in some parts of this grammar. I only list works that have a linguistic focus. Arguably the Jivaroan cultures are better known via the anthropological literature; there are numerous anthropological 12 works for Shuar, Awajun and Achuar, but less so for Wampis. Table 1.1. Previous studies on other Jivaroan languages Language Author Description Shuar Turner (1958b) Ph.D. dissertation focusing on phonology and morphology Shuar Pellizaro (1969) Sketch grammar Shuar Turner (1992) Sketch grammar Shuar Pellizaro and Náwech (2005) Dictionary with grammar notes Shuar Gnerre (2010) Description on parts of the grammar (phonology, verb, noun) and some historical comparison with other language families. Shuar Saad (2014)a Sketch grammar with focus on morphology Achuar-Shiwiar Fast et. al (1996) Dictionary for pedagogical purposes that contains a grammar sketch Achuar/Shiwiar Fast (1975a) Study of Shiwiar phonological system Achuar/Shiwiar Fast (1975b) Short study of Achuar phonology Awajun Pike and Larson (1964) Study about hyperphonemes and their distribution Awajun Larson (1978) Study of reported speech Awajun Corbera (1978) Bachelor’s degree thesis with a focus on phonology Awajun David Payne (1989; 1990a) Two studies of “accent” Awajun Corbera (1994) Doctoral dissertation, a grammar with emphasis on phonology and morphology 13 Awajun Wipio et al. (1996) Dictionary for pedagogical purposes Awajun Uwarai et al. (1998) Dictionary with an alternative orthography to that proposed by Wipio et al. (1996) Awajun Overall (2007) Ph.D. dissertation consisting of a complete grammar description Awajun Overall (2008) Article about nasal ŋ in Jivaroan languages Awajun David Payne (2008) Translation in Spanish of Payne’s 1974 M.A. thesis on nasality Awajun Overall (2011) Article about clause linking a Unfortunately, Saad’s work was published near the end of the year 2014, when my own analysis of Wampis was mostly done, so it has been di#cult to compare Saad’s ideas for Shuar with my own for Wampis. 1.5. Previous studies about Wampis In comparison with other members of the Jivaroan family, Wampis has not been much studied and remains under-documented. To the best of my knowledge, there is neither a published grammar or sketch grammar of Wampis. Modern published works that have speci"cally Wampis as their object of study consists of a vocabulary with a pedagogical focus (Jakway, & A. 1987), a lexical comparison of Wampis and Awajun (Larson 1955–1957), a (rather) short vocabulary list (Corbera Mori 1980) and a short description of the phonemic inventory of the language (Beasley & Pike 1957). A brief sociolinguistic study about identity based on lexical di$erences between Shuar, Wampis 14 and Awajun is found in Valeš (2013).11 Studies on Wampis sound-symbolism in names of "sh and birds, from an ethnobiological perspective, are found in Berlin (1992; 1994). Besides these, there are a handful of works that contain data that may be of interest to researchers of Wampis, but they do not constitute linguistic works. The most important of these works was published by the Spanish Jesuit anthropologist García-Rendueles (1996–1999). His 1996–1999 publication is a bilingual (Wampis–Spanish) compilation of mythical narratives thematically organized. Sadly, Garciá-Rendueles’ work, monumental in scope, has seen only two of the planned "ve volumes published (the "fth volume was to contain a vocabulary). Yet, the wealth of data in the published parts can o$er researchers a rather insightful view into the culture.12 11. The author bases his analysis on a questionary that asks what word speakers use for the concepts of ‘butter/y’, ‘/ute’ and ‘number one’. Unfotunately, the author does not explain the reason why only these three concepts were chosen, in which locations the interviews took place, how were interviews carried out (for instance, using Spanish?), etc. In addition, the author states that the term “Huambisa/Wampis . . . parece desarrollarse despueś del cambio de frontera en 1942 para designar a los Shuar que vivián en el territorio peruano” (“apparently developed after the borderline [of Peru and Ecuador] changed in 1942 to designate the Shuar that lived in the Peruvian territory” (Valeš 2013: 175). This is inaccurate (unfortunately, the author does not cite any source that the reader could consult). There are documents which identify the Wampis as a group since at least the very late 1700s/early 1800s, way before the 1941 war between Peru and Ecuador. See Chapter II for details on the history of the term Wampis. 12. Unfortunately, the transcription of texts is inconsistent and the translation o$ered is a very loose one. 15 Finally, the works mentioned above are not comprehensive of all materials on Wampis. A number of Summer Institute of Linguistics (SIL) missionaries and linguists, especially David Beasley, have left notes on the Wampis language, but they remained unpublished. I have not had access to these unpublished notes. Also, SIL has published a number of materials in Wampis for bilingual school education. 1.6. Socio-linguistic situation The Jivaroan peoples have a strong commitment to their identity, including of course their languages. The Wampis are no exception. The Wampis language variety is currently spoken in most contexts and domains of communication, and its presence sometimes extends to spaces where Spanish is dominant, such as radio transmissions,13 school classes and local authority meetings.14 The Wampis approved a new alphabet for their language in 2012 and are looking forward to developing the language for pedagogy, law-making and other aspects of “modern” life (such as internet use). 13. Spanish is dominantly used in radio transmissions, but I have heard Wampis spoken in programs on occasion. Messages are often delivered in Wampis. There is at least one religious program where Wampis is sometimes used. Cumbia and traditional music in Shuar and Awajun are also transmitted. There is only one radio station in the Santiago River, but on occasion, radio show transmissions from Ecuador (in Spanish and, near the borderline with Ecuador, in Shuar) are captured. 14. In formal situations involving Spanish-speaking authorities, Spanish is used. 16 The Wampis society remains one of the poorest minority groups in Peru. As migration of young Wampis people to cities increases, so does pressure from Spanish and other languages. Increasing contact with outsiders who seek opportunities to engage in logging or gold-mining adds to this pressure too. The location of the Wampis territory in a political borderline and the fact that this territory is an area of natural resource exploration (the region is rich in gold and other minerals, exploration for oil is going on) also have put the Wampis in a very sensitive situation. Hence, the Wampis are a vulnerable group and the language has been listed as having the status of “vulnerable” in the Atlas of the World’s Languages in Danger (Moseley 2010). 1.6.1. Varieties of Wampis An examination of Wampis-internal linguistic variation has not been attempted, nor was it the goal of the research leading to the writing of this dissertation. However, at "rst glance it can be stated that variation in Wampis is mostly lexical and phonetic, with some di$erences in the pronunciation of a few grammatical morphemes. There exists mutual intelligibility between Wampis speakers from di$erent areas. There are two large areas where Wampis is spoken: the Santiago river and the Morona River. The Santiago and Morona rivers run almost parallel to each other, but they are separated by 17 a mountain range, called Kampankis. I did not collect data from the Morona river area. Along the Santiago river, there are three general areas: the lower Santiago is mostly Awajun territory, the Middle and Upper Santiago is considered Wampis territory. The one variety of the Santiago area that always was suggested to me as being divergent from other varieties of the Santiago river region is the one spoken on the Caterpiza river, a tributary of the Santiago. Though I have been unable to go to the Caterpiza river, I have witnessed dialogues between people of the Middle Santiago and Caterpiza, and they can certainly communicate with each other. This suggests that the Caterpiza variety is actually not very divergent from the other dialects of Wampis,15 though it seems to be the most sociolinguistically marked16 of the dialects spoken in the Santiago basin. The study of the varieties of Wampis, as for most other Jivaroan languages, is yet to be fully explored. García-Rendueles (1996–1999) outlined his view of the 15. The Caterpiza river is located on the east side of the Santiago river. Most Wampis communities in the Middle Santiago river are located on the west side. This probably contributes to the Caterpiza dialect being perceived as di$ering from others. There is also the matter of contact: it may be the case that the Caterpiza dialect is relatively divergent, but since Wampis people move around and visit or trade with each other, they can have a certain level of understanding, simplifying their speech when they communicate. 16. By “marked” here I mean that it is perceived as di$erent. There is no strong negative sociolinguistic attitude towards the people who speak this dialect. 18 dialectal situation in the Santiago basin, which was based upon the geographical location of communities. Table 1.2 shows his proposal. My comments are in parenthesis. Table 1.2. Dialectal situation along the Santiago River according to Garciá-Rendueles (1996-1999) Garciá-Rendueles’ proposed groups Garciá-Rendueles’ comments Yakinia Shuar (‘people from up-river’) North-Santiago area, related to the Shuar of Ecuador Muraña Shuar (‘people from the mountains’) Shuar from the headwaters of the Santiago river a6uents. (Located on both sides of the Santiago.) Moronanmañan Shuar (‘People from the Morona’) Wampis speaker who migrated from the Morona area into the Santiago basin. (Located on the east side of the Santiago.) Shir-Wampis (‘Good Wampis’)a Located in the middle-Santiago (on the west side of the river.) Wampis-Awarun “Corrupted language” (spoken for people living in the transitional border between Wampis and Awajun territotry) Apach (‘mestizo, non-Wampis’) (Spanish-speaking people) from Loreto and the Andes. a (/ʃiira/) means ‘good’. Obviously, this is an evaluative term by Garciá-Rendueles that does not have any scienti"c basis. He considers this variety prescriptively, as the “purest” Wampis. The same can be said of his Wampis-Awarun group, which he considers as “corrupted”. García-Rendueles’ sketch of Wampis varieties seems relatively compatible with my observations for the most part; however, I did not "nd substantial linguistic 19 di$erences between his “Yakinia” and “Muraña” Shuar. This was corroborated by the attempted subgrouping given by my main Wampis collaborators when I asked them. It may be the case that during the last 30–40 years or so (García-Rendueles’ "eldwork took place in the 1970’s and early 1980s) new generations of speakers have come closer to Wampis as ties with the Ecuadorian side remained distant since the 1940s.17 Thus, I would propose a tentative revision of García-Redueles’ work with four main varieties18 for Wampis. The “Wampis-Awarun” variety referred to by the Spanish author needs to be studied in a di$erent light, i.e. from a languages-in-contact perspective, rather than as some sort of “corrupted language”. Therefore, for the moment it is left out of the subgrouping until its nature is explained more satisfactorily. I had occasional contact with people from Cucuasa and Papayacu who were living in or visiting Candungos but almost none with people from the other communities close to the border with Ecuador. According to García-Rendueles, Cucuasa, Onanga and Ampama communities are 17. The situation was exacerbated by two new armed con/icts between Peru and Ecuador that occurred in 1981 and 1995. 18. Notice, again, this tentative division is based on quick notes made while on the "eld. The Morona area needs to be more thoroughly investigated. The few people I interviewed claiming to know well the Morona area told me there are no major internal di$erences. However, as the Morona is actually a large area, this statement needs to be corroborated through "eld research. 20 grouped in the “Yakinia” dialect, whereas Candungos belongs to the “Muraña” subgroup. Table 1.3 shows the proposed dialectal variation for Wampis. Table 1.3. Possible dialectal variation proposed for Wampis Proposed Wampis varieties García-Rendueles’ equivalent a. Upper Santiago (iaki=ni=̃ia ʃuara, mura=́ia ʃuara)a Yakinia Shuar/Muraña Shuar b. Middle Santiago (naka=ia ʃuara)b Shir-Wampis c. Katirpisa (amaini=ia ʃuar)c Moronanmaña Shuar d. Morona (muruna=numa=ia ʃuar)d Variety spoken in the Morona basin, not included in García-Rendueles’ sub- classi"cation. There can be sub-varieties. a iaki=ni=̃ia ‘above=LOCATIVE=ABLATIVE’, the ‘people from up-stream’, mura=́ni=̃ia ‘mountain=LOCATIVE=ABLATIVE’ ‘people from the mountain’. b naka=ia (‘in front=ABLATIVE) ‘[people] from in front’) is how Middle Santiago people identify themselves (the communities of Middle Santiago are located only in the west side of the river, hence the name). c amaini=ni=̃ia ‘other bank of the river=LOCATIVE’, i.e. ‘people from the other bank’. d muruna=numa=ia ‘Morona=LOCATIVE=ABLATIVE’ , i.e. ‘people from the Morona’ Unfortunately, this proposed division is for, the moment, based on the judgements of speakers (and for the most part, based on geography) rather than on actual data. Much remains to be done on the "eld of variation for Wampis. 1.6.2. Auto-denomination The Wampis people refer to themselves simply as [wampiś] /uampisa/ ‘Wampis’ 21 or [ʃuaŕ wampiś] (ʃuara uampisa ‘person Wampis’), i.e. ‘Wampis person’. Other common denominations are [ií wampisti]́ (ii uampisa=ti ́‘1PL Wampis-PLURAL SPEECH ACT PARTICIPANT’), [huti ́wampisti]́ (huti ́uampisa=ti ́‘1PL Wampis-PLURAL SPEECH ACT PARTICIPANT’), which both can be translated as ‘we the Wampis’. Another term used is [ii ʃuarti]́ (ii ʃuara=ti ́‘1PL person-PLURAL SPEECH ACT PARTICIPANT’), i.e. ‘we the people’. Despite the negative connotations of the word Jivaro in Spanish (where it became practically a synonym of ‘savage’), some Wampis people also used this term to introduce themselves when we met, saying they were “Jivaro Wampis”. This is an interesting re-appropriation of the term, without the negative meaning, as part of their identities. However, this use of the term Jivaro is not shared by all Jivaroan groups.19 In general, the Wampis can extend the use of the term [ʃuaŕ] /ʃuara/ ‘person’ to refer to people that they can identify as Jivaroan. In some cases, ʃuara also means ‘enemy’ and is sometimes used to refer to the Awajun in narratives of their past wars (otherwise, the Wampis used [awarún] /auaruna/ to refer to said group. The term Jivaro itself is a hispanization of the proto-Jivaroan form *ʃ(i)uara from which modern [ʃuaŕ] /ʃuara/ is a re/ex (Gnerre 1973). The ethnonyms Shuar, Achuar and Shiwiar all 19. The term is particularly insulting for the Shuar of Ecuador. See Katan Jua (2011). 22 come from the same proto-form too (Achuar apparently etymologically comes from at͡ʃu ‘aguaje’+ ʃuara ‘person’). The etymology of the name Wampis most likely derives from the noun uampi ‘sabalo sp. (Salminus sp.)’ and what seems to be a phonetic reduction of ɨnt͡sa ‘river’ > - sa.20 The same reduced structure is veri"ed in some Jivaroan toponyms in Wampis territory, cf. kankasa [kaŋkás] from kanka ‘boquichico’ + sa (Spanish ), i.e. ‘boquichico stream’; t͡ʃinkanasa [tʃiŋkanás]21 from tʃinkana ‘bamboo sp.’ + sa ‘stream’ (Spanish 22, i.e. ‘bamboo stream’; t͡ʃapisa [tʃapís] from tʃapi ‘yarina palm’ + sa ‘stream’, i.e. ‘yarina stream’ (Spanish ). I was told a story (that seems to justify a folk etymology) in which a Wampis group of people were very quick to organize themselves to face an imminent attack from a rival group. As the story goes, many Wampis from di$erent parts of the Santiago sailed the river, coming “as fast as the uampi "sh” to the location where they were summoned.23 Other people were more 20. Though notice that there is also a word related to water: sasa ‘streaming water’, ‘waterfall’. 21. Some speakers pronounce it [tʃiŋkántsa], which seems to be more closely related to t͡ʃinkana + ɨnt͡sa ‘river’. 22. The complete name in Spanish is . 23. The uampi "sh is indeed regarded by the local people as a fast-moving "sh. 23 cautious and said that the name likely refers to people that was known to live in a small river tributary where uampi "sh (Salminus sp.) abounded, and hence the name. The ethnonym is probably related to the old Huambisa River, a tributary of the Santiago. The Wampis used to live in the headwaters of the Santiago River, along its tributaries, before communities were created along the main river itself. The root uampi also appears in other parts of the Jivaroan territory, for instance the river Huambi (a hispanization with a voiced stop), north of the Santiago river, in the region of Sucua, Ecuador, near the border with Peru. It is interesting to notice that an early work about Jivaroan people (Karsten 1935) locates the Wampis on the Upper Morona River, and in his map (given at the end of his cited work) the Wampis appeared north of the Santiago, in the immediacies of modern quebrada Huambi in Ecuador. However, Karsten did not visit the Wampis in Morona or Santiago. In fact, the presence of the Wampis in the Santiago River region, and as far South as the Marañon River, dates from at least the very early 19th century when they—sometimes together with the Awajun— are identi"ed as the Jivaroan group which made several raids in the gold/rubber mining towns in the lower Santiago and Upper Maranon (Guallart 1990). The Wampis refer to their language as [iiɲa ́t͡ʃit͡ʃaḿ] (iina t͡ʃit͡ʃama ‘1PL.GEN 24 language’), i.e. ‘our language’, [wampiś t͡ʃit͡ʃaḿ] (uampisa t͡ʃit͡ʃama ‘Wampis language’), [ʃuara ́t͡ʃit͡ʃam] (ʃuara ́tʃitʃám ‘person\GEN language’), i.e. ‘the language of the people’ or simply [t͡ʃit͡ʃaḿ] (t͡ʃit͡ʃama ‘word, language’). 1.7. Non-verbal communication The Wampis used to have a percussion system for long-distance communication. Large signal drums (tuntui)́ were hung from trees or walls and played for di$erent communicative needs. Communication using these drums is still practiced, though not as frequently as it was in the past. I was unable to have a "rst-hand experience of drum- communication as my research focus lay somewhere else. Karsten (Karsten 1935: 108– 110) described the making of the signal drum among the Jivaro; the use of similar drums is found in other places in Amazonia (Thiesen 1969; Neves 2001: 269). When the Wampis are in the jungle, they also communicate from a distance using buttress roots as percussion instruments. In this way, they are able to communicate the location of someone when hunting, and in the past they communicated if an enemy raiding party had been seen coming. Nowadays, the Wampis also use a horn or shell called bosina (from Spanish) or kat͡ʃu (from Quechua) to call for important meetings or events in the community. The arrival of a visitor to a host’s 25 house in the past was signalled by using a small drum, by the horn, or by putting both hands together in a tubular form and whistling into them to amplify the sound (these actions are still practiced today but mostly for especial celebrations). A whistling "language" to signal actions, especially when hunting, and to imitate birds—at which the Wampis excel (Berlin 1994)—is still present in Wampis culture. Body language is important in oratorial discourse and in ritualistic situations. Body gestures are somewhat important in story telling too, and often accompany the description of actions. Spitting on the /oor is an important part of singing anɨnta ‘magical song’ and is used at the end to conjure the power of the song. The act of spitting on the /oor when conversing or drinking is seen as normal and shows familiarity among men.24 Earlier descriptions of Wampis culture discuss how in certain speci"c contexts ritualistic body language also included certain oral elements as an integral and important part of the ritual. For example, when the Wampis went to war and had surrounded the enemy, every member of the raiding party, one after another, would have to yell a long yaa! (a type of challenge or war cry) while thumping their feet and showing their spears. They had to follow a speci"c increasing and decreasing 24. See Descola (1986: 165–168) and Boster (2003) for alternative accounts on the interpretation of the act of spitting in Jivaroan culture. 26 melody pattern when making the war cry. It was believed that if the warrior did not follow the ritual correctly (which included following the melody), he would be doomed. It is di#cult to say whether there was a more lexical meaning assigned to these cries; however, it is interesting that some speakers were able to nominalize these cries when I asked if there was a “name for the action” (e.g. [yaaḱat] (iaa-ka-ta ‘cry.yaa-INTENS-NMLZ’, i.e. ‘the crying of yaa’), suggesting it may have been treated as some type of verb in the past. Body painting is common among the Wampis. In the past, when a war raid was performed, Wampis men usually painted their bodies in black with sua ‘huito (Genipa americana)’. I have not seen much literature on this, but there seems to have been an identi"cation code associated with color: in contrast to the Wampis who used black, their usual enemies the Awajun would paint their bodies in red.25 Men still paint their faces in black and carry their spears26 when they march to protest regional or national policies perceived as contrary to their interests. Men who have achieved high status 25. A recurrent observation of war body painting can be found in some narratives in Guallart (1990: 237 and $.). 26. The spears nowadays are carried more as a symbol of Wampis identity, rather than as real weapons. 27 within the community also wear a crown made of toucan feathers [tawaś] (/tauasa/) and carry a spear [naŋ́ki] (/nanki). Women wear colorful earrings [akit́ɛi]̃ (/akitai/), usually made of toucan feathers and giant beetle wings, and put on their typical red or purple/blue dresses [tarat́͡ʃ] (/tarat͡ʃi/), which they tie up on the right shoulder. Piercing in the lower lip and nose was practiced until about three generations ago, but it is no longer practiced. Some elderly people I met still exhibit these piercings. For special occasions or celebrations, the Wampis still paint their faces with thick lines in red. Facial tattoos were also used. In fact, elderly people, still conserve their tattoos. Tattoo and face-painting patterns represent the possession of [arútam] (/ arutama/), a vision of power given by the most important Jivaroan spiritual entity (see §2.7). Tattoos were also apparently made to distinguish individuals within a family. People I asked about this did not remember the speci"c details, though. 1.8. This dissertation This dissertation constitutes the "rst attempt ever at providing a complete grammatical description of Wampis. To the best of my knowledge, there is no previously published grammatical description, or even a sketch, of Wampis. The grammar is based on extensive "eldwork carried out at di$erent moments in 2008, 28 2011, 2013 and 2014. This grammar is divided into twenty one chapters. Following this introduction, Chapter II provides brief notes on the Wampis people. Chapters III trough VI provide analyses of di$erent aspects of Wampis phonology, phonetics, morphophonology and prosody. Chapter VII constitutes an introduction to morphology and word classes. Chapter VIII provides a description of all closed word classes in Wampis. Next, Chapter IX is dedicated to adjectives. The noun, noun phrase and adpositional phrases are described in Chapters X and XI. Chapter XII is an introduction to the study of the verb, which is then followed by an in-depth study of verbal morphology: “derivational” level in Chapter XIII, and “in/ectional” level in Chapter XIV. Chapter XV looks at nominalization. Chapter XVI o$ers an analysis of the simple clause, followed by an analysis of verbless, copular and existential constructions in Chapter XVII. Chapter XVIII describes discourse-oriented and mood clitics. Complex clause constructions, including subordination, complementation and relativization, are analyzed in Chapters XIX and XX. Finally, some concluding comments are provided in Chapter XXI. 1.9. Theoretical framework The approach in this work re/ects many "ndings and discussions established or 29 now generally assumed from the development of comparative philology, Structuralism, and even some formalist syntactic frameworks. However, my understanding of language relies on an idea that does not "t within a structuralist or formalist (Chomskyan) theories. That is, that language use, grammar structure, and cognition are closely connected. Language is a dynamic system, adapted and constrained by our ways of categorizing our experience. This is a general principle that guides my approach to the study of language in general and, in particular, my analysis of Wampis. The main paradigm that informs this dissertation is the functional-typological approach, with an appreciation of historical developments in understanding the current state of Wampis. Functionalism is not one uni"ed theory, but in its various iterations it rejects formalism as an explanation of language facts. Functionalism claims that grammar is not an autonomous system, but it is part of a system of human communication composed of a cognitive representation system and a communicative coding system (Givón 2002a: 7). In that regard, the functionalist approach rejects the idea that a theory of grammar can "nd explanatory adequacy (Chomsky 1965) by circularly resorting to a purportedly autonomous language module, ultimately linked to a distinctly abstract untestable notion of Universal Grammar. Thus, it can be said that 30 the most basic epistemological distinction between functionalism and formalism is what counts as an explanation (Payne 1999b). A formal model often serves simply as a “restatement of the facts at a higher level of generalization” (Givón 1979: 5) and, in that regard, it lacks explanatory strength. Functionalism, on the other hand, seeks explanations for formal structures in the explanatory domains of semantics, grammaticalization, discourse-oriented communicative, cognitive constraints or broader cognitive pressures (such as attention and representation, for example). Within this view, language is seen as a set of tools “whose forms are adapted to their functions” (DeLancey 2001a). Hence, language structure has an adaptive motivation—i.e. it is the way it is because it re/ects constraints in language use, where language suits speakers’ needs. Functional studies often incorporate a typological domain in their analyses. Within this dissertation’s framework, linguistic typology is viewed as the study of language variation and of the limits on that variation. It is assumed in this work that there are language-speci"c constructions, but also that di$erent language structures can, from one language to another, encode the same type of function or communicative job. Cross-linguistic recurrent patterns are often functionally motivated and need to be 31 explained. In this sense, typology is not merely descriptive, but also it can be explanatory (Croft 2003: 284). In my analyses of Wampis data I incorporate typological analytical tools and try to relate the data to categories and constructions found in this language to those that are, or are not, found recurrently cross-linguistically. I try not to provide a simple inventory of features; rather, my use of a typological framework has been motivated by the necessity to understand (and ultimately seek to explain) how certain Wampis constructions work by relating them to known linguistic patterns. However, I always de"ne the categories found in the Wampis data in their own terms, explicitly characterizing them and providing examples of how formal structures are paired to functions in this language. By comparing language-speci"c structures of languages, including those of Wampis, we can eventually better propose hypotheses addressing how di$erent linguistic functions could have developed and come to be encoded in various languages. Ultimately, we can try to explain how those linguistic functions relate to human cognition in general. Often, grammatical units are not synchronically transparently motivated. Most of the time, the ontological opacity of grammatical categories is a consequence of language change. In this dissertation, as I said, language is seen as a dynamic system, 32 i.e. a system that is continuously being re-created by communicative activity. That means that, although I recognize the methodological value of di$erentiating synchrony from diachrony, I believe that maintaining the rigid boundary of this dichotomy prevents us from obtaining better understanding of how diversity in grammatical structure has arisen (see, for instance, Givón (1971; 1979), Lehmann (1985), Bybee (1985), Heine et al. (1991), Gildea (1998), among others). Therefore, I also use the framework of grammaticalization (or historical processes of change) to help explain patterns in a language. The main premise here is that “in grammaticalization, functional shifts lead, and structural change follows” (DeLancey 1994: 4). While this dissertation is grounded within the theoretical stand point provided by functional approaches, I take functionalism as a starting point, and not as an end point. It is not my goal to prove a theory or to make it look better than others, for “theory informs and shapes, but does not control” (Rice 2006: 262). Rather, I seek to build knowledge from empirical data and transmit that knowledge (of the Wampis language) as clearly and as comprehensively as I can. From that perspective, the grammatical description proposed throughout this work generally follows a tradition that has been referred to as Basic Linguistic Theory (Dixon 1997). In this approach to 33 grammar description, the researcher aims to describe a set of facts with no restrictive theoretical implications (Dryer 2006). This is especially important in analyzing an under-documented language such as Wampis, because it allows the researcher to be open to new phenomena and to formulate statements that are supported by the data as well as to test typological claims. In this regard, and from a practical standpoint, my analysis of Wampis has also bene"ted from consulting literature on writing a grammar, such as Payne and Weber (Eds.) (2006) or Ameka et al. (Eds) (2006) 1.10. Methodological considerations: !eldwork and data This dissertation is based largely on primary natural data gathered from extensive periods of "eldwork. Underlying this idea is the "rm belief that "eldwork is an indispensable methodological activitiy in writing a grammar, as this is the principal way to study a language as a holistic system, i.e. as a system that functions within a socio-cultural human network, as well as within historical and political contexts with ties to the past and, as I have been reminded many times by the Wampis, to the future. 1.10.1. Fieldwork Fieldwork was conducted in the Wampis communities of Puerto Galilea, Huabal, Boca Chinganaza and Candungos, all located in the political Department of Amazonas, 34 Peru. Additionally, I worked in the towns of Jaen and in Peru’s capital, Lima. In Jaen and Lima there are small populations of Wampis immigrants, some of whom accompanied me to the Wampis villages. Figure 1.3 shows the locations of the Wampis villages where I carried out "eldwork. I have been working with Wampis speakers since 2007, and made trips to their communities in 2008, 2011 and 2013. To go there from Lima involves at least three days in di$erent climate/altitudes, transportation by bus, car, boat; and diverse (usually not good) road conditions. Typically, the route that I followed was Lima-Jaen-Bagua-Santa Maria de Nieva. From the town of Santa Maria de Nieva one takes a boat to the Santiago River district, where the Wampis territory is located. Because it is a long trip, some extra considerations were given to equipment and lodging. I established the town of Jaen as my “base camp” for going into and arriving from Wampis territory. Jaen is located in the Department of Cajamarca, very close to the Department of Amazonas. In Jaen, I stayed with a local, non-Wampis, family, but worked with di$erent Wampis speakers who were currently living there. It takes at least one more day to arrived from there to Santa María de Nieva, including a stop in the town of Bagua. In Santa Maria de Nieva, I would spend the night and go by boat the 35 next day to the Santiago river area, where the Wampis villages are located. Figure 1.3. Map of "eldwork sites 36 Some work would be done in Jaen both before and after I entered the Wampis area for "eldwork. In Jaen I usually made contacts, backed up data, checked up the data collected and made preparations for continuing with the trip. I also worked in Lima with speakers of Wampis in-between "eld trips throughout 2013. This work would typically include transcribing, translating and re-checking texts and other materials collected, as well as doing additional recordings, backing up the data and inserting them into a Fieldworks Language Explorer (FLEx) database. During this time in the city, I would meet with my Wampis teachers a minimum of two times a week (for two-three hours of work at a time). Overall, I spent approximately 10 months in the "eld, plus approximately 2 months in Jaen and 6 months in Lima working with one or two Wampis speakers at a time. My "eldwork was supported by a Summer Research Grant from the Center for Latino/a and Latin American Studies of the University of Oregon in 2008 and an NSF Doctoral Dissertation Improvement Grant for the 2012–2014 period (Award #1226222). Fieldwork trips to Wampis communities are listed in Table 1.4, the approximate duration of each stay is indicated in parenthesis. 37 Table 1.4. Field trips to Wampis communities Date (approximate duration) Communities visited June-August 2008 (8 weeks) Puerto Galilea, Boca Chinganaza July-August 2011 (4 weeks) Puerto Galilea, Boca Chinganaza February-March 2013 (7 weeks) Puerto Galilea, Huabal, Boca Chinganaza April 2013 (4 weeks) Puerto Galilea, Boca Chinganaza May 2013 (3 weeks) Puerto Galilea, Huabal July-August 2013 (6 weeks) Puerto Galilea, Candungos October-December 2013 (8 weeks) Puerto Galilea, Candungos There are many constraints on "eldwork. Probably, time and budget are basic concerns for most "eldworkers. In addition, the "eldwork area for this research shares the following constraints with many other impoverished parts of the world: no or very restricted electricity, restricted means of communication, unreliable and sometimes unsafe transportation, no potable water/sewage system, poor sanitary conditions and minimal health assistance, and so on.27 In fact, the Santiago River area, according to the 27. At the time of writing, a restricted cellular phone network exists in Puerto Galilea. In other communities along the Santiago River there is very restricted or no access to cellular networks. Usually houses do not have phone lines, but there is a communal satellite phone in each locality. Some communities have improvised basic potable water systems carrying water from streams, but access to these systems is not general for all the population. Some villages, like Puerto Galilea, have electricity from around 6 p.m. to 9 p.m., while other communities must rely on gas generators for events needing electricity. 38 poverty map elaborated by FONCODES28 (2006), is one of the top 15 poorest districts in Peru by index of deprivation. According to the poverty map elaborated by INEI29 (2010), the Santiago River district is overall one of the poorest in Peru (placed at #64 out of 1836 districts)—approximately 89% of the population lives in poverty condition. Despite the potential problems arising from the lack of so-called “modern” commodities, I have been warmly welcomed by my Wampis hosts, have lived with them, shared at their tables, been considered as a brother to some of them and generally have been able to nurture a standing relationship of friendship, mutual respect and commitment toward Wampis society. If anything, their stance in the face of hard living conditions shows their admirable ways of adapting to their environment, an exemplary determination for sustainable development and their rich cultural persistence for seeking a “life of abundance”.30 I am of the opinion that the pretension of being a professional or scientist who only cares about his research is unviable—much less in a "eldwork situation. While my 28. Peru’s Fund of Cooperation for Social Development (Fondo de Cooperación para el Desarrollo). 29. Peru’s National Institute of Statistics and Informatics (Instituto Nacional de Estadiśtica e Informat́ica). 30. To “live in abundance” is a cultural concept in Wampis cosmovision. See §2.7 for more details. 39 research focus has always remained centered on doing a profesional job documenting and analyzing the language, it is impossible not to realize that the linguistic researcher interacts in a multidimensional socio-cultural network when in the "eld. The relation between researcher and “subject”, no matter how grand the research, is only one thread of that network. I try not to dissociate being a linguist, or a scientist, from the set of greater responsibilities that arise in human relationships (just as my Wampis friends felt they had more responsibility toward me than just teaching me their language). I have tried to act accordingly. During my stays at the Wampis communities, I was given a room within my hosts’ family house, where I was able to work and read in my “free” time. I partook in the everyday life of my host families and of the communities. I have thus participated in the most diverse activities: from bathing early in the morning and before dusk at the river, trying to help in the chacras (swidden gardens) or going "shing, to conversing familiarly around a “masatito” (manioc beer) bowl. I have accompanied people to meetings with regional authorities, met with local school teachers, participated in “social dances” (in practice, cumbia or chicha music parties), helped seek communication with a congressman in Lima to talk about the creation of a new district in the Rio Santiago area, and have even been encouraged to "lm an entire 40 football tournament event during Rio Santiago’s three-day anniversary celebrations. Through such activities, I have gradually come to have a better understanding of Wampis society. Of course, this has not been an easy job: I have made numerous mistakes and have not always found people enthusiastically receptive of all my endeavors, but for every misstep, there has always been a friendly Wampis hand to support me and to give me yet another humbling lesson in life. My experience with the Wampis people is one of the most important things to happen in my life, and it has helped me become a better person. During the "eld research time, many people worked with me and helped me understand the Wampis world. At the risk of being unfair to others who also collaborated with me, Table 1.5 on next page ventures to list the main contributors to the present work (under "role", I have tried to indicate, however inadequately, some roles that the named individual played in this project). Ethics and "eldwork is a valid and valuable object of study, and there is a great deal of good literature about it—some of it has been quite useful for me (Hale et al. (1992); Grinevald (1998); Hale (2001); Gil (2001); Fleck (2008); Czaykowska-Higgins (2009); Jansen, Underriner, & Jacob (2013)). In general, I have tried to adhere to best 41 practices in my own "eldwork, acknowledging community member collaborators as experts, incorporating local people to co-work with me and facilitating training in language documentation. In addition, the faculty at Department of Linguistics at the University of Oregon excels at linguistic "eldwork, and many formal and informal discussions with my professors and peers also taught me with useful ways to establish meaningful and collaborative connections during "eldwork research. Table 1.5. Primary Wampis teachers and expert collaborators Name Sex Age Role Community of origin Óscar Jimpikit M 44 Teacher, story-teller, historian, culture expert, translator, ethnobiology expert Candungos Dina Ananco F 30 Teacher, translator, story-teller Huabal Clara Navarro F ?? Wise woman, culture expert, story-teller Puerto Galilea Shapiom Noningo M 55 Wise man, teacher, story-teller, culture expert Puerto Galilea Juan Luis Nuningo M ?? Wise man, teacher, story-teller, historian, culture expert Puerto Galilea Atilio Nuningo M 33 Teacher, translator Puerto Galilea Josue ́Yacum M 35 Teacher, translator Boca Chinganaza Lidia Wajarai F 45 Teacher, culture expert Candungos 42 Another interesting issue that arose during my "eldwork is the importance and value of responding to suggestions and felt needs of community members in data collections and language events. One thing that I do not usually "nd in the published literature is an exercise on detailing what could be the perspective of the speakers of a language during "eldwork. Usually the speakers’ perspectives are displaced to a secondary role to give room to the researcher’s description of his/her own project (which is necessary but not the only important thing). What do speakers of a language want, speci"cally, in a language project? Answers, of course, will vary from speaker to speaker and from project to project. Perhaps because it is obvious (and we always lose track of the obvious), it somehow took me by surprise to "nd that the Wampis wanted to do things with their language, above all. Surely like any other community of people, they are very conscious of their culture and language. Most of my Wampis interlocutors agreed that to have a reference grammar for their language would be very important— it is a serious endeavor for them. But the idea of a grammar is somewhat abstract and di#cult to grasp, even for the grammarian! It was in the little struggles of everyday life, though, that I realized that there were many language-related concerns the Wampis people wanted to deal with: they want a dictionary; they have problems translating 43 laws and documents as the language has yet to develop jargon for speci"c "elds; there are people interested in recording voice or video to narrate stories for their children; they need native-language texts for school; some wanted to tell the Wampis perspective of recent events in Peruvian history so that it be on record for future generations; people would ask me to create a CD with the story they had told me or the songs they had sung so they can listen to them afterwards; some would ask me to record a message in my computer or camera to deliver to a relative in Bagua, or Jaen; and so on. Most Wampis speakers agreed to teach me and collaborate with me because they wanted to give something back to their community and to their children: they feel my research would not a$ect them because they were already adult or “old”, but it could bene"t the new generations. There was a sense of cultural awareness that needed to be passed onto the future. Thus, language is indeed a constant concern in the life of the Wampis. Their concerns have been also incorporated in the current work and have actually expanded my knowledge of Wampis. A speci"c morpheme, the allative, for example, which were not yet present (the allative is not frequently used in general) in the natural speech data I had yet recorded, appeared in one text that one of my teachers suggested we record. He wanted to have a written history of the foundation of his community, Candungos— 44 in part because his father had been one of the founders. His story was transcribed, printed and photocopied at a later time in the nearest town where there was access to printers (Santa María de Nieva), and handed back to my teacher. What is more, following up with the discussion of the story, my teacher also told me that his father had made friends with a Spanish gold miner and rubber tradesman from the late rubber boom times (1930’s-1950’s). Thus, new texts were collected and a discussion of those troublesome times from the Wampis perspective ensued. This is an example of how engaging in a community and giving a proper place to community speakers’ needs expanded the practical and intellectual horizons of this project. 1.10.2. Data, data gathering and processing The bulk of the data presented in the current work comes from the analysis of about 10 hours of transcriptions of texts in the Wampis language, plus data written down in seven notebooks. Data in notebooks usually contain vocabulary lists, notes on diverse aspects of the language (from word pronunciations to speci"c syntactic constructions), a few texts that could not be recorded with the voice-recorder, a few text transcriptions of recorded material, cultural notes and elicited data. The texts recorded belong to various genres, with the majority being 45 monologues comprising mythical and historical narrations, but there are also biographies, procedural texts, descriptions, expositions (revolving around Wampis cultural themes) and oratory. There are also conversations, jokes, songs31 and even three short tongue-twisters that I was taught “to better my Wampis pronunciation”. While this dissertation relies on the analysis of natural texts, I have also used elicitation as a methodological tool to gather additional data and to understand the data. For elicitation, I used two types of “sessions”: the "rst type were sessions dedicated to elicitation—I would sometimes reserve at least one hour of work to elicit data with my teachers. The second type was more informal, but more consistently carried out: elicitation done in the process of text transcription and translation. Invariably, we would always "nd that a certain phrase or word could be said in an alternative way. I readily wrote down these observations in a notebook as well as elicited phrases that helped us investigate where and how these alternative constructions could be used. For elicitation tasks, I tried not to rely only on translations of Spanish into Wampis, but rather would often propose a Wampis phrase and ask if it 31. I would like to thank Dina Ananco Ahuananchi, one of my Wampis teachers, for allowing me to digitalize four old cassettes with music in Wampis and Awajún that belonged to her family. Part of those cassettes’ contents were added to my database to be analyzed. 46 was good/grammatical. However, I would always make an annotation that the form was suggested by myself "rst, and the validity of that form was always double check with di$erent speakers for greater accuracy in the description of the speci"c form. Elicitation was also used to "ll gaps in my data, "ll out paradigms and to probe aspects of certain construction when necessary. By eliciting, I also found a few constructions that were not present in the recorded texts. In order to record good-quality data, particular recording equipment was used. The basic equipment taken to the "eld consisted of voice recorders (Marantz PMD660 for the 2008 and 2010 "eld trips, Zoom H4N for 2013), a condenser lavalier microphone (AudioTechnica AT831B), a head-mounted microphone (Shure SM10A-CN), a hand/table microphone, and a semi-professional camera capable of recording high de"nition videos (Panasonic DMC). Additional equipment used included studio monitor headphones, a laptop computer, an external hard-drive, notebooks, and an assortment of other devices (memories, dvd disks, cables, etc.). During the trips, the equipment was put in waterproof cases; some equipment, like the external hard drive, were chosen because they were rain and fresh water resistant. Physical back-ups were constantly made using the external hard-drive and other memory devices. Once in Lima, I also 47 backed-up the data using an on-line internet service. An additional observation concerns the language via which I communicated with Wampis people and into which the data was translated. My "rst interactions with the Wampis people were in Spanish, until I gradually became more con"dent in using Wampis at least in basic communicative situations. Although I am a native speaker of Spanish, I am aware that my variety is not the same as the regional Amazonian Spanish or the Spanish spoken by the Wampis. Many younger Wampis generations (ages 15-40 years) are also more or less /uent in Spanish. Most male Wampis between the ages of 40 to 60 speak Spanish with di$erent degrees of /uency. The number of Wampis female speakers who can speak Spanish in this age range vary, as not all of them had access to education in the past. Elder female speakers are usually monolingual in Wampis and know only a few basic phrases in Spanish. Elder male speakers' knowledge of Spanish varies depending on whether they had access to primary or in few cases secondary school. Young children (ages 0 to 5) usually do not learn Spanish until they start school (though they are exposed to some Spanish, especially from radio shows). Thus, Amazonian Spanish and Wampis language interact: the Spanish acquired by the Wampis is Amazonian but with a Wampis substrate. This Amazonian Spanish with a Wampis 48 substrate is the variety that was used for translations in most cases (one of my Wampis teacher masters the Spanish spoken in the Coast of Peru, in Lima—where she attended classes at a local university—, which is my own variety of Spanish). Especially during my "eldwork in 2013, my teachers and co-workers often helped me ask questions or explain the tasks in Wampis to other Wampis speakers. They also acted as interlocutors to make speakers feel more comfortable and generally to set a more natural environment. When I was the only interlocutor, I tried to interact with interviewee(s) in Wampis so that they did not feel they were in an “arti"cial” setting. In general, I tried to establish a recording setting in which the speaker would speak to an interlocutor and not to a microphone. Even in cases where I myself did not use Wampis properly, I managed to create a more comfortable atmosphere to work and people felt more compelled to expand their role to supplement information and teach me how to say in Wampis what I wanted to communicate. This, in turn, opened additional opportunities to ask more questions about the language. However, in cases when I was not able to convey my thoughts using Wampis and felt communication was broken, I resorted to Spanish to try to maximize understanding of the task at hand. Transcription was done always with the help of my teachers, all of them native 49 speakers of Wampis. Literal and free translations of the texts were given in Spanish by my Wampis teachers. As I grew con"dent in the Wampis language myself, I started to propose translations of parts of the texts with the supervision (and not a few corrections) of my teachers. This was done as a personal exercise to elicit data and to gain a better understanding of the language rather than to actually translate texts—my teachers’ translations were always considered to have authoritative status. I noticed that most people were comfortable with the voice recorder and did not mind the lavalier microphone, so these equipments were my "rst choice for recording in most cases. Recording of words and phrases for phonetic analysis was done mostly in the city of Lima, where I could secure a quieter environment to work in. The work/ow for processing the data was divided into several steps. After recording in the "eld, I backed up the data and asked the speaker(s) for a free translation or explanation of the just-recorded text. Translations or explanations were not always possible, as speakers would go around on their own errands, but when speakers were available I would record or write down the translation or explanation o$ered. Generally, I did not immediately transcribe the texts, but I invited the interviewee to listen to the recording, provide a free translation if possible and talk with 50 me about the content. I always took notes during these occasions. In a few cases, explanations were given in Spanish. At this time, I also wrote down notes about grammar structures or interesting language use that I captured while re-listening to the recording two or three times. Finally, I took note of phrases that speakers felt needed to be corrected because they considered they were “ungrammatical”, “did not sound good”, or because there was an error of performance. Transcription and rechecking of translations into Spanish usually happened in Jaen and Lima with other Wampis speakers. This also proved to be a useful exercise to double-check for possible errors. I used ELAN (https://tla.mpi.nl/tools/tla-tools/elan/), a free transcribing and annotating software program, for transcriptions. In ELAN, I also later translated the text into English. Then I exported the ELAN "le into FLEx (http:/ /"eldworks.sil.org//ex/), a freely available software for language documentation and analysis. The databases created in FLEx contain both a lexicon and an interlinearized collection of texts. The interlinearized texts contain a practical orthography line, a phonetic analysis, a morpheme-by-morpheme analysis, a literal translation in Spanish, and free translations in Spanish and English, plus they include additional notes explaining either grammatical or cultural points. 51 The outcomes of the project include: • A grammar of Wampis (the present dissertation). • A FLEx database, including a vocabulary that I hope to publish in the following years. • A collection of traditional songs. • A collection of /yers and pamphlets that were (and will be) distributed among community members at the requirement of the people—most of these contain narratives about di$erent events and issues in Wampis recent ethno-history. • In addition, in co-supervision with my Wampis teacher Dina Ananco, we are preparing a collection of stories for children, which we will seek to make available in audio and, if possible, video. Once the project is "nished, the material collected (video, voice recordings, ELAN transcriptions and FLEx database) will be archived in the University of Oregon Scholar’s Bank, as well as in the Ponti"cia Universidad Catolica of Peru’s Digital Archive of Peruvian Languages. 1.11. Typological pro!le: The Wampis language at a glance To "nish this introduction, here is a summary of some of the highlights of the Wampis language. These and other exciting features of Wampis will be analyzed in this 52 dissertation: • Wampis has a four vowel system typical of Jivaroan and Amazon languages: /i, ɨ, a, u/. The language distinguishes between nasal and oral vowels. • Wampis exhibits a very complex prosody. The acoustic feature of high pitch is associated with a high tone that comes from two distinct sources: one is a high tone associated with a rhythmic pattern (a “metrical high tone”) and the other is a high tone that is associated to the lexical representation of certain morphemes, including lexical roots, a#xes and clitics (a “lexical high tone”). In addition, high tone has functional load in the language: it serves to mark the genitive and vocative cases, as well as the locative. The language also exhibits complex nasal prosody (there is a nasal domain through which nasality spreads), and nasality also has functional load as it serves to mark some possession and switch-reference. • Wampis is an agglutinative and cliticizing language. The accusative case and all adpositions are clitics. There is only one pre"x. • Plurality is marked in the verb, not in the noun. When a noun needs to be speci"ed for number, a pluralized copula is used. There is no grammatical gender. • Wampis is head and dependent-marking and has a nominative/accusative pro"le. 53 • Wampis has a typologically uncommon object marking pattern: third person object noun phrases are marked when the subject is a "rst singular, third singular or third plural person; on the other hand, third person objects are not marked when the subject is a second singular, second plural or "rst plural person. First and second person objects are always marked. • Wampis verbal morphology is pretty complex. There is a set of what I call “aktionsart” su#xes that can derive new meanings and are also used in certain morphosyntatic environments to mark perfective aspect or to use the verb in the future tense. There are at least two morphological causative markers: one is the only pre"x present in the language; the other is a su#x. There is also applicative morphology. • The tense system distinguishes several points of reference to the past and future. Grammatical nominalizations are used in some of these distinctions, having formally nominalized non-in/ected verbs as the main verb of a clause. The marking of mood and modality is quite re"ned, with many distinctions made (declarative, uncertainty, inferential, mirative, sudden realization, and so on). • Wampis possesses a unique hierarchical agreement pattern, which as far as I know is only found within the Jivaroan family. In the Wampis hierarchical agreement, when a 54 third person agent acts upon a "rst plural or second plural patient, the verb agrees with the patient (the object) and not with the agent (the subject). • Wampis exhibits a feature of symmetrical objects; i.e. the object noun phrases of ditransitive verbs are both marked identically. However, only one object can be marked on the verb. • Wampis has very sophisticated hypotaxis strategies, which include a rich switch- reference system and tail-head-like constructions. Wampis possesses a set of non- canonical switch-reference markers that is cross-linguistically very rare. On the other hand, there is very little evidence of true coordination. • The kinship system is based on gender-ego and thus some kinship terms di$er for male and female speakers. 55 CHAPTER II THE WAMPIS PEOPLE 2.1. Introduction This chapter provides a brief ethnographic summary of the Wampis culture, as background for the primary focus of the entire work, which is the linguistic structure of the Wampis language. Throughout history, Jivaroan peoples have shown a remarkable sense of freedom and cultural identity. The secluded territory that the Wampis inhabit, surrounded by other Jivaroan groups, mountain ranges and rivers of di#cult access, have allowed them to resist colonization at di$erent times. Resistance against di$erent outside powers constitute a most important point of Wampis ethnic pride. At the same time, that protective ethos and relative isolation have made the Wampis history and culture little known, when not misunderstood or manipulated by outsiders. The structure of this chapter is as follows: §2.2 presents a brief ethnohistory of the Wampis; §2.3 is a description of the current situation of the Wampis people; §2.4 describes some important points of their material culture, and §2.5 brie/y notes the head-shrinking practice by which the Jivaroan peoples are well known; §2.6 describes their main economic activities; §2.7 describes their traditional spiritual beliefs; §2.8 56 describes the Wampis oral tradition; §2.9 discusses several points regarding cultural transmission and social organization among the Wampis. 2.2. Brief ethnohistory of the Wampis While the proto-history of Jivaroan peoples is little known, there is considerable evidence of prolonged occupation north of the Upper Marañon area by the ancestors of modern Jivaroan groups. Archeological studies have shown that the territory occupied by modern Jivaroan peoples witnessed early development of complex cultural processes. In the area between the Marañon and the Zamora-Chinchipe rivers, there was a local culture that already engaged in trade with cultures from the Paci"c Coast and the Andes approximately 5,000 BP32 (Valdez 2013: 71). People of Jivaroan ancestry seem to have stably inhabited both sides of today’s Ecuador and Peru border for many centuries. Rostain (2010) has linked the proto-Jivaroan culture directly with the Huapula culture in Ecuador (800–1,200 AD), and through Huapula to a successive cultural sequence in the Upper Marañon highland jungle that dates back to about 700 BC. According to Rostain, the Huapula would have migrated from the Sangay volcano area, central Ecuador, /eeing an eruption in 400–600 AD, thus spreading and 32. I use ttime scales as they appear in the referenced works. 57 populating modern Jivaroan territory. It seems likely that as these proto-Jivaroan peoples spread further towards the Marañon they engaged in contact with other cultural traditions. According to Morales (p.c. cited in Pitman et al. (2012: 289)), ceramics found in the village of Candungos, on the Santiago river, were identi"ed as belonging to the Chambira culture, which has been dated 3,500–2,000 BP (Morales Chocano 1998). Other studies have shown ancient human occupations with complex organization in the Marañon, Corrientes, Upano, Paute and Upper Santiago areas, all occupied for centuries by Jivaroan groups. Corrugated-style pottery vestiges attested from around the 8th century AD seem to indicate a singular archaeological horizon of complex social and cultural di$usion that has been directly related to the dispersion of proto-historical Jivaroan groups.33 This would connect Jivaroans with cultural processes also found in the Marañon, Ucayali and Upper Amazon; i.e. processes related with other linguistic families such as Arawak, Panoan and Tupí-Guaraní (Gu$roy 2006). It is reasonable to think that the genesis of modern Jivaroan culture is thus the contact between these traditions that concurred in the Marañon basin. This connects with the hypothesis that Amazonia saw cultural development that in/uenced the Andes and the Coastal regions 33. Myers (2005) reports corrugated style pottery currently in use by the Wampis. 58 of Peru, a theory that was put forward by Lathrap (1970). Furthermore, it is possible that before the Inca ruled in the northern Andes, Jivaroan languages served as lingua franca, as the Jivaro formed a bridge between the coastal regions of Guayaquil and Tumbes (South Ecuador/North Peru) and the Amazon (Whitten, Jr. 1976), having even reached the Andes of today’s Piura in the coast of Peru (Hocquenghem 1989; Espinoza Soriano 2004). Guallart (1990) and Regan (1999), based on the analysis of iconography and myths, have also speculated on possible contact of the Jivaroan with the Mochica, a powerful ancient culture of the Peruvian northern coast. According to said authors, the Jivaroan myth of the giant cannibal Iwa (/iua/),34 who in the stories eats the Wampis and Awajun until he is "nally killed by the Jivaroan cultural hero Etsa ‘sun’ (/ɨt͡sa/̃, is based on their memories of the war of resistance against the Mochica. This hypothesis is based on archaeological studies cited by Guallart and Regan that propose that the 34. In this chapter, I will use the Wampis alphabet to write proper and common nouns. Underlying forms using IPA symbols are provided between slash symbols / /. The Wampis alphabet has the vowels (where represents /ɨ/), and the consonants (where = /t͡ʃ/, = /h/, = /ɲ/, = /ʃ/, respectively). The glides written and are considered allophones [w] and [j] of vowels /u/ and /i/ in the present grammar. In subsequent chapters, which are dedicated to the linguistic analysis of the language, I will only use IPA symbols (not alphabetic symbols) for representing Wampis words. 59 Mochica traded gold with the Marañon region. Memories of contact with other pre- hispanic groups have also remained in Wampis folklore. For instance, they often tell stories about encounters with the Wampukus, who are said by the Wampis to be uncontacted people of the jungle. Known to the Spanish as Huambucos, they were a historical group who lived around Tomependa, in the Jaen basin. The Huambucos were decimated by wars against the Spanish, as well as by interethnic con/icts with the Wampis and Awajun. Their remnants mixed with the people that settled in Spanish- founded towns in the Upper Marañon. By the early 20th century, the Huambucos had already disappeared as an ethnolinguistic group. Before the arrival of the Spanish, the Incas had tried to conquer the Jivaroan peoples. Though the Incas imposed their rule in the northern Andes and managed to assimilate some Jivaroan groups there, they were unsuccessful when they tried to enter the jungle towards the rich regions of the Chinchipe and Santiago rivers (the Santiago is considered by the Wampis to be their ancestral land). There, Jivaroan groups allied to defend their territories. Colonial chronicles tell of at least two Incan attempts to conquer them, one by Inca ruler Tupac Inca Yupanqui and the other by his successor Huayna Capac. These attempts were frustrated by the Jivaroan coalition, which 60 delivered an unprecedented defeat to the powerful (and otherwise unbeatable) Incan army. Soon after their conquest of the Incan Empire, the Spanish organized several expeditions to the East of the northern Andes, entering through the colonial towns of Loja and Zamora (now Ecuador) and Jaen de Bracamoros, Chachapoyas and Moyobamba (now Peru).35 At that time (mid-1500s, early 1600s), this was, ethnically and linguistically, a diverse territory in which the Jivaroan were one of the main groups (Espinoza Soriano 1973; Taylor & Descola 1981; Taylor 1991; Torero 1993; Torero 2002: 273–298). There the Spanish found the Paltas, Malacatos, Xorocas, Guayacundo, Bracamoros, Rabona, Tontones, Jivaros, Giuaras36, all of whom have been described as sharing cultural traits that have become characteristic of modern Jivaroan peoples. These groups were neighbors of other ethnolinguistic groups which are extinct today: the Bagua, the Patagon, the Chirinos, the Huambucos, the Cañar,37 the Tabancales, the 35. All of these towns exist today, but their locations di$er from the "rst locations where the old colonial towns were "rst founded. 36. or is another spelling for Proto-Jivaroan *ʃ(i)uara ‘person’, though the name was later applied to the inhabitants of the lower and middle Santiago River; i.e. the ancestors of the Wampis. 37. While the Cañar language is extinct, ethnic Cañar people can still be found in today’s Ecuador and Peru. 61 Pericos, the Sacatas, among others. The term Jivaro quickly became an umbrella term to identify the ancestors of all Jivaroan groups. From the Peruvian/Ecuadorian Andean region, the Spanish organized several expeditions into the Marañon in the second half of the 16th century. News of gold, rich lands and kingdoms animated these "rst expeditions. Looking for ways to establish their domains in these new lands, the Spanish distributed “newly discovered” lands in encomiendas and later in repartimientos38 where the indigenous peoples were used as laborers and payed tribute. However, the Spanish found "erce resistance from the Jivaroans and other peoples inhabiting the region. After several unsuccessful attempts by other explorers, Juan de Salinas y Loyola led his famous entry into the Marañon in 1556. Following the downward course of the Santiago River, at the heart of the modern Wampis territory, he founded the town of Santiago de las Montañas in 1557, and then continued down into the Marañon and the Ucayali Rivers. Santiago de las Montañas was "rstly founded in the main course of the Santiago, but later Salinas decided to move the former town to near the con/uence of the Santiago and the Marañon, in a 38. The encomienda was a system that granted a Spanish person a number of “Indians” to be under his responsibility. The repartimiento was a forced labor system. In practice, both encomienda and repartimiento systems could di$er little from slavery, and there are many accounts telling of the abuse of the Spanish, one of the main reasons why the Jivaroan rebelled. 62 place called (Cuesta, S.J. 1984: 113). It is in this context that the "rst encounters of the Europeans with the ancestors of the Wampis, who had successfully blocked the way to previous expeditions, occurred. Santiago de las Montañas was the only Spanish town on the Santiago river during this time. Over the subsequent years it would be attacked by the Wampis, depopulated and completely abandoned. The Santiago and Chinchipe river basins appeared to have been particularly rich in gold, and its extraction dominated the economic activities of the region once they came under Spanish dominance (Santos-Granero 1992: 93). The Spanish founded several towns surrounding the territory as they colonized: Logroño, Sevilla del Oro, Jaén de Bracamoros, San Ignacio were located to the North and West of the Santiago, and Santa María de Nieva, San Francisco de Borja (the capital of the colonial Province of Maynas) and the aforementioned Santiago de las Montañas were established to the South. However, the Spanish e$orts to settle in these "rst towns and exploit the natural resources that existed in the region (forcing the native population to serve as laborers) faced strong opposition from Jivaroan peoples. The pre-Wampis and pre-Awajun were among the "rst to revolt in Santiago de las Montañas in 1569. Several other revolts 63 happened in the second half of 16th century, all led by the well-organized Jivaroan. These revolts culminated with the legendary destruction of the town of Logroño, near the con/uence of the Paute and the Zamora rivers (now belonging to Ecuador), northern tributaries of the Santiago, in 1599,39 by a confederation of Jivaroan groups from the Paute, Santiago and Morona rivers. A subsequent Jivaroan attack also partially destroyed the town of Sevilla del Oro. These actions stopped the Spanish attempts to conquer that territory. Incapable of conquering the Jivaroan people, the Spanish resorted to correrías (surprise raids) to “pacify” and capture children and adult men to be used as laborers through the 17th century (Santos-Granero 1992: 158). In the 17th and 18th centuries the Spanish also tried to establish reducciones, or missions, to convert the Jivaroan peoples to Christianity. In that spirit, the Jesuit entered the Marañon in 1638. However, 39. Scholars agree now that the destruction of Logroño was exaggerated by Juan de Velasco, a Jesuit and historian of the 18th century, who served as the source for the legend that made the Jivaro melt gold and pour it into the mouth of the governor of the town of Macas. According to the legend, the governor of Macas was visiting Logroño when the Jivaroan attacked and punished him for his ambition. Velasco speaks of Logroño as a city where at least 20,000 people were killed in the 1599 attack. Such “cities” in reality were no more than small settlements with a small population of Spanish. It is likely, according to Descola and Taylor (1981), that the revolt involved not only Jivaroan but other indigenous groups, specially the numerous (at that time) Cañar. 64 they were mostly unsuccessful in founding missions. The only mission established in the Santiago River was the Mission of Naranjos, established in the 17th century, but it was of ephemeral existence. Unable to move the native inhabitants into a single town or mission, the missionaries resorted to organizing raids too, using a combination of Spanish army men, settlers and warriors from friendly indigenous groups brought from the Marañon, Ucayali and Huallaga rivers. These raids were usually called cacería de jíbaros ‘hunting of Jivaros’ and sought to force relocation of the indigenous peoples into sedentary towns (Abad Gonzáles 2003: 115). Maroni (1988 [1738]) describes how, in e$orts to avoid being captured, many Jivaroans commited suicide, burned their houses or killed their children, making this one of the most ominous periods in Jivaroan history. A few elderly Wampis still remember stories about bearded people with swords and armor who came and took their children and women. The Wampis attacks against colonial towns in the Marañon and even as far as the Ucayali at that time and during the next century are probably related not only to territorial disputes with the Spanish and outside settlers, but also partly to tribal con/icts that were re-ignited during these cacerías de jíbaros. In additon, during this time, the indigenous population of the Upper Amazon, including the Jivaroan, was dramatically reduced by epidemics, virtual slavery 65 and constant war. As a result of the problems in entering Jivaroan territory, the Spanish stopped their attempts to conquer it. Jivaroan autonomy was further expanded by the suppression of the Jesuits in the Portuguese and Spanish Empire territories in 1767 and the tumultuous pre-independence times in the Peruvian viceroyalty during the last decades of the 1700s and "rst decades of the 1800s. By the beginning of the 1800s, many of the old colonial towns and missions in the Jivaroan region were abandoned or had a small mixed Spanish, mestizo (a person of combined European and Amerindian descend) and indigenous population. By the end of the 18th century the Wampis were already a well distinguished group (Taylor 1994: 91; Costales & Costales 2006: 17). Notorious for their raids against towns on the Marañon river, the Wampis appear in di$erent letters and documents by local authorities40 as well as in the accounts of notable scientist and travelers (e.g. Raimondi (1862)), throughout the 19th century. With Peruvian Independence in 1821, a new wave of colonization attempts arrived. This process was accelerated by the rubber boom (1880-1910) and Peru’s desire to secure the Amazon territory from other 40. See, for examples, the many references to the Wampis in di$erent o#cial documents and letters published by Larrabure i Correa (1905–1909). 66 emergent countries claiming access to it (Ecuador, Colombia, Brazil). The rubber boom era surely was a distressful time for the Wampis, who organized armed raids to the “frontier” towns on the Marañon (the old surviving colonial towns), such as Barranca, Santiago, Santa Teresa or Borja. The city of Iquitos, nowadays the largest city in the Peruvian Amazon, was a small village until Meztizo families from these ravaged towns arrived: they were escaping the ruination of their towns by Wampis attacks in the mid-1800s. In many of these raids the Wampis allied with the Awajun and other groups like the Kandozi. However, the Wampis did not always attack rubber traders and gold washers; rather, they tried to establish trading connections with them. The trade allowed the Wampis to obtain guns, tools and other supplies. Therefore, the Jivaroan raids at this time (and likely in colonial times as well) must be conceived of having not one but several causes: there are several examples throughout history that prove the Wampis are willing to accept outsiders among them as long as their own customs are respected.41 But the news of forced labor; severe punishments; the spread of diseases; 41. The ancestors of the Wampis surely were instrumental during Juan de Salinas’ expedition into the Marañon, and helped him found Santiago de las Montañas. They also helped Father Lucero establish the Mission of Naranjos and a century later they helped famous French explorer and scienti"c Charles de la Condamine pass through the Pongo de Manseriche, near the con/uence of the Marañon with the Santiago river. The abandonment of the missions or the attacks on towns they had surely allowed to be founded in their territory requires a more complex explanation (see Taylor & Descola (Taylor & Descola 1981), Taylor (Taylor 1994) for a 67 abuses to women, children and men; old enmities with other indigenous groups; the erosion of old Jivaroan alliances; the pressure to obtain tsantsa (shrunken heads) to exchange for guns and tools; as well as the inequality in trading all may have contributed to originate these attacks. An outstanding Wampis leader of this time was Tsamarain (/t͡samarainta/), who embodies the epitome of the great Jivaroan leader: intelligent, charismatic, fair, brave, with profound knowledge of traditions. Tsamarain was able to ally the Wampis and Awajun to "ght and eradicate the rubber patrons and their local armies. He was so famous locally that a judge and poet from Iquitos wrote a poem in Spanish in which Tsamarain personi"es a paragon that "ghts the rubber patron’s abuses. A second short-spanned rubber boom occurred during the World War II period. This coincided with the "rst Ecuadorian-Peruvian War (1941). New abuses by rubber patrons and the Peruvian army, which started to establish stations and patrol the borders in the area, occurred. Memory of the last great traditional Wampis leaders that fought the army and police to stop their abuses come from these times. One of these last great leaders was Piruch (/pirut͡ʃa/), who apparently was met in his late years by reconstruction of the historical on-and-o$ relationship of the Jivaroan with the Spanish and other groups). 68 American explorer Lewis Cotlow (1953).42 In the Wampis stories, Piruch is a leader who, tired of abuses against women and men (including his own brother, who was killed), battles the military until a peace agreement is reached. 2.3. Current situation: The Wampis today During the 1960’s and 1970’s, the Wampis entered a new era in their relationships with national Peruvian society. The increasing number of merchants, the more continuous presence of missionaries and the growing national concern to secure the borders after the 1940’s con/ict with Ecuador led to profound changes in Wampis society. Access to school and technical education was introduced through missionary work, especially Summer Institute of Linguistics (SIL), who established a bilingual education program. They were successful in introducing literacy to a large proportion of the Wampis population. From the 1960’s the "rst modern Wampis villages in the Santiago and Morona were founded as a result of the desire to establish schools and medical posts. The Wampis, who were used to living in small family nuclei scattered in the jungle, started a quick process of relative “sedentarism” along the margins of the 42. Interestingly, Cotlow mentions a Wampis leader , whom Cotlow held in very high regard. The time Cotlow published his book coincides with the time the Wampis leader apparently lived. 69 main river (the Santiago), where most new communities were established. The passing of a Peruvian legislation commonly known as Ley de Comunidades Nativas (Law of Native Communities) in 1974 declared the property of Amazonian lands as inalienable from the indigenous groups and non-seizable by outsiders. This allowed new legislation that opened the opportunity for the Wampis to buy lands and to establish their "rst political organizations—the most important of them at this time was the Consejo Aguaruna Huambisa (Awajun Wampis Council), one of the principal predecessors of Peru’s largest indigenous organization, Asociación Interétnica de Desarrollo de las Selva Peruana (AIDESEP). Over the years, the communities have worked to obtain land titles. However, new land laws passed during President Fujimori’s 1990’s government virtually ended the protected status of the indigenous lands in the Amazon. In practice, this opened the native territory to mercantilist activities focused on mineral and logging industries (Abad Gonzáles 2003: 96–100). The clash between the indigenous people’s claim to lawful recognition of their ancestral lands and the government policy of exploiting natural resources disregarding indigenous people’s place in the decision making process has led to several lawsuits and protests over the years. Confrontations escalated in 2009, when protests were held against then-President 70 Alan García’s policies that intended to privatized and accelerate concessions for the utilization of natural resources by private companies. The protests reached their peak with the “Massacre of Bagua” (known as El Baguazo in Peru), where more than 30 people died during a confrontation between the police and protesting indigenous people, mainly Wampis and Awajun.43 Thus the Wampis entered the 21st century amidst tragedy, contempt and mistreatment, a fact that opposes radically with my own experience of relatively peaceful life in their communities. 2.4. Material culture While not highly elaborated, there is an interesting tradition of handmade craftsmanship among the Wampis. Pottery, weaving, the making of weapons, the making of canoes and musical instruments constitute important crafts in their culture. 43. Accounts of the actual death toll, especially that of Wampis and Awajun protesters, and information about police abuses and how García’s government mishandled the situation di$er. Neutral third-party reports, including those of prestigious United Nations Human Rights reporter James Anaya (2009) and Belgian volunteer witnesses who were in the area when the protests occurred (CATAPA 2009), have cast very serious doubts on o#cial numbers and on the government actions before, during and after the tragedy. As an illustration of the contempt with which the government regarded the protesting indigenous people, consider these words of then- President Alan Garciá himself, who declared on June 5 2009, day the Massacre of Bagua started: “these people [the protesting indigenous people] have no crown, they are not "rst class citizens [...to tell Peruvians...] you have no right to come here [to Wampis/Awajun territory]. . . and those who think in that way want to lead us to irrationality, to a primitive regression” (Godoy & Faya 2009). 71 Some of these are considered an art form with religious connotations within the Wampis world view. Clay pottery and gourds are used as utilitarian ware, produced by women. While simple in design, ceramics are decorated red and black with complex (and beautiful) motifs related to some animals like the jaguar and the boa. The Wampis distinguish several types of bowls and jars that serve di$erent functions; e.g., pinin (/pininka/) ‘bowl (generic)’, yukun (/iukuna/) ‘bowl for washing hands’, ichinak (/it͡ʃinaka/) ‘jar (generic)’, muits (muit͡sa/) ‘large jar for manioc beer’. Wooden spoons and stirring paddles complement traditional Wampis cooking-ware. Nowadays, bowls and beautiful traditional handcrafts (mainly seed earrings, feather-work and necklaces) are sometimes sold in small street fairs within the communities. Threading, weaving and basketry is done by men.44 The making of cloths and baskets constitutes a rich tradition among the Wampis. Men weave traditional cloths itip /itipa/ ‘male skirt’, tarach /tarat ͡ʃi/ ‘dress (for women)’ and make baskets and bags for di$erent functions. Some of them are to be carried exclusively by the men themselves (e.g. wampach /uampat ͡ʃi/ ‘type of bag’, pitak (/pitaka/) ‘traveling basket’). Monkey and 44. This is a particular Jivaroan cultural pattern. 72 medium/large-sized rodent skins are cured and are used for making bags too. Men also make brooms. In the past, men who had mastered how to make dugout canoes (kanu /kanu/) were held in high regard. Making canoes is still part of Wampis culture, but, as motorboats are becoming more common, this practice is declining. Notice, nonetheless, that canoes were likely not original of Jivaroan culture. The word kanu itself is of Cariban origin. Men used to travel in rafts (paapan /paapanku/) before acquiring canoes from some other group (probably from Tupi-related groups that traded in the Upper Marañon). A knife made of a type of bamboo (chinkan /t͡ʃinkana/) is another traditional man’s tool. In the past, people attributed power to this knife, and it was said that its owner could communicate with the knife as though it were alive. The knife was also used to cut the umbilical cord of newborns also. Today, older men still keep their bamboo knife but mostly for symbolic reasons. Traditionally, the spear (nanki /nanki/) and shield (tantar /tantara/) were the weapons used for warring. Before the turn of the 20th century, however, some Jivaro had already acquired "rearms, usually shotguns and muzzleloader ri6es (generally 73 called akaru /akarú/).45 But the spear is still a symbol of Wampis identity. Inside the house, the wall in front of the main door is usually decorated with one or two spears horizontally arranged. There is a special wooden stool (chimpui /t͡ʃimpui/́) where the family head sits and which symbolizes his power. The family head sits there to receive relatives, friends or to do business while his wife provides the invitee with manioc beer. In practice, the head of the family usually allows children to sit on it and play; however male invitees and women customarily do not sit in the chimpui as it would be seen as a disrespectful action. Otherwise, people sit on wooden benches carved out of tree trunks (kutan /kutanka/) that are placed on both sides of the table and along the walls. For hunting, the traditional tool used by the Wampis is blowgun (uum /uumi/), though nowadays shotguns are preferred. Blowguns are still used to hunt small game, especially birds and monkeys. The making of a blowgun follows a ritual that involves dieting. Toy blowguns continue to be one of the favorite toys of children, who grow up learning how to use blowguns. Darts (tsentsak /t͡sɨnt͡saka/) are also made by hand; when used, the point is coated with curare poison (tseas /t͡sɨasa/).46 Shamans’ tsentsaks have a 45. The term akarú is apparently a borrowing from Spanish ‘arquebus’. This suggests that the Jivaroan were acquainted with European technology early after their "rst encounters between the 16th and 17th centuries. 46. An interesting print of how the Wampis used to engaged in an exchange network in the 74 special status: every shaman has a set of magical tsentsak used for specialized functions (for instance, to cure, to cast a protective spell, to make rain, to curse an enemy, and so forth). In the Wampis view, the shamanic tsentsak has an “invisible” counterpart that carries out the shaman’s conjure. Musical instruments are ritual objects among the Wampis. The tuntui /tuntui/́ ‘signal drum’, the tsapraka /t͡sapraka/ ‘traverse /ute’, the tuman /tumankɨ/ ‘type of jaw harp’ and the keer /kɨɨra/ ‘violin-like instrument’ occupy a special place in the Wampis world. Similarly to the making of the blowgun, the process of making musical instruments involves fasting and sexual abstinence before and during the making of the object, as well as feasting and celebrating when it is done. In the case of the tuntui, traditionally a young woman must play it for the "rst time. Afterward, it is to be played by an adult man. An interesting musical instrument of the Jivaroan is the tuman, a bow- like instrument with a thread of chambira palm leaf attached to a stick. Similarly to a jaw harp, the tuman is played by placing an end of the bow between the teeth and pulling the string with the index "nger. In general, these instruments are played by men Upper Amazon remains in the Wampis vocabulary. An alternative non-native name for curare is tikuna, the name of an ethnic group who lives near the border between Peru and Brazil. The Tikuna are well known for their curare. 75 and are used in ritual celebrations and when singing traditional drinking and magical songs (nampet /nampɨt/ and anen /anenta/, respectively). Sometimes, the tune of the /ute can replace the anen song, especially if the person is playing alone. During traditional celebrations and parties, women usually sing and dance, and accompany the music with the sound of bracelets and belts made of pieces of snail shells (kunku /kunku/); men wear a rattle-like ornament on their ankles (makich /makitʃa/), made of seeds. Though the Santiago River territory has gold and other minerals, metalworking is not part of the Wampis culture. By the early 20th century, the Wampis engaged in artisanal gold prospecting to exchange the mineral for outside goods such as machetes, guns or clothing. Some Wampis and Awajun still pan small quantities of gold manually. However, nowadays ungoverned gold prospecting and illegal mining are causing environmental damage in their territory. As rivers are polluted with mercury in the process, mining is frowned upon by Wampis society. At times, entire communities have acted to stop illegal activities and expelled outsiders—a practice that, in actuality, has been continually exercised since early colonial times. 76 2.5. Tsantsa ‘shrunken-head’ Jivaroan groups often have been described as one of the most aggressive and dangerous groups in the Amazon. Very early reports during the 17th century speak of their custom of cuting the heads of their enemies and making trophies of them, known as tsantsa (/t͡sant͡sa/). As we have seen, during the 1800s and the "rst decades of the 1900s, the Wampis were considered the scourge of the Marañon and Santiago, as they often raided the small towns and settlements along these rivers. By the end of the 19th century, western explorers and merchants started to trade tsantsa to sell them in museums and private collections. The Jivaro quickly became the focus of popular attention and the uncritical way outsiders took tsantsa contributed to their characterization as infamous savages. The fact is that the practice of head- shrinking—which involved a deep respect for the dead and a very strict ritual—was quickly transformed into an exotic object of Western culture’s fascination. The desacralization of the ritual tsantsa by westerners actually had a negative impact on Jivaroan people: scholars have found that warfare among Jivaroan groups and the death toll actually increased exponentially as trading tsantsa was encouraged by the outside market (Ross 1988; Steel 1999). 77 2.6. Economy and subsistence The main economic activities of the Wampis are hunting, swidden agriculture and "shing. The products of these activities provide them with their principal means of subsistence. Staple crops are banana, plantains and manioc. They also produce several varieties of peanuts, sweet potato, sachapapa (Dioscorea tri)da) and sugar cane. The Wampis garden actually is very complex and many varieties of useful plants (edible and medicinal) can be found there. Each adult man and woman has one or more plots that they work to meet self-sustaining needs. Consumption of meat in the Wampis diet is restricted to a limited number of species (Berlin & Berlin 1983). Communities regulate themselves and can establish periodical bans on hunting and "shing so that animal populations are not severely threatened. I was told that in recent years there have been several cases where communities have imposed a ban on hunting or "shing in certain areas due to waste material from gold miners polluting the streams. Following a traditional custom, food is always shared with the extended family. The nourishing nijamanch (/nihamat̃͡ʃi/) ‘manioc beer’ is the preferred drink for any social occasion, and it is always present in everyday life. The Wampis trade or sell game meat, skins and produce (especially bananas) in 78 the bigger nearby towns, so they "nd small markets in the mestizo town of La Poza (located on the Santiago River, very close to the Wampis town of Puerto Galilea) and in Santa María de Nieva (the capital of the Province of Condorcanqui, Amazonas), as well as in small street markets in the communities. Some sell small quantities of gold in Santa Mariá de Nieva too. The market economy is slowly entering in some local villages, and now it is possible to see one or two small corner stores in some of them. Local projects also include development of "sh farms. The Wampis have been trying to increase the production of marketable crops like cacao and co$ee, which grow naturally and productively in the area, to sell in the towns of Santa María de Nieva, Bagua and Jaen, but the lack of roads and the cost of /uvial and terrestrial transportation impose many di#culties for the moment. 2.7. Traditional beliefs and religion The Wampis have a very rich and complex world view. Singing, dieting and sexual abstinence are practiced as part of several rituals related to di$erent aspects of life, like hunting, farming, celebrating/remembering the dead, curing illnesses and obtaining knowledge through hallucinogenic plants. The Wampis consider that there are di$erent “spirit” realms that are inter- 79 connected: entsa /ɨnt͡sa/ ‘river’ is the water world; nunka /nunka/ ‘earth’, is where humans and animals live; the underworld is the world of Nunkui /nunkui/, related to beings with knowledge of plants and trees; nayaim /naiaimpi/ ‘sky’,47 is where the stars, heroes and souls of dead people live. The Wampis can communicate with the other worlds through dreams and the ingestion of sacred plants such as ayahuasca, datura and tobacco. As in most other Jivaroan groups, the central concept in the Wampis’ world view is that of Arutam /arutama/. The concept of Arutam is very complex. Arutam is a protective primeval spirit of power that ties the Wampis individual’s past, present and future. Arutam is said to own everything in the universe. The power of Arutam is related to primordial times and is usually gifted by an ancestor48 (whom the person does not necessarily know) or another powerful being through dreams. The visions of Arutama also work as an oracle in that it predicts what is going to happen in a person’s life and the person seeks to ful"ll that personal vision. Thus, Arutam o$ers explanations for the 47. Though in general /naiaimpi/ can be translated as ‘sky’, it really means ‘everything that is above the line of trees in the horizon’. Some Wampis translate this word into Spanish as ‘space’ rather than ‘sky’. 48. The term is probably related to the Proto-Jivaroan root *aru- ‘old, be late’. 80 current condition of a person and gives her guidance in life. The Wampis consider that a person who has obtained such a vision is given a certain power that can be observed through the person’s actions, way of speaking and attitude to lead. Such a person becomes a waimaku ‘person of vision’ (etymologically uai-ma-ka-u ‘see-REFL-INTENS-NMLZ’, lit. ‘one who has seen him/herself’) and obtains a status of respect within the community. The quest for possession of Arutam is therefore both a constant motive and raison d’être of Wampis life. In fact, though not a physical object, the possession of Arutam power is arguably the most important personal possession a Wampis can have. The vision of Arutam is sought after with the help of a brew of plants that possess psychotropic properties; chief among them is the ayahuasca vine (natem /natɨma/). The vision produced by the ingestion of ayahuasca does not itself give Arutam power: it is through dreams that an Arutam "nally reveals itself and the person is provided with explanations of the induced visions and then given a certain “power”. Traditionally, the Wampis sought Arutam near waterfalls, considered to be sacred places that propitiate communication with the spirit world. The place where the Wampis rested to dream after taking ayahuasca is called ayamtai /aiamatai/̃ ‘resting place’. All sort of protective entities of the woods are believed to linger in or around the ayamtai. In practice, the 81 Wampis can drink the ayahuasca brew in a place nearby their house (a clearing in the jungle, a garden or a secluded spot near the river) and come back to rest at home. They can also have the ritual in the house of a healer man or a relative who knows the rite. Traditionally, young men would be initiated in the use of ayahuasca starting before their teenage years. A close relative, usually the father or maternal uncle, and a shaman would help the youngster. Arutam have several manifestations; most of them are related to the natural world: kaya /kaia/ ‘stone’, yawaa /iauaa/̃ ‘jaguar’, ipamat /ipamata/ ‘thunder’, panki /panki/ ‘boa’, and so forth, but Arutam may also present itself as a person’s ancestor or as a supernatural being like muuk /muukɨ/ ‘head’.49 Each manifestation can give a speci"c power: for instance, it was believed that a person with Arutam kaya could not be killed by her enemies, and a person with Arutama ipamat is said to possess the power of speech. (Since oratory is very well regarded among the Wampis, a person with oratory skills is set to be a leader.) A person can have more than one Arutam at a time. Another way to obtain Arutam power is through an encounter with Arutam in the jungle, who often appears in a frightening form that the person must “defeat”.50 A 49. This arutam comes in the form of an apparition of a monstrous head with feline attributes. 50. Accounts of what it means to “defeat” in this context vary. Most Wampis say that one must 82 person can also acquire another person’s Arutam by killing him and following a strict ritual that used to involve the process of head-shrinking followed by a period of dieting and subsequent celebration. As warring is no longer part of Wampis life, this is no longer done. In general, a Wampis that dreams of an ancestor or an entity believed to be related to Arutam, or who has an “supernatural” encounter in the jungle, will seek to obtain the potential vision of power by drinking ayahuasca. Nowadays, a professional, educated person (this usually means a person that has gone to a university in a city, preferably Lima) can be regarded as having a vision similar to that of the waimaku of olden days. Wampis leaders (including political leaders) are usually considered waimaku. They combine their charisma with complex political and social networks to rally people in their favor. This is an interesting way in which the Wampis worldview has adapted to modern life, showing that their social system is not rigid. Many Wampis who have migrated to cities have still close ties with their communities of origin and feel bound to represent them, claiming to have “vision”. overcome one’s fear and touch the arutam form. I was also told that one must frighten the apparition or even try to kill it. Either way, the entity will disappear, letting the person know "rst that it is Arutam. 83 To possess Arutam means that a person will achieve a “good life” (ʃiir puhut). A good life is also related to the concept of living in abundance (tarimat /tarimata/).51 In the Wampis world view, this concept is related to the idea of leading a plentiful life within the community, in good health and with all needs covered (especially food). The concept of abundance is linked to nunkui, entities that facilitate fertility and agriculture. 2.7.1. Evangelization Missionaries (evangelical and catholic) have had an intermittent presence in the Wampis area. Since the 1950s their presence has been more stable, though not quite as in/uential as, for example, it was among the Awajun, who have a high rate of Nazarene and Catholic converts; or the Shuar in Ecuador’s Upano region, were the Salesians evangelized them and had an impact even on the language (Gnerre 2000; Gnerre n.d.). As a result, while Christianity is accepted, the Wampis are less acculturated and, thus, native religious beliefs are maintained in parallel to Christian ones by the Wampis. In general, a Wampis has no problem going from traditional to non-traditional religion. For example, it was common for the wife of one of my hosts to go to the garden early morning and sing anen ‘magical song’ dedicated to Nunkui and then go to the local 51. The term tarimata literally means “to stand stable on the ground”. 84 Nazarene church. As far as I can tell, there does not seem to be much syncretism of native and Christian beliefs, as has been the case in many native groups of South America. 2.8. Oral tradition and folklore The Wampis have a rich oral tradition, which is shared with certain variations by other Jivaroan groups. In this tradition, there was a "rst era where animals were people, before they turned into their current state. Many Wampis myths explain the current state of the world in terms of these transformations. The world in which they live is thus seen as an inverted world that mirrors the primordial world. Their myths also explain the connection between the Wampis society and nature, as well as the powers the Wampis associate with di$erent animals through Arutam. Lexically, the Wampis language does not distinguish between the human kin and most mythical characters, people who live in the “spirit” world and many supernatural beings that populate the jungle. Most of them fall under the category of shuar /ʃuara/ ‘person’. Mythical heroes, thus, are usually presented as human beings that appear in a primordial-time scenario; i.e. before they turned into animals or other natural phenomena (sun, moon, and so on). 85 Apart from oral stories that refer to a mythical past, the Wampis have narratives about the rubber boom times, the foundation of the communities and their contemporary struggle to defend their lands.52 That is to say, they maintain a memory of their own contemporary ethnohistory from at least four or "ve generations ago. 2.9. Cultural transmission, social life and organization As in many other cultures, elders are usually regarded as repositories of knowledge among the Wampis. Some elders who founded the current communities decades ago are still alive and are very respected by younger generations. Despite the fact that the elders sometimes complain about the lack of interest younger generations have for learning the traditions, most traditional beliefs of the Wampis are being transmitted to the youth. Knowledge of plants, animals and crafts are also being passed down from parents to children. Songs occupy an important place in Wampis culture. The Wampis have three traditional types of songs: anen (/anɨnta/) ‘magical song’, nampet (/nampɨta/) ‘drinking song’, ujaj (/uhaha/) ‘celebration/war song’. The ujaj was sung when men marched to war and in the victory celebration, where tsantsa ‘shrunken head’ were presented. This 52. As an example of the latter topic, see Santos-Granero and Barclay (2011). 86 celebration is no longer held, and thus it is more di#cult to "nd people that remember ujaj. On the other hand, anen and nampet are still part of the habitual life of many Wampis. Nampet are sung in family parties and communal celebrations. The native nampet co-exists with other non-native music styles favored by the Wampis, like cumbia music. Anen are very poetic and make use of a highly "gurative language usually associated with animals. They are transmitted individually and used for di$erent social functions: love a$airs, healing ceremonies, to propitiate good hunting or good agriculture, to remember a departed friend or relative, etc. There was a counselling ritual in which fathers used to wake up their children very early in the morning to give speeches that instruct the children on how to be a good person according to the Wampis worldview. Family genealogies and enemies were repeated, as well as some stories that taught children a moral code. During this counselling, the Wampis drank waiś /uaisa/ ‘Ilex guayusa’, an energetic brew deemed to cleanse the body. This counselling ceremony is no longer practiced by everyone, especially by young parents. The Wampis had a greeting ceremony whereby a visiting person would state the reason of his presence in someone else’s house. The ceremony involved a great deal of 87 performance in which the host and each of the male visitors saluted each other, speaking in a forceful manner and accompanying their gestures with a back-and-force movement of their spears. This ceremony is no longer practiced in the same contexts as it used to be. Now it is performed only on few particular occasions, for example when an outsider comes to meet the community assembly, or when a delegation from another community arrives to participate in a local festivity. Because this is no longer a common feature of Wampis society, men that know how to do the greeting ceremony are regarded as performers, and are usually called upon to enact the ceremony. The Wampis society is pretty egalitarian. In the past, they did not have a "gure- head, a nobility or an elite, at least not in the same way these concepts are conceived of in the Western world. In principle, no person is considered superior to other and all voices in the communal assembly are usually heard. Traditional work is distributed by gender: men are in charge of hunting, weaving and warring; women attend to cooking, preparing manioc beer, pottery and working in the gardens or chacra. In my experience, however, I observed men and women helping each other on many occasions, especially when hunting53 and doing agricultural labor. Women in Wampis society also act as 53. While women typically do not hunt, they often accompany the hunting group to provide food and take care of the hunting dogs. 88 leaders, go to communal assemblies and make their voice be heard. Female leaders are known as nuwa uun lit. ‘big woman’, or Spanish lideresa ‘female leader’. Each Wampis community has its own law code that sanctions or condemns potential actions within the community, such as robbery, mistreatment or violence. The Wampis used to live in big family houses scattered in the territory, separated from each other by at least a few hundred meters. A Wampis family was usually composed of a man, his wives (a Wampis man usually had two or more wives) and their children, and sometimes close relatives, including elderly people and sons-in- law. In times of war, a war party leader, the kakaram (/kakarama/ ‘valiant, powerful man’), called other people to be part of his party through the activation of family alliances. Another type of socially respected "gure is the pamuk /pamuka/, typically an experienced person who supervised rituals and o$ered counsel on di$erent matters. The shaman, iwishin or uwishin (iuiʃi-inu ‘bewitch-NMLZ’) , was an ambivalent "gure that occupied a very important place within the social group. The iwishin protected the group from enemy shamanic attacks,54 and in turn could send attacks to the enemy. Thus, an enemy iwishin was the "rst to be blamed when someone died. On the other 54. See Chaumeil (1988) for an elaborated analysis on the defensive specialization by which Wampis shamans are held in high regard. 89 hand, the shaman would also act as a healer, helping people, even strangers, recover from di$erent illnesses. Nowadays, likely due to the in/uence of the church, the iwishin is seen mainly as an evil character, though I was told there were “good” iwishin who practice “white” witchcraft. The presence of an iwishin in a community is deemed suspicious. The concept of healing falls now in the category of curandero ‘healer’ (a term borrowed from Spanish). Though scattered houses are still found in the Wampis territory, nowadays the center of Wampis social life is the village. There are around 60 villages distributed in the Santiago and Morona Rivers. National administrative institutions are present in the communities through the district municipalities and justices of the peace. Most of the authorities are native local people. There are also a few military posts near some communities and in the border with Ecuador. In the last decades, social programs and medical services have expanded their presence, though their coverage is still poor due to lack of resources. Each community also elects its own apu, a local leader. The traditional social institutions are not dead. People who are considered wise can still be chosen as pamuk by the community assembly, and their advise is sought in important matters. In addition, while the kakaram are no longer visible in Wampis 90 society, the ideal of the warrior leader and defender is very present among the Wampis. I was told, for example, that during the important 2009 protests in Bagua, pamuk were called to Puerto Galilea, capital of the district, to o$er support and advise the leaders who were coming from di$erent parts of the Santiago. There, anen were also sung to provide spiritual support for the protesters. The rapid mobilization of a good number of native people from di$erent communities during those protests can be seen as the continuation of the traditional Jivaroan kind of leadership, being capable of activating the old network of alliances that was crucial to survival in the past. Green (2009) has studied the transformation of the "gure of the Jivaroan leaders of prior days into the modern-day leaders of native political organizations. In general, it can be said that the Wampis have adapted well to new institutions and are well organized for political action. The Wampis are well aware of their need to have a greater legal presence to get support for their demands. Some of the most important regional organizations are the Federation of Wampis Communities of the Santiago River (FECOHRSA), the Wampis and Awajún Indigenous Peoples of the Kanus (OPIWAK) and the Shuar Organization of the Morona (OSHDEM). In addition, there is a growing number of regional organizations and political parties were the Wampis are 91 participating. In the last general elections (2011), for example, Eduardo Nayap, of Awajun origin, obtained a place in the Peruvian Congress with the support of the Awajun and Wampis. In the same elections, there were other Wampis and Awajun people running for Congress or for the Region of Amazonas government o#ce. The "rst schools established by SIL continue to function under the Peruvian Education System. Primary schools are present in most communities, but secondary schools are located only in some of them. Thus many children travel by river to go to school and some communities, like Puerto Galilea, have organized communal rooms to receive students from other communities. Even so, not all children have access to secondary school due to the cost of mobilization and the expenses it implies. Education is bilingual in primary schools, but in some instances the teachers are Spanish speakers and have not received suitable training for intercultural contexts. In general, the lack of resources prevents a better, more culturally-appropriate education. Some parents are able to send their children to boarding schools directed by Catholic congregations in Jaen and Santa María de Nieva. In sum, while many core traditions are maintained, it is also obvious that the Wampis society continues to adapt to contemporary times. Many traditions are 92 therefore changing or being left behind. For instance, as recently as twenty or thirty years ago, the Wampis used to have several important seasonal celebrations. In fact, the Wampis calendar was very ritually organized and festivities could last several days. Traditional festivities have been replaced by new “o#cial” or national ones (such as Peru’s Independence Day, the community foundation’s anniversary, local school anniversary, etc.). Therefore, though the Wampis maintain their culture, new traditions are born as in/uence from outside grows. One of my Wampis teachers explained this situation by saying that the old ways, the way of Arutam, were changing toward the way of “education” or the way of “the professional”, stating that “we no longer defend ourselves with weapons, we defend ourselves with words”. 93 CHAPTER III SEGMENTAL PHONOLOGY AND PHONETICS 3.1. Introduction In this chapter, I present a study of the segmental phonology of Wampis consonants and vowels, as well as their phonetic variation. There is only one previous work on Wampis phonology written by Beasley and Pike (1957). I discuss some of their claims in light of the recent data collected during my "eldwork. Notable di$erences between the present analysis and Beasley and Pike’s 1957 article include the fact that, unlike the cited authors, I do not consider glides [w] and [j]55 and the velar nasal [ŋ] as phonemes. Also, devoicing of vowels, as reported by Beasley and Pike in the same article, does not occur in my data.56 The voicing of /p/, /t/ and /k/ in the environment of a sonorant sound is not systematic in my analysis (occuring only sporadically in fast speech). Beasley and Pike based their analysis on one speaker of the dialect of the 55. There is one exception: the third person past tense morpheme morpheme -ji, where a phonemic /j/ occurs. See §3.2.7 for details. 56. Devoicing of vowels has been reported for other Jivaroan languages such as Achuar (Fast 1975b), Shuar (Pellizaro & Náwech 2005) and Awajun (Pike & Larson 1964), though Overall (2007) did not "nd evidence of devoicing in modern Awajun. 94 Wachiyacu River, in the Morona region, whereas my data comes from the Santiago region as reported in Chapter I. Therefore, apart from possible idiolectal considerations, historical changes in the last "fty years and dialectal di$erences may account for the di$erences between Beasley and Pike’s and my own analyses. The structure of this chapter is as follows: after this introduction, the consonant phoneme inventory is analyzed in §3.2. The major allophonic variations of consonants are described in §3.3. Section §3.4 describes other optional consonant allophony. In §3.5, a summary of consonants distribution in Wampis is provided. Section §3.6 is dedicated to the study of vowels and §3.7 to the main processes of vowel allophony. Section §3.8 discusses some interesting hypo-articulatory phenomena involving both consonants and vowels. Loanwords are discussed in §3.9. Finally, brief considerations on transcription are given in §3.10 3.2. Consonant phonemes The relevant articulatory parameters for Wampis consonants are place and manner of articulation; voice does not play a role in distinguishing contrastive phonological units. The Wampis phoneme inventory contains 14 consonants: 4 voiceless stops (/p/, /t/̪, /k/ and /ʔ/), three nasals (/m/, /n/ and /ɲ/), three fricatives (/s/, /ʃ/ 95 and /h/), two a$ricates (/ts/ and /tʃ/), one rothic /r/ and a glide /j/. The glottal stop /ʔ/, the nasal palatal /ɲ/ and the semivowel /j/ have a very marginal distribution in my data. The semivowel /j/ only occurs in one su#x, the third person-past tense -ji, and /ʔ/ and /ɲ/ occur in less than four words in the database. Because of the limited distribution of these phonemes, I represent them between parentheses in the table below. I consider the semivowels [w], [ɰ] and [j]—(with the one exception just described above)—to be positional allophonic realizations of /u/, /ɨ/ and /i/, respectively (see §3.7.3). Table 3.1 shows the consonant phonemes of Wampis. Table 3.1. Wampis consonant inventory Bilabial Dental Alveolar Palato-alveolar Palatal Velar Glottal Stop p t ̪ k (ʔ) Nasal m n (ɲ) Fricative s ʃ h A$ricate t͡s t͡ʃ Rhotic r Approximant (j) Whereas most consonant phonemes (except /r/, /ɲ/ and /ʔ/) occupy syllable- initial position, the distribution of consonants as syllable codas is very restricted (see 96 §3.5). Apart from vowels, only the nasals /m/ and /n/ occupy coda position underlyingly, though most other consonants may occur in that position in surface forms. All consonants occupy word-initial and word-medial positions on the surface, but underlyingly only vowels occupy word-"nal position57 and constitute the only syllabic nucleus. We will see that vowel elision a$ects both word structure and metrical high tone assignment in Wampis (see Chapter VI for a detailed analysis of the prosody of Wampis). 3.2.1. Oppositions The following examples show relevant minimal pairs and near minimal pairs that demonstrate the phonemic status of the consonants given in Table 3.1. The examples are arranged "rst by point of articulation and, secondly, by manner of articulation. More detailed description of the phonemes is given in the discussion below. By point of articulation: (1) p vs t ̪ #CV /puku-/ ‘plant manioc stem’ vs /tu̪ka-/ ‘hit target with dart or bullet’ .CV /hapa/ ‘deer’ vs /hata̪/ ‘sickness’ 57. Some onomatopoetic words and interjections do not follow this rule. It is common that onomatopoeia and interjections do not follow the systematic patterns of a language. 97 (2) p vs k #CV /pita̪ka/ ‘travelling basket’ vs /kita̪ka/ ‘thirst’ .CV /nupa/ ‘weed’ vs /nuka/ ‘leaf’ (3) t ̪vs k #CV /tu̪nta̪/ ‘quiver’ vs /kunta̪/ ‘mud’ .CV /ɨtɨ̪/ ‘wasp’ vs /ɨkɨ/ ‘yet’ (4) m vs n #NV /mai/ ‘both’ vs /nai/ ‘tooth’ .NV /numi/ ‘tree’ vs /nuni/ ‘do that’ (5) s vs ʃ #CV /sampi/ ‘shimbillo (Inga spp.)’ vs /ʃampi/ ‘lizard’ .CV /mísu/ ‘naked’ vs /míʃu/ ‘cat’ (Spanish)’ (11) n vs t͡s #CV /napa/ ‘bee’ vs /t͡sapa/ ‘calabash ’ .CV /punu/ ‘calabash sp.’ vs /put͡su/ ‘white’ (12) n vs r58 .CV /anuma/ ‘dock (V)’ vs /aruma/ ‘later’ (13) s vs t͡s #CV /sai/ ‘brother-in-law (of male)’ vs /t͡sai/ ‘tree sp.’ .CV /sasa/ ‘small water stream/waterfall’ vs /t͡sat͡sa/ ‘paternal aunt ’ (14) s vs r .CV /tu̪-sa/ ‘say-SUB’ vs /tu̪ra/ ‘then’ (15) t͡s vs r .CV /t͡sat͡sa/ ‘paternal aunt’ vs /t͡sara/ ‘thorny’ 58. The rhotic does not occur word-initially, there are very few examples of it word-initially in borrowings from Spanish. 99 (16) ʃ vs t͡ʃ #CV /ʃimpa/ ‘type of pidgeon’ vs /t͡ʃipa/ ‘tree sp.’ .CV /ti̪kiʃi/ ‘knee’ vs /ti̪kit͡ʃi/ ‘one’ 3.2.2. Stops As we have seen, the stops in Wampis are all voiceless and include a bilabial, a dental, a velar and a glottal segment. In the next subsections, I brie/y describe each one. 3.2.2.1. Voiceless bilabial stop /p/ Phonemically, the voiceless bilabial stop occurs in syllable initial position, as in example (17). Word-internally, it occurs between vowels and after a bilabial nasal, as in (18)–(19). Only at the surface level, the bilabial stop occurs syllable-"nally (20). (17) #C /paki/ → [páki] ‘collared peccari’ (18) V.CV /apa/ → [ápa] ‘father’ (19) N.CV /ampuʃa/ → [ampúʃ] ‘owl’ (20) /tu̪ntu̪pɨ/ → [tu̪n̪tú̪p] ‘back’ The only clear exception that I have found where /p/ occupies a phonemic coda position in my data seems to be sound-symbolic in nature: (21) t̪upt̪ut̪ ‘over eat’ 100 There is also a word in Jakway et al. (1987: 123) that could be another exception and is not attested in my data: (22) ‘black monkey’ The structure of this word is suspicious, though, as the written in seems to be a palatalization operating on /k/ (the palatalization process is described in §3.3.2). Therefore, we would have an underlying structure of the form /i.u.ka.pi.ka/ → [jukápkʲa]. The elision of the third vowel of the word (the /i/) is a common process in that context in Wampis, which exhibits pervasive vowel elision phenomena (see §6.4.4). I am unaware of a root for the word /iukapika/, though it may be semantically related to /iukaipi/ ‘dirty’. 3.2.2.2. Voiceless dental stop /t/̪ The voiceless stop /t/̪ is a dental consonant, not an alveolar. Not only can this be seen in an articulatory analysis but also in the fact that the nasal /n/ gets assimilated and realized as a dental [n̪] when it follows /t/̪, which indicates a dental feature received by the nasal: (23) anɨn̪tɛ̪ɛʲmʲat ̪ anɨnta̪i-ma-ta̪ heart-VBZ-NMLZ ‘mind’ 101 The distribution of /t/̪ is similar to that of /p/: underlyingly, it occurs at the beginning of the syllable, but not as a coda. Di$erent environments are shown in (24)– (26). At the surface level, on the other hand, /t/̪ can occur as a syllable coda (27). (24) #C /tu̪/ → [tú̪] ‘say’ (25) V.CV /ɨtɨ̪/ → [ɨt́ɨ̪] ‘wasp’ (26) N.CV /hinta̪/ → [hín̪tʲa] ‘trail’ (27) /kupita̪/ → [kupít]̪ ‘ant sp.’ As I said earlier, because there is no dental vs. alveolar stop opposition, in the remaining chapters of this dissertation I will transcribe the voiceless dental stop with a /t/ symbol, without marking the dental feature. 3.2.2.3. Voiceless velar stop /k/ The following examples show the velar stop /k/ in word-initial and word- internal positions. (28) #C /kanu/ → [kańu] ‘canoe’ (29) V.CV /ikama/ → [ikʲám] ‘forest’ (30) N.CV /kɨnkɨ/ → [kɨŋ́kɨ] ‘sachapapa (Dioscorea tri)da)’ (31) /ipaku/ → [ípʲak] ‘achiote (Bixa orellana)’ 3.2.2.4. Glottal stop /ʔ/ The status of the glottal /ʔ/ is more di#cult to discern than that of the other stops. The glottal stop has a very restricted distribution, occurring intervocalically only in a few words in my data. Unlike the other marginal phoneme (the palatal nasal, cf. 102 §3.2.3.3), the glottal stop does occur in native Wampis words, though most of them are interjections. Here are some examples where the glottal occurs: (32) /haʔa/ ‘yes’ (33) /ɨt͡sa ̃tɨaʔai/̃ ‘time between 1-3 p.m.’59 (34) /kiaʔai/̃ ‘at sunset’ (35) /aiʔai/ ‘beetle sp.’ All in all, the best arguments for considering the glottal as a phoneme despite its marginal distribution in the Wampis system are: a) the glottal stops in these words cannot be omitted; i.e. they are integral part of the word, or the words are otherwise considered incorrect; b) there are indeed contrastive pairs with other consonants (cf. /haʔa/ [haʔa]́ ‘yes’ vs /hata̪/́ [hata̪]́ ‘die!’); and c) its occurrence would not be explainable by other means such as allophony or contact. 3.2.3. Nasal consonants Wampis has bilabial /m/ and an alveolar /n/ nasal segments. The velar nasal [ŋ], which was considered as a phoneme by Beasley and Pike (1957), is not considered 59. This expression is composed of ɨt͡sa ̃‘sun’ and the root tɨa ‘incline’. The analysis of tɨaʔai ̃is a bit complicated as the word is already lexicalized, but it is possible that etymologically it comes from tɨa ‘incline’ and a=i ̃‘copula=Locative’; i.e. its literal meaning would be ‘when the sun is inclining’. Many words and expressions referring to the parts of the day in Wampis include a description of a position of the sun (“incline”, “vertical”, etc.). It is possible that the glottal was inserted between words, an optional process that is still observed in current Wampis. The same cluster /ʔai/̃ occurs in example (34). 103 to have phonemic status in the present analysis. The behavior of nasal consonants is a particular point of interest for Jivaroan languages. Unlike Awajun (Corbera Mori 1994; Overall 2007; Payne 2008), Wampis does not de-nasalized its nasals. That is, /m/ and /n/ in Wampis do not have the allophones [ᵐb] and [ⁿd], respectively. As will be discussed later in this chapter (§3.8.5), /m/ and /n/ do not triggered systematically the voicing of stops in Wampis either; a characteristic that has been reported for Shuar (Turner 1958b; Pellizzaro 1969; Turner 1992; Pellizaro & Náwech 2005)60 or Achuar-Shiwiar (Fast 1975a; Fast 1975b; Fast et al. 1996). 3.2.3.1. Bilabial nasal /m/ Underlyingly, the bilabial /m/ appears in word-initial, syllable-initial and syllable-"nal positions, but not word-"nally. At the surface level, [m] can occur word- "nally following vowel elision, as shown in (39). (36) #N /mɨtɨ̪ka/ → [mɨtɨ̪ḱ] ‘equal’ (37) V.NV /mama/ → [máma] ‘manioc’ (38) VN. /ampuʃa/ → [ampúʃ] ‘owl’ (39) /kaɨma/ → [kawɨḿ] ‘many’ 60. But notice that Saad (2014) states that there is no obligatory voicing of stops after a nasal consonant in Shuar. 104 3.2.3.2. Alveolar nasal /n/ The alveolar nasal has the same restrictions as /m/. Examples (40)–(43) illustrate its occurrence in di$erent positions. (40) #N /nasɨ/̃ → [násɨ]̃ ‘wind’ (41) V.NV /uɨnu/ → [wɨńu] ‘mouth’ (42) VN. /nantu̪/ → [nán̪tu̪] ‘moon’ (43) uunta̪ → [uún̪] ‘big’ 3.2.3.3. Marginal consonantal nasal phoneme /ɲ/ A palatal nasal phoneme /ɲ/ occurs only in two words in my data. Those words are /kaɲiru/ [kaɲiŕ] ‘parasitic "sh species’ and /muɲusunka/ [muɲusúŋ] ‘ammunition’. The word /kaɲiru/ does not seem of Jivaroan origin; cognates of this word are found quite widespread in the Amazon to designate several species of infamous parasitic cat"sh. It may have been borrowed through Spanish though it is also possible that it was borrowed from another Amazonian language (the equivalent Portuguese term, is apparently of Tupí origin (Houaiss & Villar 2001)). Notice that this word appears written as with an /n/ in Berlin (1994: 89): this corresponds better with the Wampis phonological system. The word /muɲusunka/ comes from Spanish ‘ammunition’. Strictly speaking, /ɲ/ is to be considered a phoneme , despite not being an original Wampis consonant. Due to its 105 restricted, non-productive occurrence, and to the non-native origin of the couple of words where it is found, I list /ɲ/ as a very marginal phoneme. This marginal /ɲ/ must not be confused with the palatal allophone of /n/ (see §3.3.2) which results from a regular palatalization pattern. 3.2.3.4. Nasal consonants in coda and word-!nal positions Previous analyses of Jivaroan languages have stated the existence of a phonemic velar nasal /ŋ/, whose occurrence would be restricted to coda and word-"nal positions. With reference to Wampis, Beasly and Pike (1957) consider this velar nasal to be a phoneme in the language. Furthermore, according to said authors, the alleged velar phoneme ŋ would have the allophone [ɲ] in certain environments. The supposed phonemic status of ŋ is problematic according to my analysis. In addition, under the analysis presented here there is no reason to posit [ɲ] as an allophone of the supposed velar nasal phoneme either. First, in the data collected for this dissertation, the nasal /n/ surfaces as a velar [ŋ] after assimilation to a velar consonant /k/. This occurs also at the end of the word but the assimilation process is obscured in the surface form because of a systematic apocope process (§6.4.5.1). For instance, Beasley and Pike 106 (1957: 2) give the following examples:61 (44) /namaŋ́/ → [namaŋ́] ‘/esh’ /titiŋ́/ → [titiŋ́] ‘scorpion’ /tat́aŋ/ → [tat́aŋ] ‘pottery board’ Examples like the ones shown in (44) are the result of a very surface analysis by Beasley and Pike. The actual phonological forms of ‘/esh’, ‘scorpion’ and ‘pottery board’ are /namankɨ/, /ti̪ti̪nki/ and /ta̪ta̪nku/, respectively. When they are in the nominative case, their pronunciations are [namaŋ́], [ti̪ti̪ŋ́] and [ta̪ta̪ŋ́], following a pervasive apocope rule. However, the full phonological forms of these words surface when another morpheme, for instance the accusative =na, is added. This is shown in (45). (45) namankɨ=na → [namaŋkɨń] ‘/esh (ACC)’ ti̪ti̪nki=na → [ti̪ti̪ŋkiń] ‘scorpion (ACC)’ ta̪ta̪nku=na → [ta̪ta̪ŋkún] ‘pottery board (ACC)’ Hence, the "nal surface velar nasal [ŋ] is explained by assimilation of /n/ to an underlying velar stop, which is then dropped on the surface in the nominative form, 61. I transcribed Beasley and Pike’s examples as they are given in the cited text. The accent marks in the underlying forms appear in the original. In my analysis, a metrical high tone can be derived through rules described in §6.4, thus, I do not use “accent” marks in underlying forms except in some necessary cases where the high tone is not metrical but lexical, i.e. when the high tone is part of the lexical representation of words (cf. §6.5). See Chapter VI for the analysis of Wampis prosody. 107 which is the form given in Beasley and Pike’s 1957 article.62 This assimilation process is entirely predictable, hence there is no need to posit the velar nasal as a phoneme. The examples of ŋ in coda position (word-internally) given by Beasley and Pike, as in (46), are also explainable by assimilation of /n/ to a following velar stop. (46) /waiŋkiu/ → [waiɲgióu] ‘found’ (Beasley & Pike 1957: 2) /ukuiŋkiata/ → [ukuiɲgiata] ‘Open it!’ (Beasley & Pike 1957: 2) Finally, the palatalized realization of the spurious ŋ, which can also be observed in (46), does not occur in my data in the context where it precedes the stop consonant. In such cases, the nasal surfaces as [ŋ] (i.e. assimilated to the next consonant).63 However, this palatal realization can be accounted for as a trace of the palatalization of /n/ when preceded by /i/, as described in §3.3.2. In sum, there is no reason to postulate a phoneme ŋ, and both [ŋ] and [ɲ] are two distinct allophones of /n/. With regard to /m/, it follows a similar process as /n/ concerning palatalization (it palatalizes when preceded by /i/—see §3.3.2). However, /m/ does not assimilate in 62. The elision of the stop consonant is due to a restriction in Wampis which prohibits complex codas. After the last vowel is dropped due to apocope, the sequence Nasal+Stop is an impossible coda in Wampis. The only consonants allowed in the margins of the coda are the A$ricates. See §4.3 for a detailed explanation of this process. 63. Notice also that it seems a bit unnatural to claim that there is no assimilation of the /n/ to the following /k/. I certainly have not examples like (46) in the data. 108 front of /k/ or /t/̪ in coda position (either word-internally or word-"nally, following an elided consonant), as shown in (47)–(48). (47) ɨɨḿkats̪a ɨɨma-ka-ta̪sa advance-INTENS-PURP ‘intending to advance’ (48) poúmtan paumita=na "sh.sp=ACC ‘palometa (ACC)’ (Wampis paumita < Spanish palometa ‘"sh sp.’) The distribution of nasal consonants in front of a voiceless stop is given in Table 3.2. Table 3.2. Distribution of nasals in front of obstruents ___/p/ ___/t]̪ ___/k/ /m/ [m] --- [m] /n/ --- [n̪] [ŋ] Overall (2008) has proposed a nasal hyperphoneme N that is realized as [m], [n], and [ŋ] in front of stops. As Overall has noticed, this is a useful characterization 109 because sometimes there is a /oating nasal phoneme that is realized as a nasal consonant N, as shown in the next examples. (Notice also that in Wampis the consonant N can surface without the need of a stop following it.) As an illustration, compare (49) with (50). (49) hɨɨ̃ ̃ń ~ hɨɨ̃ ̃ ́ hɨɨ̃ ̃ house\1PL/2PL/3.POSS ‘his/her/our/your (PL)/their house’ (50) núwɨ ̃ nuɨ ̃ woman\1PL/2PL/3.POSS ‘his wife’ Additional examples of a homorganic N surfacing in front of a stop are given in (51)–(52). (51) wariḿpʲa wari=̃pa what=Q ‘what?’ (52) niŋ́kʲa ni=̃ka 3SG=FOC ‘he’ 110 In the current analysis, I prefer to keep two types of nasals separate as they seem to be of di$erent origins. The "rst type is a lexically-speci"ed phonological segment /m/ or /n/ (as the "rst /n/ of /nua/ in example (50)). The second is an “/oating” nasal phoneme that can be realized as pure nasality on a vowel or as a nasal consonant (as in examples (49) or (51)–(52), respectively). Barring minor details in analysis, the present analysis supports Overall’s (2008) claim that the velar nasal [ŋ] does not possess phonemic status in Jivaroan languages. 3.2.4. Fricatives The fricatives consonants in Wampis are the alveolar /s/, the palato-alveolar /ʃ/ and the glottal /h/. 3.2.4.1. Alveolar fricative /s/ The alveolar /s/ occupies only syllable initial position. It occurs between vowels following an open syllable, and can occur syllable-"nally only in surface form, after apocope. (53) #C /suku/ → [súku] ‘type of basket’ (54) V.CV /mukusa/ → [mukúsa] ‘black’ (55) /t͡sɨasa/ → [t͡sɨás] ‘poison’ 111 3.2.4.2. Palato-alveolar fricative /ʃ/ The palato-alveolar /ʃ/ follows the same restrictions as /s/. /ʃ/ is the only consonant in the available data that never appears before the high central vowel /ɨ/. The following examples demonstrate the occurrence of /ʃ/. (56) #C /ʃuara/ → [ʃuár] ‘person’ (57) V.CV /ataʃu/ → [atáʃ] ‘hen’ (58) VC. /aíʃimanku/ → [ɛíʃmaŋ] ‘man’64 (59) /aíʃiru/ → [ɛíʃur] ‘my husband’ 3.2.4.3. Glottal fricative /h/ The glottal fricative /h/ is commonly produced with a more fortis realization (with more closure) in stressed syllables. However, this seems to be a rather minor variance and Wampis speakers did not seem to feel a major di$erence between fortis and lenis realizations of /h/. In syllable-"nal position the glottal fricative is usually lengthened. The next examples show the occurrence of the glottal fricative in di$erent positions: 64. One could analyze the phonological representation of this word as /aiʃmanku/, without a second vowel /i/ after /ʃ/. The reason for this is that said vowel never surfaces in the actual pronunciation of the word as a result of vowel elision (one rule of vowel elision is to drop the third vowel from the left). However, the elision of the vowel is entirely predictable and the word ‘man’ is clearly related to /aíʃi/ ‘husband’, so there is certain basis to state that the phonological form is /aiʃimanku/. In relation to cases like the example presented above, an issue that awaits future research is what the psychological reality of phonological representations of words is like in languages such as Wampis, where vowel elision is very pervasive. 112 (60) #C /hɨã/ → [hɨɰ̃ã]́ ‘house’ (61) V.CV /aha/ → [áha] ‘garden’ (62) /uɨha/ → [uwɨh́] ‘hand’ There have been reports of nasality associated with the glottal fricative /h/ in the related language Awajun (Payne 1990a; Corbera Mori 1994; Payne 2008). The relation between /h/ and nasality has been discussed in Matissof (1975) and is reported in other languages of the Amazon region such as Yagua (Payne & Payne 1990). In Awajun, /h/ comes from a merger of Proto-Jivaroan *h and *r. Overall (2007: 31 and $) has hypothesized that there was an intermediate stage in Awajun in which *r > h ̃> h; in other words, apparently in Awajun /h/ and /h̃/ were distinguished in past stages of that language. Furthermore, Overall observes that several examples show that a nasal vowel adjacent to /h/ in Awajun correspond to an oral vowel adjacent to /r/ in other Jivaroan languages such as Shuar. Further evidence in support of this "nding is given with Wampis examples in Table 3.3.65 65. The Awajun and Shuar examples are cited from Overall (2007: 35). 113 Table 3.3. Re/exes of Proto-Jivaroan *r and *h Awajun Shuar Wampis Proto-Jivaroan Gloss aha aha aha *aha ‘fell (trees)’ ah̃a ara ara *ara ‘sow seeds’ uha uha uha *uha ‘tell’ ũha uha uha *ura ‘open’ In Wampis, nasality spreads through the fricative /h/, but it is not clear that /h/ is the locus of nasality; i.e. /h/ does not seem to trigger nasalization.66 Rather, nasalization seems to be related to an underlying nasal vowel. This seems to correspond with Overall "ndings for Awajun, where the /h/ that seems to trigger nasalization is a re/ex of Proto-Jivaroan *r, which continued to be /r/ in Wampis.67 3.2.5. A&ricates Wampis possesses an alveolar a$ricate /t͡s/ and a palato-alveolar /t͡ʃ/. As we saw in section §3.2.1, a$ricates enter in phonemic contrast with their fricatives 66. There are a few examples that may suggest that this was not the case in the past. For instance, some speakers sometimes nasalize the word hu ‘proximal demonstrative’, pronouncing [h̃ũ], suggesting that /h/ may be the locus of nasality (since /hu/ is otherwise pronounced with an oral vowel in most cases). 67. But notice that there are examples in Awajun where a nasal vowel surfaces adjacent to /h/ where the cognate in other Jivaroan languages is /h/ (and not /r/). These examples are problematic as they show that in Awajun nasal vowels may or may not be a re/ex of the change *r > h. 114 counterparts. Beasley and Pike (1957) described voiced allophones for the a$ricates, but these simply do not occur in my data. As the name implies, the a$ricates are to be considered one co-articulation rather than two di$erent segmental articulations. This can be seen in certain phonetic reductions involving a$ricate consonants, as shown in (63), where /at͡sa/ is reduced to [t͡sá], but never [sá]. This indicates that /t͡s/ is considered synchronically as one segment by Wampis speakers.68 On the surface level, a$ricates are the only segments that can occur word "nally in the coda margin, after a nasal consonant. No other consonant can occur in that position: (63) /t̪ɨnt̪ɨt͡ sa/ → [t̪ɨn̪t̪ɨt͡ s] ‘Proper name’ 3.2.5.1. Voiceless alveolar a&ricate /t͡s/ The voiceless occurs at syllable-initial position and can follow a close or open syllable. The a$ricate /t͡s/ seems to be incompatible with the vowel /i/: I have not found any examples of this consonant co-occurring with said vowel. The next examples illustrate the occurrence of /t͡s/. (64) #C /t͡samarainta/ → [t͡samarɛɛ́n̪] ‘proper noun’ (65) V.CV /ɨt͡sã/ → [ɨt́͡sã] ‘sun’ (66) VN.CV /ɨnt͡sa/ → [ɨńt͡sa] ‘river’ 68. Notice also that in Wampis there are no consonant clusters syllable-initially. The sole exception is [Cr] (where C is a stop) in a very speci"c case: only after vowel elision is this cluster allowed. 115 (67) /tɨntɨt͡sa/ → [tɨ̪n̪tɨ̪t́͡s] ‘proper noun’ 3.2.5.2. Voiceless palato-alveolar /t ͡ʃ/ The distribution of /t͡ʃ/ is similar to that of /t͡s/. Syllable "nally, it only occurs after vowel deletion. (68) #CV /t͡ʃinki/ → [t͡ʃíŋki] ‘bird’ (69) V.CV /t͡ʃit͡ʃa/ → [t͡ʃit́͡ʃa] ‘speak’ (70) VN.CV /int͡ʃi/ → [ínt ͡ʃi] ‘sweet potato’ (71) /tikit͡ʃi/ → [tíkit ͡ʃ] ‘other’ 3.2.6. The rhotic /r/ The rhotic /r/ varies mostly freely between a /apped [ɾ] and an approximant [ɹ] realization from speaker to speaker, and even in the same speaker. Of these two variants, the /ap is the most common one. The rhotic does not occur at the beginning of words. Examples (72) and (73) illustrate the occurrence of this segment. (72) /pɨnkɨra/ → [pʷɨŋ́kɨɾ] ~ [pʷɨŋ́kɨɹ] ‘good’ (73) /aruta̪ma/ → [aɾúta̪m] ~ [aɹúta̪m] ‘spirit of power’ Some speakers (including at least one of my teachers) occasionally also produced a slight trill-like realization of the rhotic intervocalically, but in most cases this segment varies between [ɾ] and [ɹ]. The rhotic will be transcribed with the symbol for the remainder of the dissertation. 116 3.2.7. Marginal palatal /j/ There is only one item in Wampis where the glide /j/ can be analyzed as a phoneme: the su#x -ji ‘3.PT+DECL’. Notice that, for the most part, glides (including the palatal glide) in Wampis are completely predictable, occurring word-initially and intervocalically (cf. §3.7.3 and §4.2).69 The only case where the palatal glide is not predictable is with -ji, where /j/ can occur word-internally after a consonant: (74) maãḿji ma-̃a-́ma-ji kill-HIAF-IMM.PT-3PT.+DECL ‘He killed.’ Compare with the surface forms the ablative =ia assumes: following a consonant, the ablative surfaces as [ia] in (75); following a vowel, it surfaces as [ja] in (76): (75) huabaĺia Huabal=ia Huabal=ABL ‘from Huabal’ 69. Notice that there are no diphthongs in Wampis. Vowels in vowel clusters are considered in di$erent syllables. Thus, while the "rst vowel of a cluster after a consonant, e.g. Cia (where C=consonant), is an on-glide in terms of articulation, it is treated as a vowel for all purposes concerning syllable formation, tone assignment and metrical foot, as well as vowel elision. Therefore, they are considered vowels throughout this dissertation. For Wampis, glides are considered positional allophones of high vowels intervocalically or word-initially. 117 (76) kut͡ʃańmaja kut͡ʃa=nVma=ia lake=LOC=ABL ‘from the lake’ It is very likely that the morphophonological environment has played a decisive role in the development of the su#x -ji ‘3.PT+DECL’. In fact, the modern form -ji seems to be a reanalysis of what historically was a sequence probably related to the old past marker *ia70 and the third person perfective marker -i ̃~ -i. This sequence *ia+*-i may have collapsed to -ji. Further, because this su#x frequently occurs with the declarative - i (which “fuses” to the preceding -i), -ji has been reanalyzed as a single Person+Mood portmanteau su#x: (77) aʃi ́wɛiŋ́karu tiɲ́u aŕmaji aʃi uaina-ka-ara-u tu-inu a-ara-ma-ji all see-INTENS-PL-NMLZ say-NMLZ COP-PL-IMM.PT-3.PT+DECL ‘They saw everything.’ The best evidence for the reanalysis of -ji as a single morphological unit is that the su#x -ji is dropped completely in non-declarative contexts. This means that the -ji is considered a single unit marking 3.PT and declarative, so it has to be deleted in non- 70. Synchronically, the su#x -ia marks remote past in Wampis. 118 declaratives structures. To illustrate this, consider how the su#x -ji does not occur marking third person in the verb ‘go’ in (78) (in an interrogative clause) and the copula in (79) (where the copula is relativized by the demonstrative nu), respectively: (78) urúk puhú wɨaŕmia uruk puhu-u wɨ-ara-mia how live-NMLZ go-PL-DIST.PT ‘How did they live?’ (79) urúk . . . nahaańarukit tiɲ́u aŕma ńuna óuhmatsattahɛɛ uruka nahaana-ra-u=ki=ita tu-inu a-ara-ma nu=na how make-DISTR-NMLZ=RETH=COP say-NMLZ COP-PL-IMM.PT NON.VIS=ACC auhumatu-sa-tata-ha-i inform-ATT-DEF.FUT-1SG.SBJ-DECL ‘I am going to tell what they say [about] how . . . [people] transformed [into an animals].’ From the above discussion, the conclusion is that -ji is no longer analyzable as di$erent morphemes, and thus it must have the underlying form /ji/. Because the glide of -ji is no longer predictable, /j/ is better analyzed as a unique phoneme. 3.3. Consonant allophonic variation This section is dedicated to studying the systematic allophonic variation in Wampis. Variations related to syllable-structure and morphophonology are treated in Chapter IV and Chapter V, respectively. 119 3.3.1. Bilabial labialization The following rule formalizes the labialization of bilabial consonants /p/ and /m/ in front of a high-central vowel: C → Cʷ/ ___ɨ [+bilabial] The following examples illustrate this rule: (80) /pɨńkɨra/ → [pʷɨŋ́kɨr] ‘good’ (81) /amɨ/ → [aḿʷɨ] ‘you’ This allophonic change does not have any consequence in the prosody or grammar of Wampis, so I do not generally transcribe it, except if necessary, as in the above examples. 3.3.2. Consonant palatalization The following rules formalize the palatalization of stop and nasal consonants. Rule 1 predicts that a stop is palatalized after /i/ or /iN/, i.e. a progressive palatalization. Rule 2 states that a nasal consonant is palatalized after /i/ and before another vowel. (Notice that N=nasal consonant.) Rule 1: Palatalization of a Stop consonant C → [Cʲ]/ i(N) ___ [+stop] 120 Rule 2: Palatalization of a Nasal consonant N → [Nʲ]/ i ___ V The examples in (82) illustrate this rule when the consonant follows /i/, and (83) illustrates the palatalization of a nasal between /i/ and a vowel. (82) /ipaku/ → [ipʲaḱ] ‘achiote (Bixa orellana)’ /iḱama/ → [iḱʲam] ‘forest’ /itu̪rt ͡ʃata̪/ → [it ̪ʲ úrt͡ʃat]̪ ‘di#cult’ (83) /imai/̃ → [im̃ʲɛɛ̃]̃ ‘way over there’ /inita̪/ → [iɲit́]̪ ‘inside’ The domain of palatalization is not only local (i.e. a$ecting a consonant adjacent to /i/). Palatalization a$ects the consonant of the onset of the next syllable after /iN/. The next examples show the same the palatalization of a stop consonant after an /iN/ sequence. An interesting note is that in the case of a /iN.C/ cluster, the nasal following /i/ is not palatalized because it is assimilated to the following consonant (see also (85)– (86) below): (84) /kinta̪/ → [kiń̪t ̪ʲa] ‘afternoon’ /inta̪ʃi/ → [in̪t ̪ʲaʃ́] ‘hair’ Palatalization applies across morpheme boundaries too, as shown in (85)–(86). Notice that in (85) nasality surfaces as a nasal assimilated to the next velar stop. 121 (85) níŋkʲa ni=̃ka 3SG=FOC ‘he’ (86) pɛíŋkʲamu paina-ka-mau boil-INTENS-NMLZ ‘boiled’ Palatalization may optionally surface even in occasions where the high vowel /i/ is deleted, as in (87), where the /i/ of the future nominalizer -tinu is deleted. So, apparently palatalization applies at an intermediate level of derivation, otherwise the palatal realization on the surface is not easily explainable. (87) mãánitɲ̪um ma-̃a-́nai-tinu=numa kill-LOAF-RECIP-FUT.NMLZ=LOC ‘Where they will "ght each other.’ In accordance with the "rst and second palatalization rules given above, when there are two consecutive nasals, the second one undergoes palatalization, but not the nasal consonant that is adjacent to /i/. In practice, two consecutive nasal consonants only occur after vowel elision, as the next examples illustrate: 122 (88) sukúrti̪nɲaʃa suku-ra-tinu=na=ʃa shrink.head-DISTR-FUT.NMLZ=ACC=ADD ‘The head-shrinking too (ACC)’ (89) nahańarti̪nɲaka nahana-ra-tinu=na=ka make-DISTR-FUT.NMLZ=ACC=FOC ‘What is going to transform (ACC)’ Figure 3.1 shows an spectrogram with the acoustic realization of [nahańarti̪nɲaka] ‘what is going to transform’ from example (89). Notice the low F1 (slightly higher than the F1 of the adjacent [n]) and high F2 of [ɲ], indicated by arrows—this is typical of palatal sounds (Ladefoged 2005: 163). In addition, notice the transition of F1 and F2 of [ɲ] into the following vowel [a]. The complex word [nahańarti̪nɲaka] ‘what is going to transform’ shown in Figure 3.1 illustrates the palatalization process within one word; however, it will be seen that the same palatalization process also occurs beyond the word (the latter phenomenon is discussed below). 123 Figure 3.1. Non-local palatalization of an /n/ after an /in/ sequence of a preceding syllable Palatalization also occurs beyond the word boundary. In this context, the "nal nasal consonant is palatalized to [ɲ] if the next word begins with a vowel (recall that, per the second palatalization rule given above, a nasal consonant becomes palatalized in the context: i___V). In such cases, an interesting phenomenon of what can be called “amalgamation” frequently occurs, whereby two grammatical words are pronounced in one phonetic chain. When this happens, the surface [ɲ], which is the last segment of the 124 "rst word, e$ectively “connects” to the "rst vowel of the next word. This only occurs word "nally when /n/ follows the vowel /i/. For instance, the sequence of two grammatical words in (90)—the nominalized verb ‘eat’ and the subordinated copula (‘being’)—after vowel elision, is pronounced as a phonetic chain [jurumiɲ́aśa]̃ in the text from where it comes. (90) jurumiń + asa ̃ → [jurumiɲ́aśa]̃ iu-ru-ma-inu a-sa ̃ eat-APPL-REFL-NMLZ COP-SUB/3SG.SS ‘Being a big eater . . .’ Figure 3.2 shows the spectrogram where [jurumiɲ́aśa]̃ is pronounced. As can be seen, [ɲ] surfaces following /i/. As in the previous "gure, notice the low F1 and high F2 of [ɲ], as well as the transition of its F1 and F2 into the following vowel [a]. Note, in addition, that there is no pause between the two grammatical words.71 71. Acoustically, it seems like there is an [n] that transitions into an [ɲ]. In terms of length, the whole segment always occurs relatively long (for instance, compare with the other nasal /m/ in the spectrogram). It is possible to even consider having a sequence [nɲ] or a long palatalized [nːʲ]. Wampis speakers consider that what they pronounce is a palatal [ɲ], which is re/ected in the way how they write it: . Correspondingly, I transcribe it also as [ɲ]. 125 Figure 3.2. Spectrogram showing a surface palatal nasal [ɲ] in word boundary Quite interestingly, based purely on acoustics, one may think that [ɲ] forms a syllable with the following [a]. Nevertheless, speakers do not consider instances like the above example as one word. There seems to be a well delimited distinction between the acoustic nature of “amalgamation” of two words via palatalization and the psychological reality of “word”. While there are clear cases of cliticization and mismatches of grammatical and phonological words in Wampis (see §6.8), I do not have any attestation in which speakers considered cases like [jurumiɲ́aśa]̃ as one word. I did 126 not devote much time to research this particular phenomenon; however, I did carry short tests to look for speakers’ judgments of syllable and stress (in which we trained tapping on a table to identify number of syllables and stressed syllables). Speakers would consistently separate the grammatical words, e.g. [jurúmin]72 and [aśa]̃, and would tell me the number of syllables for each word, e.g. “three” in [jurúmin] and “two” in [aśa]̃. Additionally, I also asked directly if they consider cases like [jurumiɲ́aśa]̃ to be one or two words, and speakers would consistently tell me that there were two words, even in cases when I commented that I was hearing one word. I have no single case of the contrary attested. A plausible explanation for the speakers’ identi"cation of two distinct words in these cases is that each element in the acoustic chain surfaces with its own high tone (i.e. an acoustic high pitch), thus they are considered two phonological words for the purpose of prosodic analysis (it will be seen that, from the point of view of Wampis prosody, one de"ning notion of a phonological word in Wampis is having at least one high tone—see §6.8 for details). Finally, it seems that palatalization of stops is becoming optional among younger generations. People well in their 40s and above palatalize consonants, but I 72. When there is no vowel following, /n/ is not palatalized. 127 found that this was not always the case with younger speakers. More data is needed to evaluate this possible generational change. 3.3.3. A&ricativization of /ʃ/ The fricative /ʃ/ and a$ricate /t͡ʃ/ are neutralized following another consonant. This is formalized by the following rule: /ʃ/ → [tʃ] / C ___ Examples of the neutralization of /ʃ/ and /t͡ʃ/ are given below. In (91), the last vowel of the root is deleted, and because [ntt̪͡ʃ] is an impossible cluster in Wampis, the intermediate [t]̪ is also deleted (no stop can occur in the margin of a complex coda). (91) t͡sawánt͡ʃa t͡sauanta̪=ʃa day=ADD ‘days too’ (92) hɨɰ̃ańt͡ʃa hɨã=na=ʃa house=ACC=ADD ‘to the house too’ 3.3.4. Dissimilation of a&ricates After vowel elision, when two a$ricates are to occur next to each other, the "rst a$ricate in the sequence becomes an alveolar stop. In the next example, following 128 vowel elision,73 the /ʃ/ of =ʃa becomes [t͡ʃ] (as per the rule given in §3.3.3). When this happens, at an intermediate level, it comes in contact with the /t͡ʃ/ of tikit ͡ʃ and dissimilation occurs. (93) tikit́t ͡ʃa tikit ͡ʃi=ʃa other=ADD ‘another too’ In practice, the chances that two a$ricates occur adjacent to each other are very restricted. A limited set of items attaching to a stem that possesses another a$ricate must occur in the appropriate morphophonological context: =ʃa ‘Additive’, -t͡ʃa ‘Negative’ and the historically related form t͡ʃau ‘Negative nominalizer’. 3.4. Optional variation of consonants 3.4.1. Fricativization of /t͡ʃ/ In word-"nal environments, the a$ricate /t͡ʃ/ can be realized as a fricative, neutralizing the opposition with the alveolar /ʃ/. The following rule formalizes this variation: 73. In this case, vowel elision applies to the third vowel (from the left). Wampis deletes the last vowel of a word, and then each alternating vowel starting from the third vowel from the left. Note that some morphemes, like =ʃa above, are immune to apocope. Vowel elision is explained in Chapter VI. 129 /t͡ʃ/ → [ʃ] / ___# The neutralization of /t͡ʃ/ and /ʃ/ at the end of words can be observed comparing (94) with (95): (94) tí̪kiʃ ti̪kit͡ʃi ‘other’ (95) tí̪kiʃ ti̪kiʃi ‘knee’ As we can see, the underlying /t͡ʃ/ of ‘other’ becomes a fricative. This surface form is undistinguishable from ‘knee’ in the nominative form. Notice that the fricativization process is particular of /t͡ʃ/, as it does not happens with the alveolar a$ricate /t͡s/, as shown in (96). (96) tɨ̪n̪tɨ̪t́͡s tɨ̪n̪tɨ̪t͡sa ‘proper noun’ 3.4.2. /h/ → [f] Very occasionally, it is possible to hear the glottal fricative /h/ being realized as a labiodental fricative [f] or even as a voiceless bilabial fricative [ɸ], as in (97). 130 (97) óuftu̪kɛɛ Auhutu̪kai ‘proper noun’ This occasional change /h/ → [f] may be due to in/uence from Peruvian Amazonian Spanish. In this variety of Spanish, there is a similar permutation between /x/ and /f/ (e.g. > [fan] ‘proper noun’, > [xualta] ‘fault’) found in Amazonian Spanish,74 whose in/uence may be at the origin of the Wampis process. In Peruvian Amazonian Spanish, the change x > f ~ ɸ occurs preceding back vowels in word-initial or word-internal positions (Ramírez 2003; Vallejos 2014). However, unlike Peruvian Amazonian Spanish, I have no examples of /h/ → [f] occurring at the beginning of the words in Wampis, only word-internally. 3.5. Distribution of consonants Table 3.4 summarizes the distribution of the Wampis consonants that I have described in the preceding sections. The table refers to the possible combinations of phonemes in the underlying structure of Wampis words. Note that underlyingly, the Wampis language has a strong preference for simple open syllables, and it generally 74. This permutation is also present in the Spanish spoken by the Wampis, which is based on Amazonian Spanish. 131 avoids CC clusters not involving a nasal as "rst element. This structure, however, is completely overridden when vowel elision applies. Vowel elision processes are very pervasive in Wampis and they obscure quite considerably the rather straightforward underlying patterns of Wampis words. (Because vowel elision is associated with the prosodic structure of the language, it is analyzed in Chapter VI, see §6.4.4.) Table 3.4 is largely based on my database but I also checked Jackway et al.’s dictionary (1987). Thus, while I do not claim that the results shown in this table are necessarily exhaustive, they are the most comprehensive for Wampis to this date, giving a very good general idea of Wampis consonant restrictions and combinatory patterns. In general, no consonant is allowed word-"nally in the underlying representation of words. In fact, there are very few items (including clitics and su#xes) that end with a consonant in Wampis—and in most of those cases, a historical reconstruction with a "nal vowel in the morpheme is almost always clear. The word- "nal no-consonant restriction does not apply at the phonetic level, where consonants can occur at the end of the word after "nal vowel elision. I do not count here onomatopoetic words that can have a glottal or another consonant word "nally, as they can be regarded as a-systematic. 132 Table 3.4. Distribution of consonants in Wampis /#___/ /V.___/ /N.___/ /___./ /___#/ /___i/ /___ɨ/ /___a/ /___u/ /p/ ✓ ✓ ✓ --- --- ✓ ✓ ✓ ✓ /t/ ✓ ✓ ✓ --- --- ✓ ✓ ✓ ✓ /k/ ✓ ✓ ✓ --- --- ✓ ✓ ✓ ✓ (/ʔ/) --- ✓ --- --- --- ✓ --- ✓ --- /m/ ✓ ✓ --- ✓ --- ✓ ✓ ✓ ✓ /n/ ✓ ✓ --- ✓ --- ✓ ✓ ✓ ✓ (/ɲ/) --- ✓ --- --- --- ✓ --- --- ✓ /s/ ✓ ✓ --- --- --- ✓a ✓ ✓ ✓ /ʃ/ ✓ ✓ ✓ --- --- ✓ --- ✓ ✓ /t͡s/ ✓ ✓ ✓ --- --- --- ✓ ✓ ✓ /t͡ʃ/ ✓ ✓ ✓ --- --- ✓ --- ✓ ✓ /r/ --- ✓ --- --- --- ✓ ✓ ✓ ✓ a All of the /si/ sequences in the database occur word-internally. Turning now to preferred and dis-preferred consonant-vowel pairs, the incompatibility of the vowel /ɨ/ with the palato-alveolars /ʃ/ and /t͡ʃ/ is a signi"cant "nding. The language drastically disfavors syllable combinations between palatal a$ricates and the high central vowel. The incompatibility of /t͡s/ and the vowel /i/ is also very remarkable.75 Though it may not appear clearly from Table 3.4, the 75. Gnerre (2010) reports a similar situation for Shuar too. 133 combination /si/ is also very disfavoured in Wampis—strikingly, there is no occurrence of this combination word-initially, and word internally it is restricted to a few lexical words, most of them nouns related to /ora or fauna, or loanwords from Spanish (e.g. kirusinka ‘kerosene’). By contrast, the combinations /t͡sɨ/ and /sɨ/ are quite frequent. Thus, it seems that Wampis has a very interesting phonotactic pattern with respect to the class of sibilants: • The palato-alveolar sibilants /ʃ/ and /t͡ʃ/ favor combinations with /i/ and disfavor combinations with the closest vowel to /i/ in the vowel space, which is /ɨ/. • The non-palato-alveolar sibilants /s/ and /t͡s/ disfavor combinations with /i/ and instead favor combinations with the next closest vowel, which is /ɨ/. The pattern described above can be thought of in terms of adaptative dispersion; i.e. in terms of ease of articulation and perceptual distinctiveness. In fact, it has been found that languages favour certain combinations of consonants and vowels while restricting other sequences. For instance, Maddieson and Precoda (1992) found that some languages avoid combinations of glides with cognates vowels, as well as pointed out that some other languages avoid velars before high vowels. Likewise, in Wampis, there seems to be preference and avoidance of certain combinations with the class of 134 sibilants. In the case of Wampis, it is possible to say that there is a neutralization of /t͡ʃ/ and /t͡s/ in speci"c environments: t͡s → t͡ʃ / ___i t͡ʃ → t͡s / ___ɨ It is not possible to say that there is neutralization of /s/ and /ʃ/ because there are a few examples of /si/ sequences as explained above; however, the strong preference of /s/ to pattern with /ɨ/ and of /ʃ/ to pattern with /i/ suggests that ease of articulation76 may have played a key role in determining combinatorial patterns of sibilants and vowels in Wampis. A more thorough study that takes into account other factors, such as vowel frequency, for instance, may shed more light into the striking patterns found in Wampis, and will constitute a good of point of future research. 3.6. Vowel phonemes Wampis possesses a system with four contrastive vowel qualities, including a high central vowel /ɨ/ that is quite typical of Amazonian languages (Dixon & Aikhenvald 1999; Payne 1999a). The system consists of eight vowels: there are four contrastive oral vowels and each has a contrastive nasal counterpart. Thus, the relevant 76. Lindblom and Maddieson (1988) show that ease of articulatory e$ort seems to play a role in determining languages’ consonants sets. 135 features of the Wampis vowel system are height, frontness/backness and oral/nasal prosody. The vowel system is presented in Table 3.5. Table 3.5. Wampis vowel inventory Front Central Back High i i ̃ ɨ ɨ ̃ u ũ Low a ã 3.6.1. Vowel quality In order to accurately describe vowel quality, measurements of the "rst and second formants of vowels for thirty Wampis words were taken (details of the study are described below). The results of these measurements allow us to know the actual place of articulation of vowels over acoustic space. One Wampis woman, 29 years old, and one Wampis man, 42 years old, were recorded saying a series of thirty words. The words were recorded three times in isolation and one time within the carrier sentence [tʃitʃaḿka _____________ tah́i núwɛiti] which translates as ‘the word ___________ is what we say’ (this carrier sentence was suggested by my Wampis teachers). Measurements of F1 and F2 formants were then 136 taken using PRAAT, a free software for phonetic analysis developed by Paul Boersma and David Weenink (http://www.praat.org). The measurements of F1 and F2 were made at the mid-point of the vowel to diminish possible consonant perturbation e$ects. The mean values of the "rst and second formants for each vowel are presented in Table 3.6. Nasal vowels were not measured. Table 3.6. Means for F1 and F2 measurements of Wampis vowels Vowels F1 F2 i ɨ a u 441.586 481.88 777.873 393.645 2595.739 1597.941 1768.847 1079.221 The relationship between the F1 and F2 is summarized in a vowel space plot. This was done by matching grid references of the means of F1 and F2 of each vowel. Figure 3.3 shows the plot-chart for Wampis vowels. 137 5001000150020002500 200 300 400 500 600 700 800 900 1000 F1 (Hz) F 2 ( H z ) i ɨ u a Figure 3.3. Plot-chart of Wampis vowel acoustic space 3.6.2. Vowel oppositions The following oppositions con"rm the phonological status of /a, i, ɨ, u/. (98) i vs ɨ /int͡ʃi/ ‘sweet potato’ vs /sɨnt͡ʃi/ ‘strong’ i vs u /ti̪/ ‘intensi"er’ vs /tu̪/ ‘say’ i vs a /ta̪kita̪/ ‘hatch (N)’ vs /ta̪kata̪/ ‘work (N)’ ɨ vs u /nukɨ/ ‘only that’ vs /nuku/ ‘mother’ ɨ vs a /ɨa/ ‘look for’ vs /aa/ ‘write’ 138 u vs a /nuku/ ‘mother’ vs /nuka/ ‘leaf’ 3.6.3. Oral versus nasal contrasts The following contrastive examples show the phonemic status of oral versus nasal vowels. (99) i vs i ̃ /hii/ ‘"re’ vs /hii/̃ ‘his eye’ ɨ vs ɨ ̃ /uɨɨ/ [wɨɨ́] ‘salt’ vs /uii/̃ [wɨɨ́]̃ ‘he is gone’ u vs. ũ /uuru/ ‘cry several’ vs /ũuru/ ‘shake (earthquake)’ a vs a ̃ /sasa/ ‘small waterfall’ vs /sas̃a/̃ ‘hoatzin (Opisthocomus hoazin)’ 3.6.4. Distribution of vowel clusters As Table 3.7 shows, combinations of all vowels are possible in Wampis with the exception of /iɨ/ and /ɨi/ clusters. Table 3.7. Distribution of vowel clusters /i/ /ɨ/ /u/ /a/ /i/ ✓ --- ✓ ✓ /ɨ/ --- ✓ ✓ ✓ /u/ ✓ ✓ ✓ ✓ /a/ ✓ ✓ ✓ ✓ 139 The caveat is that clusters of phonological non-identical vowel are liable to become a cluster of glide+vowel if the "rst vowel in a VV cluster is a high vowel /i/ or /u/. In the case of the vowel /ɨ/ in an ɨV cluster, a velar approximant [ɰ] can be inserted optionally (i.e. the cluster may become [ɨɰV] on the surface form). See §3.7.3 and §4.2 for more details about glides. Vowels /i/ and /ɨ/ are neutralized in the vicinity of each other, as they do not co-occur. This pattern seems to relate to the neutralization of sibilants before these same vowels /i/ and /ɨ/ (see §3.5). 3.7. Vowel allophony 3.7.1. Assimilation of /a/ The vowel /a/ is raised when it precedes the high vowel /i/ and /u/. This pattern can be summarized with the following rule: V → V / ____ V [+low] [+high] [+high] From the rule stated above, the sequences /ai/ and /au/ are pronounced [ɛi] and [ou], respectively. Optionally, these sequences may surface as [ɛɛ] or [oo], respectively—in fact, Wampis speakers show quite a bit of alternation between these 140 possible realizations. Word-initial and word-"nal environments seem to favor the realizations [ɛɛ] and [oo], whereas word-internally the preferred realizations are [ɛi] and [ou]. Notice that this surface alternation does not happen with the sequence /aɨ/.77 The speci"c allophonic rules are presented in Table 3.8. Table 3.8. Assimilation of /a/ Speci"c allophonic rules Examples Gloss /ai/ → [ɛi] ~ [ɛɛ] /___i /au/→ [ou] ~ [oo] /___u /aɨ/ → [ə]/ ___ɨ /nai/ → [nɛí] ~ [nɛɛ́] /au/ → [óu] ~ [óo] /aɨt́a̪/ → [əɨt́]̪ ‘tooth’ ‘distal demonstrative’ ‘green (not ripe)’ The surface realizations [ɛɛ] and [oo] may be a recent development, as neither Beasley and Pike (1957) nor Jakway et al. (1987) report them. Therefore, this change seems to be following a relatively transparent direction towards the direction of the quality of a raised vowel, e.g. *ai > ɛi > ɛɛ. This change has not been reported for other Jivaroan languages. The word /auhu/ ‘common potoo’ (Nyctibius griseus) seems to be “immune” (probably because of its apparent onomatopoetic nature) to the rule of 77. This may be due to the the rather infrequent combination of these vowels in that order. There is only "ve words with the sequence /aɨ/ in the database. 141 assimilation of /a/ and it is pronounced [aúhu]. 3.7.2. Lower allophones of high vowels When they do not received high tone, the high vowels /i/, /ɨ/ and /u/ have lowered lax allophones [ɪ], [ə] and [ʊ], as illustrated in the next examples. (100) /kuhi/ → [kúhɪ] ‘kinkajou (Potos 0avus)’ /hakɨ/ → [hákə] ‘warm water’ /kuiu/ → [kújʊ] ‘guan sp.’ This allophony does not have any further consequence for the Wampis grammar. Therefore, I only transcribe these allophones when necessary to illustrate their occurrence, like in the examples above. 3.7.3. Glides The approximants [j] and [w] in Wampis can be analyzed as positional allophones of the high vowels /i/ and /u/ in initial position and intervocalically.78 In other environments they are prohibited. (101) /iauaa/̃ → [jawaá]̃ The formation of glides can be more clearly seen in sandi environments: 78. With the one exception /j/ in the su#x -ji ‘3.PT’. See §3.2.7. 142 (102) aj́a a=ia cop=REM.PT (103) pɨŋ́kɨrt͡ʃowɛit ̪ʲi pɨnkɨra-t͡ʃau=aiti̪ good-NEG.NMLZ=COP.3+DECL ‘It is not good’ In (102), when the copula a carries the remote past su#x -ia, the glide [j] surfaces. Likewise, in (103), when the base pɨnkɨra-tʃau ‘not good’ hosts the copula =aita, the glide [w] surfaces. According phonemic status to the glides would add more segments to the Wampis inventory and, therefore, a new tier of at least sixteen potential oppositions per #GV and VGV environment: ji, jɨ, ja, ju; ji,̃ jɨ,̃ jã, jũ; wi, wɨ, wa, wu; wi,̃ wɨ,̃ wã, wũ. The occurrence of the glides can be predicted with no major problems; thus, it is better to posit them as positional allophones of /i/ and /u/. 3.7.3.1. Fricativization of [w] The approximant [w], when preceding a high front vowel /i/, becomes fricativized. This change is formalized in the following rule: [w] → [β]/___i 143 The following examples illustrate the above rule: (104) /ui/ →[βií] ‘I’ /saaui/ → [saaβ́i] ‘transparent’ /uinka/ → [βiŋ́kʲa] ‘blue’ Sometimes, [w] also surfaces as a labiodental approximant [ʋ] in the same environment. This is also a minor change and is not transcribed unless necessary. 3.8. Other hypo-articulatory (casual speech) phenomena In this section, some other cases that are not generally systematic are described. These can be analyzed as cases of hypo-articulation, as they do not occur in isolation but in normal speech; they alternate with otherwise general rules. 3.8.1. De-palatalization Occasionally, speakers do not palatalize a consonant in an environment that would otherwise require palatalization. This is not frequent and may be due to hypo- articulation, as the same word generally occurs palatalized. Here are a couple of examples that in my data show both realizations. (105) miɲitt̪a̪hɛɛ ~ minitt̪a̪h́ɛɛ mini-ta̪ta̪-ha-i come-DEF.FUT-1SG.SBJ-DECL ‘I will come.’ 144 (106) kiń̪t ̪ʲa ~ kiń̪ta̪ kinta ‘Afternoon’ In addition, the "rst /n/ of /pininka/ never seems to undergo palatalization, it is pronounced [piniŋ́] and not [piɲiŋ́]. This is probable due to consonant dissimilation. 3.8.2. Palatalization of [w] and /h/ Sometimes the glide [w] and the fricative consonant /h/ are optionally palatalized after a high vowel /i/. (107) /iuant͡ʃi/ → [iwańt͡ʃ] ~ [iwʲant͡ʃ] ‘devil’ /ihu/ → [ih́u] ~ [ih́ʲu] ‘palm heart, chonta (Astrocarium chonta)’ 3.8.3. Alternation [mʷɨ] ~ [mu] In rapid speech, the sequence [mʷɨ] is sometimes realized as [mu]. (108) /amɨ/ → [amʷɨ] ~ [ámu] ‘you’ /kamɨ/→ [kamʷɨ] ~ [kámu] ‘interjection’ 3.8.4. Lenition of /ʃ/ Though very infrequent in my data, the palato-alveolar /ʃ/ may surface as a velar fricative [h] intervocalically. For example, /amɨ=ʃa/ ‘you too (2SG=ADD)’ may surface as [aḿʷɨha]. I observed this variation primarily in Candungos (Upper Santiago), but not in communities of the Middle Santiago such as Puerto Galilea or Huabal. 145 However, I was told by my teachers that the debuccalized realization may be pronounced by any Wampis speaker sometimes in rapid speech and they did not consider it a marked error (their explanation ran along the lines of “people sometimes speak like that”). Thus, it seems that this is a case of hypo-articulation rather than a dialectal variation. 3.8.5. On the voicing of stops In previous works on Jivaroan languages (Beasley & Pike 1957; Turner 1958b; Turner 1992; Gnerre 2010), stops have been described as assimilating to a preceding homorganic nasal. With regards to Wampis, Beasley and Pike state: “In the middle of words, after nasal consonants, however, voiced allophones of the stops and a$ricates appear. The /p/ may be voiced preceding /ɾ/, and /k/ may be voiced before /m/ and /ŋ/” (1957: 1). I have not found any example of voiced a$ricates in my data. As for stops, this process of voicing is not systematic and, save for a few examples in discourse (which may be considered casual speech), voiced stops generally do not occur before nasals. In elicitation, when words were pronounced in isolation, stops following a nasal consonant are never voiced. One may argue that in the past there used to be a productive voicing rule of stops following a nasal consonant, judging by geographic and 146 proper names that have survived such as , , , , , and so forth. These are, mostly, Spanish renditions of Jivaroan names (some of them dating back to colonial times), though in principle the Spanish would have been well-used to hearing the di$erence between a voiced and a voiceless stop. However, all of these names are pronounced with voiceless stops in Wampis, at least synchronically:, e.g. [kan̪tú̪nkus], [kaŋkaś], [wampiś], [t͡ʃinkanaś], and so on. It may as well be the case that these place names were based on the speech of other Jivaroan groups where voicing of stops is apparently more frequent, such as Shuar.79 Sociolinguistic factors may as well be playing a role, as my teachers sometimes regarded voicing as “not Wampis”. Voicing of stops in the environment of a nasal does happen in Wampis but it is rather infrequent; certainly far from systematic. When it does happen, the environment that is by far more relevant to the voicing of stops is /N.___r/, not just /N.___/ or /___r/. Even in these instances, though, voicing is not systematic and voiced and voiceless realizations alternate. It is interesting to notice that the voicing of stops in the 79. For Shuar, Saad (Saad 2014: 18) states that voicing of stops after nasal occurs but is optional. 147 environment of a nasal consonant was quickly corrected in transcription, with speakers stating that it was better transcribed with a voiceless grapheme (e.g.

instead of ). In the following example, which has a close phonetic transcription, the relevant part is how the nominalized verb iruna-tu-ra-mau (which is glossed as ‘get together’) surfaces. We notice that the pronunciation of this morphologically complex form surfaces as [iɾúnðɾamu] (the relevant context is underlined in the example), with the underlying /t/̪ being voiced. However, other stops that occur in the example in the immediacy of a nasal or /N.___r/ are not voiced (for instance, see the word [nawan̪tɾ̪umhɛɛ]̃ ‘with your daughter’, where /t/̪ is not voiced in spite of following /n/). Thus, it must be concluded that voicing is not systematic. (109) iɾunðramu aḿʷɨ nawan̪tɾ̪umhɛɛ ̃puhústa̪san iruna-tu̪-ra-mau amɨ nauanta̪-rumɨ=hai ̃ puhu-sa-ta-nu get.together-APPL-DISTR-NMLZ 2SG daughter-2SG=COM live-ATT-PURP-1SG.SS ‘I [want] to live with your daughter.’ 3.8.6. On devoicing of vowels Devoicing of vowels was also reported for Wampis by Beasley and Pike: “Certain vowels which in the middle of words in a restricted word list are clearly voiced, and are members of the ordinary voiced vowel system, lose their voicing when unstressed, and when the su#xes following them are dropped from the word” (1957: 6). I have not 148 found such process. It may be that since the 1950s devoicing of vowels has lead to vowel elision, which is quite spread in Wampis as some of the examples in this chapter have shown.80 Therefore, devoicing of vowels can be regarded as a previous historical step to elision. What I did "nd during acoustic analysis is that speakers may rarely pronounced the last vowel of a word with very little energy. In other words, some "nal vowels in words with more than two underlying vowels—the last vowel of a word with more than two underlying vowels undergoes apocope, as will be seen in detail in §6.4.4 (apocope is a systematic process in Wampis)—may actually occur in citation forms (during elicitation), but in a very shortened form realized with much less energy. They are hardly perceivable and in fact I noticed them primarily after looking at spectrograms generated by PRAAT. These vowels are not devoiced, they show a little voicing as illustrated with the word /uiśuta̪/ ‘ant sp.’ (which is otherwise pronounced [wiśut]̪) in Figure 3.4. 80. See §6.4.4 for an analysis of vowel elision. 149 Figure 3.4. Spectrogram of [wiśuta̪] showing a very reduced vowel at the end The phenomenon of vowels not undergoing apocope in Wampis is very rare, in discourse as well as in elicitation vowels in "nal position undergo apocope. 3.9. Loanwords 3.9.1. Spanish loanwords Two types of loanwords from Spanish coexist in Wampis: those which comply with the Wampis phonological system and those which do not. Most Wampis people are 150 generally familiar with Spanish (many of them, especially middle-aged and younger are bilinguals) and therefore are able to pronounce Spanish words without major phonological adaptation. It is plausible to think that the borrowings that do not undergo phonological adaptation are more recent than the ones that have been adapted, though some recent borrowings may have been adapted due to their importance for language policy, such as papí ‘book’ (from Spanish ‘paper’). Therefore it is possible to "nd common Spanish words as part of a Wampis speaker’s everyday speech, such as [polisía] ‘policeman’ (from Spanish ), and Spanish loanwords that have been reinterpreted following Wampis rules, such as /suntara/ → [suntáɾ] ‘soldier’ (from Spanish ). Both types of loanwords are fully integrated into the Wampis language rules when not in careful speech, i.e. they are bound to Wampis rules such as high tone shift when they receive morphemes that have an e$ect in their phonetic realization, cf. for /bála/ ‘bullet’ (from Spanish ), the pronunciations is identical to Spanish: [bála]. However, when this item receives the accusative =na it changes the placement of the high tone, just like regular Wampis nouns do (cf. §6.6): [balán] (/bala=na/ ‘bullet=ACC’). Table 3.9 presents a sample of Spanish borrowings used in Wampis. 151 Table 3.9. Sample of Spanish loanwords in Wampis Wampis Spanish Gloss Phonological adjustment /mat͡ʃu/ [mát͡ʃu] /panka/ [páŋka] /uru/ [úru] /iskupíta̪/ [iskupít]̪ /papí/ [papí] /arusa/ [arús] /iusa/ [júsa] /iumunka/ [jumúŋka] /irinku/ [iríŋku] ‘male’a ‘bread’ ‘gold’ ‘shotgun’ ‘book’b ‘rice’ ‘god, church’ ‘lime’ ‘white person’ o>u C#+-ka o>u e>i; o>u e>i o>u; rr>r; C#+-a C#+-a; dio> iu C#+-ka; liu>iu #Cr>#VrV; o>u; g>k Non-adapted loanwords /este̪/ /kartu̪t͡ʃo/ /bala/ /lin̪te̪rna/ /gasolina/ /kanta̪-ma/ c ‘pragmatic marker’ ‘cartridge’ (for gun) ‘bullet’ ‘/ashlight’ ‘gas’ ‘sing’ a In Wampis, this is a familiar term to refer to a son. b In Spanish, means ‘paper’, not ‘book’. c This is a Spanish demonstrative that means ‘this’, but in Peruvian Spanish it is used also as a hesitation device. The Wampis used this word with the latter function. Perhaps the most interesting feature of the examples presented in Table 3.9 is that in Wampis there is a paragogical vowel that by default seems to be /a/. In words ending with a nasal consonant, the paragogical syllable /-ka/ is added. Whenever the Wampis requirement that underlyingly words must end in a vowel (V#) seems to be at risk of being violated by the loanword, these paragogical segments, /a/ and /ka/ are 152 inserted to preserve the Wampis word structure requirement. 3.9.2. Quechua loanwords Though not too numerous, Quechua loanwords are not uncommon in Wampis. It is uncertain when the contact between Wampis and Quechua started. Most borrowings probably are very old, originating from at least the 17th century or later. Quechua has never been spoken in the Santiago area, thus it is conceivable to think that Quechua words could have been incorporated into Wampis as a result of the expansion of the Quechua language by missionaries working in the Amazon in colonial times. As we saw in Chapter II, the Wampis were never fully colonized or fully evangelized, therefore it may be the case that loans from Quechua were incorporated as Quechua gained notoriety in the larger Ecuadorian and Peruvian Amazon area as some sort of lingua franca. However, the Wampis (or pre-Wampis) may have been in contact with Quechua- speaking peoples prior to the above pointed time, as they traded with Andean groups before the Spanish conquest. Some of these groups, for instance the Cañari (and even proposed Jivaroan groups such as the Guayacundos), were probably already quechuanized by the time the Spanish arrived. In fact, the Inca had expanded Quechua to the northern Andean territories as they expanded their territory, and were pushing to 153 conquer the Upper Marañon region before the arrival of the Europeans.81 Therefore, while some borrowings may have come as a result of the expansion of Quechua in colonial or more modern times, it is also quite possible that some of them originated in pre-Hispanic times. Table 3.10 shows examples of Quechua borrowings that occur in Wampis.82 Most loans from Quechua in my database are nouns or adjectives. As we can see, there is little re-adjustment changes of some of the original Quechua words into Wampis. One interesting fact is the insertion of the vowel /a/ to comply with the Wampis (C)V(N) pattern, cf. ‘town’. There are some words that come ultimately from Quechua but were most likely borrowed through regional Amazonian or even standard Peruvian Spanish (which has been in/uenced by Quechua) in recent times. For instance, Wampis kant͡ʃa ‘sports "eld’ is a pan-Spanish borrowing from Quechua; the word apu ‘community chief’, which is also present in Spanish, designates a socio-political institution that was introduced in the 20th century in many communities in the Amazon as a way to have a visible leader 81. Notice that the Quechua variety that was expanded by the Incas was not the same as the one used by missionaries later (Cerrón-Palomino 1987). 82. Stark and Muysken (1977) and Cerrón-Palomino (1994) were consulted for Quechua words. 154 in the newly created towns. Table 3.10. Sample of Quechuan loanwords in Wampis Wampis Quechua Gloss Phonological adjustment /sɨnt͡ʃi/ [sɨńt͡ʃi] /iaakata̪/ [jaákat]̪ /ima/ [iḿa] /pínkui/ [piŋkui] /kut͡ʃa/ [kút͡ʃa] /kat͡ʃu/ [kat́͡ʃu] /ʃaa/ [ʃaá] /apu/ [aṕu] /kant͡ʃa/ [kańt͡ʃa] /iat͡ʃa/ [jat́͡ʃa] ‘strong’ ‘town’ ‘Intensi"er’a ‘/ute’ ‘puddle, lake’ ‘horn (to call)’ ‘corn sp.’ ‘chief’ ‘sports "eld’ ‘wise, clever’ i>ɨ a>aa; V-insertion u>∅ tʃ>ʃ; r>∅ a In Quechua, this word is polysemous. It is an intensi"er for exclamative expressions, as well as an interrogative word ‘What?’, it also means ‘thing’. Muysken (2010; 2013) has noticed certain similarities in the way the copula clitics work in Ecuadorian Quichua and Jivaroan languages (particularly, Shuar), though his analysis focuses on the form, not the diverse functions and contexts where the copula occurs, which seem to di$er between the Shuar copula and the Ecuadorian Quichua copula. 3.9.3. Loanwords from other languages There is a number of words in Wampis that seem to proceed from other 155 Amazonian languages, most notably from Cariban and Tupi-Guarani languages. The exact way how these words entered into Wampis is unknown yet (and it is theoretically possible that they come from an original unidenti!ed proto-source). However, the fact that many of these words are also found in other Jivaroan languages may indicate that they rather were present or borrowed in past states of Jivaroan. Gnerre (2010) speculates that some words in Jivaroan languages may come from other Amazonian languages. Table 3.11 provides a few examples of these words and their possible sources. Table 3.11. Some Wampis words that may come from other Amazonian languages Wampis Gloss Possible source (according to Gnerre (2010)) /tu̪ná/ [tu̪na]́ /t͡sukanká/ [t͡sukaŋka]́a /iauaa/̃ [ɲawãá] /kanú/ [kanú] ‘waterfall’ ‘toucan’ ‘dog (also jaguar)’ ‘canoe’ Cariban Tupi ́ Tupi ́ Arawak a It is possible that the sequence -ka that occurs in certain loanwords from Spanish is forming this word too, thus we would have the structure tsukan-ka. In addition, Gnerre (2010) has argued for a potential genetic connection of Jivaroan with Arawak, and Wise (2014) has presented evidence of contact, rather than genetic relationship, between Jivaroan and Arawak. Much of the discussion is yet 156 insu#cient to make any de"nitive statement. 3.10. Notes on transcription Throughout this dissertation I follow IPA conventions for representing the Wampis consonants and vowels. However, for practical reasons, some decisions have been made regarding transcription of examples. Because the dental stop /t/̪ does not have any other non-dental counterpart, I have used the dental symbol ̪ in this chapter, but I will represent it simply with in the remaining of the dissertation (this a practical decision already made by Overall (2007) for Awajun). I use the symbol to represent the Wampis rothic phoneme whose realization may vary between [ɹ] and [ɾ]. For practical convenience, other “minor” allophonic variation is not represented in the remaining of the dissertation (unless a close phonetic transcription is necessary to prove a statement). For instance, the allophone [mʷ] of /m/, which only occurs before /ɨ/, or the allophone [β] of the glide [w], which only occurs before /i/, are not represented, since they occur in very speci"c environments. Other phenomena that is widespread in the language, such as palatalization, is represented in the phonetic transcription of examples. In addition, I will mark the most likely locus of nasalization with a tilde Ṽ, and an 157 underlying high tone with an accent mark V́, following the observations made throughout Chapter VI, which is dedicated to the prosody of Wampis. 158 CHAPTER IV ALLOPHONIC GLIDE DERIVATION, EPENTHETIC GLIDE INSERTION AND PHONOTACTICS 4.1. Introduction This chapter discusses the formation of glides in Wampis, as well as the syllable and word structure of the language. Glides are important in Wampis because they are treated as consonants for the purposes of word-internal vowel elision (cf. §6.4.5). Once glides are formed, syllables become relevant for the purpose of establishing rhythmic feet and vowel elision, which is very pervasive in Wampis. Hence, the structure of syllables are treated also in this chapter, before analyzing the morphophonology and prosody of the language, which will be seen in Chapters V and VI, respectively. In this chapter, as well as in Chapters V and VI, I will be adopting a rather rigorous formalism for greater illustration at the moment of describing word derivations. There are two main advantages in positing several steps in derivation of words: "rst, it is useful to describe very complex processes; second, it is also useful to show possible deep forms, especially useful for those seeking to do comparative work with other Jivaroan languages, and as a helpful reference for internal reconstruction. However, I do not 159 make any claim in terms of the cognitive processes that the several steps described in some examples here may represent. There are several theories that can help understand the link between the mental representation of words, their storage in our lexicon and their actual realization. The relevance of a language like Wampis for those theories is yet to be explored. The structure of the chapter is as follows: §4.2 o$ers an analysis of the formation of glides, and §4.3 describes the phonotactic patterns of Wampis, including an analysis of the syllable structure and combinatorial restrictions. 4.2. Derivation and insertion of glides There are three phonetic glides in Wampis: [w], [j] and [ɰ]. Glides in Wampis are of two classes: "rst, [j] and [w] are positional allophones of the phonemes /i/ and /u/, respectively. As stated in §3.7.3, the glides [j] and [w] are not considered as part of the underlying phonemic inventory of Wampis83 on grounds of analytical parsimony. Second, there are epenthetic glides [j], [w] and [ɰ] that are optionally inserted after a homorganic vowel. In this section, I analyze the derivation and insertion of glides in Wampis. 83. With the exception of the /j/ that occurs in the su#x -ji ‘3.PT’ as explained in §3.2.7. 160 The surface realization of glides is predictable for reasons that have to do with syllable constraints. What is more, it is not necessary to establish a di$erence between Vowel + Glide sequences (VG or GV) and Vowel + Vowel sequences; i.e. a contrastive opposition of [aj] vs. [ai], or [wa] vs. [ua] is never found in Wampis. Glides are formed very early in the phonetic derivation of words. Once created, they count as consonants for purposes of word internal vowel elision (but not apocope) and metrical high tone assignment, as will be seen in §4.2.2 (see also §6.4). In general, glide formation can be seen as the result of the interplay with syllable constraints that: (a) disfavor non-identical vowel sequences (i.e. two-mora) of the form V1[+high]V2; (b) prefer onsets over codas; and, (c) prefer simple onsets over complex onsets. From the list of constraints just mentioned, (a) favors the formation of glides in a V1V2 environment where V1 is a high vowel, and (b), (c) govern the mapping between segments and onset and coda positions. The interaction of these constraints can be seen at work in the next examples with simple words. (1) /nua/ → [nú.wa] ‘woman’ (not *[nwa]) /t͡ʃui/ → [t͡ʃú.wi] ‘paucar (Icteridae sp.)’ (not *[t͡ʃwi]) 161 /sua/ → [sú.wa] ‘huito (Genipa americana)’ [not *[swa]) /hɨã/ → [hɨ ̃.́ɰa] ‘house’ (not *[hɰa]) /kuiu/ → [kú.ju] ‘Spix’s guan’ As a consequence of (a)–(c) above, an important principle of word-formation in Wampis is that glides are allowed word-initially and intervocalically. Accordingly, the following rules apply in Wampis for the formation of glides. Rules in A and B explain the derivation of glides as allophones of /i/ and /u/. Rule C explains the epenthetic glides. A. Allophonic glides [j] and [w] are formed in intervocalic position. Two rules can be associated with this type of derivation. (A.i). Any vowel cluster involving a VV1V(V) sequence, where V1 is the high vowel /i/ or /u/, becomes V.GV(V). V V1 V (V) → V.GV(V) | {i, u} The examples in (2) illustrate this rule. (2) /auanki/ → (awaiŋki) → [awɛíŋki] ‘give back’ /asɨuai/ → (asɨwai) → [asɨẃɛɛ] ‘adze’ (from Spanish ) /kauaikama/ → (kawaikama) → [kawɛíkʲam] ‘loro machaco viper’ (Bothriopsis bilineata) /iuant͡ʃi/ → (iwant͡ʃi) → [iwańt͡ʃ] ‘mythological character’ 162 /kaiuka/ → (kajuka) → [kajúk] ‘black agouti’ /naiampi/ → (najaimpi) → [najɛím] ‘sky’ /uuɨha/ → (uwɨha) → [uwɨh́] ‘hand’ (A.ii). Any sequence of the form VVV1V, where V1 is the high vowel /i/ or /u/, becomes VVGV. The following rule captures this pattern: V V V1 V → VV.GV | {i, u} The rule in (A.ii) is similar to the previous rule (A.i), but the important di$erence is in the position of V1 (which represents the high vowel /i/ or /u/). This rule serves to explain (infrequent but attested) instances like the ones in (3). (3) /inaiia/ → (inaija) → [iɲɛíja] ‘stop, leave’ /ankaiia/ → (ankaija) → [aŋkɛíja] ‘wasp sp.’ B. High vowels /i/ and /u/ (but not /ɨ/) in word-initial position become glides [j] and [w], respectively, when they are followed by a di$erent vowel. The following rule formalizes this change. V1 V2 → GV/#____ | {i, u} The examples in (4) illustrate the above rule. (4) /iapi/ → [jaṕi] ‘face’ /iumi/ → [júmi] ‘water, rain’ /uakani/ → [wakań] ‘soul’ 163 /uɨɨ/ → [wɨɨ́] ‘salt’ /uinka/ → [wiŋ́kʲa] ‘blue’ /uaura/ → [woúr] ‘act.in.frenzy’ It is important to note that the high vowel /ɨ/ never becomes an approximant [ɰ]; so it never occurs in word-initial position as a glide. The velar approximant [ɰ] in Wampis only occurs intervocalically and it is purely epenthetic, as illustrated by the examples in (5). (5) /ɨakama/ → [ɨɰaḱma] ‘hunt’ /hɨã/ → [hɨɰ̃a] ‘house’ More details on the velar approximant [ɰ] are given in §4.2.4, see also next point, rule C. C. Epenthetic glides [j], [w] and [ɰ] are optionally inserted. Any sequence of the form V1V2, where V1 is a high vowel and is not word-initial, may insert a homorganic glide intervocalically. Examples: (6) /uiaitahai/ → (wijaithai) → [wij́ɛithʲɛɛ] ‘I am’ /sua/ → [súwa] ‘huito (Genipa americana)’ /uɨahai/ → (wɨɰahai) → [wɨɰ́ahɛɛ] ‘I am going’ 4.2.1. Importance of the ordering of rules Rules in A and B must apply in that order; that is, glides must be formed intervocalically before they are created at the beginning of the word. Consider (7): 164 (7) /iia/ → [ija]́ ‘fall’ /uuɨha/ → [uwɨh́] ‘hand’ /uui/ → [uwi]́ ‘peach-palm (Bactris gasipaes)’ /ɨuɨkama/ → [ɨwɨḱma] ‘make go, guide’ If we apply rule B in all of the above examples, the derivations of those words would be incorrect, e.g. *[jia] for ‘fall’, *[wuɨha] for ‘hand’, [wui] for ‘peach-palm’, and so on. For further illustration, consider the following examples with the forms [iẃa] ‘giant mythological cannibal’ and its metathesized variant [úja]. In both cases we have to posit intervocalic glide formation "rst, otherwise we would derive the words wrongly. (8) /iua/ → [iẃa] (not *[jua]) /uia] → [úja] (not *[uja]) The formation of allophonic glides must apply at the level of the morpheme, because once the allophonic glides are formed, they stayed the same for most other subsequent derivational and in/ectional processes. Consider a complex word like ‘I am eating’: (9) júawɛɛ iu-a-ua-i eat-IPFV-3.SBJ-DECL ‘He is eating.’ 165 In (9), we have a word that underlyingly contains several vowels: /iu-a-ua-i/. So following the rule A discussed in the previous section, the derivation would be [iwawɛɛ]84 (intervocalic glide formation applies "rst, so the second /u/ becomes [w]). This is not the correct output, though. Thus, we have to explain the surface form [júawɛɛ] as derivations occurring in the root /iu/ ‘eat’ and the su#x /ua/ ‘3.SBJ’, respectively, rather than on the complex word.85 However, allophonic glides can be reverted to their underlying vowel form following vowel elision if the derived glide is incompatible with the well-formedness of the word. This is illustrated in (10)–(11). First, let us state that in Wampis a third vowel of a word (from left to right) undergoes elision if it is in a CV sequence (vowel elision is analyzed in detail in Chapter VI). In the examples, the vowel to be deleted is underlined. (10) kut͡ʃańmaja kut͡ʃa=numa=ia puddle=LOC=ABL ‘from the puddle’ 84. Recall from §3.7.1 that the sequence /ai/ is pronounce [ɛi] or [ɛɛ]. 85. Morphophonological processes in Wampis are fairly complex. Many morphemes exhibit particular behaviors. Morphophonology is analyzed in Chapter V. 166 (11) ikaḿia ikama=ia forest=ABL ‘from the forest’ In (10), the ablative /ia/ surfaces as [ja], since it is in intervocalic position. However, in (11), it surfaces as [ia]. This occurs because, after vowel elision, the glide is between a consonant and a vowel, which contradicts the Wampis principle that glides are allowed word-initially and intervocalically. As a result, the palatal [j] reverts to [i]. This means that in Wampis there must be re-checking rules that ensure that rules of well-formedness are not violated. The following ordered derivation illustrates the interplay between vowel elision, glide formation and re-checking rules: (12) Input: /ikamaia/ ‘from the forest’ Glide formation: ikamaja Vowel elision: ikamja Re-check: ikamia ([j] reverts to [i]) Output: [ikaḿia] With regard to rule C, which explains the insertion of optional epenthetic glides, it must be preceded by rule B (derivation of glides word-initially). The examples in (13) show that if rule C applies before rule B, the output of the words would be incorrect (the incorrect derivations are given in parenthesis). (13) /iakuma/ → [jakúm] ‘howling monkey’ (not *[ijakma]) 167 /uais/ → [waiś] or [wajiś]86 ‘Ilex guayusa’ (not *[uwais]) /uari/̃ → [wari ̃]́ ‘what’ (not *[uwari]̃) /ui/ → [wií]87 ‘1SG’ (not *[iwi]) /iutai/̃ → [jútɛi]̃ ‘food’ (not *[ijutɛi]̃) 4.2.2. Properties of glides Glides exhibit the following properties: • Glides are consonant-like for the purpose of word-internal vowel elision. For instance, in a word like ‘rest’ in (14) the glide is treated as a consonant: the third vowel from the left (which is to be deleted) is the third underlying /a/ (if the glide counted as a vowel, then the third vowel would be the second /a/ from the left): (14) /aiamatana/ → a.ja.ma.ta.na → [ajaḿat] ‘rest (ACC)’ • From the above point, it follows that glides are do not count as moras. Evidence of this it that words that have only one mora on the surface lengthen their vowel to comply with a two-mora restriction for words in Wampis—it will be seen in Chapter VI that the prosodic word in Wampis requires two-moras. A word like /ui/ ‘1SG’ derives a glide, then it must lengthen its vowel /i/: /ui/ → [wií]. In addition, because glides are 86. The second realization, [wajiś], is the realization with the optional epenthetic [j] inserted. 87. Mono-vocalic roots lengthen their vowel to comply with a two-mora per prosodic word restriction in Wampis. 168 not moraic, they are not tone bearing units. • Interestingly, however, glides are not treated as consonants for the purpose of word- "nal vowel elision (apocope). In (15), just like the last vowel /ɨ/ is deleted in the word ‘root’, it would be expected that the last vowel /a/ of the word ‘from the puddle’ be deleted; however, it is not deleted. (15) /kankapɨ/ → (kaŋ.ˈka).pɨ → [kaŋkaṕ] ‘root’ /kutʃanumaia/ → (ku.ˈt͡ʃa).(nu.ˈma)ja → [kut͡ʃańmaja] ‘from the puddle’ 4.2.3. Epenthetic versus allophonic glides A potential problem of analysis arises around the distinction between surface epenthetic glide and allophonic glide forms of an underlying vowel. In some cases, if the nature of the underlying vowel is not determined, the phonological derivation may result in unprecedented word formations. For instance, it can be posited that the word [júwi] ‘squash’ can have two possible underlying representations: /iui/ or /iuui/. In the "rst hypothesized case, /iui/, rules of initial glide formation (rule B) and C (medial glide insertion) could derive [júwi]. But notice that rules must be applied in order, thus rule A.i (intervocalic glide formation) must apply "rst. However, in that case, the derivation should be [iẃi], not [júwi] as the word occurs. Hence, the only way to derive [júwi], with the rules as claimed to be ordered above, is by positing an underlying form 169 /iuui/ which results in [júwi] after applying the rule in A.ii and B: (16) /iuui/ ‘squash’ Intervocalic glide formation (Rule A.ii): iuwi Initial glide formation (Rule B): juwi Output: [júwi] The following examples show that regarding glides, following the rules is important for the correct derivation of similar surface forms: (17) a. /ui/ → [wií] ‘1SG’ b. /uui/ → [uwi]́ ‘peach-palm’ (18) a. /ia/ → [jaá] ‘who’ b. /iia/ → [ija]́ ‘fall down’ (19) a. /iua/ → [iẃa] ‘mythological character’ b. /iuua/ → [júwa] ‘cousin of female’ (20) a. /t͡ʃiua/ → [t͡ʃiẃa] ‘trumpeter (bird sp.)’ b. /t͡ʃuanka/ → [t͡ʃuaŋ́] or [t͡ʃuwaŋ́]88 ‘buzzard’ In (17), if the word for ‘peach-palm’ were /ui/ phonemically, we would expect 88. The realization [t͡ʃuwaŋ́] shows an optional epenthetic glide [w]. 170 its phonetic realization to be [wií], just like the form for the "rst person singular pronoun. Instead, we have [uwi]́. Likewise, in (18), /iia/ ‘fall down’ di$ers from /ia/ ‘who’ in the number of moras underlyingly, otherwise it would be expected that both would have the same phonetic realization. The same holds for the pair /iua/ and /iuua/ in (19): the di$erence is in the underlying moras; whereas (20) shows a phonetic near minimal pair between a word that has an allophonic [w] (‘trumpeter’) and a word that has a epenthetic [w] (‘buzzard’). Therefore, an underlying (V)V1V1V2 sequence must be posited to account for the realizations [uwi]́, [ij́a] and [júwa] respectively. Thus, the di$erence here is that no glide insertion (i.e. ∅ → G) is claimed for these words; rather, it is posited that a change V → G occurs (in this case, {i, u} → G/V___V). For (20), it is clear that in the word ‘trumpeter’ there has to be an underlying /u/, otherwise the glide [w] cannot be explained. On the other hand, in ‘buzzard’ the glide [w] is optional, as evidenced by the fact that there are two possible realizations of the word, [t͡ʃuaŋ́] or [t͡ʃuwaŋ́]. Epenthetic glides can always be optionally inserted in the environments u___i, u___a, and ɨ___a in rapid speech. (21) /kuiʃi/ → [kuwiʃ́] ~ [kuiʃ́] ‘ear’ /muit͡sa/ → [muwit́͡s] ~ [muit́͡s] ‘type of jar’ 171 /nua/ → [núwa] ~ [núa] ‘woman’ /tsɨasa/ → [t͡sɨɰaśa] ~ [t͡sɨás] ‘dart poison’ /kuitama/ → [kuwit́am] ~ [kuit́am] ‘look after’89 With relation to this, epenthetic glides are actually more frequently not inserted, and this can be used to help decide whether a glide is epenthetic or not. Therefore, whereas it is possible to "nd realizations such as /nua/ → [núwa]~[núa] ‘woman’, that same alternation does not occur with words such as /iuui/ ‘squash’; i.e. *[yúi] never occur as an alternate realization of [júwi]. This suggests that, for Wampis, the underlying segments have to be mapped into syllable positions, but non-underlying (epenthetic) segments do not. Finally, in some other cases, there is also incidental evidence that helps distinguish between epenthetic and allophonic glides. For instance, the word /uuɨha/ [uwɨh́] ‘hand’ has the variant pronunciation [ɨwɨh́] (i.e. there are two variant underlying lexical forms). Thus, underlying representations such as /uuɨha/ and /ɨuɨha/ must be posited, otherwise—i.e. if we analyze the word as /uɨha/~/ɨɨha/ in its lexical underlying form—in the second case (/ɨɨha/), there would be no need of inserting a glide, as sequences of two identical vowels are perfect in Wampis. Hence, we must 89. This is a borrowing from Spanish ‘look after’. 172 propose /uuɨha/~/ɨuɨha/ as the lexical representations for these variants, both being derived by applying the same rule in A.i given at the beginning of §4.2. 4.2.4. On the approximant [ɰ] Unlike the other high vowels, the high central vowel /ɨ/ does not have an approximant allophone in intervocalic position,90 though, as we saw in the previous section, a velar epenthetic glide can be inserted next to it. There are no examples of VɨV sequences in roots in my data. Most examples of such sequences come from morphologically complex words. In these cases, /ɨ/ does not change to [ɰ] but it may insert an homorganic glide, or just form a triphthong, as illustrated in (22)–(23). (22) kɨɨámu ~ kɨɨɰ́amu (not *[kɨɰ́amu]) kɨɨ-a-mau burn-IPFV-NMLZ ‘"re’ (‘what is burnt’) (23) nakúɨamu ~ nakúɨɰamu (not *[nakúɰamu]) nakuɨ-a-mau extend.arm-IPFV-NMLZ ‘arm’s extension’ The epenthetic insertion of [ɰ] is completely optional when /ɨ/ is followed by 90. This contrasts with Awajun, where [ɰ] is an allophone of /ɨ/ (Overall 2007). 173 /a/, but [ɰ] is never inserted when /ɨ/ is followed by /u/:91 (24) /sɨuka/ → [sɨúk] ‘type of heron’ /ɨḱɨu/ → [ɨḱɨu] ‘young child’ The velar glide is the most infrequent of the glides and in many cases it is not inserted at all. The velar glide, apparently, is becoming an unproductive synchronic phone. Sometimes, a fricative [ɣ] surfaces instead of [ɰ] when followed by /a/: (25) /ɨakamata/ → [ɨakmat] ~ [ɨɰaḱmat] ~ [ɨɣaḱmat] ‘hunt’ /hɨã/ → [hiã] ~ [hɨɰ̃a] ~ [hɨɣ̃a] ‘house’ /pɨaka/ → [pɨaḱ] ~ [pɨɰaḱ] ~ [pɨɣaḱ] ‘bed’ At this point in the derivation of words, where glides are formed, the notions of phonotactics and syllable structure become relevant for some morphophonological processes, as well as for explaining the rhythmic pattern of Wampis, on which stress, vowel elision and metrical tone assignment depend. Thus, now the discussion turns to the phonotactics and the syllable in Wampis. 4.3. Phonotactics In this section, the syllable patterns found in Wampis are discussed. Syllables constitute the minimal structure where phonemes are organized into sequences, which 91. There are very few examples of this sequence of vowels in the data, so perhaps the apparent bias against the insertion of [ɰ], in this case, may be the result of very low frequency in discourse. 174 in turn form higher units to such as morphemes and words. Categories such as vocalic and consonantal segments can be distinguished and de"ned in terms of their distribution within syllables. Syllables are also important for determining constraints on clusters and which sequences are valid for building words, as well as for determining the rhythmic patterns of the language. For methodological reasons, I will suggest that a hypothetical syllable structure is needed at an intermediate level of derivation. This hypothetical syllable structure provides a template where syllables are mapped onto rhythmic feet. After syllables and feet are mapped, vowel elision applies. Vowel elision causes that, on the surface, words in Wampis present a complex structure that obscures a rather quite simple underlying structure. 4.3.1. Syllable structure in Wampis Underlyingly, words in Wampis exhibit a rather very simple pattern that can be formalized as: (C)V(N) (C=consonant, V=vowel, N=Nasal consonant) where the only indispensable element is the syllable nucleus, which is always a vowel. There are no syllabic segments (i.e. elements that can occur in the “V” slot) other than 175 vowels in Wampis. A strong lexical-level restriction of Wampis is that, underlyingly, consonants never occur at the end of the word—in that environment only vowels are allowed (see §3.5). At an intermediate stage of derivation, when glides are derived and feet are formed, before vowel elision, for analytical purposes, I propose that Wampis exhibits the syllable structure illustrated by Figure 4.1 (where σ=syllable, O=onset, R=rhyme).92 σ O R Nucleus Coda All consonants and glides, except /ʔ/ all vowels nasal consonants ! Figure 4.1. Syllable structure at an intermediate stage (before vowel elision) Note: ɰ, r and the marginal palatal ɲ do not occur word-initially. Representing the hypothetical syllable structure at an intermediate level of derivation is important for understanding the formation of rhythmic feet on which vowel elision depends in Wampis. Table 4.1 provides a provisional list of combinations 92. This syllable structure will be rede"ned later as the analysis progresses. 176 of vowels and consonants based on the structure provided in Figure 4.1 (V=vowel, C=consonant (other than nasal), G=Glide, N=nasal consonant). Some new syllable types will be added when re-syllabi"cation occurs after vowel elision (this is discussed later in this same section). Table 4.1. Provisional distribution of syllable types in Wampis Word-Initial Word-Medial Word-Final V, VN, CV, NV, GV, CVN, NVN, GVN V, VN, CV, NV, GV, CVN, NVN, GVN V, CV, NV, GV Word-initially and word-medially, the syllable structures at this intermediate stage of derivation are the same, with the caveat that [ɰ], /r/ and the marginal nasal consonant /ɲ/ do not occur at the beginning of words, only word-medially. For now, let us consider the examples of the syllable types (the relevant portions are bolded).93 Word-initial and word-medial (26) V /amɨ/ → a.mɨ → [aḿɨ] ‘you’ 93. Two things may be noticed by the reader when looking at the examples: "rst, on the surface forms, there are examples of consonants at the end of words; second, I consider sequences of two vowels (i.e. two moras) in di$erent syllables. This two facts are discussed below, in this same sub-section. 177 /iḱama/ → i.ka.ma → [iḱʲam] ‘forest’ /uruna/ → u.ru.na → [urún] ‘poor’ /hiãt͡ʃi/ → hɨ.̃a.t͡ʃi → [hi ̃át͡ʃ] ‘small house’ /sɨuka/ → sɨ.u.ka → [sɨúk] ‘type of heron’ (27) VN /ankanta/ → aŋ.kan.ta → [aŋkań] ‘free’ /ɨnt͡sa/ → ɨn.t͡sa → [ɨńt͡sa] ‘river’ /ampuʃa/ → am.pu.ʃa → [ampúʃ] ‘owl’ /t͡saankuna/ → t͡sa.aŋ.ku.na → [t͡saaŋkún] ‘tobacco (ACC)’ (28) CV /tukɨ/́ → tu.kɨ ́ → [tukɨ]́ ‘always’ /hakɨ/ → ha.kɨ → [haḱɨ] ‘warm water’ /paki/ → pa.ki → [paḱi] ‘peccary’ (29) NV /nuku/ → nu.ku → [núku] ‘mother’ /mukusa/ → mu.ku.sa → [mukúsa] ‘black’ /miʃu/ → mi.ʃu → [miʃ́u] ‘cat’ /nanamata/ → na.na.ma.ta→ [nanaḿat] ‘/ight’ (30) GV /uaua/ → wa.wa → [waẃa] ‘type of palm’ /iapi/ → ja.pi → [jaṕi] ‘face’ /iama/́ → ja.ma ́ → [jama]́ ‘just now, newly’ (31) CVN /kant͡sɨ/ → kan.t͡sɨ → [kańt͡sɨ] ‘type of plant’ /pampa/ → pam.pa → [paḿpa] ‘sandstone’ 178 /iurankɨ/ → ju.raŋ.kɨ → [juraŋ́] ‘fruit’ (32) NVN /nunka/ → nuŋ.ka → [núŋka] ‘land, earth’ /namankɨna/ → na.maŋ.kɨ.na → [namaŋkɨń] ‘meat (ACC)’ (33) GVN /iant͡se/ → jan.t͡sɨ → [jańt͡sɨ] ‘frog sp.’ /aiumpana/ → a.jum.pa.na → [ajumpań] ‘rooster (ACC)’ Word-)nal (34) V /nui/̃ → nu.i ̃ → [nui ̃]́ ‘there’ /hɨã/ → hɨ.̃a → [hɨ ̃á] ‘house’ /aa/ → a.a → [aá] ‘outside’ (35) CV /apa/ → a.pa → [aṕa] /panki/ → paŋ.ki → [paŋ́ki] ‘boa’ (36) NV /iama/́ → ja.ma ́ → [jama]́ ‘newly’ /mama/ → ma.ma → [maḿa] ‘manioc’ (37) GV /kaia/ → ka.́ja → [kaj́a] ‘stone’ /t͡ʃiua/ → t͡ʃi.wa → [t͡ʃiẃa] ‘turmpeter (bird)’ Some of the previous examples have presented some data that includes: a) 179 consonants other than nasal occurring in coda position and at the end of the word, b) sequence of vowels being analyzed in di$erent syllables. Now the discussion turns to explaining this two important points. The "rst phenomenon—consonants other than a nasal consonant in coda position, including word-"nal position—occurs after vowel elision leads to re- syllabi"cation. The syllable structure provided previously in Figure 4.1 helps us understand how syllables are mapped onto rhythmic feet in Wampis. After glide formation, feet are formed (glides are treated as consonants for word-internal elision, which depends on the rhythmic stress pattern of the language). A thorough analysis of the rhythmic pattern of Wampis is provided in Chapter VI, but let us state for the moment that Wampis forms iambic feet from left to right.94 Once feet are formed, the last vowel of the word is deleted if this vowel is in a CV syllable. Then, counting from left to right, the third vowel and every other alternating vowel in a CV syllable is also deleted. After the processes of vowel elision (word-"nal and word-internal vowel elision), re-syllabi"cation occurs. After re-syllabication, the form of the word is identical to its surface output form. The following examples illustrate re-syllabi"cation 94. It will be seen in Chapter VI that a metrical high tone is assigned to the most prosodically prominent foot (which is the "rst from the left) if the word does not have a lexical tone. 180 in Wampis—I include the stages of glide formation, feet formation, vowel elision and re- syllabi"cation for greater illustration (vowels to be deleted are underlined—see the line labeled “feet”): (38) Input: /uuɨha/ ‘hand’ Glide: u.wɨ.ha Feet: (u.ˈwɨ).ha Vowel elision: u.wɨ.h Re-syllabi"cation: u.wɨh Output: [uwɨh́] (39) input: /iurumakat͡ʃi/ ‘small cooked manioc’ glide: jurumakat͡ʃi feet: (ju.ˈru)(ma.ˌka).t͡ʃi vowel elision: ju.ru.m.ka.t͡ʃ re-syllabi"caiton: ju.rum.kat͡ʃ output [jurúmkat͡ʃ] In (38), the lexical representation of the word is /uuɨha/. After vowel elision, the word re-syllabi"es (there are two syllables in the surface form), so the consonant h is found at the end of the surface form of the word. In (39), a similar process of derivation occurs with /iurumakat͡ʃi/. After vowel elision, m re- syllabi"es and closes the second surface syllable, in turn, the a$ricate t͡ʃ does the same and closes the last surface syllable. It is crucial for the derivation of the 181 word ‘small cooked manioc’ to posit that glide formation must occur "rst, because only after the glide is formed we can count three vowels from the left (the third vowel is the /a/ of the sequence /ma/) when vowel elision occurs. In other words, only after glide formation feet are parsed and vowels are deleted, and the surface forms of words can be derived correctly, as shown in (39). It is necessary now to rede"ned the syllable structure given previously in Figure 4.1. On the surface, the maximal syllable in Wampis is now de"ned as: (C) (R) V (C) (T) (where C=consonant, R=rhotic, V=vowel, T=a$ricate). Figure 4.2 illustrates the structure of the maximal syllable in Wampis. σ Onset Rhyme C R Nucleus Coda C T All consonants, glides r all vowels all consonants affricates (except /ʔ/) (except glides, ɲ) Figure 4.2. Maximal syllable structure in Wampis 182 Note: ɰ, r and the marginal palatal ɲ do not occur word-initially. Only stops occur with r in complex codas. Finally, we need to add some new syllable types to account for surface forms of words. Table 4.2 lists all syllable types found on Wampis words. Table 4.2. Distribution of syllables types in Wampis Word-Initial Word-Medial Word-Final V, VN, CV, NV, GV, CVN, NVN, GVN V, VN, CV, NV, GV, CrV, CrVC, CrVN, CVN, CVC, NVN, GVN, V, VN, CV, NV, GV, CrV, CrVN, CrVC, CVN, CVC, NVN, CVNT, GVNT Word-initially, the same syllable types as given in Table 4.1 are maintained. Word-medially and word-"nally there are new types added because of vowel elision. The new syllable types are: word-medially: CrV, CrVC, CrVN, CVC; and word )nally: VN, CrV, CrVC, CrVN, CVN, CVC, NVN, CVNT, GVNT. The only complex onset allowed is Cr. Importantly, in this case only stops occur in the position of “C”. The only complex coda allowed is NT and occurs 183 only word-"nally. In theory the syllable type CrNT is possible, but is not attested in the data. Here are examples of the newly-added syllable types, the relevant portions are in boldface (for examples of the rest of syllable types, see (26)–(37) above): Word-medial (40) CVC /kakaramana/ → ka.ka.ra.ma.na → ka.kaŕ.man ‘powerful man (ACC)’ /iankipikit͡ʃi/ → jaŋ.ki.pi.ki.t ͡ʃi → jan.kiṕ.kit ͡ʃ ‘collared peccari (DIM)’ (41) CrV /nauanturumina/ → na.wan.tu.ru.mi.na→ na.wań.tru.min ‘your daughter (ACC)’ /huakarika/ → hu.a.ka.ri.kʲa → hu.a.́kri.kʲa ‘if we stay’ (42) CrVC /aiantaratasa/ → a.jan.ta.ra.ta.sa → a.jań.trat.sa ‘intending to turn around’ /atukaratukati/ → a.tu.ka.ra.tu.ka.ti → a.tú.krat.ka.ti ‘May he help us!’ (43) CrVN /amikurunaʃa/ → a.mi.ku.ru.na.ʃa → a.mi.krun.ʃa ‘my friend too’ /kautramkaji/ → ko.u.tu.ra.mɨ.ka.ji → ko.ú.tram.ka.ji ‘They came for you.’ Word-)nal (44) VN /t͡saanku/ → t͡sa.aŋ.ku → t͡sa.aŋ́ ‘tobacco’95 /paanta/ → pa.aŋ.ta → pa.ań ‘clear’ 95. The "nal syllables ku of ‘tobacco’ and ta of ‘clear’ are deleted because after vowel elision, /k/ and /t/ cannot be in the margin of a complex coda (only a$ricates are allowed in that positions, 184 (45) VNT /iuant͡sɨ/ → ju.an.t͡sɨ → ju.ańt͡s ‘reed’ (46) CrV /t͡sɨnt͡sakari/̃ → t͡sɨn.t͡sa.ka.ri ̃ → t͡sɨn.t͡sa.́kri ̃‘his dart’ /piripiri/ → pi.ri.pi.ri → pi.ri.́pri ‘plants used for healing and bewitching’ (47) CrVC /t͡ʃaaparaka/ → t͡sa.a.pa.ra.ka → t͡sa.a.́prak ‘type of traverse /ute’ (48) CrVN /nakanturana/→ na.kan.tu.ra.na → na.kań.trun (ACC) ‘ant sp.’ /sɨɨturuna/ → sɨ.ɨ.tu.ru.na → sɨ.ɨ.́trun (ACC) ‘cedar’ (from Spanish ) (49) CVN /ankanta/ → aŋ.kan.ta → aŋkań ‘free’96 /nauantu/ → na.wan.tu → nawań ‘daughter’ (50) CVC /kantuta/ → kan.tu.ta → kan.tút ‘type of bird’ /tɨmaʃi/ → tɨ.ma.ʃi → tɨ.maʃ́ ‘lice’ as de"ned in Figure 4.2). 96. The syllables ta of ‘free’ and tu of ‘daughter’ are deleted because after vowel elision, /t/ cannot be in the margin of a complex coda (only a$ricates are allowed in that positions, as de"ned in Figure 4.2). 185 (51) NVN /kanampa/ → ka.nam.pa → ka.naḿ ‘stone axe’ /kuhant͡ʃama/ → ku.han.t͡ʃam.a → ku.hań.t͡ʃam ‘opossum’ (52) CVNT /makant͡ʃi/ → ma.kan.t͡ʃi → ma.kańt͡ʃ ‘reed’ /marunt͡ʃi/ → ma.run.t͡ʃi → ma.rúnt͡ʃ ‘prawn’ (53) GVNT /naiant͡sa/ → na.jan.t͡sa → na.jant͡s ‘ocean’ /iuant͡ʃi/ → i.wan.t͡ʃi → i.wańt͡ʃ ‘devil’ The second important point to consider is that sequences of vowels are analyzed as forming di$erent syllables. There are several pieces of evidence that suggest that sequences of vowels are better analyzed as belonging to di$erent syllables (actually, in Wampis, the only relevant unit for metrical tone assignment, vowel elision, reduplication and other processes is the vowel): • Metrical tone assignment: metrical high tone is assigned to the most prominent foot (which is the "rst from the left in an iambic pattern)—see §6.4 for a detailed analysis. The mapping of syllables with the iambic parsing is crucial. In the next example, the prosodically prominent portion of the foot is in bolds and the vowel to be deleted is underlined. Thus: 186 (54) /naiant͡sa/ → (na.ˈjan).t͡sa → [najańt͡s] ‘ocean’ Now consider the following parsing for the word ‘small’, which derives the correct output in the language, following the iambic pattern: (55) /iairat͡ʃi/ → (ja.ˈi).(ra.t͡ʃi) → [jɛiŕat͡ʃ] ‘small’ If we consider a di$erent parsing, for instance, if we consider that the portion jai is one syllable and we parse (jai.ˈra).t͡ʃi, then we would have to explain why the /i/ receives the high tone instead of /a/ (as the correct output in (55) shows): according to the (incorrect) analysis (jai.ˈra).t͡ʃi, the a of ra should receive the metrical high tone. We would have to proposed ad-hoc rules to explain this and other possible deviations from the rule of metrical high tone (§6.4.2), with the additional problem that then the rule of metrical tone assignment in Wampis would become irrelevant to explain otherwise pretty regular patterns. As I say, this is a most crucial point in the analysis of Wampis syllable structure, because all relevant processes related to this structure: parsing of iambic feet, vowel elision, metrical high tone assignment are otherwise very di#cult to explain in terms of regularities. In turn, these regularities explain the form of surface words and the richly complex prosody of the language. • Partial reduplication: Partial reduplication in Wampis copies all the phonetic material until the second mora after glide formation. There is a su#x -kaua ‘Reduplicative’ that 187 triggers partial reduplication when it is received by the verb and gives a semantics of repetitive action (this su#x is analyzed in §19.9.4). Consider the following examples (the material to be reduplicated is in boldface): (56) juá juákua ̃ iua iu-a-kaua ̃ REDUP eat-IPFV-REDUP\3SG.SS ‘[she/he] eating and eating’ cf. /iuakaua/ → ju.a.ku.a ̃ (57) puhú puhúiɲakua puhu puhu-ina-kaua REDUP living-PL.IPFV-REDUP/3.SS ‘[they] living and living’ cf. /puhuinakaua/̃ → pu.hu.i.ɲa.ku.a In (56), the material to be reduplicated (until the second mora) is [júa]. In (57), there is a cluster with the vowels /ui/ of /puhu/ ‘live’ and the plural imperfetive /ina/. To obtain the right output in reduplication (which is the portion [puhú]), the sequence /ui/ must be parsed u.i, because the vowel [i] is not reduplicated. This is a strong evidence that /ui/ is not treated as forming a syllabic unit. The fact that said vowels belong to di$erent morphemes underlyingly is irrelevant to partial reduplication, because in (56) vowels from di$erent underlying morphemes are copied without problem (/iu/ ‘eat’ and the imperfective /a/, [júa]). 188 • Reduction of vowel sequences: Some vowel sequences can be reduced if the vowel is in a prosodic position where it undergoes elision.97 In the next example, the second /a/ is deleted. Parsing the word as is shown in (58) derives the correct output; once again, that is only possible if it is hypothesized that vowels form di$erent syllables: (58) /iaiñaita/ → (ja.i)̃(na.i)ta → [jɛi ̃ɲ́it] ‘to help each other’ • Optional insertion of epenthetic glides: An additional evidence comes from the fact that in Wampis epenthetic glides can optionally be inserted between V1V, where V1 is a high vowel, as explained in §4.2. This suggests that there is a bias in the language to parse those sequences of vowels into di$erent syllables. 4.3.2. Phonotactic restrictions As a summary of the previous discussion, the following points characterize the structure of syllables and words in Wampis. • Because of the allophonic derivation of glides or of epenthetic glide insertion, glides occur in syllable onset position. • As a result of vowel elision, re-syllabi"cation occurs. When a vocalic nucleus is 97. Vowel elision is explained in Chapter VI. 189 deleted, consonants that used to be in onset position in an open syllable (CV) become the coda of a previous di$erent syllable. Consequently, all consonants are permitted in simple coda position in the phonetic derivation (with the exceptions of glides and the palatal ɲ). • Complex onsets and codas can occur on the surface form, but are restricted to speci"c combinations of phonemes. The only complex onset is the combination Stop+r. And the only complex coda is the combination Nasal+A$ricate. As a consequence, all other consonant clusters in Wampis are heterosyllabic. The following restrictions happen at the level of the word: • The rhotic r, and the marginal consonants glottal ʔ and palatal ɲ do not occur word- initially. • Word-initially, the sequence si is unattested—the sequence si is disfavored in other positions as well (very few examples of it in the language). See §3.5 for details. • The only combinations of vowels that is prohibited is iɨ or ɨi. See §3.6.4 for details. • No consonant cluster is allowed at the beginning of the word (the clusters Stop+r and Nasal+A4ricate only occur word-internally and word "nally, respectively). Once the combinatorial patterns and phonotactic restrictions of Wampis words 190 have been de"ned, the conditions are set for the analysis of morphophonological processes and the prosody of the Wampis language. The next two chapters discuss these topics at length. 191 CHAPTER V MORPHOPHONOLOGY 5.1. Introduction This chapter describes general patterns of morphophonemic change that operate at the morpheme boundary. The morpheme boundary is the chief morphological environment that triggers phonological processes in Wampis. Most allophonic rules, discussed in Chapter III, also apply when morphemes interact with one another. For issues concerning shift of placement of metrical high tone induced by the presence of a bound morpheme, see Chapter VI. In other words, this chapter is dedicated to the analysis of phenomena that occur at the morpheme boundary that do not have to do with allophony or induced metrical hight tone shift. The structure of the chapter is as follows: §5.2 provides a brief introduction to general processes observed in morpheme boundary; §5.3 and §5.4 look at sandhi processes where mora reduction occurs; §5.5 describes morphemes that trigger vowel harmony. Next, §5.6 provides an analysis of several morphemes that exhibit di$erent forms according to the morphophonological environment where they occur. Finally, other phenomena associated with morphophonology are analyzed in §5.7 and §5.8. 192 5.2. Morpheme boundary Morpheme boundary phenomena are really complicated in Wampis; a good number of speci"c conditions or rules must be hypothesized to understand them. In most cases, morphophonological phenomena in Wampis must be dealt with at the level of individual morphemes; i.e. they concern the behavior of particular morphemes (or processes triggered by speci"c morphemes) rather than to classes of morphemes. However, certain generalizations can be established so as to describe morpheme juncture phenomena in a more orderly fashion. Table 5.1 shows these general patterns. Table 5.1. Morphophonological processes in Wampis Process Cross-reference Vowel (mora) reduction: • reduction of the last vowel of the stem §5.3 • reduction of the "rst vowel of the su#x/clitic. §5.4 Vowel harmony §5.5 Other phenomena: • degemination §5.7.1 • optional gemination of consonants §5.7.2 • optional metathesis §5.7.3 • Vowel-switching for marking possession §5.8 193 5.3. Reduction of the last vowel of a stem The last vowel of a stem or root is deleted when it is followed by a "rst vowel of certain morphemes if certain conditions are met. Such conditions can be generalized as being of two types: a) an immediately preceding vowel of a stem of a speci"c quality is deleted when the stem receives the su#x; b) an immediately preceding vowel of the stem, regardless of its quality, is deleted when the stem receives the su#x. Generally, the types of processes described above can be formalized with the following rule: Vi + Vj → Vj where “+” indicates certain morpheme boundaries. Table 5.2 summarizes the morphemes that trigger the reduction of the last vowel of the stem. The examples in the next sections will sometimes represent derivation of words in several steps, as necessary. Vowels that undergo a morphophonological process will be in boldface type (I will label the line where morphophonological process occur as “sandhi”). Vowels that undergo elision based on the rhythmic pattern of the 194 language will be underlined. Table 5.2. Morphemes that trigger reduction of last vowel (mora) of the stem Morpheme (§functional discussion) Gloss -inu (§5.3.1) ‘Set I nominalizer’ -u (§5.3.2) ‘Set II nominalizer’ -a(́u) (§5.3.3) ‘High a$ectedness aktionsart’ -i (§5.3.4) ‘Low a$ectedness aktionsart’ -a (§5.3.5) ‘Imperfective’ -i ̃~ -ɨ ̃(§5.3.6) ‘Third person perfective’ 5.3.1. Set I nominalizer -inu A "nal high vowel /u/, /i/ or /ɨ/ of a stem is always lost to the "rst vowel /i/ of the Set I98 nominalizer -inu ‘Agentive nominalizer’. A "nal low vowel /a/ of a stem is lost to the "rst vowel of -inu if, after glide formation, the stem contains more than two moras. If the stem has less than two moras, the last vowel /a/ of the stem is not lost. In the following examples, I provide "rst a morphological analysis of the word to illustrate the two (or more) morphemes that form the word. Recall that the relevant environments where the morphophonological process occurs are written in bolds. 98. The are two sets of nominalizers in Wampis, -inu belongs to Set I. See Chapter XV for an analysis of nominalization. 195 (1) Morphological analysis: ɨt͡sɨru-inu announce-NMLZ ‘announcer’ Derivation: /ɨt͡sɨruinu/ glide: none syllable: ɨ.t͡sɨ.ru.i.nu.na sandhi: ɨ.t͡sɨ.ri.nu feet: (ɨ.ˈt͡sɨ).(riˌ.nu) vowel elision: ɨt͡sɨrin output: [ɨt͡sɨŕin] (2) Morphological analysis: hintina-karatu-inu teach-1PL.OBJ-NMLZ ‘teacher’ (i.e. ‘someone who teaches us’) Derivation: /hintinakaratuinu/ glide: none syllable: hin.ti.na.ka.ra.tu.i.nu sandhi: hin.ti.na.ka.ra.ti.nu feet: (hin.ˈti).(na.ˌka).(raˌti)nu vowel elision: hintinkartin output: [hintiŋ́kʲartin] With regard to a preceding low vowel /a/ of a stem, when the nominalizer -inu is used: a) the low vowel /a/ of a stem is reduced if the stem possesses more than two 196 moras as in (3), and b) the low vowel /a/ of the stem is not reduced it the stem possesses less than two moras as in (4). (3) morphological analysis: uɨka-inu walk-NMLZ ‘walker’ derivation: /uɨkainu/ glide: wɨkainu (two-mora stem: wɨka ‘walk’) syllable: wɨ.ka.i.nu sandhi: does not apply feet: (wɨ.ˈka).(i.ˌnu) vowel elision: wɨkain output: [wɨkɛín] (4) morphological analysis iuruma-inu eat.cooked.manioc-NMLZ ‘(big) eater’ derivation /iuramainu/ glide: jurumainu (three-mora stem: juruma ‘eat cooked manioc’) syllable: ju.ru.ma.i.nu sandhi: jurumin feet: (ju.ˈru).(mi.ˌnu) vowel elision: jurúmin output: [jurúmin] 197 5.3.2. Set II nominalizer -u The last vowel /u/ of a stem is reduced when preceding the set II99 “subject nominalizer” -u. (5) morphological analysis hintina-tu-u teach.PFV-APPL-NMLZ ‘one who has taught.’ derivation /hintinatuu/ glide: does not apply syllable: hin.ti.na.tu.u sandhi: hin.ti.na.tu feet: (hin.ˈti).(na.ˌtu) vowel elision: hintintu output: [hintińtʲu] That the last vowel of the root is the one that undergoes reduction is shown by the fact that -u is never lost when the last vowel of the stem is di$erent than /u/: with a preceding /i/ or /ɨ/100 all vowels (i.e. moras) are maintained: 99. Again, there are two sets of nominalizers in Wampis, -u belongs in Set II. See Chapter XV for details on nominalization. 100. There is an apparent exception: when the verb uɨ ‘go’ receives the nominalizer, often times speakers seem to pronounce [wúu] (other times the pronunciation is an expected [wɨú]). Two facts may contribute to the apparent assimilation of /ɨ/ to /u/. First, the phonetic environment in which /ɨ/ is found (between [w] and [u]) seems to favor assimilation. Second (not necessarily opposed to the previous point), the nominalized form of ‘go’ is very common (in terms of frequency), so its high frequency in speech may have led to assimilation. 198 (6) iṕatiu iṕati-u shoot+LOAF-NMLZ ‘he was one who shot’101 (7) pɨpɨŕɨu pɨpɨrɨ-u turn.around-NMLZ ‘he turning around’ (Lit. ‘an around-turner’) From the above examples, it is preferable to state that the nominalizer -u is not lost in morpheme boundary, rather the last vowel /u/ of the stem is lost. Otherwise, it would be problematic to explain why the /u/ of the nominalizer -u is reduced sometimes while some other times it is not reduced. If the vowel preceding the nominalizer -u is /a/, then two possible outcomes may occur depending on whether the vowel /a/ is in an position to be deleted or not: 1) the vowel /a/ of the stem is not reduced if it is in a position where it is not marked for deletion, or 2) the vowel /a/ of the stem is reduced if the vowel is in an position where it must be deleted. 101. The past tense interpretation derives from the use of the aktionsart su#x -i ‘Low a$ectedness’. Aktionsart su#xes are used in perfective contexts (see §13.3). 199 For instance, for the word in (8): (8) arakama-u plant-NMLZ ‘planter’ the following relevant foot parsing is established, and the ouput as indicated in (9). (9) derivation for /arakamau/ ‘planter’ feet: (a.ˈra).(ka.ˌma).u vowel elision: arakmau output: [araḱmoo]102 It can be seen that the "nal vowel /a/ of the stem to which the nominalizer -u attaches is not deleted. The vowel that is in the position to be deleted is the previous vowel /a/ (the third counting from left to right). Now, compare with the next example, including the relevant part of the word’s derivation: (10) tɨpɨśu tɨpɨ-sa-u lie.down-ATT-NMLZ ‘one who sat’ derivation for /tɨpɨsau/ ‘one who sat’: feet: (tɨ.ˈpɨ).(saˌ.u) vowel elision: tɨpɨsu output: [tɨpɨśu] In (10), the vowel /a/ of the attenuative su#x -sa is in a position where it is 102. Recall from §3.7.1 that the sequence /au/ is pronounce [ou] or [oo]. 200 deleted according to the metrical rhythmic pattern of Wampis, which deletes the third mora (if in a CVC cluster) counting form left to right. Thus, said vowel is deleted and the result is [tɨpɨśu]. In Chapter VI, It will be seen that only a few morphemes (among them, the Set II nominalizer -u) allow reduction of a mora in a CVV cluster (the usual environment for vowel elision is CVC). Thus, the fact a mora is reduced in the presence of the nominalizer -u is seen as a property of -u. 5.3.3. High a&ectedness -a(́u) A "nal vowel /a/ of the verb stem is reduced when the stem receives the high a$ectedness -a(́u). The following example illustrates this behavior: in (11) the verb has a perfective sense of ‘just done action’ and uses the 3 person past su#x -ji. This perfective use is only done with the perfective stem of the verb, which is formed with aktionsart su#xes (see §13.3.2). By contrast, in (12) there is an imperfective stem: this is evident by the used of the 3 person non-past su#x -ua which occurs with the imperfective (and in future contexts) (see §14.3.2). (11) hiãj́i hɨã-́ji arrive+HIAF-3.PT+DECL ‘He just arrived’ 201 (12) hɨaẃɛi hɨã-ua-i arrive+IPFV-3SBJ-DECL ‘He is arriving’ When the "nal vowel of the stem is reduced preceding the high a$ectedness su#x -a(́u), no further vowel elision takes place after the rhythmic feet are stablished. (13) ahuntóu ahunta-́u add.water+HIAF-NMLZ ‘She was one to add water.’ derivation: syllable: a.hun.ta.a.́u sandhi: a.hun.ta.u feet: (aˈhun).(ta.́u) vowel elision: (blocked by previous mora reduction at sandhi) output: [ahuntóu] In (13), it would be expected that the third vowel of ahunta ́‘add.water+HIAF’ is deleted by the rule of word-internal vowel elision once feet are parsed, as it is the third vowel and is in a position to be deleted. What is more, recall from §5.3.2 that the nominalizer -u allows for the elision of the preceding mora. However, in this case, vowel elision is blocked, as there has been a previous reduction of a mora at the “sandhi step” (morpheme boundary). This is a particular behavior that occurs with the high 202 a$ectedness -a(́u),103 because other su#xes allow vowel elision to occur. Compare (13) with the derivation of the word in (14) with the nominalizer -inu (cf. §5.3.1). In (14), the vowel /u/ of the stem that precedes -inu is deleted. Then, the vowel /i/ of -inu is in a position to be deleted, so it undergoes elision. This does not occur in (13) with the high a$ectedness aktionsart. (14) t͡suaḱratɲun t͡sua-karatu-inu=na heal-1PL.OBJ-NMLZ=ACC ‘the doctor (acc)’ (Lit.: ‘one who heals us’) derivation: /t͡suakratuinuna/ syllable: t͡su.a.ka.ra.tu.i.ɲu.na104 glide: t͡su.a.ka.ra.ti.ɲu.na feet: (t͡su.ˈa).(ka.ˌra).(ti.ˌɲu).na vowel elision: t͡suakratɲun output: [t͡suaḱratɲun] Some verbs in the database seem not to lose their "nal vowel when they carry the high a$ectedness su#x. Table 5.3 lists these verbs. 103. And also with the non-imperfective plural -ara and the copula clitcs, as explained in §5.4.2 and §5.6.5, respectively. 104. Recall from §3.3.2 that /n/ becomes [ɲ] in an i__V environment. 203 Table 5.3. Verbs that never reduce their "nal mora when receiving the high a$ectedness aktionsart su#x -a(́u) Verb Gloss akima ‘unfasten’ ɨnkɨma ‘enter’ ma ̃ ‘kill’ paka~pɨka ‘put in a line’ ta ‘arrive’ tankuma ‘domesticate animals’ The long form -aú of the high a$ectedness su#x occurs before the apprehensive -(a)i and the non-imperfective plural marker -ara. Examples: (15) mantuaẃɛipʲa ma-̃tu-aú-ai-pa kill-1SG.OBJ-HIAF-PL-APPR-PROH ‘Don’t kill me!’ (16) mantuaẃaru ma-̃tu-aú-ara-u kill-APPL-HIAF-PL-NMLZ ‘killers [of my brother]’105 An alternative analysis is to propose that a glide [w] is inserted at the 105. In the text where the example comes from, the expression refers to the killers of the brother of the protagonist of the story. In particular, the applicative refers to the fact that the action is being done to the detriment of the protagonist. 204 morpheme boundary between the high a$ectedness aktionsart and the apprehensive -ai or the plural -ara. This glide would be obligatory, as it always occurs between these morphemes (possibly to distinguish vowels that are alike in quality). The high a$ectedness aktionsart su#x -a ́and the imperfective -a (§5.3.5) are homophonous, but the high affectedness aktionsart always carries a high tone (i.e. it has a lexical high tone), whereas the imperfective -a does not. 5.3.4. Low a&ectedness -i A "nal vowel of a verb stem is deleted when preceding the low a$ectedness su#x -i if the stem has more then two moras at the point of glide derivation. If, at the point where glides are derived, the stem has two or less moras, the last vowel of the stem is not deleted. As an illustration, compare (17)–(18). (17) wɨkɛímiaji uɨka-i-mɨ walk-LOAF-2SG.SBJ+DECL ‘You just walked.’ derivation: /uɨkaimɨ/ glide: wɨkaitasa ̃ syllable: wɨ.ka.i.mɨ sandhi: does not apply (stem wɨka has two moras) feet: (wɨ.ˈka).(i.mɨ) vowel elision: does not apply output: [wɨkɛímɨ] 205 (18) ʃiiḱmimɨ ʃiikima-i-mɨ blow.nose-LOAF-3.SBJ+DECL ‘He just blew his nose.’ derivation /ʃiikimaimɨ/ glide: does not apply syllable: ʃi.i.ki.ma.i.mɨ sandhi: ʃi.i.ki.mi.mɨ feet: (ʃi.ˈi.(ki.ˌmi).mɨ vowel elision: ʃiikmimɨ output: [ʃiiḱmimɨ] There are two exceptions to the statement that words with less than two moras do not lose their "nal vowel preceding the low a$ectedness su#x -i. The verbs tu 'say' and maa 'bathe' lost their last vowel when receiving the low a$ectedness. In (19) there is an example with ‘say’—I “skip” the step of foot formation, as it is not really relevant for illustrating what happens to tu+i (no vowel elision based on rhythmic pattern applies in this case): (19) titʲah́ɛɛ ti-ta-ha-i say+LOAF-IMM.FUT-1SG.SBJ-DECL ‘I will say.’ derivation: /tuitahai/ glide: does not apply syllable: tu.i.ta.ha.i sandhi: ti.ta.ha.i output: [titʲah́ɛɛ] 206 Next, (20) presents an example with ‘bathe’. The root maa is represented in a “fused” form with the low a$ectedness aktionsart su#x -i. (20) mɛit́hʲɛɛ mai-ta-ha-i bathe+LOAF-IMM.FUT-1SG.SBJ-DECL ‘I will bathe.’ derivation: /maaitahai/ glide: does not apply syllable: ma.a.i.ta.ha.i sandhi: ma.i.ta.ha.i feet: (ma.ˈi).(taˌ.ha).i vowel elision: maithai output: [mɛit́hʲɛɛ] There is at least one root with fewer than two moras in the database that does not merge with -i due to glide creation: au ‘parboil’. (21) awit́hʲɛɛ au-i-ta-ha-i parboil-LOAF-IMM.FUT-2SG.SBJ-DECL ‘I will parboil [the manioc]’ The low a$ectedness never occurs with a verb ending in /ɨ/. Recall that there is a restriction in Wampis that prohibits vowel clusters of the form /ɨi/ or /iɨ/ (§3.6.4). In general, it is very di#cult to decide whether the low a$ectedness -i is lost to a previous vowel /ɨ/ of a stem, because it is not known if the low a$ectedness is really present in 207 the "rst place (in addition, there are also few verbal roots that end in a vowel /ɨ/ and that can carry the low a$ectedness su#x). Some examples from elicitation suggests that with a verb ending in /ɨ/, the low a$ectedness -i is not reduced but is assimilated: (22) mɨhɨɨ́tmɨ mɨhɨ-ɨ-ta-mɨ stink-LOAF-IMM.FUT-2SG.SBJ+DECL ‘You will stink.’ 5.3.5. Imperfective -a The imperfective -a is never deleted. In morphological complex verb forms, a "nal vowel /u/ of the stem is reduced when the verb receives the imperfective -a. For instance, in (23) the verb stem is complex, consisting of the root hintina ‘teach’ and the applicative -tu. When the verb receives the imperfective -a, the /u/ of the applicative is reduced. (23) hintińtawai hintina-tu-a-ua-i teach-APPL-IPFV-3.SBJ-DECL ‘He teaches [him/them].’ derivation for /hintinatuawai/ ‘He teaches [him/them]’ glide: hintinatuawai syllable: hin.ti.na.tu.a.wa.i sandhi: hin.ti.na.ta.wa.i.ti feet: (hin.ˈti).(naˌ.ta).(wa.ˌi).ti vowel elision: hintintawaiti output: [hintińtʲawɛiti] 208 In addition, a "nal vowel of a morphologically simple verb form belonging to the Imperfective conjugation II106 is always reduced when the verb receives the imperfective -a. (A discussion of verb conjugation patterns is found in §12.8.) For instance, puhu ‘live’ belongs to the Imperfective conjugation II, thus its last vowel /u/ is reduced and occurs as puha in its imperfective form. Following a Jivaroan tradition initiated by Overall (2007), for practical purposes I gloss this (and other similar examples) using a “+” symbol to signal that there are identi"able morphemes occurring in a “fused” form. (24) puhaẃɛi puha-ua-i live+IPFV-3.SBJ-DECL ‘She/he is living.’ This only occurs with verbs that belong to the Imperfective conjugation II. For instance, the verb amu ‘"nish’ belongs to a di$erent imperfective conjugation, so it does not loose its "nal vowel to the imperfective -a. (25) amúawɛi amu-a-ua-i "nish-IPFV-3.SBJ-DECL ‘She/he is "nishing.’ 106. The Imperfective conjugation II is characterized by losing the last mora of the root or stem when it carries the imperfective -a. See §12.8.1.1 for details. 209 Because the patterns of conjugations are not always predictable, they are treated in a separate part of the grammar, in §12.8. 5.3.6. Third person perfective -i~̃-ɨ ̃ A "nal vowel /i/ of the stem is reduced when it receives the third person perfective su#x -i~̃-ɨ.̃107 (26) pati ̃ ́ pati-i ̃ sound.shot+LOAF-3.PFV ‘It just sounded (i.e. a sound of a gun shooting).’ The third person is not reduced with a preceding vowel of a quality di$erent than /i/. (27) at͡ʃiḱʲarɛi ̃ at͡ʃi-ka-ara-i ̃ grab-INTENS-PL-3.PFV ‘They just grabbed him.’ The third person perfective provokes the occurrence of the long form of the associated motion su#x -ki(ni) (see §5.6.3). 5.4. Vowel (mora) reduction of the su0x\clitic at morpheme boundary Certain su#xes reduce their "rst vowels when they are in contact with a 107. In theory, the third person perfective -i~̃-ɨ ̃should reduce a vowel /ɨ/ of the stem, but there are no examples of -i~̃-ɨ ̃occurring preceding a vowel /ɨ/ as the morphological conditions for that to happen are minimal (no verbal su#x preceding the potential occurrence of the third person perfective ends in an /ɨ/). 210 preceding vowel of the stem. Similarly to what has been described in the previous sections, the vowel of the stem can be required to be of a speci"c quality for the vowel of the su#x or clitic to be deleted; or in some other cases the "rst vowel of the su#x or clitic will be deleted regardless of the quality of the preceding vowel of the stem. The general rule that formalizes this type of vowel reduction of the su#x or clitic at morpheme boundary is: Vi + Vj → Vi where “+” indicates certain morpheme boundaries, as explained in the following sections. The following morphemes presented in Table 5.4 lost their "rst vowel when in contact with a preceding vowel of a stem. They are explained in the following sub- sections. Table 5.4. Morphemes that reduce their "rst vowel (mora) in morpheme boundary Morpheme (functional description) Gloss -ina (§5.4.1) ‘Plural imperfective’ -ara (§5.4.2) ‘Non-imperfective plural’ -i (§5.4.3) ‘Declarative’ -ai (§5.4.4) ‘Apprehensive’ =api (§5.4.5) ‘Sudden realization’, ‘Tag question’ 211 5.4.1. Plural imperfective -ina The initial /i/ of the plural Imperfective -ina is reduced with an immediately preceding "nal /i/ or /ɨ/ of the stem or root. With a "nal stem vowel /a/ or /u/, nothing happens and the /i/ of -ina is not reduced (30). Examples (28) and (29) illustrate the behavior of -ina with a preceding vowel /ɨ/ and /i/, respectively. (28) wɨńawɛɛ uɨ-ina-ua-i go-PL.IPFV-3.SBJ-DECL ‘They are going’ (29) umiɲ́awɛɛ umi-ina-ua-i complete-PL.IPFV-3.SBJ-DECL ‘The are "nishing.’ (30) tuiɲ́awɛɛ tu-ina-ua-i say-PL.IPFV-3.SBJ-DECL ‘They are saying.’ 5.4.2. Non-imperfective plural -ara The initial vowel /a/ of the non-imperfective plural su#x -ara is reduced with a preceding vowel of the same quality. 212 (31) wɛiŋ́kʲaru uaina-ka-ara-u see-INTENS-PL-NMLZ ‘There [were] ones to see.’ derivation: /uainakaarau/ glide: wa.i.na.ka.a.ra.u syllable: wa.i.na.ka.a.ra.u sandhi: wa.i.na.ka.ra.u feet: (wa.ˈi).(naˌ.ka).(ra.ˌu) vowel elision: wainkaru output: [wɛiŋ́kʲaru] Similarly to what happens with the high a$ectedness -a(́u), the reduction of a mora blocks further vowel elision of the remaining vowel when feet are parsed. In the case of -ara, once the initial vowel of /a/ is reduced, the remaining /a/ of the stem is not deleted. In the next example, the vowel /a/ of -ara is lost to a previous vowel /a/ of the stem. The remaining vowel /a/ would be in a position (third mora from the left) to be deleted, but it does not undergo elision. Here another phenomenon associated to the fascinating prosody of Wampis must be introduced: when a vowel in a position to be deleted is not deleted, the marking of vowels for elision is pushed one mora to the right (and then every other alternating vowel is deleted). 213 (32) umaŕartatuak uma-ra-tata-ua=ka drink.PFV-DISTR-DEF.FUT-3.SBJ=Q ‘are they going to drink?’ derivation: /umaraaratatauaka/ glide: u.ma.ra.a.ra.ta.ta.wa.ka syllable: u.ma.ra.a.ra.ta.ta.wa.ka sandhi: u.ma.ra.ra.ta.ta.wa.ka feet: (u.ˈma).(raˌ.ra).(ta.ta).(wa.ka) re-check vowel elision: (u.ma).(ra.ra).(ta.ta).wa.ka (vowel to be deleted pushed one mora to the right)108 vowel elision: umarartatuak output: [umaŕartatuak] 5.4.3. Declarative -i The declarative su#x is reduced with a preceding stem-"nal /i/ or /ɨ/ vowel. Again, I use the practice of glossing a “+” symbol, which means that there are more than one discernible morpheme “fused” together after a morphophonological process. (33) tah́i ta-hi say+IPFV-1PL.SBJ+DECL ‘We are saying’ (34) taḿɨ ta-mɨ say+IPFV-2SG.SBJ+DECL ‘You are saying’ 108. Notice that the vowel a before the glide [w] is deleted, so the [w] reverts to [u] in the derivation. 214 The declarative su#x -i surfaces with stems that end in vowel di$erent than /i/ or /ɨ/, as shown in (35). (35) tah́ɛi ta-ha-i say+decl-1sg.sbj-decl ‘I am saying.’ 5.4.4. Apprehensive -ai The apprehensive -ai losses its "rst vowel /a/ next to a preceding vowel /a/ of the stem. (36) naḱi ɛípʲa naki a-ai-pa lazy COP-APPR-PROH ‘Don’t be lazy.’ (37) at͡ʃiḱɛipʲa at͡ʃi-ka-ai-pa grab-INTENS-APPR-PROH ‘Don’t grab it.’ 5.4.5. Sudden realization and tag question =api The vowel /a/ =api, which is used to mark sudden realization and in tag questions (cf. §18.9), is reduced when it follows a vowel /a/ of the stem. (38) jatsúrnapi iat͡su-ru=na=api brother-1SG=ACC=SUD.REALZ ‘[They are killing] my brother!’ 215 5.5. Vowel harmony The restrictive =kɨ (cf. §18.3) and the su#x -ki(ni) ‘do action while moving’ (cf. §13.3.2.6) undergo processes of vowel harmony. Vowel harmony involves the vowels /i/ and /ɨ/. A particular pattern in Wampis (as will be seen from the discussion in §5.5.1 and §5.5.2) is that, regardless of order, the high front vowel /i/ dominates over the high central vowel /ɨ/ for purposes of vowel harmony. It is interesting that vowel harmony between /i/ and /ɨ/ somehow “re/ects” the phonotactic restriction that prohibits clusters /iɨ/ or /ɨi/ (§4.3.2). 5.5.1. Restrictive =kɨ The vowel of the restrictive =kɨ always undergoes vowel harmony with a last high front vowel /i/ of the stem. In the case of =kɨ, it is the vowel of the clitic which changes: ɨ → i. Compare the realizations of =kɨ (39) as opposed to (40). (39) nui ̃ŋ́ki109 nu=i=̃kɨ non.vis=LOC=RESTR ‘there only’ 109. One of the surface manifestations of underlying nasality is the occurrence of a nasal manifestations of nasality is the occurrence of [n] after the vowel that is lexical locus of nasality when some morphemes are added to he root. See §6.10 for more detailed discussion of nasal prosody. 216 (40) húkɨ hu=kɨ PROX-RESTR ‘this only’ Interestingly, the harmony process does not occur if the underlying vowel of the root is not a high front vowel, even when the underlying vowel of the root is deleted on the surface: (41) wampiśkɨʃa uampisa=kɨ=ʃa Wampis=RESTR=ADD ‘The Wampis only’ There are a few examples where sometimes the no application of vowel harmony (when it is expected) is problematic to explain. It is very frequent that a narrative ends with the following prosodic word: (42) nút͡ʃikɨti nu-t͡ʃi=kɨ=iti NON.VIS=DIM=RESTR=COP.3+DECL ‘That only is.’ The non-application of vowel harmony in (42) is only apparent. The 3 person copula clitic =iti reduces its "rst vowel /i/ when it is preceded by a vowel /ɨ/ of the stem. The relevant parts of the derivation would be: (43) derivation for /nut͡ʃikɨti/ syllable: nu.t͡ʃi.kɨ.i.ti sandhi1: mora reduction: nut͡ʃikɨti 217 sandhi2: vowel harmony: blocked output: [nút͡ʃikɨti] However, there are similar examples where vowel harmony applies: (44) nui ̃ŋ́kiti nui=̃kɨ=iti NON.VIS=RESTR=COP.3+DECL ‘There only is.’ In this case, the reverse order of morphophonological process must apply. This behavior is not very well understood. It may be the case that nut͡ʃikɨti ‘that only is’ is an idiosyncratic form, perhaps a consequence of the relative high frequency with which it occurs in narratives. One important observation from (43), however, is that when mora reduction occurs between the stem and the copula clitic, other vowel elision processes are blocked. 5.5.2. Do while moving -ki(ni) The associated motion su#x -ki(ni) ‘do while moving’ (§13.3.2.6) triggers optional vowel harmony with a high central vowel /ɨ/ of a preceding syllable of a verbal stem. In practice, this optional process of vowel harmony is infrequent, as there are few verb stems that contain an ɨ and at the same time are able to receive the associated motion su#x -ki(ni). 218 (45) pɨit́ki wɨḿaji pɨɨtu-ki wɨ-ma-ji /ash.lightning-WHILE.MOVING go-IMM.PT-3.PT+DECL ‘Flashing of lightning went on.’ (46) waki ́tkin uakɨtu-ki-inu return-WHILE.MOVING-NMLZ ‘one who returned (i.e. a returner)’ The root of the verbs in the above examples possess an /ɨ/, as evidenced in the next examples where there is no vowel harmony. (47) pɨɨt́awɛɛ pɨɨta-ua-i /ash.lightning+IPFV-3.SBJ-DECL ‘It is /ashing (lightnings).’ (48) wakɨt́ɛiɲawai wakɨta-ina-ua-i return.IPFV-PL.IPFV-3.SBJ-DECL ‘The are returning.’ 5.6. Di&erent forms of su0xes or clitics Some su#xes and clitics have di$erent forms: most of these di$erent forms vary between a long and short form, but at least in the case of the locative =nVma the variation in form has to do with an intra-morphemic vowel. The use of the distinct forms of these morphemes is morphophonologically conditioned, except for =nVma 219 where no apparent condition is found. However, for practical purposes =nVma is presented in this section, as =nVma shares with the other morphemes the fact that it can occur with di$erent forms. Table 5.5 lists these morphemes. Table 5.5. Morphemes that exhibit di$erent forms Morpheme (functional description) Gloss or Category =nVma ‘Locative’ a(́u) ‘High a$ectedness aktionsart’ -ki(ni) ‘Do while moving’ =(n)i ̃ ‘Locative’ -(n)i ̃ ‘Di$erent subject’ =aita~=ita; =aiti~=iti Copula clitics 5.6.1. Locative =nVma The realization of the locative =nVma varies between [=nama] and [=numa]. One may be tempted to suggest that such variation possibly arose from a process of vowel harmony: some noun roots that end with a vowel /a/ tend to prefer the form [nama], whereas others ending in a high vowel tend to appear with the form [numa]: (49) kajańam wampiśɲum kaia=nama uampisa=numa stonce=LOC Wampis=LOC 220 nuŋkańam t͡sukińɲum nunka=nama t͡sukinta=numa earth=LOC corner=LOC There are, however, plenty of examples where a potential vowel harmony hypothesis is not veri"ed. Many words may appear with either form [nama] or [numa] regardless of the "nal vowel of the stem, as shown in (50). (50) ikᶨamnam ~ ikʲaḿnum ɨnt͡sańam ~ ɨnt͡sanum ikama=nVma ɨnt͡sa=nVma forest=LOC river=LOC 5.6.2. High a&ectedness -a(́u) The high a$ectedness su#x -a(́u) occurs with the form -aú before apprehensive - ai and non-imperfective plural -ara. The details are described in §5.3.3. 5.6.3. Do while moving -ki(ni) The su#x -ki has a long form -kini that occurs when preceding the 3 person perfective -i ̃~ -ɨ.̃ The next example shows both realizations of this su#x (the example comes from two consecutive lines of the same text). In the example, -ki(ni) occurs with the long form "rst, when “fused” with the third perfective su#x, and with the short form in the next line:110 110. There is an underlying nasalization in the su#x -ki(ni). Nasalization sometimes surfaces as a high pitch. See §6.10 for details. 221 (51) tikit́͡ʃik t͡ʃankiń ʔɛiḿkʲamuanúna hurukni ̃ ́. . . núu út͡ʃi huruki ́wɨsatahkama ̃ ́ tikit ͡ʃiki t͡ʃankina aima-ka-mau=a=nú=na one basket "ll.up-INTENS-NMLZ=COP=NON.VIS=ACC hu-ru-kini ̃́ take-APPL-WHILE.MOVING+3.PFV nu ut͡ʃi hu-ru-ki ̃ wɨ-sa-tahkama ̃́ NON.VIS child take-APPL-WHILE.MOVING\3SG.SS go-ATT-FRUSTR\3SG.SS ‘[The child] took one basket that was "lled. . . that child when he wanted to go, having taken [the basket]. . .’ 5.6.4. Locative =(n)i ̃and Di&erent subject -(n)i ̃ The locative =(n)i ̃occurs with a long form [ni]̃ following a vowel /i/ of the stem. The di$erent subject -(n)i ̃is historically related to the locative homophonous form, and follows exactly the same pattern. Compare (52), where the form of the locative is =i,̃ with (53), where the long form is used. (52) hui ̃ ́ hu=i ̃ PROX=LOC ‘here’ (53) hintiɲ́i hinti=ni ̃ trail\1PL/2PL/3.POSS=LOC ‘In his trail.’ 222 The following are examples of -(n)i ̃as a di$erent subject marker, showing the same alternation: (54) puhaḱui ̃ puha-ku-i ̃ live+IPFV-SIM-DS ‘While I am living...’ (55) puhaḱmin puha-ku-mi-ni ̃ live+IPFV-SIM-2SG-DS ‘While you are living...’ 5.6.5. Copula clitics Copula clitics in Wampis have a long and short form as shown in Table 5.6. Table 5.6. Forms of copula clitics in Wampis Speech act participant 3 person aita ~ ita aiti ~ iti The use of the long or short form depends on several criteria. After glide derivation: • With a preceding cluster of vowels in the root or stem, the long forms are always use. A glide (either allophonic or epenthetic) is always present in these cases. The next examples illustrate this behavior. 223 (56) aróuwɛithɛɛ ara-u=aita-ha-i plant-NMLZ=COP-1SG.SBJ-DECL ‘I am a planter.’ (57) waa ̃ɲ́ɛiti111 uaa=̃aiti hole=COP.3+DECL ‘It is a hole’ (58) jamɛíjɛiti iamai=aiti now=COP.3+DECL ‘It is now.’ • With a single "nal vowel /a/ of the stem, the short forms are used.112 (59) pɨŋ́kɨrɛiti pɨńkɨra=iti good=COP.3+DECL ‘It is good.’ 111. The palatal [j] nasalizes to [ɲ] as nasalization spreads from the nasal vowel of the root to the glide. 112. One could ask why the short form and not the long form is used. It is possible to propose that there is vowel reduction of the stem, rather than insertion of the short form. But notice that this makes it harder to explain examples like (57), where there is no deletion of the vowel of the stem. 224 (60) jɛit́mɨ ia=ita-mɨ who=COP-2SG.SBJ Who are you? glide derivation: /iaitamɨ/ → jaitamɨ (the stem [ja] has a "nal single vowel /a/) (61) kanúsiɛithɛɛ kanusa=ia=ita-ha-i Santiago.river=ABL=COP-1SG.SBJ-DECL ‘I am from the Kanus (area).’ • With a single "nal vowel /i/ and /u/ of the stem, the long forms are used if the stem has two or less moras.113 An epenthetic vowel is inserted if there is no allophonic glide: (62) ni ̃j́ɛiti ni=̃aiti 3SG=COP.3+DECL ‘It is him.’ (63) kusújɛiti kusu=aiti murky.water=COP.3+DECL ‘It is murky water.’ • With a single "nal /u/ of the stem, the short forms are used if the stem has more than 113. There are some exceptions to this. The form ni=̃kɨ=iti ‘3sg=restr=cop.3’ (‘he only is’) which surfaces as [ni ̃ŋ́kiti], only has two moras but uses the short form of the clitic. 225 two moras. (64) karaṕuiti karapu=iti nail=COP.3+DECL ‘It is a nail’ (karapu ‘nail’ from Spanish ) • With a single vowel /i/ of the stem, the short forms are used if the stem has more than two moras—but in this case, the mora of the "rst vowel /i/ of -iti is reduced: (65) tikit́͡ʃiti tikit ͡ʃi=iti other=COP.3+DECL ‘It is another one.’ • With a preceding vowel /ɨ/ the stem, the short forms are used and the "rst mora of /i/ is always reduced: (66) núkɨti nu=kɨ=iti NON.VIS=RESTR=COP.3+DECL ‘That only is.’ (67) nɨrɨt́i nɨrɨ=iti fruit=COP.3+DECL ‘It is a fruit.’ 226 (68) nukaṕɨthi nukapɨ=ita=hi many=COP=1PL.SBJ+DECL ‘We are many.’ Notice that because there is reduction of a mora to a preceding vowel /i/ or /ɨ/, as we saw previously with the high a$ectedness -a(́u) (§5.3.3) and the non-imperfect plural -ara (§5.4.2), further elision of vowels is blocked. Compare (68), where the vowel /ɨ/ is not deleted, with (69), where /ɨ/ is deleted: (69) nukaṕt͡ʃouwɛithi nukapɨ-t͡ʃau=aita-hi many-NEG.NMLZ=COP-1PL.SBJ+DECL ‘We are not many.’ • Finally, if the vowel that precedes the copula clitic receives a high tone (or has a high tone in its lexical representation), then a glide is inserted and the long forms are used, regardless of number of moras or vowel quality. (70) unt͡surij́ɛiti unt͡suri=́aiti numerous=COP.3+DECL ‘They are numerous.’ 227 (71) t͡sunaj́ɛiti t͡suna=́aiti T ͡suna=́COP.3+DECL ‘It is Tsuna.́’114 5.7. Other phenomena 5.7.1. Degemination of stops Roots or stems lose their surface "nal stop consonant with an identical initial consonant of an immediately subsequent morpheme. Thus the consonant cluster is reduced according to the pattern illustrated by the following rule: Ci + Ci → Ci where C=stop, and “+” indicates morpheme boundary. Example (72) shows an instance of this process (the vowels that are to be deleted are underlined); the consonants that merge in the surface form are bolded. (72) óohmatin áuhumatu-tinu inform-FUT.NMLZ ‘one who will inform’ The degemination of consonant stops strictly occurs in the context of a morpheme boundary; it does not occur if the consecutive consonants belong to the same 114. Tsuna ́is a proper noun. 228 morpheme. For instance, the de"nite future -tata does not merge its initial consonant when the "rst /a/ is deleted: (73) óohmatsattahɛɛ auhumatu-sa-tata-ha-i inform-ATT-DEF.FUT-1SG.SBJ-DECL ‘I am going to tell.’ 5.7.2. Optional gemination of consonants The locative =nVma can optionally insert a homorganic initial consonant when the root that receives it appear ends in a vowel: (74) limańnumia Lima=nVma=ia Lima=LOC=ABL ‘from Lima’ 5.7.3. Optional metathesis Metathesis is not systematic in Wampis. However, there are some examples of metathesis in the language. It seems like morphemes that possess an /r/ favor metathesis. The "rst person possessive su#x -ru tend to be particularly susceptible to undergoing optional metathesis. For instance, consider the word nukut ͡ʃi ‘grandmother’ (that the underlying representation of ‘grandmother’ is nukut͡ʃi is evident from (75)). In (76), the vowel /u/ of the 1 singular possessive -ru surfaces in its “normal” position: 229 (75) nukut͡ʃiń nukut͡ʃi=na grandmother=ACC ‘my grandmother (ACC)’ (76) nukút͡ʃrun nukut͡ʃi-ru=na granmother-1SG=ACC ‘my grandmother (ACC)’ However, examples from texts and casual speech show that speakers metathesized forms can occur: (77) nukút͡ʃur nukut͡ʃi-ru grandfather-1SG ‘my grandmother’ Here is another example of metathesis with a di$erent su#x, the distributed action -ra (so, again, the common element seems to be the /r/ of the su#xes)—this -ra is a verbal su#x (cf. §13.3.2.4 for an analysis of -ra): (78) nɨkapaŕ warukú timaji nɨkapɨ-ra ̃ ua-ru-ka-u timaji feel-DISTR\3.SS climb-APPL-INTENS-NMLZ NARR ‘Having felt her, he climbed toward her.’ Metathesis in lexical roots does not occur, with very rare exceptions. The only instances of apparent metathesis that I have found with lexical roots come from 230 alternations of one noun and one verb (plus possible derivations of each of them). Interestingly, these two items are homophonous—whether they are historically related is not known, but synchronically they do not seem to be related. The noun for ‘devil’ seems to be based on the noun for the mythological character Iwa (/iua/) ‘Mythological giant cannibal’) and the verb ‘last’ seems to be related to ‘be alive’.115 The items are shown in (79). (79) /iua/ [iẃa] ~ /uia/ [úja] ‘Iwa (a mythological character)’ /iuantt͡ʃi/ [iẃant͡ʃ] ~ /uiant͡ʃi/ [újant͡ʃ] ‘devil’ (also an alternate name for the mythological character Iwa) /iua/ [iwa]́ ~ /uia/ [uja]́ ‘be alive’ /iuantu/ [iwańtu] ~ /uiantu/ [ujańtu] ‘last’ (probably related to ‘be alive’) 5.8. Vowel-switching for marking possession A subclass of nouns marks a noun as possessed by a 1pl/2pl/3 person by switching the vowels /a/ or /u/ of the root in the last surface vowel into ɨ ̃or i.̃ The patterns of this vowel-switching phenomenon for marking possession are explained in detail in §10.4.1. 115. Probably a nasal variation iua ̃triggered the occurrence of the nasal consonant /n/ before - t͡ʃi (this -t͡ʃi very likely is the diminutive (which has the same form in current Wampis)), and also before -tu (maybe the applicative su#x -tu?). 231 CHAPTER VI PROSODY 6.1. Introduction This chapter explores the prosody of Wampis in the following way: §6.2 gives an overview of preliminary works in other Jivaroan languages and provides some important de"nitions to understand the phenomenon of stress and tone in Wampis. §6.3 de"nes the mora in Wampis, §6.4 is dedicated to stress. Next, §6.5 describes the tone system of Wampis, followed by a description of phenomena of high tone placement shift in §6.6 and a summary of stress and tone in §6.7. In §6.8, the prosodic word in Wampis is de"ned. §6.9 and §6.10 provide notes on utterance-level prosody and nasal prosody in Wampis. 6.2. Previous works and preliminary notions Other Jivaroan languages have been varyingly described as having one or more of the following categories: • Stress (Wampis) (Beasley & Pike 1957) • Stress, Tone, “Tone-stress” (Shuar) (Turner 1958b) • Pitch-accent (Shuar) (Gnerre 2010) 232 • Stress, Accent (also mention of Tone) (Achuar) (Fast 1975a) • Accent/Pitch Accent (Awajun) (Payne 1990a; Corbera Mori 1994; Overall 2007) As we can see, there has been some confusion in the way how previous works (especially early works) have described the prosody of Jivaroan languages. The most authoritative studies of the prosody of other Jivaroan languages are provided by Payne (1990a) and Overall (2007), both for Awajun. These authors treat the Awajun prosodic system in terms of pitch accent. Overall provides the following rule of thumb for said language: “The principle of “one word per accent, one accent per word” is an important criterion and diagnostic of wordhood” (2007: 96). We will see Wampis does not exactly "t within this criterion. The Wampis system has remained practically undescribed until the present work, with the only exception being passing comments about “stress” in Beasley & Pike (Beasley & Pike 1957). The analysis proposed here follows the lead of Hyman’s call for a property-driven account of prosodic systems (Hyman 2006; Hyman 2009), rather than necessarily trying to assign the language to one versus another “type”, given the considerable typological variation found across languages. Hyman proposes that there is two prototypes of prosodic systems: Stress and Tone, and claims that there is no “Pitch 233 accent” prototype. Prototypical stress systems (e.g. Czech) and prototypical tonal systems (e.g. Madarin Chinese) are relatively well understood, but there are languages that are less-studied that appear to combine characteristics of stress and tone system. Wampis is an example of these type of languages that do not exhibit a classical tone or stress system, but are “somewhere” in the middle. These languages have been traditionally labelled “pitch-accent” systems and have posed descriptive and theoretical problems for linguists (Hyman 2006). Several studies have cast serious doubts about “pitch-accent” as a typological concept. In fact, Hyman (2009; 2010) demonstrates that the label “pitch-accent” has been used to describe very di$erent phenomena; thus, there is no uni"ed concept of pitch-accent: “pitch-accent is not a coherent notion, rather a ‘pick and choose’ among the properties that characterize prototypical tone vs. stress- accent systems” (Hyman 2006: 172). On the other hand, applying Hyman’s property- driven approach to word-prosody typology to Iquito (a Zaparoan language spoken in Peru), Michael (2011) shows that it is possible, and more elegant, to describe so-called “pitch-accent” languages in terms of the two prototypes of stress and tone suggested by Hyman. I will propose that the prosodic system of Wampis exhibits a stress system that is 234 distinguishable from a tone system. By proposing a system that distinguishes stress and tone in a parsimonious fashion, a better understanding of the Wampis system is gained, and (otherwise unconnected) phenomena of vowel elision and tone assignment "nd their motivation, at least partially, in their association with a “covert” stress system. The Wampis word prosodic system is complex, it exhibits a stress system which depends on a rhythmic pattern, and a tonal system which is partially dependent on stress. Before going into analytical details, some terminology and de"nitions need to be introduced. With regard to the two prototypes of stress and tone, what I understand as a language with stress is one in which there is an indication of word-level metrical structure that meets two criteria: obligatoriness and culminativity. Obligatoriness is understood as the criterion by which every lexical word has at least one syllable marked for metrical prominence. Culminativity is understood as the condition by which every lexical word has at most one syllable marked for prominence (Hyman 2006: 168). With regard to tone, I follow Hyman’s broad de"nition: “a language with tone is one in which an indication of pitch enters into the lexical realization of at least some morpheme” (2006: 167). I will propose that Wampis exhibits a low-density tone system. I understand by 235 “low density tone system” a prosodic system in which many syllables or moras do not carry a lexical tone (Gussenhoven 2004). However, we will see that one of the de"ning requirements for prosodic words in Wampis is to have at least one high tone (i.e. an acoustic feature of high pitch). When Wampis words do not have a lexically-speci"ed tone, they ful"ll the tone requirement via a metrical tone. High tone in Wampis is contrastive. I will distinguish between three types of high tones: • Metrical high tone: A metrical high tone is a high tone that is dependent on the metrical system of the language. • Lexical high tone: A lexical high tone in Wampis is a high tone that is present in the lexical representation of roots, a#xes and some clitics. • Grammatical high tone: A grammatical high tone is a high tone that expresses a grammatical category (i.e. a toneme). High tone is represented by an accent marked above the vowel V́. What I understand as prosodic word is a constituent that is higher than the syllable and foot but lower than the intonational phrase (Hogg & McCully 1987). In Wampis, the prosodic word is de"ned as the domain where stress, tone assignment and a minimal two-mora requirement are satis"ed. 236 6.3. The mora and mora requirement The mora is the unit that determines syllable weight. In Wampis, the mora equals the vowel, codas are irrelevant for stress and tone assignment (see David Payne (1990a) for a similar treatment of the mora in Awajun). In Chapter VI, I analyzed the metrical feet as being formed at an intermediate stage of derivation where the only consonants allowed in coda position are the nasals /n/ and /m/. Notice how the second vowel (in an open syllable) attracts the high tone in the words in (1), even though there are contiguous closed syllables, and even though the structure of the syllable changes from open to close in [arútam] (nominative, where ru is an open syllable) and [arútman] (accusative, where rut is a closed syllable). The changes to the surface syllable structure do not a$ect stress and high tone assignment. (1) kakaŕam kakarama ‘powerful person’ arútam arutama ‘power vision’ (nominative) arútman arutama=na power.vision=ACC ‘power vision (ACC)’ 237 The mora is the tone bearing unit (TBU) in Wampis. When there is an induced shift of tone placement by the presence of another morpheme that triggers the shift (like the locative =nama in (2)), the high tone changes to another vowel: (2) wa ̃á waãńam uaã uaã=nama hole hole=LOC Figure 6.1 and Figure 6.2 show the di$erence in tonal contour in the above words. The sharp contrast in pitch contour indicates that the high tone falls on the vowel; which is considered to be the tone bearing unit. Time (s) 0 0.5663 0 5000 Fre qu en cy (H z) w ãá Figure 6.1. Spectrogram showing pitch contour of [wa ̃á] ‘hole’ 238 Time (s) 0 0.7147 0 5000 Fre qu en cy (H z) w ãá n a m Figure 6.2. Spectrogram showing pitch contour of [waãńam] ‘in the hole’ Prosodic words in Wampis are minimally bimoraic. Words that only have one mora in their lexical representation lengthen their vowel to comply with this two-mora restriction. Compare the realizations of hu ‘Proximal demonstrative’ and ni ̃‘3SG’ when they occur alone and when they occur with other morphemes. (3) /hu/ [húu] ‘Proximal demonstrative’ /huka/ [húka] (hu=ka ‘PROX=FOC’) /ni/̃ [ni ̃í] ‘3SG’ /niña/ [ni ̃ɲ́a] (ni=̃na ‘3SG=ACC’) An important note to remember from Chapter IV is that moras are counted after 239 glide derivation; for instance, a word like ui ‘1SG’ lengthens its vowel /i/ after the glide [w] is derived, because glides do not count as moras. (4) /ui/ [wií] ‘1SG’ 6.4. Stress in Wampis This section presents an introduction to the study of stress and tone in Wampis. It is generally recognized that stress is the manifestation of a rhythmic structure (Hayes 1995). Consequently, the aim of this study is to show that stress in Wampis follows a rhythmic pattern and that the complex interaction with tone can be partially associated with this pattern. Stress in Wampis is not completely predictable. Nevertheless, I propose that a great proportion of words in Wampis follows a rhythmic pattern that is manifested in an iambic foot structure in words with more than two moras. A high tone is attracted by the most prominent constituent of the metrical parsing unit (i.e. the foot) and assigned to a mora following strict rules. Though restricted to certain combinations (which are not infrequent in actual speech), there can be more than one high tone in a prosodic word (if a metrical high tone interacts with a lexical high tone), thus the principle of “one accent per word”, which is how typical so-called pitch-accent systems are described, is not an accurate description for Wampis. Regarding stress, it is important to 240 distinguish between words that have two moras (the minimum allowed in the language) and words that have more than two moras. 6.4.1. Stress and tone in bimoraic words in Wampis Wampis treats morphologically simple bimoraic words di$erently than other words (recall that there are no monomoraic prosodic words). Morphologically simple bimoraic words are uncontroversial in terms of their prosody: a high tone is always assigned to the "rst mora of the word. This pattern can be described in terms of extra- metrical restriction: in Wampis there appears to exist a general extra-metrical restriction: extra-metricality of the last mora. There are almost no exceptions to this pattern of assigning a high tone to the "rst mora in bimoraic words, only roots with lexically-speci"ed high tone (and a few interjections and ideophones) do not follow this rule. The words in (5) illustrate the pronunciation of morphologically simple bimoraic words in Wampis: (5) /apa/ [aṕa] ‘father’ /ɨt͡sa/ [ɨt́͡sa]̃ ‘sun’ /hɨã/ [hi ̃á] ‘house’ /paki/ [paḱi] ‘white-lipped peccary’ /kanka/ [kaŋ́ka] ‘boquichico (Prochilodus sp.)’ /númi/ [númi] ‘tree’ /núku/ [núku] ‘mother’ /ui/ [wií] ‘1SG’ 241 Most morphologically complex words have more than two moras underlyingly (though on the surface they may occur with two moras because of vowel apocope). When underlyingly bimoraic words carry additional morphology (and therefore become words with more than two moras), they become susceptible to shift in the placement of their high tone, which is induced by some su#xes and clitics (§6.6). 6.4.2. Stress in words with more than two moras Unlike underlyingly bi-moraic words, the prosody of words with more than two underlying moras in Wampis is complicated. Pervasive vowel elision processes apply in words with more than two moras, posing di#culty for the analysis of the prosody. In this study, I propose that the vowel elision processes of Wampis are related to a historical rhythmic pattern. My hypothesis is that this rhythmic pattern is still visible in the language, but it is somehow obscured precisely because of vowel elision. However, I claim that vowel elision can be seen as a manifestation of the stress pattern in current Wampis, rather than as an element that obscures the rhythmic pattern. The proposed hypothesis of development of a rhythmic pattern in Wampis is formalized in the following terms: • In past stages of Wampis, prosodic prominence was found in a rhythmic iambic 242 pattern. • Unstressed elements were unvoiced; in later historical stages devoicing of unstressed vowels turned into a process of vowel elision. There is some evidence of voiceless vowels in past stages of Wampis. Voiceless vowels are mentioned in Beasley & Pike’s study of Wampis phonology: “certain vowels which in the middle of the words in a restricted word list are clearly voiced, and are members of the ordinary voiced vowel system, lose their voicing when unstressed, and when the su#xes following them are dropped from the word” (1957: 4). There is no evidence of voiceless vowels in current Wampis, but see §3.8.6 for a possible related phenomenon. The rhythmic pattern of Wampis that I propose consists of left-to-right parsed iambic feet, i.e. feet that are right-headed, as illustrated in Figure 6.3 (where W=Weak position in the foot, S=Strong position in the foot). Left-to-right iamb (W ˈS)⎯ ⎯ ⎯ ⎯ → Figure 6.3. Parsing directionality in Wampis A possible historical scenario for the hypothesis outlined above would have an 243 ideal foot parsing with CV syllables as presented below (we will see that this analysis for the most part is applicable synchronically for current Wampis): (CV. ˈCV) (CV. ˌC.V) (CV. ˌCV) CV . . . Prosodic prominence was established for the "rst foot from the left. Therefore, the strong position of the "rst metrical foot was assigned a metrical high tone. If we abstract the moras (which are the relevant TBU), we have the second mora as the most prominent for metrical high tone assignment: (V V́) (V V). . . At some point, Wampis speakers presumably started to de-voice vowels in weak positions, which led to vowel elision. In turn, vowel elision led to re-syllabication: CV.CVC.CVC.CVC. . . Having outlined a possible scenario of development of the Wampis system, the discussion now turns to the more detailed analysis of Wampis data for words that have more than two moras. I use the symbol ˈ to represent the primary stress of the word (the most prominent prosodic position in the metrical feet). As in previous chapters, vowels that are deleted on the surface are represented with an underline. High tone is represented with an accent on top of the vowel (V́). With words that possess more than two moras, the Wampis system forms 244 metrical feet at an intermediate stage of derivation after glide derivation (§4.2) and morphophonological processes such as mora reduction at morpheme boundary (§5.3). As mentioned previously, Wampis parses iambic feet of the form (W S). As an illustration, consider the derivation of the root maa ‘bathe’ in [mɛit́hʲɛɛ]. In the derivation, we see that at the level of sandhi the second /a/ of root is reduced in the presence of the low a$ectedness aktionsart su#x -i. Only then the feet are formed and word can be derived with the correct output. (6) mɛíthʲɛɛ maa-i-ta-ha-i bathe-LOAF-IMM.FUT-1SG.SBJ-DECL ‘I will bathe.’ derivation: /maaitahai/ glide: does not apply syllable: ma.a.i.ta.ha.i sandhi: ma.i.ta.ha.i feet: (ma.ˈi).(ta.ha).i vowel elision: maithai output: [mɛíthʲɛɛ] The next examples illustrate the iambic foot. For ease of representation, I do not include intermediate stages if they do not apply (for instance if there is no vowel that becomes a glide, I “skip” the glide derivation step). 245 (7) /namaka/ ‘"sh’ syllable: na.ma.ka feet: (na.ˈma).ka vowel elision: namak output: [namaḱ] (8) /t͡sauanta/ ‘day’ glide: t͡sawanta syllable: t͡sa.wan.ta feet: (t͡sa.ˈwan).ta vowel elision: t͡sawan (/t/ is also dropped, because it cannot occur in coda margin) output: [t͡sawań] (9) /paantama/ ‘plantain’ syllable: pa.an.ta.ma feet: (pa.ˈan).(ta.ma) vowel elision: paantam output: [paańtam] (10) /maata/ ‘to bathe’ (maa-ta ‘bathe-NMLZ’) syllable: ma.a.́ta feet: (maˈa).ta vowel elision: maat output: [maat́] (11) /hintinakaratuinu/ ‘teacher’ (hintina-karatu-inu ‘teach-1PL.OBJ-NMLZ’) syllable: hin.ti.na.ka.ra.tu.i.nu sandhi: hin.ti.na.ka.ra.ti.nu feet: (hinˈti).(na.ka).(ra.ti.)nu vowel elision: hintinkartin output: [hintiŋ́kʲartin] 246 (12) /pɨaka/ ‘bed’ syllable: pɨ.a.ka feet: (pɨ.ˈa).ka vowel elision: pɨak output: [pɨaḱ] (13) /kuhant͡ʃama/ ‘opposum’ syllable: ku.han.t͡ʃa.ma feet: (ku.ˈhan).(t͡ʃa.ma) vowel elision: kuhant͡ʃam output: [kuhańt͡ʃam] (14) /t͡ʃit͡ʃastasanu/ ‘in order to speak’ (t͡ʃit͡ʃa-sa-tasa-nu ‘speak-ATT-PURP-1SG.SS) syllable: t͡ʃi.t͡ʃa.sa.ta.sa.nu feet: (t͡ʃi.ˈt͡ʃa).(sa.ta).(sa.nu) vowel elision: t͡ʃit͡ʃastasan output: [t͡ʃit͡ʃaśtasan] (15) /tikit͡ʃiki/ ‘one’ syllable: ti.ki.t͡ʃi.ki feet: (ti.ˈki).(t͡ʃi.ki) vowel elision: tikit͡ʃik output: [tikit́͡ʃik] From the above examples, the next rule of metrical high tone assignment can be posited: Rule of metrical high tone assignment in Wampis: A metrical high tone is given to the left-most prominent constituent in the iambic structure. 247 This rule predicts that a metrical high tone occurs in the "rst foot counting from the left. The foot is right-headed, thus the second mora receives the metrical high tone ((V ˈV́) (V V) . . .). The words in (16) further illustrate the rule of metrical high tone in use: (16) /aiamarukatasanu/ → Glide: ajamarukatasanu → [ajaḿruktatasan] ‘in order for me to defend’ (aiamaru-ka-tasa-nu ‘defend-INTENS-PURP-1SG.SS’) /ɨakamau/ → [ɨaḱmoo]116 ‘hunter’ /kaaʃapa/ → [kaaʃ́ap] ‘ray ("sh)’ /kakarama/ → [kakaŕam] ‘powerful person’ /kusɨa/ → [kusɨá] ‘sabalo ("sh sp.)’ /kuiʃi/ → [kuwiʃ́] ‘ear’ /kunt͡ʃai/ → [kunt͡ʃɛí] ‘type of fruit’ /namaka/ → [namaḱ] ‘"sh’ /tutupini/ → [tutúpin] ‘straight’ /uɨnunt͡ʃi/ → Glide: wɨnunt͡ʃi → [wɨnúnt͡ʃ] ‘clavicle’ To complete the analysis of Wampis stress, the acoustic correlates of stress and the processes of vowel elision are presented next. 6.4.3. Acoustic correlates of stress Measurement of length in vowels as well as in pre- and post-nucleus consonants was done in order to establish a possible acoustic correlate to stress. From the acoustic analysis, two factors seem to correlate to stress: 116. Recall that the sequence /au/ is pronounced [ou]~[oo], and [ai] is pronounce [ɛi] ~ [ɛɛ] (§3.7.1). 248 • Stressed vowels are lengthened (they are longer than unstressed vowels) • Onset stop consonants of stressed syllables are lengthened. To analyze the acoustic correlates of stress, a short study was carried out. This study consisted in the examination of vowels and consonants in "ve words repeated three times in isolation plus 1 time in a carrier phrase (the token given in the carrier phrase was the one analyzed). Three speakers were recorded saying the following words: kunkukakɨ [kuŋˈkúkak] ‘only Ungurahui (Oenocarpus bataua)’ tutupini [tuˈtúpin] ‘straigth’ tikit ͡ʃikiʃa [tiˈkiʃ́kiʃa] ‘one more’ ʃukukakɨ [ʃuˈkúkak] ‘only the shell’ akutata [aˈkútat] ‘to hurt’ These items were chosen because they preserve a structure where at least two vowels and two stops occur word-medially, so they can be measured and compared. The mean results of the measurements are presented in Table 6.1. Table 6.1. Measurement of vowel and consonant length in pre-tonic, tonic and post- tonic positions Pre-tonic vowel Tonic vowel Post-tonic vowel 0.57 s 0.88 s 0.62 s Pre-tonic consonant (onset of stressed syllable) Post-tonic consonant 0.142 s 0.102 s 249 As can be seen, tonic vowels are overall longer than pre-tonic and post-tonic vowels. In addition, the pre-tonic consonant (the onset of the stressed syllable) is considerably longer than the post-tonic consonant. For the study, the post-tonic consonants were the onset of the next unstressed syllable, thus they are in comparable position with the pre-tonic consonants, which were the onset of stressed syllables. Note that for the word tikit ͡ʃikiʃa [tiˈkiʃ́kiʃa], the post-tonic consonant measured was the second surface [k], not the [ʃ] in the surface coda. An illustration of di$erence in length measurements is given in Figure 6.4 with the word tikit ͡ʃikiʃa [tiˈkiʃ́kiʃa] ‘one more’, where the tonic vowel is the second /i/. Correspondingly, this vowel is longer in duration (0.82 s) than the other two unstressed vowels (pre-tonic /i/ = 0.40 s, post-tonic /i/ =0.61 s).117 Similarly, the pre-tonic consonant (the "rst /k/ in the example) is also lengthened (0.140 s). Compare with the second /k/, which is shorter (0.93 s). 117. There is an underlying vowel that comes after the second /i/ (the tonic vowel) but is deleted and does not occur on the surface form. 250 Figure 6.4. Spectrogram of framed token of [tikiʃ́kiʃa] ‘one more’ 6.4.4. Vowel elision The preceding discussion has described the stress system of Wampis as well as provided acoustic evidence for it. Now I discuss the processes of vowel elision, which are one of the most interesting and complex phenomena in Wampis. The basic principles of vowel elision have been introduced already in previous chapters (cf. §4.2.1). Vowel elision is pervasive in Wampis, as it is in other Jivaroan languages 251 (Payne 1990a; Corbera Mori 1994; Overall 2007). Vowels are systematically elided word-internally and word-"nally if certain conditions are met. 6.4.5. Word-!nal and word-internal vowel elision Word-"nal (apocope) and word-internal (syncope)118 vowel elision apply strictly in that order in Wampis. Generally speaking, they apply in words of more than two moras.119 They operate on both simplex and morphologically complex words. Apocope: Word-"nally, vowels in a CV sequence (where C is any stop except a glide) are deleted. Syncope: After vowel apocope, the third vowel starting from the left, and every other alternate vowel in a CV.C cluster are deleted. There are some peculiarities and exceptions to the above rule, which will be described in the following sections, but these rules apply to a large portion of words in Wampis. I argue that apocope and syncope are for the most part another manifestation of 118. In previous works on Awajun, the term syncope have been used to mean word internal vowel elision in general (not only of unstressed vowels) (Payne 1990a; Overall 2007). I use syncope in the same sense. 119. Underlyingly the last syllable in a word is phonologically either CV or V in Wampis, since there can be only vowels in word-"nal position. This was described in Chapter IV. 252 the rhythmic pattern of Wampis described in the previous section. The reason for this argument is that the wovels that are deleted in Wampis precisely correspond to vowels that word-internally are in a weak position in the iambic feet. Vowel apocope can be described by invoking extrametricality of the last mora, i.e. this particular element is ignored by the stress principles of Wampis. Consider the word arutama, which deletes the last vowel /a/: (17) /arutama/ ‘Power vision (a spiritual concept)’ feet: (aˈru)(taˌma) output: [arútam] When morphology is added to the word, and the vowel /a/ which was deleted is no longer in word-"nal position, he vowel is again incorporated into the rhythmic pattern. Now the second /a/ from the left is deleted by syncope, as it is the mora in a weak position, and the third /a/ surfaces: (18) /arutamakɨ/ ‘only Arutam’, =kɨ ‘restrictive’ feet: (aˈru)(taˌma)kɨ output: [arútmak] The regular process of apocope deletes the vowel of a "nal CV syllable occurs "rst; that is, this rule applies prior to syncope. The examples in (19) show instances of 253 the rule of apocope in simplex words. (19) /ipaku/ → (iˈpa)ku [ipʲaḱ] ‘Bixa orellana’ /t͡ʃan.ki.na/ → (t͡ʃanˈki)na [t͡ʃaŋ́kin] ‘basket’ /muit͡sa/ → (muˈi)t͡sa [muit́͡s] ‘type of jar’ /arutama/ → (aˈru)(tama) [arútam] ‘spirit of power’ /himara/ → (hiˈma)ra [himʲaŕ] ‘two’ /uuɨha/ → (uˈwɨ)ha [uwɨh́] ‘hand’ The next examples show the process of apocope in morphologicallly complex words. (20) takat́ taka-ta work-NMLZ ‘work’ (21) t͡ʃankiɲań tʃankina=na basket=ACC ‘the basket (ACC)’ (22) jatst͡úr iatsu-ru brother-1SG.POSS ‘my brother’ A GV syllable in "nal position is immune to apocope. In (23), words with a "nal GV syllable do not delete their last vowels. The "rst word, ‘branch’ is morphologically 254 simple; the second word, ‘from the lake’, is morphologically complex. Compare with maatai ̃‘place to bathe (N)’ where the last vowel is not deleted either. For the process of syncope, glides are relevant and treated as regular consonants in onset position. (23) kanauɨ → [kanawɨ] ‘branch’ kutʃa-numa-ia → [kutʃanmaja] lake-LOC-ABL ‘from the lake’ maatai ̃ → [maat́ɛɛ]̃ ‘place to bathe’ Syncope is also a recurrent process in Wampis. This is a di$erent process that deletes the third moraic vowel counting from the left (i.e. from the beginning of the word) and then every other second vowel. (24) akahiki=na ‘male braid hairstyle=ACC’ feet: (aˈka)(hiˌki)na vowel elison: akahkin output: [akah́kin] (25) tutupini-t ͡ʃi ‘straitgh-DIM’ feet: (tuˈtu)(piˌni)t͡ʃi vowel elision: tutupnit͡ʃ output: [tutúpnit͡ʃ] 255 The following example illustrates syncope with a complex verb form. Apocope does not apply (the last vowels form a VV cluster) but syncope still applies. (26) taka-sá-tata-ha-i work-ATT-DEF.FUT-1SG.SUBJ-DECl ‘I am going to work’ feet: (taˈka)(saˌta)(haˌi) vowel elision: takaśtahai output: [takaśtahɛɛ] Syncope occurs only when the output of the apocope process allows the correct syllable formation, as explained in Chapter V. For instance, the "nal vowel in ipakukɨ ‘only Huito’ is deleted, but if syncope were to subsequently apply, it would produce an impossible coda for Wampis. Therefore, the process of syncope is blocked, and the derived word is [ipakuk], not *[ipakk]. (27) Ipaku=kɨ ‘huito (Genipa americana=restr) feet: (iˈpa)(kuˌkɨ) vowel elison: ipakuk output: [ipʲaḱuk] The same “blocking” of syncope occurs in certain word-internal con"gurations related to N.CV sequence. The next subsection describes this in detail. 256 6.4.5.1. Elision of CV in N.CV sequence If vowel elision occurs in a N.CV sequence, an NC cluster where C is any one stop (/p/, /t/, /k/) would result, but such a cluster is disallowed in Wampis, as a stop consonant cannot map into the margin of a syllable coda (see §4.3). This constitutes an important characteristic of the Wampis re-syllabi"cation structure: to avoid the unaccepted sequence, entire open syllables—both C and V together—are deleted after a nasal consonant. The consonant-deletion rule is formalized by the following rule, which applies word-"nally and word-internally. /C/ → ∅/N___. | [+stop] or [+nasal] The next examples illustrate the above rule in word "nal position. (28) /uunta/ → [úun] ‘big’ /nauantu/ → [nawán] ‘daughter’ /paapanku/ → [paápaŋ] ‘raft’ /kunampɨ/ → [kunám] ‘red squirrel’ Elision of CV syllables in N.CV clusters is also possible word-internally. First, consider /paantama/ ‘plantain’ in the nominative form (which is unmarked). This word deletes its last vowel (via apocope), which yields the intermediate form [paantam]. If the rule of syncope applies, that would yield the surface form [paantm], but the 257 syllabications [pa.ant.m] or [pa.antm] are impossible in Wampis. Thus, the process of syncope does not apply to preserve the correct word structure: (29) paantama feet: (paˈan)(taˌma) vowel elision: paantam output: [paańtam] However, when other morpheme is added, syncope applies. In (30), apocope applies deleting the last mora, then syncope syncope applies deleting the third mora. Here is a fascinating detail in Wampis re-syllabi"cation process: since a stop cannot be on the margin of a complex coda, and cannot form a complex onset with a nasal (as we saw in Chapter IV, the only complex onset allowed in Wampis is [Stop+rhotic] and the only complex coda allowed is [Nasal+A$ricate]), the solution of the language is to completely delete an entire word-internal syllable to ensure that the word formation is derived. (30) paantama=na ‘plantain=ACC’ feet: (paˈan)(taˌma)na vowel elision: paańtman rechecking: paańman (/t/ is in impossible position, gets deleted) output: [paańman] The only occasion when the underlying stop in a N.CV cluster is not deleted is when, because of vowel elision, the stop forms a sequence with the rhotic. Compare 258 (31) and (32). (31) nauantu ‘daughter’ feet: (naˈwan)tu vowel elision: nawant (/t/ is impossible in coda margin, gets deleted) recheck: nawan output: [nawań] (32) nauantu-ru=hai ̃‘daughter-1SG=COM’ (‘with my daughter’) feet: (naˈwan)(tuˌru)(haˌi)̃ vowel elision: nawantruhai ̃ output: [nawantruhɛɛ]̃ Finally, another point to notice is that a$ricates can occur in surface "nal complex codas, as shown in the next examples in which said consonants are not deleted when their phonemic syllabic nucleus is elided: (33) /aunt͡su/ → [aúnt͡s] ‘Spix’s guan’ /marunt͡ʃi/ → [marúnt͡ʃ] ‘prawn’ /tɨntɨt͡sa/ → [tɨntɨt͡s] ‘proper name’ 6.4.5.2. Special vowel elision in CVV sequences Some sequences of the form CVu and CVi allow internal vowel elision if the vowel is in a weak position where it can be deleted. Bound morphemes associated with this behavior are listed in Table 6.2. The vowel that can be elided in these morphemes is their "rst vowel (/a/ in all cases). 259 Table 6.2. Bound morphemes that allow internal vowel elision Morphemes Gloss -mau Set II nominalizer “non-subject” nominalizer -t͡ʃau Negative nominalizer -nai Reciprocal -nau Benefactive In (34), the second vowel /a/ from the left (the vowel /a/ of -nai ‘reciprocal) is in a weak position in the metrical foot (third mora), thus it is deleted. Compare with (35), where the same vowel is not in a weak position in the metrical foot, thus it is not deleted. (34) uha-nai-ta ‘inform-RECP-NMLZ’ (‘to communicate among several’) feet: (uˈha)(naˌi)ta vowel elision: uhanit output: [uhańit] (35) kahɨra-nai-ta ‘hate-RECP-NMLZ’ (‘to become enemies’) feet: (kaˈhɨ)(raˌna)(iˌta) vowel elision: kahɨrnait output: [kahɨŕnɛit] Other examples are given in (36). 260 (36) t͡ʃit͡ʃa-mau → [t͡ʃit͡ʃaḿu] ‘speak+IPFV-NMLZ’ (‘what is spoken’) t͡ʃit͡ʃa-t͡ʃau → [t͡ʃit͡ʃat͡ʃu] ‘speak+iPFV-NMLZ’ one who does not speak’ Dina-nau → [ðińanu] ‘Dina-BEN’ (for Dina) 6.4.5.3. Immunity to elision Some word-"nal and word-internal vowels of certain morphemes are apparently immune to elision. These lexically speci"ed exception vowels occur both in lexical roots, enclitics and su#xes (cf. Overall (2007: 92) for details on a similar phenomenon in Awajun). It is also possible that these vowels (moras) that are immune to elision are treated as heavy syllables, possibly forming their own foot, which would prevent their deletion. More data is needed to test this hypothesis. Interestingly, when a vowel is lexically marked as “immune” to elision, the next mora is automatically deleted. That is, the rhythmic feet are re-structured after the immune mora, and new feet are formed (in the new feet, the "rst mora following the immune mora is always in the weak position in the iamb). For example, the verb root usuma ‘paint face’ never loses its third mora (the vowel /a/), so the next one is deleted. (37) usúmakhɛi usuma-ka-ha-i paint.face-INTENS-1SG.SUBJ.DECL ‘I painted my face’ 261 A possible feet formation for this word would be (uˈsu)(ˌma)(kaˌha)i, where the syllable /ma/ that contains the immune mora forms its own foot. There are also su#xes and enclitics that do not delete their vowel in a position where it is expected to be deleted. For instance, compare the behavior of the su#x -tá ‘imperative’ and the su#x -ta ‘action nominalizer’. The imperative does not undergo apocope, but the nominalizer does. (38) [uhaktá] uha-ka-tá inform-INTENS-IMP ‘inform (her/him)!’ [uhat́] uha-ta inform-NMLZ ‘to inform’ Table 6.3 lists the grammatical morphemes that do not delete a vowel when they are in a position where they would be expected to be deleted. Some of the morphemes always occur in "nal position of the word. Some of the morphemes have a lexically speci"ed high tone. In the table, A=Apocope, S=Syncope. 262 Table 6.3. Bound morphemes immune to vowel elision in Wampis Morpheme Gloss Comment =ka =kɨ =ni =tí -tá -ʃa -mi -ti =api -tasa Focus Restrictive Allative Plural Speech Act participant Imperative Additive120 Hortative Jussive Tag question Purpose Immune to A (always "nal) Immune to S, not to A Inmmune to S Inmmune to A and S Immune to A and S Immune to A (always "nal) Immune to A (always "nal) Immune to A (always "nal) Immune to A and S Last vowel immune to A and S, "rst vowel not immune Selected examples with some of these morphemes are given in (39). (39) Focus =ka ʃuara=ka → [ʃuaŕka] (‘person=FOC’) Plural Speech Act participant =ti ́ uampisa=ti ́ → [wampisti]́ (Wampis=SAP ‘We the Wampis’)) Imperative -ta ́ uha-ka-ta ́ → [uhakta]́ (inform-INTENS-IMP ‘Tell him!’) Additive =ʃa t͡sunki=ʃa → [t͡súŋkiʃa] (‘water.being=ADD ‘the water being too’) 120. The vowel of this su#x sometimes underwent apocope in the speech of one of my teachers from Boca Chinganaza in contexts where other speakers would not delete the vowel. Thus ausha 263 Apparently, when apocope is blocked syncope still occurs. In arutama=na=ka, the vowel of =ka ‘focus’ is marked for non-elision; after elision does not apply, syncope applies in the word as expected, deleting the third mora: (40) arutama=na=ka spirit.power=ACC=FOC ‘to Arutam’ feet: (aˈru)(taˌma)(naˌka) vowel elision: arutmanka output: [arútmaŋka] However, syncope does not occur with vowels that are immune to elision. In the example below, the last vowel /a/ would be liable to be deleted per the process of apocope, but since the vowel of =ʃa ‘additive’ is “immune” to elision, it is not deleted. When it is not deleted, syncope applies and it would be expected that the vowel /ɨ/ of =kɨ ‘restrictive’ is deleted. However, the vowel of this morpheme is also immune to vowel elision, so deletion does not happen at all. (41) hapaḱɨʃa hapa=kɨ=ʃa deer=RESTR=ADD ‘deer only too’ 264 6.5. Tone in Wampis In Wampis words, the alternations in pitch are associated with a prosodic feature of high tone. In the previous section, we have seen that high tone in Wampis is partially predictable and associated with the stress pattern exhibited by the language—this what I called the metrical high tone in §6.2. However, there are words and morphemes that are lexically-speci"ed for tone—that is, what I de"ned as a lexical high tone in §6.2. Wampis also possesses what I de"ned in §6.2 as grammatical high tone: in Wampis, the vocative and the genitive cases are marked with a high tone on the last vowel of the word. Thus, in Wampis there is metrical tone, dependent on the rhythmic stress patterns of the language; and there is lexical tone, which is unpredictable. In addition, some morphemes trigger the shift of the high tone to a di$erent place in the word. In this section, a general view of these patterns are presented and the analysis presented in the preceding section is further re"ned. All prosodic words in Wampis have at least one high tone, and if a given prosodic word lacks lexically speci"ed high tone (which is very common), a high tone is assigned to the syllable bearing primary stress, according to the rules given in §6.4. Hence, the tonal system depends in part on the stress system of the language for 265 meeting the requirement of at least one high tone per word. Acoustically, the feature of high pitch is interpolated from the left margin of the word and rises to the pitch peak. After the peak, there can be abrupt drops in pitch and contours are formed. This section examines the tonal system, and its interaction with the stress system, in greater detail. The tone bearing unit in the Wampis language is the mora, as stablished in §6.3. The tone system in Wampis is a privative H/∅. Phonetically, because the mora (and not the syllable) is the tone bearing unit, contours over VV clusters may create H/L or L/H patterns, but there is no evidence of phonemic low tone as far as I have investigated. Rather, moras in Wampis can be speci"ed as bearing a high tone or not. Wampis does not exhibit a prototypical tone system, i.e. a system in which every syllable or mora has a high tone in their lexical representation. Some roots, clitics and a#xes have a lexical high tone, but many other do not. Evidence of contrastive high tone in Wampis comes from tonal minimal pairs, such as the ones presented in (42). (42) /hu/ [hũú] ‘type of moss’ vs /hũ/ [hú̃u] ‘Proximal’121 121. Speakers vary the pronunciation of this word between a nasal and an oral vowel. 266 /kaapi/ [kaaṕ] ‘tamshi vine’ vs /kaapɨ/ [kaáp] ‘/y’ /tukɨ/ [túkɨ] ‘whole’ vs /tukɨ/́ [tukɨ]́ ‘always’ The spectrograms in Figure 6.5 and Figure 6.6 show the pitch track of the words [túkɨ] ‘whole’ and [tukɨ]́ ‘always’. This indicates that, in e$ect, high tone is contrastive in Wampis. Time (s) 0 0.6137 0 5000 Fre qu en cy (H z) t ú k ɨ Figure 6.5. Spectrogram showing the pitch track for [túkɨ] ‘whole’ 267 Time (s) 0 0.5422 0 5000 Fre qu en cy (H z) t u k ɨ ́ Figure 6.6. Spectrogram showing the pitch track for [tukɨ]́ ‘always’ I have stablished previously in this chapter that there is a metrical high tone and there is a lexical high tone in Wampis. A lexical high tone is insensitive to the metrical patterns of the language, and never shifts the placement of their high tone. Certain morphemes induce tone shift one more to the right (see next section). The accusative =na, for instance, always induces this shift. Compare the realizations of kankat͡ʃi ‘little boquichico (Prochilodus sp.) in the nominative and accusative form with the realizations 268 of /aúhu/ and /pɨńkɨra/ ‘good’ in identical cases: (43) Nominative Accusative kankat͡ʃi → [kaŋkat́͡ʃ] kankat͡ʃi=na → [kaŋkat͡ʃiń] aúhu → [aúhu] auhu=na → [aúhu] pɨńkɨra → [piŋ́kɨran] pɨnkɨra=na → [piŋ́kɨra] From the above examples, aúhu and pɨńkɨra are considered to have a lexical tone: their high tones are unpredictable and they never shift their placement (i.e. the vowel with which they occur is lexically speci"ed to bear them). I mark a lexical high tone with an accent mark V́ in the lexical representations of words that have been identi"ed to bear them with some exceptions: most underlyingly monomoraic verb roots in the database have a lexically-speci"ed high tone, so it is not necessary to write accent marks for them. 6.6. Induced metrical high tone shift in non-predicates The placement of metrical high tone may change if certain morphemes from a subset (see Table 6.4 below) are added. These morphemes have an e$ect on the distribution of metrical high tone in Wampis nouns, adjectives and some adverbs where they can attach. Two and three syllable nouns are, in particular, sensitive to the e$ect of these morphemes for the purpose of the distribution of tone. 269 6.6.1. Two and three mora stems Most nouns with two and three underlying moras are subject to shift their metrical high tone placement when some morphemes are attached. In general, the shift occurs one mora to the right. Table 6.4 lists the morphemes that trigger tone shift. The morpheme =ʃa only a$ects two mora words. Table 6.4. Morphemes that induce high tone shift in two and three mora words in Wampis Morpheme Gloss =hai ̃ Commitative =na Accusative =nVma Locativea -nau Possessive -ru 1sg Possessor -mɨ 2sg Possessor -ri ̃ 1pl/2pl/3 Possessor =kɨ Restrictive =t͡su Inferential =ʃa Additive a Triggers tone shift one mora to the next in bimoraic words. With three and four moraic words it can attract tthe high tone to itself. 270 The examples in (44) demonstrate the change in the placement of tone when these su#xes are added. (44) Nominative Stem + Morpheme Gloss nú.ku [núku] ‘mother’ nuku-ru [nukúr] ‘mother-1SG’ uam.pi.́sa [wampiś] ‘Wampis’ uampisa=na [wampisań] ‘Wampis=ACC’ há.pa [haṕa] ‘deer’ hapa=́ʃa [hapáʃ] ‘deer=ADD’ hɨ.́a ̃[hɨá]̃ ‘house’ hɨa=̃kɨ [hɨɰák] ‘house=RESTR’ sú.a [súwa] ‘female name’ sua=hai ̃[suwáhãi]̃ ‘Sua=COM’ ɨń.t͡sa [ɨńt͡sa] ‘river’ ɨnt͡sa=nama [ɨnt͡sánam] ‘ɨntsa=LOC’ Contrast with examples of the same words when the carry a morpheme that does not conditions the shift of high tone, for instance, the focus =ka. (45) Nominative Stem + =ka ‘focus’ nu.ku [núku] ‘mother’ nuku=ka [nukúka] u.un.ta [úun] ‘old’ uunta=ka [uúnka] ha.pa [haṕa] ‘deer’ hapa=ka [haṕaka] hɨ.a ̃[hɨá]̃ ‘house’ hɨa=ka [hɨák̃a] su.a [súwa] ‘female name’ sua=hai ̃[súwaka] ɨnt͡sa [ɨńt͡sa] ‘river’ ɨnt͡sa=nama [ɨnt͡sánam] 6.6.2. Special tone e&ect with -t ͡ʃi ‘diminutive’ and =nVma ‘locative’ In words with two moras, the diminutive -t͡ʃi lengthens the vowel of the second 271 mora (which immediately precedes it). If we count the lengthened vowel as two moras, the high tone is placed on the second mora. Therefore, the diminutive has the e$ect of shifting the high tone not one but two moras to the right, as indicated in the next example. (46) /mama/ ‘manioc’ → [maḿa] vs /mama-t͡ʃi/ ‘little manioc’ → [mamaat́͡ʃ] Figure 6.7 and Figure 6.8 show the spectrogram with the relevant pitch contours for the words in (47). Time (s) 0 0.5528 0 5000 Fr eq ue nc y (H z) m á m a Figure 6.7. Spectrogram with pitch track for [maḿa] ‘manioc’ 272 Time (s) 0 0.7559 0 5000 Fr eq ue nc y (H z) m a m aá t͡ʃ Figure 6.8. Spectrogram with pitch track for [mamaat́͡ʃ] ‘little manioc’ The e$ect of the diminutive -t͡ʃi does not occur when the underlying mora immediately preceding the diminutive is toneless. In this case, -t͡ʃi does not change the placement of the high tone. (47) /uampisa/ ‘Wampis’ → [wampiś] vs /uampisat͡ʃi/ → [wampiśat͡ʃ] ‘little manoc’ The locative =nVma induces tone shift rightward in two-mora words, but in 273 words with three underlying moras or more it can attract a high tone to itself. In this case, it allows for two high tones to exist in a single word, as it does not suppress the metrical high tone on the stem, but the high tone of =nVma is more prominent Notice that =nVma does not have a lexical tone, as evidenced by the fact that it does not bear a high tone when occurring on two mora words. Figure 6.9 shows the waveform of the word wampisa-numa=ka [wampísnúmka] ‘in Wampis’. The two high tones are indicated by arrows in the second vowel (in the stem) and in the third vowel (in the su#x =nVma). Figure 6.9. Spectrogram with pitch track for [wampiśnúmka] ‘in Wampis’ 274 6.6.3. Four mora stems As far as I can tell, stress and high tone placement in nouns with four underlying syllables or more are mostly predictable with the rules of stress and tone given in the discussion above. No tone placement shift of the kind observed with two and three mora stems is noticed, except if a morpheme with lexical tone is added. The next examples show derivations with four syllable words. Notice that =na does not induce tone shift in these words, unlike what was described in §6.6.1 for two and three mora stems (48) arutama=na ‘spirit of power=ACC’ feet: (aˈru)(taˌma)na vowel elision: arutman output: [arútman] paantama=na ‘manioc=ACC’ feet: (paˈan)(taˌma)na vowel elision: paantam output: [paańtam] 6.6.4. Nominal morphemes that have lexical tone Nominal bound morphemes that are lexically speci"ed for high tone surface as the most prominent phonetically. However they do not completely block the metrical high tone of the stem; i.e. the metrical high tone and the lexical high tone interact. 275 There is a nice phonetic e$ect in which a high tone accommodates to the presence of another high tone: the right most (the lexical high tone of the added morpheme) achieves a higher high pitch, and the one to the left (the metrical stem tone) achieves an intermediate pitch. This is demonstrated in Figure 6.10 with the word ut͡ʃitíʃa ‘we the young too’. The root ut͡ʃi has two moras, so it receives a high tone in the "rst mora [út͡ʃi] (cf. §6.4.1). The Plural Speech Act participant =ti ́(see §11.5.6) bears a lexical high tone. The other morpheme added is =ʃa ‘additive’, which does not have lexical tone. It can be observed that there is one "rst high pitch in the "rst vowel, and then a second pitch in the third vowel that corresponds to =tí. This is realized with a relatively higher pitch than the "rst one. By comparison, the second vowel does not have tone, therefore its pitch is rather low. The last vowel of the word apparently surfaces with a high pitch too, but in this case it is an acoustic e$ect of occurring after a high tone, and it indeed falls through afterward. The actual Hertz estimates for the "rst three vowels in Figure 6.10 are 120.4Hz, 65.1Hz and 145.9Hz, respectively.122 122. Measured at approximately 3/4 of vowel periodicity. 276 Time (s) 0 0.7495 0 5000 Fr eq ue nc y (H z) u ́ t͡ʃ i t í ʃ a utʃi-tí-ʃa ‘child- -Additive’‘We the young too’ 01 2 Figure 6.10. Spectrogram of the word [ut͡ʃitiʃ́a] ‘we the young too’ In the next example, the morpheme =a ́‘First’ is added to =ti.́ =a ́also bears a lexical high tone. The noun stem to which both =ti ́and =a ́are added is [hintińkartin] ‘teacher’, which has a metrical high tone in its second mora, as per the rule of metrical high tone (§6.4.2). The complete word surfaces as [hintińkartintij́a]́ with the most phonetically prominent high tone being the right most (the lexical high tone of =a ́ ‘First’). Figure 6.11 shows that actually more than one high tone can occur in a word. 277 The high tones are indicated by arrows. Note that there is a glide inserted between =ti ́ and =a.́ Time (s) 0 1.969 0 5000 Fr eq ue nc y (H z) h i n t i n k a r t i n t i j a Metrical T Lexical T1 Lexical T2 /hintina-karat-inu-tí-á/ (hin ˈtí)(naˌka)(raˌti)(nuˈtí)á → [hintínkartintíjá]Figure 6.11. Spectrogram with pitch track for [hiti kartintija] ‘we the teachers "rst’ 6.6.5. Grammatical high tone The distribution of high tone is further complicated by the use of autosegmental morphemes that themselves involve the use of a high pitch in the last vowel of the noun. They are shown in Table 6.5. 278 Table 6.5. Autosegmental tone morphemes in Wampis Morpheme Gloss V́# Locative v#́ Genitive Shifts tone one mora rightward Vocative There are at least two purely autosegmental morphemes in Wampis: the locative and the vocative. The vocative blocks "nal-vowel deletion and assigns a high tone to this vowel. (49) Nominative Vocative jat͡súr jat͡surú iat͡su-ru iat͡su-rú ‘brother-1SG.POSS’ ‘brother-1SG.POSS\VOC’ Like the vocative, the genitive is marked with a high tone in the last vowel of the word. (50) Nominative Genitive ʃuar ʃuara ́ahari ̃ ʃuara ʃuara ́ aha-ri ̃ person person\VOC farm-1PL/2PL/3.POSS ‘the person’s farm’ The locative high tone moves the location of the tone one mora rightward, regardless of the number of syllables in the root: 279 (51) Nominative Locative ɨńt͡sa ɨnt͡sa ́ ‘river’ ‘in the river’ 6.6.6. Distribution of tone in predicates The distribution of tone in verbs generally follows the same pattern of tone assignment described so far; but in comparison with nouns, there is a relative increase in the number of verbs with lexical tone. Most verbal roots with one underlying mora and with four underlying moras (or more) have a lexical tone in my data. Bimoraic and trimoraic verbal roots vary, and a good proportion of them have no lexical high tone. As with non-predicates, there are also some a#xes with lexical high tone as well as a#xes that e$ect a shift in the assignment of tone. The su#xes that have been identi"ed as carrying lexical high tone are presented in Table 6.6. Table 6.6. Verbal su#xes that bear a lexical high-tone Su#x Gloss -a(́u) High a$ectedness aktionsart -ta ́ Imperative -hkama ́ Terminative -tahkama ́ Frustrative 280 In addition, some su#xes appear to carry a lexical high-tone, because sometimes their surface occurrence with a high tone is unpredictable. However, other times they surface with not high tone. This behavior is not well understood at the moment. These su#xes are presented in Table 6.7. Table 6.7. Verbal su#xes that sometimes surface with an unpredictable high tone Su#x Gloss -sa Attenuative -mi Hortative -ti Jussive The following principles characterize high tone assignment in Wampis verbs: a. If a verb is monomoraic, it usually carries lexical high tone which surfaces as the most phonetically prominent. Compare the realizations of uɨ ‘go’ which bears a lexical high tone when it occurs with the imperative -ta ́which also carries a lexical high tone. Unlike what occurs with nouns, this time the lexical high tone that surfaces as the most phonetically prominent is the left-most lexical tone. Compare uɨ ‘go’ with uha ‘tell, inform’. The verb uha does not have a lexical high tone, so when it carries the imperative, the imperative surfaces as the most prominent. 281 (52) wɨśata vs. uhakta ́ uɨ-́sa-ta ́ uha-ka-ta ́ go-ATT-IMP tell-INTENS-IMP ‘Go!’ ‘Tell him!’ b. If a verb has two or more moras, it receives metrical tone unless the verb has a lexical high tone (which is not common in verbs of two or three moras): (53) uhakamaji ‘she/he told’ feet: (uˈha)(kaˌma)ji vowel elision: uhakmaji output: uhaḱmaji c. Certain su#xes have lexically speci"ed high tone: e.g. -tá ‘Imperative’: (54) jɛintá jaina-ta ́ help-IMP ‘Help her/him!’ Similar to what occurs with nouns, in cases like (54) both the metrical and the lexical high tones can co-occur. In this case, the lexical high tone is phonetically more prominent than the metrical high tone. Figure 6.12 shows evidence for this statement. 282 Time (s) 0 0.7233 0 5000 Fr eq ue nc y (H z) j a í n t a ́ metrical lexical Figure 6.12. Spectrogram with pitch track for [jɛinta]́ ‘Help her/him!’ 6.7. Summary of stress and tone It seems evident from all the previous discussion that the prosodic system of Wampis does not really "t well in the so-called category of “pitch accent”: stress in Wampis is not only realized via high pitch (i.e. high tone)—this is a traditional de"nition of “accent” (one high pitch per word); rather stress in Wampis is a mix of articulatory cues related to vowel and consonant lengthening in addition to high pitch; 283 stress is also manifested in the vowel elision patterns described in the preceding sections. On the other hand, I have shown that in Wampis there can be more than one phonetic realization of high tone in a word; what is more, some roots, a#xes and clitics of di$erent nature (verbal, nominal, etc) have a lexically-speci"ed high tone. Lexically- speci"ed high tone disregards the stress pattern of the language. In addition to that, high tone has functional load in the language, as it marks case. So, in light of the facts just mentioned, how “pitch-accent” is Wampis? I argue that an answer to that question is circular in principle, because we would have to start by giving a very ad-hoc de"nition of “pitch-accent”, which would not "t any well known cross-linguistic prototype. The present attempt at analyzing Wampis prosody, on the other hand, shows that Wampis can be analyzed in terms of two very well known prototypes of prosodic system: stress and tone. Wampis is not a prototypical stress language nor is it a prototypical tone language, it has features of both ends of the continuum. 6.8. The prosodic word From the discussion above, the minimal prosodic word in Wampis is de"ned as a word with at least two moras (i.e. two TBU), one high tone, one foot and one primary stress. The requirement to have at least one high tone per prosodic word in Wampis is 284 achieved by metrical rules or by lexically specifying the high tone. Wampis codas have an interesting behavior in the sense that they are apparently weightless for the minimal word restriction (i.e. only vowels=TBUs are relevant). This is important for the current analysis, as the stress pattern is centered around the tone bearing unit (i.e. vowels). There are some processes of cliticization where the copula and a demonstrative occur together (in this way the demonstrative relativizes the copula), as in (55). The copula itself can become cliticized to another word when relativized by a demonstrative, as in (56). These constitute interesting cases where the grammatical word and the prosodic word do not match. (55) núu ʃuaŕ anú nu ʃuara a=nu NON.VIS person COP=NON.VIS ‘that who is a person’ (56) núu ʃuaŕanú nu ʃuara=a=nu NON.VIS person=COP=NON.VIS ‘that who is a person’ 285 6.9. Notes on utterance-level prosody 6.9.1. Declarative intonation contour Generally, declarative sentences show a decrease in pitch at the end of the utterance. This decreasing in pitch usually is not dramatically deep—I have only rarely found strong falling contours at the end of utterances, except of course when there is a relatively long pause after the utterance. In this occasion, the last syllable may actually show a deep falling contour and the vowels are pronounced very short and with low intensity. Figure 6.13 shows an instance with a sentence that was uttered as an introduction to a story, therefore there was a long pause before the story began. The sentence is [húu óuhmattɛɛŋ̃ka aúhu tutɛiɲɛiti] ‘this is story is called Auhu’. Note in particular the last sequence of the utterance ([ti]), which can be barely seen in the spectrogram (it is indicated with an arrow). There is also a creaky voice in the transition between the words [óuhmattɛɛŋ̃ka] and [aúhu]. Whether creaky voice is a regular phenomenon in the prosody of words or not remains to be examined more carefully. 286 Figure 6.13. Prosodic contour of ‘this story is called "Auhu"’ 6.9.2. Non-declarative intonational contour Interrogative utterances do not show a "nal rising in pitch, but the question word receives the highest level of pitch. Imperative utterances may have a "nal rising pitch if the tone of the imperative su#x -tá surfaces; i.e. if there is no other lexical tone in the verbal root. Otherwise, a command utterance does not necesarily end with a high 287 contour pitch. A similar pattern is shown in hortatory utterances with the su#x -mí. An example of an interrogative intonation contour is given with a question [ií urúk puhúmɛinkit] ‘how can we live?’ in Figure 6.14. The highest pitch is received by the pronoun íi ‘1pl’ (which is pragmatically focused in this sentence) and then by the question word uruka [urúk] ‘how’. Time (s) 0 1.039 0 5000 Fre qu en cy (H z) íi urúk puhúmɛinkit Figure 6.14. Prosodic contour of ‘how can we live?’ 288 6.9.3. Prosody of the anemamu, the protocolar salutation ceremony of the Wampis The anemamu (anɨma-mau ‘salute-NMLZ’) is the intense formal salutation ceremony of the Wampis. It is a fascinating event that is still perform among these people in special occasions such as in calendar festivities (Peru's independence day, celebration of the community's foundation, etc) or when a stranger (i.e. non-Wampis) comes to the community and ask for permission to stay. Usually, each community has a performer or performers that are actually admired because of their skills in the performance and their knowledge of the traditional protocol. In the past, this ceremony was held when the men of a war party were invoked to the house of the party leader or when someone came to visit or ask for help. The ceremony put together the leader of the visiting party and their host. They spoke in turns to each other with their mouths covered by the left hand, moving their spears with the other hand towards each other and stomping with their feet rhythmycally, speaking in a forceful manner and at a fast rate. Genealogies were recited and then a discourse stated the motive of the visit. After that, each men in the visiting party took their turns to salute the host and the same ceremony was repeated. The analysis of an anɨmamu speech that was rehearsed for me shows that, unlike 289 declarative and non-declarative speech, the intonational pitch contour of the greeting ceremony is maintained through the utterance, with shorter pauses. There is no signi"cant rising of the pitch at the phrase level, except if an element of it is pragmatically highlighted. Figure 6.15 shows the phrase [wiḱʲa kaḿɨ aḿinkʲa sɨɨ ́ tah́amɨ] ‘I tell you thanks’ in the salutation speech. We can see that the pitch contour does not fall through at the end. The highlighted word is sɨɨ ‘thanks’. Time (s) 0 3.108 0 5000 Fre qu en cy (H z) wíkʲa kamɨ ámiŋkʲa sɨɨ ́ táhamɨ Figure 6.15. Prosodic contour of ‘I tell you thanks’ 290 6.10. Nasal prosody in Wampis As we saw in Chapter III, Wampis possesses nasal vowels that contrast with their oral counterparts in roots and a#xes. (57) nuɨ → [núwɨ] ‘clay’ nuɨ ̃ → [núwɨ]̃ ‘his wife’ ũha → [ú̃ha] ‘open’ uha → [úha] ‘tell’ As described for Awajun by Overall (2007: 250 and $.), the domain of nasality in Wampis is de"ned by a sequence of nasal vowels, glides and nasal consonants. It is possible that the velar fricative /h/ also belongs in the nasality domain, though this is not very well understood yet. Nasality may originate in a root or an a#x, and then spreads over vowels, nasals and glides rightward and leftward until it "nds an obstruent (non-glide, non-nasal). Examples (58) and (59) illustrate this point with close phonetic representations. (58) [nũiɲ̃ã]́ nu-i-̃ia NON.VIS=LOC=ABL ‘from there’ 291 (59) [ɲó̃õ] jaũ ‘yesterday’ In (58), the locus of nasality is the locative =i.̃ It spreads its nasality over the morphemes located to the right and to the left. The "rst vowel of the ablative -ia is in a position to create a glide. That this glide is nasalized is corroborated by the fact that it surfaces as the palatal nasal [ɲ]. In (59), the nasality of the vowel of the root spreads leftward since that is the only direction where it can spread. Notice that the word phonemically is /iaũ/ and therefore the "rst vowel becomes a glide. The glide is then nasalized following the same pattern observed in (58), thus surfacing as [ɲ]. 6.10.1. Alternation between Ṽ and VN and between Ṽ and V One of the most fascinating problems of de"ning the nasal domain in Wampis is the fact that nasal vowels can alternate with oral vowels and with VN sequences. The perplexing alternations between nasal and oral prosodies have also been mentioned for other Jivaroan languages (Turner 1958b; Pike & Larson 1964; Overall 2007; Payne 2008). I have found that the alternations between oral and nasal vowel productions do not correspond to speci"c groups (i.e. female vs male speakers, dialectal variation) but rather depend on the individual speaker: some speakers frequently nasalize vowels, 292 some not, and not infrequently some speakers just vary between producing nasal and oral realizations. Nasality is di#cult to de"ne in the environment of a nasal consonant, but the evidence at hand almost shows that when a nasal consonant surfaces as a correlate of phonemic nasality, it suppresses other features associated with nasality, such as spread. This can be seen in derived words, compare: (60) [jóũnt͡ʃuk] iaúnt͡ʃukɨ ‘long ago’ (61) [ɲó̃õ] iaṹ ‘yesterday’ In (60) we have the word iaúnt͡ʃukɨ ‘long ago’ which is historically related to iaũ ‘yesterday’. Notice that in iaúnt͡ʃukɨ nasality does not spread over contiguous vowels and glides, as occurs with iaũ in (61): the surface glide of iaunt͡ʃukɨ never becomes [ɲ]. Nasality seems to have the capacity to surface in di$erent forms regardless of the morphological environment. This alternation seems to be pretty idiosyncratic. It is very possible that this idiosyncrasy has a regular historical explanation that is not well understood from the synchronic data. In general, issues concerning nasality, its surface 293 realizations and correlates (e.g. spread) need more detailed research. 6.10.2. Nasality as high pitch Another fascinating and typologically interesting feature of Wampis nasality is the fact that an underlying nasal morpheme can be realized as an oral vowel with high pitch. This seems to be a corollary of nasality apparently losing its functional load, thus the need of the language to assign an alternative acoustic cue to phonemic nasalization. Thus, third person singular possessor and third person same subject switch reference morphemes are marked with a phonemically nasal autosegmental morpheme. In the next example, we can see how they surface with a high pitch: (62) [naŋkí hu-kí] nanki ̃ hu-ki ̃ spear\1PL/2PL/3.POSS carry-WHILE.MOVING+3SG.SS ‘he having carried his spear […]’ The underlying nasal of this su#x surfaces when other morphemes are added. For instance, the focus =ka triggers the surface realization of the nasal as [ŋ] with nanki ‘spear’ (the relevant part is underlined): [naŋkiŋ̃kʲa] ‘his spear’; and the conditional =ka likewise triggers the surface realization of a nasal consonant with the verb stem huki: [huki ̃ŋ́kʲa] ‘if he had taken . . .’ 294 Di$erent works on nasality in Jivaroan have proposed a nasal archi-phoneme N to account for the alternations described above (Turner 1958a; Overall 2007; Payne 2008). This may be an excellent way for understanding the historical development of the alternations. Nevertheless, it is not very signi"cant for accounting for the synchronic behavior just observed: it does not completely describe all the di$erent expressions of nasality in Jivaroan languages and the issue of their partial unpredictability still remains to be more thoroughly explored in Wampis. In summary, nasality in Wampis is a domain that may "nd phonetic substance as a nasal itself, a sequence of oral plus nasal (VN), sometimes even an oral vowel, or a high pitch. 295 CHAPTER VII INTRODUCTION TO MORPHOLOGY 7.1. Introduction This brief chapter provides an overview of the types of morphological units found in Wampis, including the discussion of certain issues that pertain to multiple parts of speech. The discussion in this chapter gives the appropriate background for the detailed analyzes of word classes and their properties presented in the upcoming chapters. This chapter has the following structure: §7.2 describes the morphological pro"le of Wampis in typological terms. This is followed by a discussion about the levels of the root, stem and word in Wampis in §7.3. §7.4 discusses relevant notions related to word classes and roots. In §7.5, there is a brief discussion about clear a not-so-clear cut boundaries between word classes. §7.6 presents types of morphemes according to their bound or free nature. Next, §7.7 looks at processes of reduplication, that involve copying complete or partial portions of words. §7.8 describes compound words. Finally, §7.9 o$ers a brief discussion of particles and how they will be treated in the grammar. 296 7.2. Wampis morphological typology According to Comrie (Comrie 1989), the morphological typology of words in a language can be described with reference to an index of synthesis and an index of fusion. The index of synthesis concerns the number of morphemes that can occur in a word. The index of fusion concerns the degree of merging or overlay of meanings in one morpheme, and how easily morphemes can be separated from each other. Wampis major word classes, nouns and verbs, vary a little in regard to these two parameters. Concerning the index of synthesis, Wampis nominal words are polysynthetic and predicate words are highly polysynthetic. With regard to the index of fusion, they are agglutinative with some degree of fusion. Closed classes of words that somewhat share some morphology with nouns (but not totally, as we will see in Chapter VIII), such as pronouns and determiners, are generally polysynthetic and agglutinative. Adjectives in Wampis exhibit some morphology but they rarely contain more than two or three morphemes (i.e. the root plus a bound morpheme) at any given time.123 Adverbs are mostly isolating in Wampis, though a small subclass of manner adverbs receives person 123. It will be seen in Chapter IX that adjectives can host case and postpositional clitics. Because of this, adjectives look like they have more complex morphology than they really have. Clitics are hosted by the last element of the NP (which usually happens to be the adjective, as they occur post-nominally). 297 marking that agrees with the grammatical person of the verb present in the clause. It is shown in this chapter that Wampis has a number of items that clearly are a#xes and some other forms that are clearly independent words (e.g. nouns). In addition, Wampis possesses clitics, including postpositional forms that operate at the level of the phrase and are phonologically dependent on a host. One of the salient features of Wampis is the copious morphology with which a word can occur, which is sometimes “obscured” by the pervasive processes of vowel elision. The language has many a#xes (mostly su#xes) as well as a good number of clitics, so long words standing as utterances are not infrequent. Here is only an example of the beautiful and complex ways in which the morphology of the language works to create long expressions in Wampis: (1) t͡ʃit͡ʃaḿrumɛint͡ʃouwɛithʲɛi t͡ʃit͡ʃa-ma-ru-mai-inu-t ͡ʃau=aita-ha-i speak-REFL-APPL-POT-NMLZ-NEG.NMLZ=COP-1SG.SBJ-DECL ‘I cannot advice [him]’ (Lit: ‘I cannot be a speaker myself for him’). 7.3. Root, stem and word There is a long tradition in linguistics to recognize di$erent elements within words. For instance, within the American structuralist tradition, Bloom"eld (1933) recognized the importance of distinguishing between roots and other a#xes. 298 Subsequent works have distinguished between the analytical levels of root, stem and word. In this dissertation, a root is de"ned as an unanalyzable, morphologically simple form that contains the main portion of the semantic content of the word (Payne 1997). Examples of verbal and nominal roots are presented in (2) and(3), respectively. Note that verbs need a#xation to be used in utterances,124 except when they occur in auxiliary constructions, whereas noun roots do not necessarily need some type of a#xation to be pronounceable words. (2) uaina ‘see’ uɨ 'go' taka 'work' kanu 'sleep' (3) ʃuara ‘person’ kɨnkɨ 'wild potato' hɨmpɨ 'hummingbird' panki 'boa constrictor' A stem is a root or a root plus any derivational su#xes that can further receive in/ectional morphology. For instance, in Wampis the verbal root uaina ‘see’ can carry the causative pre"x V-; the resulting stem can then receive in/ectional su#xes before 124. A "nite verb requires to be marked for aspect, tense, person and mood. 299 being used in an utterance. (4) i-uaina CAUS-see ‘make discover, show’ A word in Wampis is made up of one morpheme or a combination of morphemes. The stem i-uaina in (4) needs to occur in a fully in/ected word (in the case of verbs, they need obligatory aspect, tense, person and mood morphology) to be pronounceable. An example is given in (5), where the verb carries the intensive -ka125 and the person marker -hamɨ ‘1SG>2SG’ (where the declarative -i is reduced and “fused” to the preceding vowel ɨ of -hamɨ). (5) iwɛińkattahmɨ i-uaina-ka-tata-hamɨ CAUS-see-INTENS-DEF.FUT-1SG>2SG+DECL ‘I am going to showing you.’ In Wampis, a word can consist of a single root, a stem, or an in/ected stem. For instance, ʃuara ‘person’ is monomorphemic and can be used in a phrase in a form where root, stem and word all converge: 125. The intensive -ka is an aktionsart su#x. Aktionsart su#xes are used to the marking of aspect in Wampis. This property is analyzed in Chapter XIV. 300 (6) núu ʃuár ah́aku nu ʃuara a=hak-u NON.VIS person COP=HAB.PT-NLMZ ‘That used to be a person.’ It is important to state that certain morphosyntactic categories in Wampis are marked by the absence of morphology. For instance, exclamative mood is realized with the stem verb plus person and tense markers, but it lacks additional phonetic material (7), unlike declarative mood which is marked with -i ‘declarative’ (8): (7) ɛiḿkʲamah aima-ka-ma-ha-∅ "ll-INTENS-REC.PT-1SG.SBJ-EXCLAM ‘I "lled [the basket]!’ (exclamative) (8) ɛiḿkʲamhɛɛ aima-ka-ma-ha-i "ll-INTENS-REC.PT-1SG.SBJ-DECL ‘I "lled [the basket]’ (declarative) Other categories in Wampis are marked by failure of vowel elision in a stem, when it would be otherwise expected (see §6.4.4 for details on vowel elision). In most cases, the non-elision of the vowel is accompanied by either of two supra-segmental phenomena: high tone or nasalization. For instance, the vocative is characterized by blocking vowel elision and attracting a high tone to the last vowel of a word. For comparison, a simple nominative possessed form is shown "rst in (9), followed by the 301 vocative in (10): (9) jatsúr iatsu-ru brother-1SG ‘my brother’ (nominative) (10) jatsurú iatsurú brother-1SG\VOC ‘my brother!’ (vocative) 7.4. Grammatical categories and roots There is a good deal of literature about grammatical categories. The early structuralist tradition emphasized that word classes should be de"ned purely in distributional terms (Bloom"eld 1926). To quote Jespersen: “some grammarians. . . maintain that the only criterion should be the form of the words” (Jespersen 1992 [1924]: 60). The morphosyntactic distribution of words are diagnostic, but distribution alone does not su#ce to explain why a language possesses a certain category (Payne 1999b). What is more, a distributional analysis may reveal “a myriad of classes, and gives us no method for deciding between parts of speech and minor syntactic categories” (Croft 2001: 83). Thus, while distributional morphosyntactic properties are useful to identify categories of words, I also take a more functional approach, relating 302 the categories found in Wampis to conceptual categories construed via cognitive models of perception and memory: “the lexical and grammatical resources of a language are therefore not semantically neutral—inherent to their nature is the structure of conceptual content for symbolic reasons” (Langacker 1988: 63). Haspelmath summarizes a relatively long tradition of typologically-oriented works: “languages cannot be compared directly on the basis of their grammatical categories. We need a tertium comparationis that is not language-particular, but is universally applicable” (2012: 114). The most universally predictive criteria to establish word-class membership seems to be semantic (Givón 2002a), especially the criterion of temporal stability for nouns (as opposed to verbs).126 According to Croft (2001), words that are lexicalized as prototypical nouns refer to concepts that occupy the most time-stable end of a continuum of temporal stability, whereas prototypical verbs occupy the other end. Prototypical nouns often belong to the semantic class of objects, whereas prototypical verbs belong to the semantic class of actions. The semantic class of property concepts 126. The concept of time-stability is less useful with certain semantic sub-classes of verbs (e.g. stative verbs like ‘live’ are conceivable as predicating a very stable process) and adjectives (or property words if there is no adjective category in the language). 303 (Dixon 1982) is somewhat in between the ends of the continuum, and are often (though not universally) lexicalized as a distinct word class of adjectives. “Object roots” are de"ned as roots that typically denote an animate or inanimate object. “Action roots” are de"ned as roots that denote a prototypical volitional action. “Property roots” are de"ned as roots that denote a property such as age, dimension or value. Following Croft's’s 2001 ideas, roots in Wampis can be classi"ed as “object-roots”, “action-roots” and “property-roots”. Continuing with Croft’s ideas, he identi"es an additional functional parameter relevant to parts of speech, namely the propositional act functions of reference, modi"cation and predication. According to said scholar, OBJECT, ACTION and PROPERTY enter in a relationship with the propositional act-types reference, predication and attribution. The semantic categories of objects, properties and actions are, respectively, the "typological prototypes of referring, attributive, and predicating constructions" (Croft 2001: 87). The combination of these concepts will help de"ne the prototypical members of part of speech classes for Wampis. Table 7.1 shows simple examples of these act-types and root classes. 304 Table 7.1. Examples of roots and propositional act-types in Wampis Reference Modi"cation Predication Object nua ‘woman’ Property pɨńkɨra ‘good’ Action uɨ ‘go’ Thus, “object-roots” are prototypically used referentially, “property-roots” are prototypically used as attributes, and “action-roots” are prototypically used as predicates. In (11), the “object-root” nua has a referential function (it refers to the object ‘woman’), the “property-root” pɨńkɨra has a modifying function (it modi"es ‘woman’), and the “action-root” uɨ has a predicating function (note it has to be in/ected to be used): (11) óu núwa pɨŋ́kɨr wɨáwɛɛ au nua pɨńkɨrauɨ-a-ua-i DIST woman good go-IPFV-3.SBJ-DECL ‘That good woman is going.’ Of course, in Wampis as in many other languages, not all nouns denote physical objects (there can be abstract objects/concepts), not all verbs denote volitional actions, and not all adjectives denote properties (or certain properties are not denoted by adjectives). These terms refer to prototypes of each major category. Understanding grammatical categories in terms of prototypes allows us to choose key lexemes to start 305 checking the morphosyntax of a language. This does not mean that I will pre-judge that categories such as “verb” or “adjective” exist in Wampis a priori; rather, I take this approach to grammatical categories as both methodologically helpful to establish prototypical members of major classes and theoretically grounded in the study of human cognition (Langacker 1976; Miller & Johnson-Laird 1976; Langacker 1987). However, the grammatical categories of a language need to be determined by morphosyntactic properties, to quote Croft: “categories in a particular language are de"ned by the constructions of the language” (Croft 2001: 85). In that regard, word classes often cannot be equated across languages. Grammatical categories de"ned for di$erent languages depend on language-speci"c criteria (Croft 1984; Lazard 1992; Croft 2001; Dryer 2006; Haspelmath 2007). Accordingly, in the next chapters, for each major word class, I will present a discussion in terms of their prototypical members, but I will also provide the relevant morphosyntactic evidence that have helped de"ned such categories in Wampis. As an example, we have just seen that the concepts of ‘woman’ and ‘good’ have the prototypical functions of object reference and property modi"cation, respectively. There is no special morphology associated with these prototypical functions, as shown in (11) 306 before. However, when ‘woman’ or ‘good’ occur in predicative function, they occur in a di$erent construction. In other words, when they are used in a function that is di$erent from their prototypical one, it is “marked” somehow by the language. In this speci"c case, a copular construction occurs to allow nouns and adjectives to function predicatively: (12) óu núwɛiti au nua=iti DIST woman=COP.3+DECL ‘That is a woman.’ (13) óu pɨŋ́kɨrɛiti au pɨńkɨra=aiti DIST good=COP.3+DECL ‘That is good.’ Thus, the fact that nouns and adjectives need to be morphosyntactically treated to become “similar” to verbs (by adding a copula) to be used predicatively, provides a "rst piece of evidence for stating that there is a basic di$erence between verbs, and nouns and adjectives in Wampis (a verb does not need a copula to be used predicatively). Detailed evidence with plenty of examples will be presented in the respective chapters where I de"ne the Adjective (Chapter IX), the Noun (Chapter X) and the Verb (Chapter XII) in Wampis. In addition, Wampis also possesses minor classes, 307 which are analyzed in Chapter VIII (I include an analysis of Adverbs in this chapter, see §8.6). 7.5. Fuzzy boundaries and clear-cut boundaries As often happens in languages around the world, the boundaries between certain categories are sometimes untidy in Wampis. For instance, nat͡sa ‘single (a person without a partner)’ can be used referentially, but it can also be used attributively, in which case it means ‘young’. The same root is used in the verb nat͡sa-ma ‘be timid, be embarrassed’, although in this case it carries the verbalizer -ma (see §12.6.2). Table 7.2 provides a sample of roots that exhibit “untidy” boundaries, i.e. roots which can be used in more than one way (referentially, attributively or predicatively) without marking the root with special (i.e. derivational) morphology. Examples like the ones given in Table 7.2 actually do not abound in Wampis. In fact, at least with reference to the two cross- linguistically major categories of word classes, nouns and verbs, ambiguous or labile categoriality is usually not an issue in Wampis. In this language, certain roots are usually used predicatively and certain other roots referentially: there is not a lot of cross-over or overlap. In more informal words, object-naming roots are not as freely used in predicative function—or vice versa—unlike, for instance, in English. 308 Table 7.2. Examples of roots used in di$erent propositional act functions Root Reference Predication Attribute uunta ✓ ‘adult’ ✓ ‘big’, ‘old’ nanki ✓ ‘spear’ ✓ ‘"sh with hook’ wɨɨ ✓ ‘salt’ ✓ ‘add salt’ tanku ✓ ‘domestic animal’ ✓ ‘tamed’ munt͡su ✓ ‘breast’ ✓ ‘be breastfed’ sɨnt͡ʃia ✓ ‘brave one’ ✓ ‘strong’ hɨã ✓ ‘house’ ✓ ‘arrive’ a Also used as an intensi"er adverb. Adapting the terms proposed in Lehmann (2008), it can be said that Wampis is a language with fairly high root categoriality with regard to nouns and verbs. Though zero derivation exists, it is not a common process in Wampis, and usually derivational morphemes (nominalizers and verbalizers) mark the change to a di$erent grammatical class. On the other hand, the distinction between the adjective and noun categories is more subtle. Many of the words that are used in attributive function in Wampis are actually formal nominalizations. Therefore, it is important to distinguish between roots that have undergone a derivational process (namely nominalization) as umi-inu in (14) 309 but can in that nominalized form be used attributively, from roots that are used for attributive functions without the need of a derivational morpheme, as pɨńkɨra in (15), which can be considered a “true” adjective. (14) út͡ʃi umiń wɨáwɛɛ ut͡ʃi umi-inu uɨ-a-ua-i child complete-NMLZ go-IPFV-3.SBJ-DECL ‘The dutiful child is going.’ (umi-inu ‘someone who completes’, translated as ‘dutiful’) (15) út͡ʃi pɨŋ́kɨr wɨáwɛɛ ut͡ʃi pɨńkɨra uɨ-a-ua-i child good go-IPFV-3.SBJ-DECL ‘The good child is going.’ More detailed analysis of the structural di$erences between nouns and adjectives is presented in Chapter IX. There, I propose that adjectives constitute a di$erent class from nouns and verbs in Wampis. Some roots can be used either attributively or adverbially. For instance, sɨnt͡ʃi can be used either attributively as in (16) or as an adverb modifying a predicate as in (17). The only way to know when sɨnt͡ʃi works attributively or adverbially, in this case, is by looking at the syntactic context where sɨnt͡ʃi occurs. (16) t͡samarɛɛ́n maãńin sɨńt͡ʃikʲa ah́aku t͡samarainta maã-́nai-inu sɨnt͡ʃi=ka a=hak-u Tsamarainta kill-RECP-NMLZ strong=FOC COP=HAB.PT-NMLZ ‘Tsamarain used to be a strong warrior.’ 310 (17) sɨńt͡ʃi júwawɛɛ sɨnt͡ʃi iu-a-ua-i a.lot eat-IPFV-3.SBJ-DECL ‘He eats a lot.’ In (16), sɨnt͡ʃi functions attributively, modifying the noun. An evidence that it is modifying the noun (apart from the semantic interpretation) is the presence of the focus marker =ka, which attaches to the last element of the NP (so sɨnt͡ʃi is inside the NP ‘a strong warrior’, modifying the noun head). On the other hand, in (17) sɨnt͡ʃi functions adverbially, modifying the predicate—this is clear in the example, as there is no overt NP in which sɨnt͡ʃi could be taking place, it necessarily modi"es the verb. Though the semantically and propositional-act based theory of parts of speech proposed by Croft (1984; 2001) is helpful to establish prototypical elements of major parts of speech that can be compared cross-linguistically, as we saw earlier, more structural (morphological and syntactic) criteria need to be invoked to attest and de"ne major word classes in a language's own terms. In other words, we need operational de"nitions by which to connect the theoretical categories to the variation found in the data. Table 7.3 summarizes some important morphosyntactic features that are related to prototypical members of each major word category in Wampis. 311 Table 7.3. Morphosyntactic properties of major word categories in Wampis Properties Noun Adjective Verb Adverb Take copula clitics when functioning as predicate Yes Yes No Yes Head of NP Yes No No No Replaced by pro-forms Yes No No No Modify by uɨantu ‘Group’ Yes No No No Can be possessor Yes No No No Take Tense, Person and Mood morphology No No Yes No Take aspectual and aktionsart morphologya No No Yes No Take morphological plural marking No No Yes No Take switch-reference marking No No Yes No Can be verbalized with -ma ‘verbalizer’ Yes Yes No Someb Can be nominalized with Set I or Set II nominalizers No Noc Yes No Can be adjectivalized with -rama ‘Adjectivalizer’d Yes No Yes Yes It is gradable (occurs in comparative constructions, superlative, etc) No Yes No No a Aspectual morphology occurs at a di$erent level of the verb piece (more towards the root). Tense, Person and Mood occur in the periphery of the verbal piece (away from the root). b Mostly restricted to temporal adverbs. c Some adjectives derive their antonyms using the negative nominalizer -t͡ʃau. d The adjectivilazer -rama (§9.8.3) is not very productive synchronically (there are plenty of examples of it in the lexicon, though). 312 Note that adverbs are also included in the table, since it is a major category in other languages, although in Wampis the category of adverbs is a semi-closed "not- noun, not-verb, not-adjective" sort of category. Detailed analysis and criteria for each of these (and other) categories are found in the subsequent chapters. 7.6. Bound and free morphemes Bound morphemes are morphemes that cannot occur as words on their own. Free morphemes are morphemes that can occur on their own. Bound morphemes in Wampis include some roots, a#xes and clitics. With regard to major root classes in Wampis, noun and adjective roots are free. Most verb roots in Wampis are bound in most cases, because they cannot occur in the clause without certain morphology. Finite verbs need to be marked for aspect, person, tense and mood. There are some exceptions to this, most notably speci"c (and very infrequent) auxiliary constructions where the auxiliary verb carries the in/ection and the main verb appears in its bare root form. 7.6.1. A0xes: derivation and in2ection There is only one pre"x in Wampis (the causative V-, see §13.2.2.1.1); the rest of a#xes are su#xes. Following a traditional linguistic distinction, a#xes can be divided 313 into derivational and in/ectional formatives. Derivational a#xes create an in/ectable stem from a root or another stem. In/ectional su#xes are required by the syntactic environment where the root appears and add a particular grammatical function to the latter. The di$erence between in/ection and derivation, while useful in some languages, is not always clear-cut in others (Payne 1985). In Wampis this distinction is not clear in some parts of the grammar. Certain Wampis verbal su#xes probably constitute the best example of morphology where the borderline between derivation and in/ection becomes fuzzy. Following Overall (Overall 2007), I will label these su#xes as “Aktionsart” (see §13.3.2 for details). Table 7.4 lists these su#xes. Table 7.4. Aktionsart su#xes in Wampis Su#x Gloss -a(́u) ‘High a$ectedness (of Patient or Location)’ -i ‘Low a$ectedness (of Patient or Agent of instransitive verb)’ -ka ‘Intensive’ -ki(ni) ‘Do while moving’ -ra ‘Distributed action’ -sa ‘Attenuative’ -ri ‘Do in proximity/while coming’ -u ‘Do in another location’ 314 The su#xes presented in Table 7.4 have a derivational function: they create new stems with (sometimes subtle) di$erent meanings from verbal roots. The following examples illustrate the derivational use of some of the above su#xes: (18) wɨ ‘go’ → wɨ-ka ‘walk, travel’ (with intensive -ka) t͡ʃit͡ʃa ‘speak’ → t͡ʃit͡ʃa-ra ‘advise’ (with distributed action -ra) ihu ‘stab’ → ihu-ra ‘grind, squeeze’ (with distributed action -ra) hapi ‘pull, drag’ → hapi-ra ‘pull up’ → hapi-ra-ki ‘pull up to carry’ (with dis- tributed action -ra + do while moving -ki) ahũ ‘throw in the water’ → ahũ-á ‘knock down’ (with high a$ectedness) Most verbs very frequently occur with one of these su#xes to appear in certain morphosyntactic contexts: with most past tense markers and in the future, the aktionsart su#xes are usually required. And to mark perfective aspect, the verb plus aktionsart stem is obligatory. For instance, the verb kankɨ frequently occurs with the intensive su#x -ka and must occur with this su#x for ‘just-done actions’ marked with the perfective (note that there is no apparent new meaning127 added in these cases: the 127. But this is open to question. It is very di#cult to know for sure whether the aktionsart is adding some meaning (other than grammatical) to the root in this case. Not all aktionsart su#xes are available for every verbal root, but sometimes the verb can take a di$erent aktionsart su#x. For instance, when adding the attenuative -sa to kankɨ, the meaning of the action was explained to me as being a gentle or slow rolling. 315 aktionsart su#x is "lling an morphosyntactic requisite for the verb to be used in perfective sense): (19) kaŋkɨḱɛi ̃ kankɨ-ka-i ̃ roll-INTENS-3.PFV ‘It just rolled.’ cf *kankɨ-i ‘roll-3.PFV’ But in imperfective environments, the aktionsart su#xes cannot occur: (20) kaŋkɨáwɛi kankɨ-a-ua-i roll-IPFV-3.SBJ-DECL ‘It is rolling’ cf. *kankɨ-ka-a-ua-i ‘roll-INTENS-IPFV-3.SBJ-DECL’ So it seems that aktionsart su#xes, apart from a derivational function, have a more grammatical function akin to in/ection, in the sense that they are obligatory in certain morphosyntactic environments, or otherwise the expression is agramatical. In turn, aktionsart su#xes are in complementary distribution for instance with the imperfective su#x, and their co-occurrence is prohibited. From the preceding discussion, it can be seen that some morphemes in Wampis (of which the aktionsart su#xes provide an illustration) pose problems with regard to their classi"cation as either derivational or in/ectional. Table 7.5 summarizes the behavior of the aktionsart su#xes with regards to some features that are traditionally 316 associated with canonical behavior of in/ectional and derivational su#xes (based on Doris L. Payne (1985) and Haspelmath (2002)) Table 7.5. Canonical in/ection and derivational properties with relation to aktionsart su#xes Canonical In/ectional Canonical Derivational Comments related to aktionsart su#xes Correlates with something elsewhere in the syntactic structure Does not correlate with something elsewhere in the syntactic structure Do not correlate with something elsewhere in the syntactic structure (e.g. not agreement markers) Obligatory Optional Obligatory in certain morphosyntactic contexts Yield same concept as base Yield new concept Both (sometimes subtle new concept) Relatively abstract meaning Have non-predictable, more concrete meaning “Obscure”, abstract meaning in some cases Participate in paradigm of oppositions Do not participate in a paradigm of oppositions In complementary distribution with imperfective Do not change category of word May change category of word Do not change category of the word Result in new lexical terms Result in new lexical terms They can derive new stems (with new meaning) Occur towards root Occur towards word edge Occurs towards root Non-recursive Recursive Can be recursive Thus, in this dissertation what is meant by “derivation” versus “in/ection” is 317 really a continuum (cf. Bybee (1985)). While I will try to characterize morphemes by referring to their derivational or in/ectional properties, it must be borne in mind that certain elements simply may be used as both derivational and in/ectional in certain contexts. In relation with the above idea, an important note to be considered is that in Wampis there is a good number of nominalization processes that partake in the grammatical structure of clauses. What is more, grammatical nominalizations have a signi"cant role in the grammar of Wampis. Some forms that can be considered the output of “derivational” processes have acquired grammatical status in modern Wampis. For instance, the nominalizer -u can derive agentive nouns from a verb stem: (21) nɨkóu nɨka-u know+IPFV-NMLZ ‘wise person’ (‘someone who knows’) The same nominalizer occurs in a (historically) complex past tense where it occurs in combination with the habitual past =hak: (22) wampiśhɛɛ ̃awarúnhɛɛ ̃maãńihakaru uampisa=hai ̃ auaruni=hai ̃ maã-́nai=hak-ara-u Wampis=COM Awajun=COM kill-HIAF-RECP=HAB.PT-PL-NMLZ ‘The Wampis and Awajun used to "ght’ 318 7.6.2. Clitics Clitics attach to a host and form a phonological word with it. Several phonological, morphological and syntactic parameters have been proposed for de"ning clitics, and for di$erentiating them from a#xes (Zwicky 1977; Zwicky & Pullum 1983; Aikhenvald 2003; Anderson 2005; Schiering 2006). Within this work, a clitic is de"ned as a grammatically independent but phonologically bound morpheme that functions at the level of the phrase. As opposed to a#xes, clitics do not operate only over single words. As opposed to particles, which can occur on their own, clitics do not phonologically occur on their own, although cross-linguistically it is common that clitics have free-standing counterparts, especially clitics historically derived from pronouns. Clitics do not constitute a single word class in Wampis in terms of prototypes, morphological distribution, grammatical function or semantics—they are, at most, a phonological class of items. However, for sake of completeness, they are mention in this section, as they are bound morphemes. Table 7.6 lists morphemes analyzed as clitics within this dissertation, with a cross-reference to the sections of the grammar where they are analyzed. 319 Table 7.6. List of clitics found in Wampis Case clitics Cross-reference =na Accusative §10.4.6.2 Postpositional clitics =nVma Locative §11.5.1.1 =(n)i ̃ Locative §11.5.1.2 =ia Ablative §11.5.2 =ni Allative §11.5.3 =hai ̃ Comitative §11.5.4 =a ́ First §11.5.7 =ti ́ Plural speech act participant §11.5.6 =á A$ective §11.5.8 Mood and discourse-related clitics =kɨ Restrictive §18.3 =ka Focus §18.4 =ʃa Additive §18.5 =ʃa Speculative §18.6 =t͡su Inferential §18.7 =ka Interrogative §18.8 =api Sudden realization, Tag question §18.9 =hama Mirative §18.10 Copula clitics =aita~=ita Copula (SAP) §17.5.3.2 =aiti~=iti Third person copula §17.5.3.2 320 7.7. Reduplication As the name indicates, reduplication is the process by which the root or stem (or a part of it) is repeated. Reduplication of verbs and adjectives is relatively productive in the Wampis language. In addition, numerals use partial reduplication to convey a distributional meaning (see §8.8.1). Reduplication can be interpreted as in/ection (to convey a grammatical function such as plurality) and as derivation (to create new words). The basic principles of reduplication in Wampis are: • The reduplicated material usually forms its own prosodic word separate from the base word.128 • The reduplicated material always appears before the base word in partial reduplication. • Nothing ever stands between the reduplicated portion and the base word. Two types of reduplication can be distinguished in Wampis: total reduplication and partial reduplication. Partial and total reduplication apply to both verbs and nouns; but there is one type of partial reduplication that is triggered by a su#x that is 128. Some forms show that reduplicated words form one prosodic word; for instance tɨpɨtpɨtu ‘decline’ clearly is a reduplication from the root tɨpɨ ‘lie.down’. Such examples are already lexicalized words in the language. 321 particular to verbs. Total reduplication, as the name indicates, copies the entirety of the surface realization of a base word: (23) aŋkán aŋkán ankanta ankanta free free ‘dispersed’ Partial reduplication copies a portion of the base word. The portion that is copied is all the phonetic material until the second mora. The following example shows a partial reduplication with a numeral: (24) Underlying lexeme: tikit ͡ʃiki Output: tikí tikít ͡ʃik tiki tikit ͡ʃiki one one ‘each one’ There is a su#x -kaua that triggers reduplication with verbs and that I gloss as ‘reduplicative’. This su#x occurs in subordinate verbs and indicates that the action of the verb is repetitive (see §19.9.4 for details). Reduplication with -kaua is based on the same idea of partial reduplication, but the su#x -kaua is added to the verbal stem. The next examples illustrate reduplication with -kaua: 322 (25) awá awátkawa ̃ aua auatu-kaua ̃ REDUP hit-REDUP\3SG.SS ‘hitting and hitting. . .’ (26) miɲí miɲíɲakua ̃ mini mini-ina-kaua ̃ REDUP come-PL.IPFV-REDUP\3.SS ‘coming and coming (many people)’ Partial reduplication of verbs without the su#x -kaua is also possible, and it is used to intensify the meaning of the verb. (27) iʃa ́iʃaḿak iʃa iʃama-kũ REDUP be.afraid+IPFV-SIM\3SG.SS ‘while being very afraid . . .’ There is also total reduplication of adverbs and ideophones. They typically convey an intensifying sense. (28) iḿʲɛɛ iḿʲɛɛ iḿai iḿai far.there far.there ‘very far away’ (29) t͡sɨút t͡sɨút t͡sɨút t͡sɨút ideo:pull.o$.leave ideo:pull.o$.leave ‘Pulling o$ many leaves’ Reduplication in Wampis is used predominantly for grammatical purposes, 323 whereas it is not very productive for derivation of new words. Grammatically, reduplication serves the functions of distributive sense as in (24), intensity as in (27), re-iterative or repetitive action as in (26). However, some lexicalized items show that reduplication may have been also productive derivationally (i.e. used to create new words) in Wampis. For instance, number four is aintukV aintukV. The original meaning of this root is not know although the word looks like a complex word,129 but in any case it is an instance of reduplication. Examples of derivational reduplication in words do not always follow the rule given above for partial reduplication. Some of them are included in the list below (all of them are already lexicalized and most are arguably ideophonic in nature): (30) t͡sɨkɨt͡skɨtu ‘agile’ < cf. tsɨkɨ ‘run, jump’130 t͡sɨt͡sɨ ‘dust’ (also ‘garbage’, a modern sense?) t͡sɨt͡sɨka ‘cold’ piripiri ‘species of plants said to have magical powers’ pɨnt͡sɨpɨnt͡sɨ ‘monkey sp.’ In addition, there are some (rather infrequent) examples in the data where reduplication is used in a seemingly derivational function. For instance, in one text, the 129. The last -kV portion is probably the su#x -kɨ ‘Restrictive’. 130. Notice that an internal vowel has been deleted but was there historically: t͡sɨkɨt͡sɨkɨtu (the underlined vowel is the one deleted). 324 mouth of the main character is broken and then sewn in such a way that she speaks without being able to move her lips very much. To transmit the way the character of the story speaks, the root mɨna ‘left’ is reduplicated and it was translated as ‘out of the side of her mouth’ (‘de costadito’ in the Spanish translation provided by my teachers). (31) mɨnátʃ mɨnánt͡ʃ tʃitʃá mɨna-t͡ʃi mɨna-t͡ʃi tʃitʃa-ã left-DIM left-DIM speak-HIAF\3SG.SS ‘Having spoke out of the side [of her mouth]. . .’ 7.8. Compounds A compound word is a word that is formed by two di$erent words. The criteria for determining compound words can be semantic or formal (Payne 1997: 92). Formally, in Wampis compound words, both elements retained a high tone and are susceptible to the vowel elision processes seen in Chapter VI. However, when they receive in/ection, the in/ection only attaches to the second element, showing that they constitute a single unit. For instance, normally the accusative =na appears on all elements of a noun phrase when a determiner is present: (32) núna sɨńt͡ʃin ʃuarań nu=na sɨnt͡ʃi=na ʃuara=na NON.VIS=ACC strong=ACC person=ACC ‘to that strong person’ However, in a compound word, like hapa_iauaã ‘puma’ (hapa ‘deer’ + iauaã 325 ‘dog’), in a similar morphosyntactic environment, the accusative attaches to the second root of the compound (iawaã) only, not to each element as in (32). Thus, hapa_iauaã is treated as a single word, as shown in the next example. (33) núna hapa ́ɲawaáñ nu=na hapa_iauaa=̃na NON.VIS=ACC puma=ACC ‘to that puma’ In addition, compare (33) with (34) below, where the word tura ‘and,’ is added. In (34), hapa ‘deer’ and iawã ‘dog’ appear as distinct words and therefore each of them receives the accusative =na. Notice that it is impossible for that phrase to have the reading ‘and to the puma’. (34) núna hapán túra ɲãwãań nu=na hapa=na tura iawã=na NON.VIS=ACC deer=ACC and dog=ACC ‘to the deer and the dog’ Syntactically, compound words always show a [Modi"er-Head] structure, whereas adjectives in noun phrases can come before or after their noun head (the latter is the most frequent order). Place names uniformly show a [Modi"er-Head] structure, though in many cases the roots are phonetically reduced: (35) t͡ʃinkanás < t͡ʃinkana+ɨntsa bamboo.sp+river ‘River of bamboo’ (for the community named t͡ʃinkanás, Spanish ) 326 Semantically, a compound word’s meaning is more particular or even completely di$erent (i.e. unpredictable) from the sum of its parts. The next examples illustrate this point for Wampis. (36) hapa ́ɲawaá ̃ hapa_iauaa ̃ deer_dog ‘puma’ (37) uún ɲawaã́ uunta_iawã big_dog ‘jaguar’ (38) wákɨ mɨsɨŕ wakɨ_mɨsɨr stomach_ruin ‘sad’ In Wampis, compounds can be created by the following combinations: Noun+Noun (as in example (36) above), Adjective+Noun (as in (37)), and more rarely Noun+Verb (as in (38)). There are also examples of the intensi"er ima+Noun: (39) imʲa ́nápi ima_napi INTS_snake ‘venomous snake sp.’ 327 7.9. Particles A particle is generally understood as an invariant element (i.e. it does not receive in/ection) that is phonologically independent. In this dissertation, I will not refer to particles as a di$erent word class on their own. Rather, I use the terms Interjection, Ideophone and Grammatical Particle to refer to elements that can be seen as having the properties of being phonologically independent and being invariant. These three categories constitute di$erent classes in the present analysis. Very brie/y: Interjections are words that constitute complete utterances by themselves (they are described in §8.11). Ideophones are words that evoke an idea in sound and function mostly adverbially in Wampis but can replace predicates (they are described in §8.12). Grammatical particle is a more restricted term: a grammatical particle is an invariant element that provides some grammatical information (e.g. aspect) to the clause. There are only two grammatical particles in Wampis, has and hak (see §8.13). 328 CHAPTER VIII CLOSED WORD CLASSES 8.1. Introduction This chapter is dedicated to closed word classes, which are those that are composed of “a "xed and usually small number of members” (Schachter & Shopen 2007: 3). The closed classes of Wampis are: pronouns, demonstratives, interrogative words, non-numeral quanti"ers, numerals, interjections and grammatical particles. There are two words that do not "t in any category and constitute their own classes: tikit ͡ʃi ‘other’ and uɨantu ‘group’. In addition, due to their morphosyntactic properties I also discuss in this chapter two word classes that are not strictly closed: ideophones and adverbs. Ideophones are an open class in the sense that the most simple way to create a new ideophone is by imitating any sound, so the inventory of ideophones can always be expanded in theory. The reason for considering adverbs in this chapter is that the only adverbializer in Wampis (which only derives time-adverbs) is no longer productive in the language; hence, in practice Adverbs constitute a semi-closed class. The word classes are presented in this way: §8.2 describes pronouns, §8.3 provides an analysis of determiners, which is followed by a description of the word tikit ͡ʃi ‘other’ in §8.4. 329 Question words are analyzed in §8.5. Adverbs are analyzed in §8.6. Sections §8.7 and §8.8 describe non-numeral and numeral quanti"ers, respectively. Section §8.9 discuses the word uɨantu ‘group’, and §8.10 discusses the (little) evidence for conjunctions. Interjections and ideophones are described in §8.11 and §8.12, respectively. Finally, §8.13 comments on two grammatical particles found in Wampis. 8.2. Pronouns Pronouns constitute a class of words that are free forms and that can "ll the position of an NP in a clause. Pronouns are referential devices that need an anaphoric or cataphoric referent in the discourse. In Wampis, pronouns generally have most of the distributional and morphological properties of nouns. However, di$erently from nouns, pronouns cannot be possessed and cannot be modi"ed. Pronouns do not receive vocative case either. 8.2.1. Personal pronouns Personal pronouns are related to the distinction of grammatical person. Wampis distinguishes between singular and plural personal pronoun forms. Table 8.1 shows the personal pronouns found in Wampis. 330 Table 8.1. Personal pronouns in Wampis Person\Number Singular Plural 1 ui [wií] ii [ií] ~ huti ́[huti]́ 2 amɨ [aḿɨ] atumɨ [at́um] 3 ni ̃[ni ̃í] nita [nit́ʲa] When they occur without additional morphology, the 1SG and 3SG pronouns lengthen their vowels to ful"ll the two-mora restriction for phonological words that operates in the language (see §6.8). For "rst person plural, there are two forms, ii and huti.́ The pronoun huti ́historically appears to be a combination of the morphemes hu ‘Proximal demonstrative’ and =tí ‘Plural speech act participant’. The di$erence between ii and huti ́is not very well understood, if there is any at all. For Awajun, Overall suggests a distinction between the cognate Awajun 1PL forms ii and hutii: according to him, ii is speci"c and usually excludes the 2 person, whereas hutii is non- speci"c and may or may not include the second person (2007: 156). On the other hand, also for Awajun, Corbera (1980: 186) suggests an inclusive (ií) vs exclusive (hutií) distinction. I have not found either of those possible distinctions in Wampis. Both huti ́ and ii can exclude or include the addressee. However, most examples of hutí in Wampis exclude the 2 person, and it is usually used to refer to the Wampis as a group. For 331 instance, the next example comes from a conversation between a Wampis speaker and myself as the addressee. The speaker has "nished telling me what the Wampis elders hope for their future generations, so he is speaking about the Wampis as a group and the addressee is not part of the intended referent of “our way of thinking”: (1) núwɛiti huti ́anɨńtɛiŋkʲa nu=aiti huti ́ anɨnta-i=̃ka NON.VIS=COP.3+DECL\GEN 1PL thought-1PL/2PL/3.POSS=FOC ‘That is our way of thinking.’ With that said, in practice ii and huti ́are more or less interchangeable—and ii is much more frequent in texts. In (2), for example, ii also refers to the Wampis and excludes the addressee. This example comes from a narrative in which the speaker tells me how he was healed after su$ering an accident while hunting. He explains in the last lines: (2) iíkʲa iḿanisrikʲa ðokto ́rnumka wɨát͡shi ii=ka imani-sa-ri=ka doktor=numa=ka 1PL=FOC do.much-SUB-1PL.SS=FOC doctor=LOC=FOC wɨ-a-t͡su-hi go-IPFV-NEG-1PL.SBJ+DECL ‘We, doing so much, do not go to the doctor.’ Example (3) clearly shows that ii may exclude the addressee: (3) iikʲa kuntiɲan ɨaḱmakrin, aḿɨka hɨãḿin wakɨtkitʲa ́ ii=ka kuntina ɨakama-a-ku-ri-ni ̃ 1PL=FOC animal look.for.game-IPFV-1PL.DS 332 amɨ hɨã-mi=ni ̃ wakɨtu-ki-ta ́ 2SG house-2=LOC return-WHILE.MOVING-IMP ‘While we go hunting, you return home.’ On the other hand, in (4), ii includes the addressee, so it means ‘you+I’. (4) jat͡surú haiḿito miɲ́a jaaḱta soleðad́ iíkʲa himʲara ́wɨsat́hi iat͡su-rú haimito mina iaakata Soledad brother-1SG Jaime (DIM) 1SG.GEN town Soledad ii=ka himara ́ wɨ-sa-ta-hi 1PL=FOC two\ITER go-ATT-IMM.FUT-1PL.SBJ+DECL ‘My brother Jaime, we (i.e. you and I) will go two times to my town Soledad.’ In sum, there does not seem to be a clear distinction between the 1PL pronouns ii and huti.́ While in most instances in the text data huti ́references the idea of a general group (usually meaning ‘‘we—of my group, usually meaning the Wampis’’), ii is also used in that way, and both pronouns can include or not the addressee. In terms of frequency, ii is much more frequent in texts. This is perhaps because of the relatively recent development of huti ́as a pronoun, which is still clearly associated with hu ‘Proximal demonstrative’ and =ti ́‘Plural speech act participant’. Both pronouns are easily interchangeable and there seems to be a competition between the older (and still more frequent) form ii and the more recently innovated huti.́ Perhaps more conversational data and analysis of speci"c contexts where they occur may reveal di$erent discourse functions of ii and huti.́ 333 8.2.1.1. Combining forms of 1sg and 2 personal pronouns 1SG and 2 personal pronouns have special allomorphs that occur only with the following morphemes: accusative =na, benefactive/possessive -nau, locative =(n)i ̃and comitative =hai.̃ Table 8.2 shows the combining forms of the personal pronouns in Wampis. Table 8.2. Combining forms of personal pronouns Person: Stem: 1SG ui~mi 2SG ami 2PL atumi Acc =na [miɲ́a]~[wiɲ́a] mina~uina [aḿiɲa] amina [atúmin] atumina Loc =(n)i ̃ [miɲ́ɛi]̃~[wiɲ́ɛi]̃ minai~̃uinai ̃ [aḿiɲi] amini ̃ [atúmin] atumini Com =hai ̃ [mihʲɛɛ]̃~[wihɛɛ]̃ uimi~mihai ̃ [aḿihɛɛ]̃ amihai ̃ [atúmhɛɛ]̃ atumihai ̃ Ben -nau [miɲóo]~[wiɲ́a] minau~uinau [aḿiɲu] aminau [atúmɲoo] atuminau As can be seen, in the case of the "rst person, the use of the combining form is optional: the pronoun can occur as the regular form ui or as mi. The locative form of the 1SG is minai ̃and not mini,̃ as would be expected. A probable explanation is that the base on which the locative marker occurs may be the accusative stem (mina~uina). 334 8.2.1.2. Genitive forms of personal pronouns The genitive forms of the personal pronoun are formed with the accusative =na for most grammatical persons. 1PL form huti,́ 2PL and 3PL pronouns mark their genitive forms with a high tone on their last vowel. In the case of the 2PL pronoun, there are two possible genitive forms: one with the accusative =na and one with the high tone. Table 8.3 shows the genitive forms of the personal pronouns. Table 8.3. Genitive forms of personal pronouns Person Genitive Form 1SG mina~uina [miɲ́a]~[wiɲ́a] 2SG amina [aḿiɲa] 3SG nina [niɲ́a] 1PL iina [iíɲa] 1PL huti ́ [huti]́ 2PL atumí~atumina [atumi]́~[atumiń]131 3PL nitá [nitʲa]́ 131. As can be seen, this form always surfaces as [atúmin] and the "nal /a/ is dropped following the rule of apocope in Wampis seen in Chapter VI. By contrast, the 2SG genitive form amina [aḿiɲa] never deletes its last vowel, contravening the rule of apocope which deletes the last underlying vowel of the word. 335 8.2.1.3. Use of 3 person pronouns as de!nite articles The third person pronouns every now and then can act as articles to specify grammatical de"niteness of a noun already introduced in the discourse. In the Wampis data, a pronoun that functions as a determiner never introduces a new participant in the discourse. Notice the use of the 3SG in the next example. The pronoun does not replace the NP; rather, it speci"es that its noun head is de"nite. In the text from where the example comes, the character Mikut is a referent already established in the discourse. He has been preparing himself to obtain certain powers by dieting and taking hallucinogenic plants. In the narrative, there are two people named Mikut, so the pronoun is used to identify one of them: (5) ni ̃í miḱut iḿatikawa núna iḿatiksan umiḱ . . . ni ̃ Mikuta imatika-u=a=nu=na imatika-sa ̃ 3SG Mikut do.much-NMLZ=COP=NON.VIS=ACC do.much-SUB\3SG.SS umi-ka ̃ complete-INTENS\3SG.SS ‘The Mikut that did all that, doing that much, having "nished . . .’132 132. A possible analysis of (5) is that the relative clause headed by the proper noun Mikut is in an appositive construction with the pronoun ni;̃ i.e. ‘He, [the] Mikut that did all that, . . .’. However, in the example, the whole NP ‘The Mikut that did all that’ falls in one single intonation contour and is specifying a de"nite referent in the discourse. The interpretation of ni ̃as an article was also double-checked with my Wampis teachers. 336 8.2.2. Cataphoric hesitation pronoun naa The hesitation pronoun naa is a cataphoric device: it precedes the mention of a referent that is to occur immediately or almost immediately afterward in the discourse. The pronoun agrees with its subsequent referent in the marking of some derivational and most case, adposition and discourse-related clitics. Pragmatically, naa usually expresses a hesitation, and its functional motivation seems to be associated with lexical access; in other words, it stands in discourse in lieu of the noun that the speaker is trying to retrieve. The pronoun naa is transparently related to the interjection naa that expresses hesitation or doubt (cf. §8.11). However, the pronoun naa is clearly grammaticalized as a participant-reference form in current Wampis, as it establishes obligatory case agreement with its noun referent, as is shown below. In (6), the hesitation pronoun naa receives the same marking (accusative =na and focus =ka) as its referent ‘wall’. (6) iṕatɨamuʃa hiãŋ́ka túkɨ naaŋ́ka taniʃ́naka jumpúat͡ʃu iṕatu-a-mau=ʃa hɨã=na=ka tukɨ shoot-IPFV-NMLZ=ADD house=ACC=FOC INTERJ naa=na=ka taniʃa=na=ka iumpu-a-t͡ʃau HESIT.PRO=ACC=FOC wall=ACC=FOC break-IPFV-NEG.NMLZ ‘The shots at the house did not break umm... the wall.’ 337 In the next example, naa stands in place of ‘relative’; both naa and its referent agree in the marking of the comitative. (7) naah́ɛɛ ̃pataíh̃ɛɛ ̃wɨaŕuka naa=hai ̃ patai=̃hai ̃ wɨ-a-́ara-u=ka HESIT.PRO=COM relative\1PL/2PL/3.POSS=COM go-HIAF-PL-NMLZ=FOC ‘Those who went with umm... their relatives.’ In (8), naa receives the locative =numa, agreeing with its noun co-referent in the marking of focus =ka too. (8) awarún ɛɛ́ɲaʃa pɨkaḿar naańmaka jɛiḱminmaka auaruna a-ina=ʃa pɨkama-ara Awajun COP-PL.IPFV=ADD form.line.PFV-PL/3.SS naa=numa=ka iakimi=numa=ka HESIT.PRO=LOC=FOC sand=LOC=FOc ‘The Awajun too, having formed a line in, umm, in the sand.’ The morphemes in Table 8.4 are attested to occur on naa as a pronoun. The diminutive and the restrictive can be considered ‘derivational’,133 the accusative and the nominative are case-markers,134 the locatives, the ablative, allative and comitative are 133. Of course the restrictive =kɨ has a discourse-level function too. See §18.3 for a description of this morpheme. 134. Notice the absence of the genitive from the list. The genitive is marked with a high tone on nouns, so probably due to its phonologically-bound nature it cannot be present in naa. As we saw in §8.2.1.2, personal pronouns form their genitive form with the accusative =na, not with a high tone. However, naa, which appears to be a relatively recent reanalysis of a hesitation interjection (§8.11), does not possess a genitive form. The historical relationship between the 338 post-positional clitics; and the additive and focus markers are discourse-level morphemes. Table 8.4. Morphemes attested as marking agreement between the pronoun naa and its noun co-referent Morpheme Gloss -t͡ʃi Diminutive -nau Benefactive/Possessive ∅ Nominative =na Accusative =nVma Locative =(n)i ̃ Locative =ia Ablative =ni Allative =hai ̃ Comitative =kɨ Restrictive =ʃa Additive =ka Focus 8.3. Demonstratives Wampis has four demonstratives: proximal hu,135 medial aanu,136 distal au, and genitive and the accusative is explained in §10.4.6.3.1. 135. Some speakers pronounce the proximal hu as a nasal hũ. 136. The source for the medial aanu is probably aa ‘outside’ plus the non-visible demonstrative nu. 339 non-visible nu. Wampis demonstratives constitute a moderately complex set that is based on one general parameter: VISIBILITY, and within visibility, there is a subsystem based on DISTANCE. As shown in Figure 8.1, the distance-oriented subsystem is based on the speaker’s point of view, it does not take the addressee into account. Figure 8.1. Distance-oriented scheme of Wampis demonstratives Demonstratives possess most nominal morphosyntactic properties: they are able to receive some nominal morphology and replace noun phrases. In addition to this pronominal function (10), demonstratives also function as determiners (9). In both functions, they indicate deixis. (9) núka arútmaka kaḿɨ jóunt͡ʃuk naŋkamaś nu=ka arutama=ka kamɨ iaunt͡ʃukɨ nankama-sa ̃ NON.VIS=FOC power.vision=FOC INTERJ long.ago begin-SUB/3SG.SS ‘That [belief in] Arutam starting long ago . . .’ (10) húna uhaḱtathamɨ hu=na uha-ka-tata-hamɨ PROX=ACC inform-APPL-DEF.FUT-1SG>2SG+DECL ‘I am going to inform you this.’ EGO Proximal Medial Distal 340 Table 8.5 summarizes the forms of the demonstratives with the attested case, post-positional and discourse-related clitics with which they occur. Forms with an (E) (=Elicited) were only collected in elicitation and are not attested in the natural texts consulted for this study. The locative forms of the demonstratives, especially hui,̃ ai ̃and nui,̃ serve as locative adverbs too. Table 8.5. Demonstratives with case, post-positional and discourse-related clitics Category Visible Non-visible Proximal Medial Distal Nominative hu aanu au nu Accusative hu=na aanu=na au=na nu=na Locative hu=i ̃ aani ̃ a=i ̃ nu=i ̃ Ablative hu=i=̃ia [hui ̃ɲ́a] aanu=ia [aania] au=i=̃ia (E) [aẃiɲ̃a] nu=i=̃ia [nui ̃ɲ́a] Allative hu=ni aanu=ni au=ni nu=i=̃ni [nui ̃ɲ́i] Comitative hu=hai ̃ (?)a aanu=hai ̃(E) au=hai ̃ nu=hai ̃ Restrictive hu=kɨ aanu=kɨ au=kɨ nu=kɨ Focus hu=ka aanu=ka au=ka nu=ka Additive hu=ʃa aan=ʃa au=ʃa nu=ʃa a Speakers’ judgments of this form (only given in elicitation) were ambivalent. While the form is possible, apparently it is not used (and there is no attestation of it in texts in the database), so speakers tend to judge it as marginally acceptable. While demonstratives share some morphological and syntactic distribution with 341 other nominal and pronominal classes, demonstratives constitute their own class on the basis of the following morphosyntactic properties: • Demonstratives do not receive genitive marking, unlike nouns or personal pronouns. • Demonstratives can bear the locative =(n)i,̃ as shown in the locative forms in Table 8.5 above. However, demonstratives do not receive other locative forms such as =nVma or V́, whereas nouns and personal pronouns do (see §11.5.1 for details on the locative markers). • The presence of demonstratives in the NP causes obligatory agreement between the demonstrative, the noun and any other attributive modi"er (i.e. adjectives and relative clauses). This is a unique property triggered by demonstratives. • Demonstratives receive the similative su#x -mamtin, which in my data does not occur on nouns or pronouns.137 The "rst two points refer to morphological impossibilities of demonstratives (in comparison to nouns and pronouns) and are self-explanatory. The other two points are explained in the next sub-sections. 137. The only other words where the similative mamtin appear in the data are not nominal: mɨtɨka-mamtin ‘equal-similative’ and tu-mamtin ‘like.that-similative’ (tu is a demonstrative manner adverb (§8.6.1). 342 8.3.1. Demonstratives and agreement Agreement marking is established between the demonstrative, the noun and any attributive modi"er of the noun head (including relative clauses) only when a demonstrative is present in the NP. This is an interesting pattern in Wampis, as nominal agreement seems to be based on the presence of the demonstrative, rather than on the noun. Table 8.6 lists the morphemes that exhibit this agreement behavior when a demonstrative is present. Table 8.6. Morphemes that display agreement in the presence of a demonstrative Category Morpheme Nominative ∅ Accusative =na Locative =nVma, =(n)i,̃ V́a Focus =ka Additive =ʃa a The demonstratives only bear the locative form =(n)i ̃; nouns can carry all three locative forms listed above. The following discussion illustrates how the presence of a demonstrative triggers agreement within the NP. First, it must be noted that nouns alone do not establish obligatory overt agreement. The marking of case and other clitic categories occurs most 343 frequently in the last element of the NP. For instance, in (11) the accusative is hosted by the adjective ‘good’, and in (12) it is hosted by the relativized copula in the relative clause that is modifying the head noun: (11) ʃuaŕ pɨŋ́kɨran wɛiŋ́kʲamhɛɛ [ʃuara pɨńkɨra]=na uaina-ka-ma-ha-i person good=ACC see-INTENS-REC.PT-1SG.SBJ-DECL ‘I saw a good person.’ (12) ʃuaŕ ʃiiŕ maãńin ańuna wɛiŋ́kʲamhɛɛ [ʃuara [ʃiira maã-nai-inu a=nu]]=na uaina-ka-ma-ha-i person very kill-RECP-NMLZ COP=NON.VIS=ACC see-INTENS-REC.PT-1SG.SBJ-DECL ‘I saw a person that was a such a killer.’ On the other hand, when a demonstrative is present, the marking of the categories listed in Table 8.6 changes. As stated previously, in such a case, the demonstrative, noun head and attributive modi"ers agree in case or focus/additive marking. Thus, compare (11)–(12) above with (13)–(14): (13) núna ʃuarań pɨŋ́kɨran wɛinkʲamhɛɛ nu=na ʃuara=na pɨnkɨra=na uaina-ka-ma-ha-i NON.VIS=ACC person=ACC good=ACC see-INTENS-REC.PT-1SG.SBJ-DECL ‘I saw that good person.’ (14) núna ʃuarań ʃiira maãńin ańuna wɛiŋ́kʲamhɛɛ nu=na ʃuara=na ʃiira maã-́nai-inu a=nu=na NON.VIS=ACC person=ACC very kill-RECP-NMLZ COP=NON.VIS=ACC 344 uaina-ka-ma-ha-i see-INTENS-REC.PT-1SG.SBJ-DECL ‘I saw that person that was such a killer.’ As can be seen, unlike (11)–(12), in (13)–(14) the accusative is borne by the demonstrative, the noun, and the adjective or relative clause in modifying function. Notice that the structures in (13)–(14) do not constitute an appositive construction, as evidenced by the fact that the accusative is marked in all modi"ers of the noun and in the noun itself. If it were an apposition, the accusative would be expected to be marked in the demonstrative and in the last element of the second NP. As an illustration, if there were an appositive construction in (13), we would expect a construction like [nu=na], [ʃuara pɨnkɨra]=na ‘that, a good person’; but not [nu=na ʃuara =na pɨnkɨra=na] ‘that good person’, which actually occurs in the above example. The previous examples o$ered instances of agreement with the accusative =na. Following are some examples with the locative =nVma, the focus marker =ka and the additive =ʃa. Notice that morphemes can occur stacked, in which case they are also repeated in the other element(s) of the NP, as in (15). (15) huiʃ̃a apat́͡ʃ puhútnumʃa unuimaru hu=i=̃ʃa apat ͡ʃi puhu-ta=numa=ʃa unuima-ra-u PROX=LOC=ADD mestizo live-NMLZ=LOC=ADD learn-DISTR-NMLZ ‘one that is educated in this mestizo’s [way] of life too’ 345 (16) óoka ʃuaŕka wɨḿaji au=ka ʃuara=ka uɨ-ma-ji dist=FOC person=FOC go.PFV-REC.PT-3.PT+DECL ‘That person went away.’ (17) núʃa ɛíʃmaŋt͡ʃiʃa kanaŕui nu=ʃa aíʃimanku-t͡ʃi=ʃa kana-ra-u=i NON.VIS=ADD man-DIM=ADD sleep-DISTR-NMLZ=COP.3+DECL ‘That little man too fell asleep.’ Though very rarely, demonstratives can apparently even trigger (optional) agreement on adverbials. In the data, this is very rare indeed, as adverbial words in general do not receive case or post-positional marking. This property is illustrated in (18). As can be seen, the intensifier ı́manis agrees with the ‘non visible’ demonstrative nu in the marking of accusative and focus. (18) núkap arútmaka irúnɨawɛɛ núnaka iḿanisnaka anɨákɨat͡shɛɛ nukapɨ arutama=ka irunɨ-a-ua-i several power.vision=FOC be.together.IPFV-IPFV-3.SBJ-DECL nu=na=ka iḿanis=na=ka anɨakɨ-a-t͡su-ha-i NON.VIS=ACC=FOC INTS.DEM.ADV=ACC=FOC remember-IPFV-NEG-1SG.SBJ-DECL ‘There are several Arutam, that much I don’t remember.’ 8.3.2. The similative constructions with -mamtin The su#x -mamtin ‘Similative’ attaches to a demonstrative and adds the meaning of ‘similar to X’. There are two constructions where the similative -mamtin occurs. The "rst 346 construction has the following structure: Similative construction 1: [NP=ABL DEM-mamtin] In this construction, the NP is a full lexical noun or pronoun that occurs preposed and marked with the ablative, whereas the demonstrative co-refers with the NP head and is marked with -mamtim. (19) tɨntɨt́saja numámtin Tɨntɨtsa=ia nu-mamtin Tɨntɨtsa=ABL NON.VIS-SIMIL ‘similar to Tɨntɨts’ (20) ɲawãája numámtin iauãa=ia nu-mamtin dog=ABL NON.VIS-SIMIL ‘similar to a dog’ The second construction where the similative occurs has the following structure: Similative construction 2: [NP=COP DEM-mamtin] In Similative construction 2, the noun occurs with a cliticized copula. Though the semantics of this construction is similar to Similative construction 1, in this case the reference is more inde"nite. In the following example, the speaker is talking about Nunkui, beings that are related to agriculture, hunting and fertility. People sing magical songs to Nunkui, who can give powers and propitiate good agriculture. Notice that the 347 speaker uses Similative construction 2 and he does not specify what kind of power Nunkui has: (21) núka núŋkuikʲa poðeŕa aánmamtin . . . nu=ka nunkui=ka [podera=a aanu-mamtin] NON.VIS=FOC Nunkui=FOC power=COP MED-SIMIL ‘that Nunkui, [she has] like a power.’138 In relation to the above description, the similative may also occur attached to the demonstrative with no co-referent noun. Similar to the previous example, in this case the similative also adds an inde"nite sense: (22) aánmamtin míɲa nukút͡ʃruka tújaji turáʃa ʃiír anɨát͡shɛɛ aanu-mamtin mina nukut͡ʃi-ru=ka tu-ia-ji MED-SIMIL 1SG.GEN grandmother-1SG=FOC say-REM.PT-3.PT turaʃa ʃiira anɨ-a-t͡su-ha-i but well remember-IPFV-NEG-1SG.SBJ-DECL ‘My grandmother said something like that, but I don’t remember well.’ Wampis does not possess inde"nite pronouns of the sort English has with some/ any-pronouns; however, one way to achieve the function of inde"niteness is with the similative. The form of the morpheme -mamtin must have been historically complex, 138. Literally translated by my teachers into Spanish as: como poder algo asi ́‘like power, something like that . . .’. 348 possibly having the form -mamVtinV. This assumption is based on the odd phonological structure of -mamtin, which possesses the sequence /mt/, when we would expect /nt/. Hence, the sequence /mt/ must have been obtained after vowel elision from an old sequence *mVt. In addition, underlyingly, morphemes in Wampis never end in a consonant, so it is more than likely that the su#x -mamtin used to have a "nal vowel that is synchronically “unrecoverable” in current Wampis. However, because of the form, the last portion of the similative, /tin/, may be related to the su#x -tinu ‘Attributive’. The attributive likely is also part of an old unproductive adjectivilizer su#x -patinu (see §9.8.4), that carries a similar meaning to -mamtin: ‘propensity to’ or ‘similar to’, e.g. ʃakarpatin ‘loose’ < *ʃaka ‘watery’, ‘watered-down’, *-ri ̃‘1PL/2PL/3.POSS’, *-pa (?), *-tinu ‘Attributive’. 8.3.3. Other functions of demonstratives Apart from their use as determiners and pronouns, there are two other important extended functions of demonstratives: they can relativize a copula, and they serve as textual anaphora (this is a typical sub-function of demonstratives, but merits a brief mention here because of the high frequency with which some demonstratives occur in that function in texts). 349 Examples of the relativization property of demonstratives are given in (23), where there is a “headless” relative, and in (24), where the relative clause is externally- headed. (23) óohumatsattahɛɛ wiɲ́a puhútruanuna auhumatu-sa-tata-ha-i [uina puhuta-ru=a=nu=na] tell-ATT-DEF.FUT-1SG.SBJ-DECL 1SG.GEN life-1SG=COP=NON.VIS=ACC ‘I am going to tell you about my life’ (Lit.: ‘. . . that which is my life’) (24) ni ̃íʃa núu tiḱit͡ʃ ʃuaŕ ańu ut͡ʃiri ́huki ́ ni=̃ʃa [nu [tikit ͡ʃi ʃuara a=nu]] 3SG=ADD NON.VIS other person COP=NON.VIS ut͡ʃi-ri ̃ hu-ki ̃ child-1PL/2PL/3.POSS take-WHILE.MOVING\3.SS ‘He too, that who was another person, having taken his child . . .’ Constructions that follow the same relativization pattern in adverbial function usually serve to introduce new information: (25) wií t͡ʃit͡ʃaḿ kaḿɨ húu puhútahui ̃ańis óo asaḿtɛɛ ̃ ui t͡ʃit͡ʃama kamɨ [hu puhuta=a=hu=i]̃ 1SG problem INTERJ PROX life=COP=PROX=LOC aani-sa ̃ a-u a-sa-matai ̃ do.that-SUB\3SG.SS exist-NMLZ COP-SUB-1SG/3.DS ‘Because, thus, I have a problem here in this life. . .’ A more detailed analysis of relative clauses is presented in Chapter XX. With regard to the discourse anaphoric function, the non-visible demonstrative 350 nu functions prominently and frequently as an anaphoric pronoun in narratives. However, note that all other demonstratives also function anaphorically in texts, especially the proximal hu and the distal au; but in comparison with nu they are much less frequent in narratives. This is not unexpected, as narratives usually relate stories whose protagonists are not present or visible, hence the preference for the ‘non-visible’ demonstrative -nu. The anaphoric function of demonstratives occurs frequently in direct speech reports. In (26), the demonstrative nu refers to the whole expression ‘May they not su$er!’ (notice that nu cannot be the object of ‘su$er’ because it does not bear the accusative marker =na). (26) iíɲa ut͡ʃiri ̃ ́huaŕar wɛit́sarɛi núu iina ut͡ʃi-ri ̃ huara-ara [uaitu-sa-ara-i] [nu] 1PL.GEN child-1PL/2PL/3.POSS stay.PFV-PL.SS su$er-ATT-PL-APPR NON.VIS ‘Our children having stayed, “may they not su$er!”, that.’ In the following example, the speaker talks about the foundation of several modern Wampis villages. He introduces some of them for the "rst time using their names (Ampama, Onanga, Kukuasa), and then uses the distal au to refer to all of them, while maintaining the topic. What follows immediately is a direct speech report construction. Notice that when the speaker says ut͡ʃita ́‘children (VOC)’, he is referring to 351 the people who were going to create the villages just mentioned before. These people did not want to come down-river to found a di$erent village, so they are told to found their own villages: (27) takasa ́ puhusmi ́ túsan ampaḿan úunkan kukúasan óona ut͡ʃitʲa ́ atúmʃa hui ̃ ́ tat́aka nakit́ʲarmɨ aŋkań aŋkań jaaḱat nahańarum atúmʃa núŋka ajamruktaŕum taka-sa puhu-sa-mi tu-sa-nu work-ATT live-ATT-HORT say-SUB-1SG.SS [Ampama=na Uunaka=na Kukuasa=na] [au=na] Ampama=ACC Onanga=ACC Kukuasa=ACC DIST=ACC ut͡ʃita ́ atumɨ=ʃa hui ̃ child\VOC 2PL=ADD here ta-ta=ka nakita-rumɨ arrive-NMLZ=FOC not.want+IPFV-2PL.SBJ+DECL ankan ankanta iaakata nahana-rumɨ REDUP free town make.PFV-2PL.SS atumɨ=ʃa nunka ́ aiamaru-ka-ta-́rumɨ 2PL=ADD land\ACC defend-INTENS-IMP-2PL.SBJ ‘[I] saying “Let’s work” to Ampama, Onanga, Kukuasa, to those: “Children, you do not want to come here, [then] each one having created your town, defend the land!”’ The most common way to "nish a narration is with the formula in (28), where the demonstrative nu refers to the whole narrative that has been told: 352 (28) núkɨti nu=kɨ=iti NON.VIS=RESTR=COP.3+DECL ‘That is it.’ (Lit.: ‘That only is.’) To conclude this section on other functions of demonstratives, it is important to mention that the distal au is frequently used instead of the third person pronoun ni ̃in conversations. The distal au is also used in comparative constructions (see §9.7.3.1), and as an interjection (see §8.11). 8.3.4. Resumptive verbs from demonstratives All demonstratives and the intensi"er ima receive the verbalizers -ni and -tika and become resumptive verbs. See §12.6.4 for a description of resumptive verbs. 8.4. Tikit ͡ʃi ‘other’ The word tikit ͡ʃi ‘other’ functions as a demonstrative: it can determine NPs as in (29) or it can occur pronominally in place of an NP as in (30). (29) tiḱit ͡ʃ númi pɨŋ́kɨr at́iɲun araḱmakmiaji tikit ͡ʃi numi pɨnkɨraa-tinu=na araka-ma-ka-mia-ji other tree good COP-FUT.NMLZ=ACC plant-VBZ-INTENS-DIST.PT-3.SBJ+DECL ‘She planted other trees that will be good.’ (30) tikit́͡ʃin óuhumatsathɛɛ tikit ͡ʃi=na auhumatu-sa-ta-ha-i other=ACC tell-ATT-IMM.FUT-1SG.SBJ-DECL ‘I am going to tell another one (i.e. another story).’ 353 However, tikit ͡ʃi does not exhibit agreement with the noun like demonstratives do. In (31), the demonstrative nu and the noun ʃuara agree in the marking of accusative, but not tikit ͡ʃi. (31) núna tikit͡ʃ ʃuarań atakʃa w#́maji nu=na tikit ͡ʃ ʃuara=na atakʃa uɨ-ma-ji NON.VIS=ACC other person=ACC again go-REC.PT-3.SBJ+DECL ‘That other person went away again.’ Furthermore, tikit ͡ʃi can be modify by determiners (32), showing that it does not belong to the latter category. Notice that when tikit ͡ʃi functions as a pronoun, it does establish agreement with the demonstrative, should a demonstrative occur in the same NP. (32) húna tikit́͡ʃin óohmatsathɛɛ nu=ka tikit ͡ʃi=na aúhumatu-sa-ta-ha-i NON.VIS=ACC other=acc tell-ATT-1SG.SBJ-DECL ‘I am going to tell this other [story].’ As a "nal note, it should be mentioned that some speakers have a free variant t͡ʃikit͡ʃi (with a palatal as "rst consonant), instead of tikit ͡ʃi. 8.5. Interrogative words Interrogative words are based on derivations of the roots tu ‘where, which’, uru ‘how’ and itu ‘make’; plus the interrogative pronouns ia ‘who’, uari ̃‘what’. Interrogative words are listed in Table 8.7. The semantic "elds associated with these roots are also 354 provided. When there is a lexical source or identi"able formatives in historically complex interrogatives words, I list them in the respective column. Table 8.7. Interrogative words in Wampis Interrogative Words Gloss Possible source or Analysis Associated Semantics tua ‘Which’ tu+a ‘Cop’ (?) Type ia ‘Who’ Human uari ̃ ‘What’ uari ‘thing’ Non-human, Type tu ‘Where’ Location uruka~uruku ‘How’ uru+=ka ‘Q’ Manner itura ‘How’, ‘In which way?’ ituru ‘make’ Manner uruku=numa ‘How + Where’ uruka+=numa ‘Loc’ Location+ Manner uruka+-matai ̃ ‘Why’ uruka+-matai ̃‘1sg/3 DS’ Reason uruti~urutia(n) ‘When’ uru+-ti ‘Time’ +a ‘Cop’ (?) Time urutima ‘How many’, ‘How much’ uruti+*-ma (?) Amount As can be seen, some question words (especially the ones derived with the roots uru) are historically complex. Words with the roots uru occur with the formatives -ka (in 355 ‘how’), -ti (in ‘when’) and -ma (in ‘how many’); however, these su#xal formatives are already fossilized and their meanings are not easily identi"able. The formative *-ka is very likely a fossilized version of the interrogative clitic =ka that is otherwise productive in the Wampis language (see §18.8). Based on semantics and phonetic similarity, the formative *-ti that occurs with uru to form the time question word ‘when’ could be related to the derivational su#x -tin ‘time’ (which can be shortened to -ti ́(see example (90) later in this chapter), that is used to derive adverbials that express ideas of date and season (see §8.6.2.1). I have found no evidence of a possible source for the formative *-ma. In addition, uruka ‘how’ has a variant uruku that seems to be an old nominalized form bearing Set II nominalizer -u ‘Subject nominalizer’ (§15.4.5). This form uruku, in the data, only occurs when the interrogative word bears the locative =numa (see example (42)), the inferential =t͡su or in combination with a cliticized copula (see (71), down below); intriguingly, these three elements cliticize to nouns (or noun phrases).139 The next examples illustrate the use of interrogative words in action: 139. In only one example in the data uruku occurs with the past habitual =huk, which occurs with verbs. 356 (33) túwa wakɨŕam tua uakɨra-mɨ which want+IPFV-2SG.SBJ ‘Which one do you want?’ (34) jɛítʲam ia=ita=mɨ who=COP=2SG.SBJ ‘Who are you?’ (35) arútmaʃa wari ̃ḿpɛitʲa arutama=ʃa wari=̃pa=ita power.vision=ADD what=Q=COP ‘What is Arutam?’ (36) tui ̃ ́juwiʃ́a tu=i ̃ iuui=ʃa where=LOC squash=ADD ‘Where is the squash?’ (37) tuiɲ̃a ́wińam tu=i=̃ia uina-mɨ where=LOC=ABL come-2SG.SBJ ‘Where do you come from?’ (38) nihaḿt͡ʃiʃa urúk noúwɛitʲam nihamat̃͡ʃi=ʃa uruka nau-u=aita-mɨ manioc.beer=ADD how chew-NMLZ=COP-2.SBJ ‘How do you prepare manioc beer?’ 357 (39) kaʃ́iʃa itʲúrsarik maã ́jútaiɲ̃ɛiti kaʃi=ʃa itura-sa-ri=ka maã iu-tai=̃aiti night=ADD how-SUB-1PL.SS=Q kill eat-NMLZ=COP.3 ‘How do we hunt at night?’ (Lit.: ‘At night, how do we normally eat killing?’)140 (40) urútian iṕatɨarat urutian iṕatu-ara-ta when shoot.PFV-PL-IMM.FUT ‘When are they going to shoot?’ (41) urútmak humɛíɲɛiti urutima=ka hu-mai-na=iti how.many=Q take-POT-NMLZ=COP.3 ‘How many can be taken?’ (42) urúkunmak wahaś uruku=nVma=ka waha-sa ̃ how=LOC=Q stand-ATT/3SG.SS ‘How is it where he is standing?’ 8.5.1. Morphosyntactic properties of interrogative words Interrogative words constitute a class that is a little bit uneven. Some interrogative words are clearly pronominal (for instance ia ‘who’) while others possess 140. The construction in which a verb is nominalized with the Patient/Location nominalizer -tai ̃ plus a 3 person copula is a normative construction in which the semantic subject is understood as a 1PL (hence the 1PL.SS marker -ri occurring on the interrogative word). 358 some adverbial and verb-like properties (like the words based on uru and itura). The basic morphosyntactic properties that serve to group interrogative words together are: • The presence of interrogative words mark the clause as a question. Interrogative words specify the information that is being requested. • Syntactically, interrogative words remain “in situ” depending on the gramamtical role they ful"ll and are rarely moved to the front of the clause. That question words occur in their “normal” position is apparently a frequent pattern of predicate-"nal languages (Payne 1997: 301). • Two /oating clitics occur only in the presence of interrogative words, =ki ‘Rhetorical question’ and =pa ‘Question marker’. • Ability to take case and post-position clitics, as well as other nominal-related morphology in a similar way that nominals do. • In addition, interrogative words derived with the (semantically more adverbial) roots uru and itura can occur with subordinators and have a special set of person markers. In what follows, I illustrate the morphosyntactic properties outlined above. • Interrogatives words mark the clause as a question The presence of an interrogative word imparts content interrogative mood to the 359 clause where it occurs. This can be seen in the fact that no other mood su#xes occur on the verb. Compare (43) with (44): (43) wií napiń wakɨŕahɛɛ ui napi=na uakɨra-ha-i 1SG snake=ACC want+IPFV-1SG.SBJ-DECL ‘I like snakes.’ (44) wiʃ́a wariń wakɨŕah ui=ʃa wari=na uakɨra-ha 1SG=ADD what=ACC want+IPFV-1SG.SBJ ‘What do I like?’ In (43), the declarative mood is marked with -i on the verb. On the other hand, the morphological position for mood (which is the last position in the verb piece) in (44) is not phonologically "lled. In other words, interrogative mood is marked by the presence of the interrogative word and the absence of other mood su#xes on the verb. • Interrogative words remain “in situ” and are rarely moved to the front In Wampis, an interrogative word is rarely moved to the front of the clause; rather, interrogative words usually remain in their “normal” order where the constituent replaced by the interrogative would be expected to occur. The canonical order of major clauses elements in Wampis is Subject Object Verb (or APV/SV). 360 Adverbial complements normally precede the verb. In (45), the interrogative word ‘who’ occurs in the normal position that subjects occupy, preceding the object. (45) A P Verb ya aḿiɲa awat́irmam ia amɨ auati-rama-ma who 2SG hit+LOAF-2.OBJ-IMM.PT ‘Who hit you?’ In (46), the complex subject ‘Ipak and Sua’ occurs first, and the interrogative word ‘how’ occurs in the position where the adverbial complement would be expected to occur. (46) S Verb ipʲaḱ suwah́ɛɛ ̃uruḱ nahańaruwait Ipaku Sua=hai ̃ uruka nahana-ara-u=aita Achiote Huito=COM how make.PFV-PL-NMLZ=COP.3 ‘How did Ipak and Sua transform?’ In (47) we have an example with all arguments of the verb ‘give’. Wampis has a symmetrical object system in term of case marking (both objects of a ditransitive verbs are marked with =na ‘Accusative’). The subject and the two objects precede the verb. The interrogative word ‘who’ is in the place usually occupied by the recipient in 361 ditransitive constructions. (47) A T R Verb aḿɨ tawasań jańaʃa susat́tam amɨ tauasa=na ia=na=ʃa su-sa-tata-mɨ 2SG feathered.crown=ACC who=ACC=ADD give-ATT-DEF.FUT-2PL.SBJ ‘To whom are you going to give the tawas?’ In (48) the question word ‘how’ precedes the verb, as normal adverbial complements do. (48) S Verb nitʲa jaaḱtanam uruḱ puhúiɲawa nita iaakata=nama uruka puhu-ina-ua 3PL town=LOC how live-PL.IPFV-3.SBJ ‘How do they live in the city? • Presence of 0oating clitics =ki ‘Rhetorical question’ and =pa ‘Question marker’ Two /oating clitics only occur in interrogative clauses when an interrogative word is present: =ki ‘Rhetorical question’141 and =pa ‘Question marker’. Both =ki and =pa can occur either in the interrogative word, as in (49) and (51); or in some other constituent of the clause, as in (50), (52). The common element is that the interrogative 141. This =ki also frequently occurs attached to ia ‘who’ in contexts where there is no rhetorical question. This may be related to a hypothesized old function of =ki as a focus-marker: in Awajun, a cognate form -ki is apparently used as a focus marker (Overall 2007: 182). This function is not clearly present in Wampis. 362 word needs to be present in the clause for these clitics to occur. (49) arútmaʃa wariḿpɛit arutama=ʃa uari=̃pa=ita power.vision=ADD what=Q=COP ‘What is Arutam?’ (50) puertaʃ́a wari ̃ ́tutɛímpɛit puerta=ʃa uari ̃ tu-tai=̃pa=ita door=ADD what say-NMLZ=Q=COP ‘How do you say door (i.e. how do you say door in Wampis?)?’ (51) ajaḿkaʃa at͡sóo jaḱi ajamrukat aiamaka=ʃa a-t͡sa-u ia=ki aiamaru-ka-ta right=ADD exist-NEG+IPFV-NMLZ who=RHET defend-INTENS-IMM.FUT ‘[In the old times] There was no rights either, who was going to defend them?’142 (52) ií urúk puhúmɛiŋkit ii uruka puhu-mai-inu=ki=ita 1PL how live-POT-NMLZ=RHET=COP ‘How can we live?’ • Ability to take case, postpositions and other morphology related to the noun Interrogative words receive case and postpositional clitics, as well as 142. The speaker is talking about how the Wampis sometimes had to "ght back Peruvian’s soldiers or police troops that sometimes had abusive behavior toward them. The question is rhetorical; what the speaker tries to say is that there was no other way to stop the abuse but defending themselves. 363 derivational morphology associated to the Noun. This is a property that interrogative words share with other pro-nominal classes (see Table 8.4 for case and postpositions that can be occur on pronouns). The next examples illustrate interrogative words with di$erent morphology associated with nominals (diminutive -t͡ʃi, benefactive -nau, accusative =na, locative =(n)i,̃ ablative =ia, and comitative =hai,̃ respectively). (53) jat͡ʃit́mɨ ia-t͡ʃi-ita-mɨ who-DIM=COP-2SG.SBJ ‘Who are you?’143 (54) húʃa janówɛiti hu=ʃa ia-nau=aiti PROX=ADD who-BEN-COP.3 ‘For whom/whose is this?’ (55) warı́n wɛı́nkʲajam uari=̃na uaina-ka-ia-mɨ what=ACC see-INTENS-REM.PT-2SG.SBJ ‘What did you see?’ (56) tuiɲ̃aítmɨ tu=i=̃ia=ita=mɨ where=LOC=ABL=COP=2SG.SBJ ‘Where are you from? 143. The diminutive is used to be polite. 364 (57) wiíʃa kut͡ʃi ́wari ̃h́ⁱɛɛ ̃sumaŕmaktahɛɛ ui=ʃa kut͡ʃi ́ uari=̃hai ̃ suma-ru-ma-ka-ta-ha-i 1SG=ADD pig\ACC what=COM buy-APP-REFL-INTENS-IMM.FUT-1SG.SBJ-DECL ‘With what am I going to buy me pigs?’ • Interrogative words receive a special set of person markers and can carry subordinators Wampis interrogative words uruka and itura have some unusually complex person marking. These interrogative words employ unique person su#xes that mark the person of the subject of their clause (Table 8.8). These person markers di$er from the set of person markers used with "nite declarative verbs. Question words can also occur with some subordinating su#xes (see Table 8.9 below): the person markers occur optionally (but frequently) when the interrogative word bears the subordinating su#xes, but they occur rarely when the interrogative word stands alone. For comparison, person markers of declarative forms are also provided in Table 8.8 (for analysis of person marking on verbs, see §14.3). To the best of my knowledge, these unique interrogative-based person markers (as well as the subordinators) do not occur with ia ‘who’, tu ‘where, which’ and uari ̃‘what’. 365 Table 8.8. Person markers that occur on question words uruka and itura Category Interrogative person markersa Interrogative word ~ Subordinate Declarative person markers 1SG -na ~ -nu -ha 2SG -mɨa ~ -mɨ -mɨ 3SG -nia ~ -ni -ua (non-past), -ji (past) 1PL -ria ~ -ri -hi 2PL -rumɨa ~ -rumɨ -rumɨ 3PL -ara (with -sa) ~ ∅ -∅ a For 3PL and 1PL, the plural imperfective -ina can also occur. Nevertheless, in general -ina marks number, rather than person (§13.3.3). The person markers on the left hand within the “Interrogative person markers” column in Table 8.8 mostly occur with interrogative words: (58) urúkakrumɨa aʃaḿarmɨ uruka-ku-rumɨa aʃama-rumɨ how+IPFV-SIM-2PL be.afraid+IPFV-2PL.SBJ ‘Why are you (PL) afraid?’ There are rare examples of the person markers listed in Table 8.8 occurring on the verb instead of on the interrogative word: (59) urúkak nɨkaṕɨamɨa uruka=ku nɨkapɨ-a-mɨa how+IPFV=SIM feel-IPFV-2SG ‘How are you feeling?’ On the other hand, the person-marking forms on the right hand within the “Interrogative person marker” column occur in subordinate verbs and optionally in 366 question words as a variants of the su#xes on the left hand144 (see Chapter XIX for a complete analysis of subordination in Wampis). From Table 8.8, it seems that interrogative-word person markers are historically complex, containing an erstwhile person su#x and the formative *a. This vowel a is probably related to the copula a (see §17.5) that has been reanalyzed as part of the person marker. It is possible that the copula a was “recycled” after itself being dropped at a previous historical stage, e.g. *uruka a-sa-mɨ ‘how COP-SUB-2SG > uruka ∅-sa-mɨ > uruka-sa-mɨ a ‘how-SUB-2SG COP’ > uruka-sa-mɨa ‘how-SUB-2SG’. In fact, constructions with a morphologically simple interrogative word and a separate subordinated copula can still be found in the language: (60) óo urúk asań nuwɨḿ aańia au uruka a-sa-ni nuɨ-mɨ aani-a DIST how COP-SUB-3SG wife\POSS-2SG do.that-IPFV ‘Why is your wife like that?’ As stated in the preceding discussion, the interrogative words derived from uru and itura can also carry some subordinating morphology, as well as the switch-reference -matai.̃ The subordinating and switch-reference morphemes found on interrogatives are indicated in Table 8.9. 144. Except the 3SG -ni that only occurs in question words (in subordinate verbs, third person is marked with nasalization on the last vowel of the stem). 367 Table 8.9. Attested subordinating and switch-reference morphology associated with interrogative words Morpheme Gloss -sa Non-temporal -ku Simultaneous -tasa Purpose -mataiã 1SG/3 Di$erent subject a -matai ̃is used everywhere else in sequential subordinate clauses, but it seems like it is becoming fossilized in question words (see detailed explanation below). The interrogative itura occurs mostly with the subordinator -sa, as in (61). (61) kaʃ́iʃa itʲursariak maã ́jutaiɲ̃ɛiti kaʃi=ʃa itura-sa-ria=ka maã iu-tai=̃aiti night=ADD how-SUB-1PL=Q kill eat-NMLZ=COP.3 ‘How do we hunt at night?’ (Lit.: ‘At night, how do we normally eat killing?’) On the other hand, uru-based interrogative words occur with any of the four su#xes presented in Table 8.9. The meaning of uruka ‘how’ apparently changes according to the marker it receives: with -tasa it means ‘what for’; with -sa and -ku, it most frequently means ‘how’ but sometimes speakers interpret it as ‘why’; with -matai,̃ it means ‘why’. The simultaneous -ku in question words is used when a related action in the discourse is seen as ongoing. In (62), uruka ‘how’ occurs with the simultaneous -ku (notice in this case the 2PL person marking on the question word). The question in the example is asked with a reason sense of ‘why is this happening to you?’ (a free 368 translation equivalent would be: ‘What’s up with you?’). Notice that the imperfective marker -a occurs with the simultaneous -ku (this happens regularly in subordinate verbs (§19.6). (62) húʃa urukaḱrumɨa hu=ʃa uruka-ku-rumɨa PROX=ADD how+IPFV-SIM-2PL ‘How is this?’ (i.e. ‘Why is this happening to you?’) (63) húʃa urúkakrik iíʃa húniah hu=ʃa uruka-ku-ri=ka ii=ʃa huni-a-hi prox=add how+IPFV-SIM-1PL=Q 1PL=ADD do.this-IPFV-1PL.SBJ ‘How are we doing like this?’ The reason I analyze the form uruka as ‘how+IPFV’ is that the vowel /a/ of uruka does not undergo elision in spite of being in a place to be deleted (Wampis deletes the third vowel from the left, as explained in Chapter VI). So, my analysis is that the imperfective fuses with the last vowel /a/ of uruka, blocking vowel elision (this is a regular pattern in Wampis, as explained in §5.3.5). Compare with the surface form of uruka in (64), where the vowel /a/ is deleted. So, morphologically, it appears as if the interrogative word is treated like a verbal stem. The facts that uruka can take subject markers, subordinators and apparently imperfective aspect makes them “clausal” (or verbal) in behavior. 369 Example (64) illustrates the occurrence of uruka with the non-temporal subordinator -sa (notice in this example there is no person marking on the interrogative word because of the presence of the copula, which takes the person marking). (64) ij́akur wɨkah́i urúksakit túsar ii-a-ku-ri wɨka-hi see-IPFV-SIM-1PL.SS walk+IPFV-1PL.SBJ+DECL uruka-sa=ki=ita tu-sa-ri how-SUB=RHET=COP.3 say-SUB-1PL.SS ‘We go on observing, thinking “how is it?”’ (Lit.: ‘. . . saying: how is it?’) With the switch-reference -matai,̃ the interrogative word uruka acquires the meaning of ‘why’. The su#x -matai ̃is a 1SG/3 di$erent-subject switch-reference marker that is used everywhere else in sequential subordinate clauses (§19.6.2). The DS function can be seen in examples like (65), where the subjects of the two clauses are di$erent: (65) urúkamtɛiñia nuhińʃa atsa ́aḿɨ wɛiŋ́kamam uruka-matai ̃ nuhinta=ʃa a-tsa amɨ uaina-ka-ma-mɨ how-1SG/3.DS egg=ADD exist-NEG+IPFV 2SG see-INTENS-REC.PT-2SG.SBJ ‘Why are there no eggs, [if] you have seen [that there were].’ However, it appears that for many speakers -matai ̃is being reanalyzed as an integral part of uruka, to the point where the word urukamatai ̃has come to be 370 considered an unanalyzable word meaning ‘why’.145 Thus, it is not unusual to hear expressions like the one in (66), where matai ̃is not marking 1SG/3 person or di$erent subject.146 (66) urúkamtai ̃wɨám aḿɨ juwat́͡ʃu urukamatai ̃ uɨ-a-mɨ amɨ iu-a-́t͡ʃau why go-IPFV-2SG 2SG eat-HIAF-NEG.NMLZ ‘Why are you leaving [when/if] you have not eaten?’ Undoubtedly, the oddities of grammaticalization have led to di$erent kinds of morphosyntantic behavior in interrogatives. As we have seen, in some cases, it appears that some complex expressions are being grammaticalized and behave already as synchronically unanalyzable question words. As a "nal point, notice that uruka and itura can replace not only NPs but also predicates. When replacing predicates, these question words take all the morphological markers of the replaced verb. 145. Perhaps a further step in the grammaticalization of urukamatai ̃is the phonetic reduction that the word frequently undergoes: in casual speech, speakers reduce the pronunciation of urukamatai ̃to [uraḿtɛɛ]̃. 146. One could argue that the answer to the whole question necessarily involves a di$erent grammatical subject, as the interlocutor will respond using the "rst person. Regardless, everywhere else -matai ̃is not used with a 2SG subject in the subordinate clause, -matai ̃normally marks that a 1SG/3 subject of a subordinate clause and the subject of the main clause are di$erent. 371 (67) ií urúk puhúmɛinkit urúkmɛinkit ii uruka puhu-mai-inu=ki=ita uruka-mai-inu=ki=ita 1PL how live-POT-NMLZ=RHET=COP how-POT-NMLZ=RHET=COP ‘How can we live? How can we [do]?’ In the context where the above example comes from, the second clause was translated as <¿cómo podemos hacer?> ‘how can we do [it]?’. There even seems to be some lexical content to the interrogative word. This and the previous examples suggest that it is possible to consider some question words as interrogative verbs, as they are able to receive some categories associated with verbs. In fact, some examples suggest that at least uruka could be treated as a question verb. In (68), the interrogative word uruka receives the future marker -ta and the 1PL marker -hi (which is taken from the set that occurs only with "nite declarative verbs), just like “normal” lexical verbs in declarative mood can. Notice also that the translation suggests some lexical meaning to the interrogative word, as there is no elided verb “do” in the Wampis structure. (68) urúkathik jamɛíʃa iíŋkʲa aʃi ́amútmaktathiapi uruka-ta-hi=ka iamai=ʃa ii=na=ka aʃi how-IMM.FUT-1PL.SBJ=Q now=ADD 1PL=ACC=ADD all amu-tama-ka-tata-hi-api "nish-1PL.OBJ-DEF.FUT-1PL.SBJ-SUD.REALZ ‘How are we gonna do? Now they are going to kill us all!’ 372 While these examples are not numerous, they are not quite infrequent. In fact, similar examples evidence that the morphosyntactic possibilities of interrogative words in Wampis are unusually complex. Future research with additional data may shed more light on the complex properties of Wampis interrogatives. 8.5.2. Interrogative words in non-interrogative constructions In many languages, interrogative words are similar or identical to pronouns, especially demonstrative pronouns. Thus they often further function as relativization pronouns or complementizers, a common extended function of demonstratives. In Wampis, interrogative words are not clearly synchronically associated to any demonstrative, as far as I can tell. However, interrogative words can occur in di$erent non-interrogative constructions such as relative clauses, suggesting a possible historical link to a demonstrative. With regard to relative clauses, ia ‘who’, tui ̃‘where’ and wari ̃ ‘what’ can occur in “headless” relatives clauses. All examples of relative clauses involving the interrogative ia ‘who’ show it in a fossilized form iaki < ia=ki ‘who-RHET’. (69) jaaḱmi jaḱi ima ́pɨŋ́kɨran jaaḱat nuniḱuka kaḿɨ núka mɨsɨt́numʃa ʃiír wɨt́in jaa-ka-mi [iaki ima pɨŋkɨra=na jaa-ka-ta nuni-ku]=ka jaa-INTENS-HORT who INTS good=ACC jaa-INTENS-NMLZ do.that-SIM=FOC 373 kamɨ nu=ka mɨsɨta=numa=ʃa ʃiira wɨ-tinu INTERJ NON.VIS=FOC war=LOC=ADD well go-FUT.NMLZ ‘Let’s do (the challenge with) the yaa!,147 [to whom is very good doing the yaa!], to that (person), it will go well in the war.’ Example (70) shows a relativization with tui ̃‘where’. (70) tuiń kúntin t͡ʃit́͡ʃa ́ańu wɨkat́usar ɨaḱir [tui ̃ kuntina t͡ʃit͡ʃa a=nu] wɨka-tu-sa-ri where animal speak COP=NON.VIS walk-APPL-ATT-1PL.SS ɨa-ki-ri look.for-WHILE.MOVING-1PL.SS ‘[Where the animals make sounds], we having walked, having looked around . . .’ In addition, uruka ‘how’ occurs in adverbial simile constructions. In simile constructions, uru-based interrogative words are treated as verbal bases in that they have to be nominalized and received a copula in order to function as adverbials. It is possible that this is a syntactic calque from Spanish, where ‘how/like’ also functions in simile constructions. An example of a simile construction with uruka is given in (71), with the nominalized variant uruku: (71) ʃuaŕ maãńin urúkuwɛit núnis hiɲaẃɛɛ ʃuar maã-nai-inu uruku=aita nuni-sa ̃ hina-wa-i person kill-RECP-NMLZ how=COP.3 do.that-SUB\3SG.SS die.PL+IPFV-3.SBJ-DECL ‘They died like warriors.’ (Lit.: ‘Like is a person warrior, doing that, they died.’) 147. The exclamation jaa is what warriors used to yell as a ritual challenge before "ghting. 374 8.6. Adverbs Adverbs as a lexical class di$er from prototypical open classes like nouns and verbs in that they have very little morphological possibilities. Adverbs have a very strong preference for occurring before the predicate, and only locative adverbs show some relative freedom to occur after the predicate.148 Adverbs are considered a semi- closed class: there is only one element that derives temporal adverbs, -tin ‘time’, that creates words referring to dates and seasons; but it is not very frequently used anymore. Some adverbs seem like grammaticalizations bearing the non-temporal subordinator -sa. For this grammar of Wampis, what I call adverbs cannot be characterized semantically with a single prototype. However, I de"ne adverbs as a functional category by contrasting this group with the functional prototypes of other categories. Thus, unlike adjectives, adverbs do not function attributively; unlike verbs, adverbs do not function predicatively or carry "nite morphology; and unlike nouns, adverbs do not 148. In a count of 120 di$erent clauses where an adverb was present, there were only 8 instances where adverbs occurred after the predicate (and two of those instances were problematic to analyze, as the adverb occurred after a relatively long pause). I did not take into consideration clauses with locative adverbs for this count, as the relative freedom that they have to occur before or after the predicate may not re/ect the syntactic behavior of adverbs in general. 375 have a referential function.149 Morphosyntactically, unlike nouns, adverbs do not head NPs, they cannot be possessed or be a possessor, and they cannot be modi"ed attributively; unlike adjectives, adverbs do not occur in comparative or superlative constructions. Rather, what I call adverbs modify verbs and adjectives, and impart temporal, locative, or circumstantial information at the clausal level. According to Schachter and Shopen, “adverbs function as modi"ers of constituents other than nouns” (2007: 20). However, in Wampis some of the items identi"ed as adverbs by the criteria above can also modify nouns. Semantically, adverbs are divided in four categories: manner (with several sub- categories), time, location, and the adverbs aía and aíatɨkɨ which form a small sub-class. Non-numeral quanti"ers can also function as adverbs, but apart from their adverbial function, they have determiner and pronominal functions. Non-numeral quanti"ers are treated in §8.7. 8.6.1. Manner adverbs Manner adverbs present how an action is carried out. Wampis manner adverbs modify an action by adding information including: modality, demonstrative manner, 149. Except for demonstrative adverbs and time adverbs, which somewhat provide deictic information. 376 intensity, speed, aspect of an event (e.g., repetition) or other manner category. Table 8.10 presents a semantic classi"cation of manner adverbs in Wampis. Adverbs that are marked with a star * form a subclass that has as its common property that the items in/ect for person (this is discussed further on in this same section). Table 8.10. Semantic organization of manner adverbs in Wampis Modality Gloss (Possible analysis, if available) aantara* ‘in vain’ nankamasa* ‘just in any way’, ‘disorderly’ nɨkasa*,a ‘truly’ ( ‘strong, valorous’ iurata ‘too much’, ‘excessively’, ‘extremely’ hɨɨpata ‘too much’, ‘excessively’, ‘extremely’ Speed iaitasa* ‘slowly’ uarib ‘quickly’ uaamakɨ ‘quickly’ Aspect atakʃa ‘again’ arumakɨ ‘immediately’ (< aruma ‘later’+=kɨ ‘RESTR’ ) arurumɨn ‘constantly’ t͡sɨkɨn ‘suddenly’, ‘rapidly’ Other ʃiira ‘well’ (also functions as an intensi"er) kaɨmas ‘squatting (like a parrot)’ (2SG+DECL ‘I am telling you truly.’ (82) aańtrarumɨ wɨkaśmarmɨ aantara-rumɨ wɨka-sa-ma-rumɨ in.vain-2PL.SS walk-ATT-REC.PT-2PL ‘You (PL) travelled in vain.’ The marking of person on the person-marking adverbs seems to be optional, 383 because there are examples where they occur unin/ected. (83) aańtra wahat́ nakit́ʲahɛɛ aantara uaha-ta nakita-ha-i in.vain stand-NMLZ not.want+IPFV-1SG.SBJ-DECL ‘I do not want to be standing in vain.’ In light of this variation, I suggest that it is possible that person-marking adverbs, which constitute a small class, are being reanalyzed by analogy to the other adverbs which do not receive person marking. It is possible that these adverbs were old verbs and thus inherited some person marking of subordinate verbs (however, synchronically this is only clear with nɨkasa ‘truly’; there is no corresponding verb for the other adverbs listed). Overall (2007: 170) suggests that a similar set of adverbs152 in Awajun may be treated as defective verbs. The second morphologically-de"ned sub-class of manner adverbs would consist of those adverbs that do not in/ect for person (and, actually, they do not receive any morphology at all). All adverbs listed in Table 8.10 that are not marked with a star * belong in this subclass of manner adverbs. 152. None of the forms that Overall lists coincide with the forms I provide for Wampis. But there is some coincidence in the meaning (not form) of two Awajun forms: diipasa ‘slowly’ and wainaka ‘in vain’. 384 8.6.2. Time adverbs Time adverbs add temporal information to their predicates or predications: (84) miɲ́a papaŕ úhatin aj́aji nuiḱ út͡ʃi asaḿtai ̃ mina papa-ru uha-tu-inu a-ia-ji 1SG.GEN father-1SG inform-APPL-NMLZ COP-REM.PT-3.SBJ+DECL nuiki ut͡ʃi a-sa-matai ̃ time.ago child COP-SUB-1SG/3.DS ‘My father told me (this story) years ago, when I was a child.’ Time adverbs are quite frequent in texts. Often times, time adverbs occur in a relative construction. This frequently occurs when the temporal information is new to the discourse. (85) núu t͡sawańta nui ̃ŋ́ka muuka ́sukuaḿuka at͡súsuiti nu [t͡sauanta=a nui]̃=ka muuka ́ suku-a-́mau=ka NON.VIS day=COP there=FOC muuka\́ACC shrink.head-HIAF-NMLZ=FOC a-t͡su-sa-u=iti exit-NEG-ATT-NMLZ=COP.3+DECL ‘That day there (Lit.: ‘that where it was a day’), there was no head-shrinking.’ Table 8.13 presents a list of time adverbs identi"ed in the Wampis data. I have arranged the adverbs taking into account the possible presence of a shared root. In most instances, these roots are semantically opaque. Thus, roots like *aru (probably related to aru ‘be late’), *ia, and *kaʃi can be seen in di$erent historically derived words. Furthermore, it is possible to see the presence of old formatives *-ma (e.g. aru-ma 385 ‘later’), and *-i and *-ni ̃(e.g. aruma-i, iama-i, kaʃi-ni)̃ probably somehow related to the locative =(n)i.̃153 In addition, it is probable that there is an old formative *pɨ in the word ahampɨkɨ ‘at midnight’ (, o#cially , is the Peruvian currency. Wampis people also use colloquial Peruvian Spanish to name the currency: ‘sol (colloquial)’, ‘half a sol (colloquial)’. 160. However, beyond ten they almost always turn to Spanish. 398 multiplication of a base, and 3) body parts. Table 8.17. Wampis numerals Numeral Expressiona Number Meaning tikit ͡ʃikib 1 ‘one’ himara 2 ‘two’ kampatuma ~ mɨnaintu 3 ‘three’ ipak usumak ~ aintuk aintuk 4 ‘four’ uuɨhana aʃi amua ́ 5 ‘all hand "nished’ uuɨhana tikit͡ʃikinakɨ iraku 6 ‘to the hand adding one only’ uuɨhana himarana iraku 7 ‘to the hand adding two’ uuɨhana kampatuma iraku 8 ‘to the hand adding three’ uuɨhana aintuk aintuk iraku ~ uuɨhana ipak usumak iraku 9 ‘to the hand adding four’ uuɨhana mai amua ́ 10 ‘both hands "nished’ uuɨhana mai amua ́tikit͡ʃkiɲakɨ (uuɨha) iraku 15 ‘adding one (hand) to both hands "nished’ uuɨhana mai amua ́nauɨna mai amua ́ 20 ‘both hands "nished, both feet "nished’ a A short free translation of some words to help understand complex expressions in numerals 5 to 20: uuɨha ‘hand’, aʃi ‘all’, amua ́‘complete’, iraku ‘adding’, mai ‘both’. See detailed explanation below. b There is a free variant of ‘one’ that starts with a palatal a$ricate instead of the dental /t/: t͡ʃikit͡ʃiki ‘one’. Here are some interesting notes that help explain the form and meaning of some Wampis numerals: 399 • The numeral tikit ͡ʃiki ‘one’, is historically decomposable into the formatives *tikit͡ʃi ‘other’ and *kɨ ‘Restrictive’. The numeral himara ‘two’ has a root *hima, which also features in the modern root for ‘have twins’ (hima-ma ‘two?-VBZ’). • There are two forms for number ‘three’, kampatuma and mɨnaintu. The form kampatuma appears to be long enough to be historically complex, but I have no evidence for a possible lexical source in current Wampis. The form mɨnaintu [mɨnɛín] is not very frequent and it is likely related to the root mɨna ‘left’, cf. also mɨna ‘be narrow’, mɨnanV ‘put aside, dodge’. • There are two forms for number ‘four’, the reduplication aintuk aintuk (<*aintukV) and the complex form ipak usumak. There is no clear source for aintuk aintuk, but compare the roots ai-na ‘fold’, ai ̃‘be with, accompany’. The reduplication itself can be iconically associated with plurality, though it is not semantically transparent how it correlates with the number ‘four’ speci"cally. The form ipak usumak is related to the cultural practice of face-painting (<*usu-ma-a-ku ‘paint.face-REFL-IPFV-SIM’) with a tint obtained from the achiote fruit ‘Bixa orellana’ (ipaku in Wampis), which is done with the index "nger (the fourth "nger if we count, as the Wampis do, starting from the little "nger). 400 • The verb used for expressing the ‘complete’ hand is based on the root amu ‘"nish’ plus the high a$ectedness aktionsarten su#x -a;́ whereas the verb used for numbers above ‘"ve’ when the hand is not complete (e.g. ‘six’ through ‘nine’) is iru ‘accumulate, reunite, get together’. • Interestingly, historically the verb iru seems to have been subordinated with the simultaneous action -ku added to an imperfective stem: iraku <*iru-a-ku ‘accumulate- IPFV-SIM’. Thus the meaning of the counting expression with this verb would be ‘adding NUMBER to the hand’. • Because numbers above ‘four’ are based on verbal phrases (e.g. ‘"nished hand’), the word ‘hand’ (which is an argument of the verb) takes the accusative =na: uuɨhana < uuɨha=na [uwɨhań]. It is unclear whether this analysis is synchronically “live”, as such numeral expressions seem already conventionalized. In complex expressions where the ‘hand’ is not complete, both the word for ‘hand’ and the added number are marked with the accusative; for instance a historical analysis for number ‘six’ would be: (110) uwɨhań tikit́͡ʃkiɲak iŕaku uuɨha=na tikit ͡ʃiki=na=kɨ ira-ku hand=ACC tikit ͡ʃiki=ACC=RESTR accumulate+IPFV-SIM ‘to the hand adding one only’ • Above number ‘four’, speakers actually employ di$erent expressions for those just 401 described above, which suggests that the system beyond number ‘four’ is relatively new and likely a consequence of modern contact with Spanish (likely through trade) and/or of the introduction of mathematic concepts associated with school. However, all di$erent expressions are based on the same idea of ‘complete hand’ and ‘adding to the complete hand’. For instance, in number ‘"ve’, the term aʃi ‘all’ is omitted by some speakers, or an alternative expression tikit ͡ʃiki uuɨhan amua ́[tikit́͡ʃik uwɨhań amua]́ ‘one hand "nished’ is used. For numbers ‘ten’ or ‘twenty’, and so on, the word mai ‘both’ can be replaced by himara ‘two’. • Notice that for ‘hand’ there is a form ɨuɨha [ɨwɨh́] that varies freely with uuɨha [uwɨh́]. • There are no ordinal numbers and cardinal numbers do not function as ordinal numbers (except for very rare examples with tikit ͡ʃiki ‘one’) in Wampis. Speakers use tura ‘and’, nui=̃ia (<‘there=ABL’) ‘moreover, in addition’ or employ sequential subordination (§19.7) instead of enumerating a sequence of propositions one after the other (e.g. instead of saying First he went to the river; second he got in the canoe; third he started the motor, the Wampis expression would be ‘Then we went to the river; having done that, he got in the canoe; from there he started the motor’. There is also a su#x -a ́that means ‘First’ (§11.5.7). 402 Numerals do not receive case or post-positional clitics, and like non-numeral quanti"ers, they may occur before or after the noun. The following examples illustrate Wampis numerals in use. (111) nańtu kampat́um ɨhɨḿu tiḿaji nantu kampatuma ɨhɨ-mau timaji moon three make."nish-NMLZ NARR ‘Three months passed.’ (112) ɨwɨjań mɛí amua ́t͡sawań puhusań nui ̃ ́ʃiír haśmiajɛi ɨuɨha mai amu-a ́ t͡sauanta puhu-sa-nu hand both "nish-HIAF day live/be-SUB-1SG.SS nui ̃ ʃiira has-mia-ha-i there well become-DIST.PT-1SG.ABJ-DECL ‘Being ten days, there, I healed up.’ 8.8.1. Morphological properties of numerals: distributive reduplication and -a ́ ‘Iterative’ There are two unique morphological processes associated with numerals ‘one’ through ‘three’ in Wampis.161 The "rst process is partial reduplication, by which a distributive meaning is imparted. Examples (113)–(114) illustrate this process. 161. As seen earlier, the other numerals are complex expressions, which is probably the reason why they do not undergo the processes described in this section. 403 (113) núna nukɨ ̃ń tiki ́tikit́͡ʃik tiʃiṕiahi nu=na nukɨ=̃na tiki tikit͡ʃiki tiʃipi-a-hi nu=ACC leaf\1PL/2PL/3.POSS=ACC REDUP one peel-IPFV-1PL.SBJ+DECL ‘We strip the leaves o$ one by one.’ (114) himʲa ́himʲaŕ namaŋkɨń susaḿiahrumɨ hima himara namankɨ=na su-sa-mia-ha-rumɨ REDUP two meat=ACC give-ATT-DIST.PT-1SG>2PL+DECL ‘I gave two pieces of meat to each one of you.’ In a brief related note, the word ankanta ‘free, dispersed’162 is typically an adjective, but it can undergo reduplication and acquire a distributive meaning similarly to numerals.163 A reduplicated form of ankanta means ‘each one’ or sometimes more speci"cally ‘each one dispersed (‘one here, another there...’)’: (115) hintińtʲawɛiti aŋkań aŋkań út͡ʃi mat͡sat́u ah́akaru ańuna hintina-tu-a-u=aiti ankanta ankanta ut͡ʃi teach-APPL-IPFV-NMLZ=COP.3+DECL free free child mat͡satu a=hak-ara-u a=nu=na live.together COP-HAB.PT-PL-NMLZ COP-NON.VIS-ACC ‘He taught each children that used to lived [there].’ Table 8.18 shows the form of reduplication of Wampis numerals. 162. ‘Free’ in the traditional Wampis sense, i.e. living separated from each other, each one in their own space in the forest. Ankanta also means ‘single (not married)’. 163. Strictly speaking, the reduplication of ankanta is not partial reduplication (partial reduplication in Wampis is de"ned by copying all the phonetic material up to the second vowel), but the word acquires the same distributive meaning so it is mentioned here. See §7.7 for details on reduplication. 404 Table 8.18. Reduplication of numerals ‘one’, ‘two’, ‘three’ Root Distributive reduplication tikit ͡ʃiki tiki tikit͡ʃiki himara hima himara kampatum kampa kampatuma The other morphological process associated with lower numerals in Wampis concerns the occurrence of the su#x -a ́‘Iterative’. The iterative only occurs with numerals and gives the idea of ‘repetition x times’, where x=number. Examples: (116) hintʲa ́tikit͡ʃkia ́kanaŕaru hinta ́ tikit ͡ʃiki-a ́ kana-ra-ara-u trail\LOC one-ITER sleep-DISTR-PL-NMLZ ‘They slept on the road for one [day].’ (117) wií núna himara ́ɨt͡sɨŕkamhamɨ ui nu=na himara ́ ɨtsɨ-ru-ka-hamɨ 1SG NON.VIS=ACC two+ITER inform-APP-INTENS-1SG>2SG ‘I just informed you that two times.’ 8.9. Noun modi!er uɨántu ‘group’ The word uɨántu constitutes a unique class in Wampis. This word always comes after the head of an NP and adds a meaning of ‘group’ or ‘class’ represented by the head noun. 405 The following example is part of an enumeration of beings and entities related to Wampis worldview. The tihai are tall human-like beings that live in the jungle. (118) tihɛɛ́ wɨántuʃa aẃɛɛ tihai uɨántu=ʃa awai tihai group=ADD exist-3.SBJ-DECl ‘There are the Tijai (i.e. as a group) also.’ The next example was translated into Spanish as ‘sachamango (Gustavia subperba) and its varieties’: (119) apɛí wɨántun araḱmakuiti apai uɨántu=na arakama-ka-u=iti sachamango group=ACC plant-INTENS-NMLZ=COP.3+DECL ‘She planted sachamango and its varieties.’ In a way, the word uɨántu could be considered a derivational-like item, because it changes the meaning of the noun. On the other hand, it could be considered an in/ectional-like item because it adds a sense of plurality. Unlike other Wampis derivational or in/ectional categories, however, uɨántu is not an a#x. Semantically, uɨántu does not "t into the prototype of other word classes such as adjectives, adverbs or other noun modi"ers: it does not have attributive function like adjectives, it does not add manner or other type of circumstantial information like adverbs, it does not have deictic or pronominal functions like pro-words or demonstratives, and it does not have 406 a referential function like nouns. Syntactically, unlike other noun modi"ers, uɨántu always comes after the noun it modi"es. In addition, uɨántu does not agree with the noun when there is a demonstrative in the NP (unlike nouns and adjectives, see §8.3.1). The source of this word could be related to an old root *uɨa.̃ The word uɨa in current Wampis means ‘father in law’, which does not seem semantically relatable to the idea of ‘group’. However, apparently in Achuar-Shiwiar the form (/uɨa/) means ‘relative’ (Fast et al. 1996: 322), in Shuar the form (/uɨatu/?) means ‘ancestor’ (Pellizaro & Náwech 2005: 489) and in Awajun ̃ (/uɨã/̃) means also ‘ancestor’ (Uwarai Yagkug et al. 1998: 53). It may be the case that the idea of group is a semantic extension of a derivation of *uɨa ̃ ‘ancestor’. In the Wampis vocabulary by Jakway et al. the entry is given with the meaning ‘descendant’ (but there is no single entry for ) (1987: 69). 8.10. Conjunctions Conjunctions are words that are used to connect phrases and clauses. There is actually little basis to claim that there are true conjunctions as a word class in Wampis. On the other hand, the language makes plentiful use of hypotactic strategies for connecting clauses. I de"ne “conjoining” in terms of coordination: two clauses (and in 407 the case of tura also phrases) that are not syntactically dependent one on another. In other words, conjunctions join two or more elements of the same rank. At the inter- sentential level, conjunctions join independent verbs which can otherwise stand alone as an assertion (Longacre 1970; Longacre 2007). Table 8.19 lists the words that assume a conjoining function. Table 8.19. Words that function as conjunctions in Wampis Word Gloss Connects tura ‘and’, ‘then’ (occasionally also ‘but’) phrases, independent clauses nuiĩa [nuiɲ̃a]́ ‘moreover’, ‘also’, ‘apart from that’ independent clauses ant͡sa ~ ant͡su ‘but’, ‘rather’, ‘on the other hand’ independent clauses These three words come from di$erent sources and show very little in common in terms of semantic prototypes or morphosyntactic properties (the only element in common is that they assume a conjoining function). The word tura can join phrases or clauses. In (120), tura coordinates a complex NP. (120) awarún túra wampisti ́ auahuna tura uampisa=ti ́ Awajun and Wampis=SAP ‘We the Awajun and the Wampis’ Examples (121)–(122) illustrate tura connecting clauses. Notice that both joined 408 clauses are independent clauses. For instance, in (121) both clauses occur with the declarative su#x, which indicates that they are independent (subordinate verbs do not receive mood markers (see Chapter XIX for an analysis of subordination)). (121) wikʲa júakun puhah́ɛɛ túra ni ̃ŋ́ka kanaḱ tɨpaẃɛɛ ui=ka iu-a-ku-nu puha-ha-i tura 1SG=FOC eat-IPFV-1SG.SS live/be+IPFV-1SG.SBJ-DECL and ni=̃ka kana-kũ tɨpa-ua-i 3SG=FOC sleep+IPFV=SIM\3SG.SS lie.down-3.SBJ-DECL ‘I am eating and you are sleeping.’ (122) arańtusa iit́ʲɛi ̃túra ɲawaã ́arútmaʃa ah́aku tiḿaji arantusa ii-tai ̃ tura with.respect see-NMLZ and iauaã ́ arutama=ʃa a-hak-u timaji jaguar power.vision=ADD exist-HAB.PT-NMLZ NARR ‘[The Arutam] we normally see with respect, and there used to be Arutam-jaguar also’ The word tura, with the addition of the concessive -ʃa, acquires the meaning of ‘but’: (123) iíkʲa kuntiń aj́ahi turaʃ́a ií jamɛíkʲa ʃuaŕɛithi ii=ka kuntina a-ia-hi tura=ʃa 1PL=FOC animal COP-REM.PT-1PL.SBJ+DECL and=CONCESS ii iamai=ka ʃuara=ita-hi 1PL now=FOC person=COP-1PL.SBJ+DECL ‘We were animals, but now we are people.’164 164. This utterance may be a little confusing. The speaker is referring to the idea, within the 409 The derived form turaʃkuʃa (< *tura-ʃa-ku?=ʃa? ‘then-CONCESS-SIM?=ADD?) means ‘or’, ‘or if not’. (124) ʃuaŕkɨʃa maãńin turaʃ́kuʃa umit́makanʃa haḿɛin ʃuara=kɨ=ʃa maã-nai-inu turaʃkuʃa enemy=RESTR=ADD kill-RECP-NMLZ or.if.not umitVma-ka=̃ʃa ha-mai-inu damage-INTENS\3SG.SS=ADD die-POT-NMLZ ‘The enemy could kill him, or if not, having hurt himself, he could die.’ Tura must have been an old verb, because it sometimes occurs with switch- reference markers and can be nominalized with Set II nominalizer -mau ‘Non-subject nominalizer’ (cf. §15.4.6). In the following example, tura occurs with the switch- reference -matai ̃‘1SG/3.DS’: (125) nuwańt͡ʃa t͡ʃit͡ʃaŕu hakúiti t͡ʃit͡ʃaŕkowɛiti turaḿtai ̃tikit͡ʃɲaʃa atiḱsan nua=na=ʃa t͡ʃit͡ʃa-ru hak-u=iti woman=ACC=ADD speak-APPL HAB.PT-NMLZ=COP.3+DECL t͡ʃit͡ʃa-ru-ka-u=aiti speak-APPL-INTENS-NMLZ=COP.3+DECL tura-matai ̃ tikit ͡ʃi=na=ʃa atika-sa ̃ and-1SG/3.DS other=ACC=ADD DO-SUB\3.SS ‘[he] used to give advice to the women, he gave them advice, and doing that to the oth- er (i.e. the men) too . . .’ What is more, in a few examples tura is translated as ‘happen so’ or ‘do so’: Wampis worldview, that in the mythical past animals used to be people and people used to be animals. 410 (126) túrakɛɛ ̃nańtu juatahkama ́tuhiń tura-ka-i ̃ Nantu iu-a-tahkama ́ tuhintu and+IPFV-SIM-DS Moon eat-IPFV-FRUST be.unable ‘[and] while that was happening, Nantu wanting to eat, he could not.’ When tura is nominalized with Set II nominalizer -mau, it acquires the more temporal meaning ‘afterward’ (perhaps from an old nominalization meaning ‘done so’ or ‘happened so’?). (127) túramu Iwʲańt͡ʃ wɨú tiḿaji tura-mau iuant͡ʃi uɨ-u timaji and-NMLZ Iwanch go-NMLZ NARR ‘Afterward, Iwanch went away.’ The word nuiĩa is derived from the non-visible demonstrative nu plus the postpositions =i=̃ia ‘=LOC=ABL’. Literally it means ‘from there’, but as a conjunction it is interpreted as ‘also’, ‘moreover’, ‘apart from that’, depending on the context. Unlike tura, which can function at the level of the sentence or phrase, nuiĩa works at the inter- clausal level. (128) sɨaḿu tiḿaji núu núwa mama ́kunturi ̃á óona nuiɲ̃a ́paańmant͡sint͡ʃa sɨaḿu tiḿaji [sɨa-mau timaji nu nua plead+IPFV-NMLZ NARR NON.VIS woman mama ́ kuntu-ri=̃a au=na] nu=i=̃ia manioc\GEN arm-1PL/2PL/3.POSS=COP DIST=ACC NON.VIS=LOC=ABL [paantama-t͡ʃi=na=ʃa sɨa-mau timaji] plantain-DIM=ACC=ADD plead+IPFV-NMLZ NARR ‘That woman pleaded with the manioc’s branches (i.e. for food), moreover, she pleaded with the little plantain too.’ 411 The word ant͡sa~ant͡su does not in/ect or receive any morphology. It marks contrast between the two clauses it joins. Semantically, ant͡sa~ant͡su assumes di$erent meaning associated with the concept of “contrast”: ‘but’, ‘rather’, ‘on the other hand’. (129) túra ańt͡su wiḱʲa maat́ahjamań ʃiír tuhińɲɛithɛɛ tura ant͡su ui=ka ma-̃a-́tahkama-́nu and but 1SG=FOC kill-HIAF-FRUST-1SG.SS ʃiira tuhintu=aita-ha-i very be.unable=COP-1SG.SBJ-DECL ‘[My wife is good at "shing...] and, on the other hand, when I want to "sh, I never can.’ (Lit.: ‘I am very unable’.) 8.11. Interjections Interjections are “words that can constitute utterances in themselves, and that usually have no syntactic connection to any other words that may co-occur with them” (Schachter & Shopen 2007: 57). Interjections usually exhibit special phonological status, and sometimes may disregard the language’s normal phonological principles and restrictions. For instance mah ‘surprise’, ‘admiration’, does not have the two-mora requisite of phonological words in Wampis. Morphologically, the particularity of interjections is that they do not take any a#x or clitic. Syntactically, interjections usually are not part of clauses, but they behave like a whole utterance on their own. Table 8.20 provides a list of frequent interjections in the data. 412 Table 8.20. Common interjections in Wampis Speaker attitude or emotion Gloss aiu ‘okay’, speakers agrees haʔhai expression of laughter when something is funny (frequent in female register) mah speaker is surprised, admiration t͡ʃai ‘boo!’ (expression to scare someone) t͡ʃúa ‘that’s bad!’, ‘dang!’ sɨɨ́ ‘what a shame’, ‘what a pity’ kut͡ʃi ́kurún ‘found you!’ (children playing hide-and-seek) Phatic expression au speaker responds to someone calling his/her name ha ‘true!’ ‘I agree’ ɨɨ́ ‘uh-huh’ sɨɨ ́ ‘thanks’ t͡ʃii ‘yes’, ‘I see’ haa ‘right?’, ‘huh?’ (frequent in female register) Answers to polar questions at͡sa,́ t͡sa ́ ‘no’ haʔa ́ ‘yes’ ɨɨ́ ‘yes’ t͡ʃaʔa ́ ‘I don’t know’ Idea of command hasta ́ ‘Wait!’ pai ‘Ready!’, ‘Stop (enough)!’ iamaikika ‘(Wait) just a little bit’ 413 ɨɨss speaker scolds a dog Discourse-oriented kamɨ ‘so’, repair naa ‘hesitation’ nɨkas ‘truly’ maa ‘hesitation’ tukɨ ‘well’, ‘ok, then’ 8.12. Ideophones Ideophones are sound-symbolic words, i.e. words that provide a “vivid representation in sound . . . in respect to manner, colour, sound, smell, action, state or intensity” (Doke 1935: 118) and whose “meanings are in the domain of sensory imagery” (Dingemanse 2014: 387). Ideophones are widespread in languages of the Amazonia (Payne 1999a) and they are certainly very important in Wampis discourse. I include onomatopoetic words (words that mimic phonetically the source of sound they try to describe) in the class of ideophones. Ideophones may present irregular phonological structure. However, an interesting "nding in Wampis is that many ideophones are formed based on the template CVCV(C) (see Table 8.21 for examples). Ideophones in Wampis can function adverbially (an important presence in that 414 function), modifying predicates. The following examples illustrate this pattern. In (130), the ideophone tapit́ modi"es the verb ‘grab’. Tapit́ is an ideophone that conveys the meaning of grabbing something or someone suddenly or forcefully. It frequently occurs with at͡ʃi ‘grab’. (130) wɨnɨ ̃ńa tapit́ at͡ʃiḱ jui ̃ ́juaḿ haḿa wɨnɨ=na tapit́ at͡ʃi-ka ̃ mouth\1PL/2PL/3.POSS=ACC IDEO grab-INTENS\3SG.SS hui ̃ju-a-́mɨ hama here eat-HIAF-2SG.SBJ MIR ‘[He] having grabbed suddenly (tapit́!) her mouth, [he said] “Here you ate!” . . .’ In (131), the ideophone used is t͡sakɨt́, which conveys the idea of thrusting or driving an object into the soil. (131) naŋki ́hukiár hiiɲ́um kaḿɨ t͡sakɨt́ t͡sakɨt́ akunaŕ nanki ́ hu-ki-ara hii=numa kamɨ spear\ACC take-WHILE.MOVING-3PL.SS "re=LOC INTERJ t͡sakɨt́ t͡sakɨt́ akuna-́ara IDEO IDEO embed+HIAF-3PL.SS ‘[They] Having taken spears, around the "rewood t ͡sakɨt! [they] having embedded [the spears]’ Ideophones di$er from adverbs in that ideophones are not only some sort of descriptive or circumstantial modi"er but they can occur without a verb and still depict 415 an event, action or situation, i.e. they are capable of recreating “aspects of sensory imagery, much like quotations recreate aspects of the experience of the quoted speech” (Dingemanse 2014: 388).165 For instance, in (132), the ideophone pɨh, which conveys the idea of jumping, su#ces to recreate the action that it depicts; without any co- occurring verb: (132) naŋki ́at͡ʃiḱ huni ́pɨh́ huni ́pɨh́ nanki ́ at͡ʃi-ka ̃ hu=ni pɨh́ hu=ni pɨh́ spear\ACC grab-INTENS\3SG.SS PROX=ALL IDEO PROX=ALL IDEO ‘Having grabbed the spear, he jumped over here, over there.’ (Lit: ‘Having grabbed the spear, over here, pɨh́!, over here, pɨh́!’) In the same vein, in (133) the ideophone parat́, which depicts an object falling into the "re, alone su#ces to convey that meaning: (133) ampúʃ hiiɲ́um parat́ hiiɲ́um parat́ ampuʃ hii=numa parat hii=numa parat owl "re=LOC IDEO "re=LOC IDEO ‘The owl fell into the "re!’ (Lit.: ‘The owl into the "re parat!’) When used in discourse, ideophones that imitate or are associated to the sounds 165. Nuckolls (1996) also calls attention to the performative nature of ideophones, which involve speakers’ iconic and indexical gestures and particular intonational dynamics (raising or lowering in pitch, lengthening of sounds, and so on). A study of ideophones involving all of these aspects is beyond the scope of this dissertation and may constitute an exciting topic for future endeavors. 416 of animals usually occur as complements of a speech verb: (134) hm̃m kaḿɨ t͡ʃit͡ʃɛíɲak h̃mm kamɨ t͡ʃit͡ʃa-ina-kũ IDEO INTERJ speak-PL.IPFV-SIM\3.SS ‘“hm̃m!” [the jaguar] was saying. . .’ (135) paḱi taś taś taḱũ awaḱou tiḿaji ut͡ʃiń paki taś taś ta-kũ peccary IDEO IDEO say+IPFV-SIM\3SG.SS awaka-u timaji ut͡ʃi=na scare.PFV-NMLZ NARR child=ACC ‘The peccary was one to scare the young man saying “taś taś”.’ The Wampis excel at making imitations of animal’s voices. What is more, they usually “call” animals using the onomatopoeia speci"cally associated with the relevant animals (for instance, hunting of birds and peccaries is frequently done by uttering the speci"c call). Berlin (1994) and Berlin and O’Neill (1981) have shown the importance of sound-symbolism in the formation of animal names in Wampis, from an ethnobiology perspective. Table 8.21 provides a sample of Wampis ideophones organized by semantic "elds. The relationship of sound-symbolism with other parts of the Wampis grammar, as well as the lexicon, is a "eld of research that remains open for detailed exploration, and 417 it must be added that sound symbolism is indeed very rich and important in Wampis. Table 8.21. Sample of Wampis ideophones Actions/Movements haah́at with no breath haã ̃t́ pulling o$ a leaf kut͡ʃat́ hit suddenly (in just one blow) kusuií spitting kuɨŕ animal or human pack running mut͡ʃiḱ moving without stopping in a reduced area pakɨt́ sound of applauding or breaking (ceramics, hard-shell fruits, etc) pɨh́ throwing a stone or a small projectile quickly, jump pɨɨ́h making something disappear quickly takɨt́ cutting with machete tapit́ grabbing forcefully or suddenly turit́ tearing a piece of cloth t͡sakɨt́ driving something (e.g. a spear) into the soil t͡sɨút pulling o$ a palm leaf t͡ʃakɨat́ t͡ʃakɨat́ swimming (also applies to "sh and other animals like river otter, capybara, etc.) Sensorial (not including sound) hiŕa hiŕa shining, blazzing sɨŕsɨŕ smell of blood t͡saṕt͡saṕ soft kúntua kúntua pain in some body part (arm, leg, etc.) or bone (may related to kuntu ‘arm’?) 418 Human or body sounds hat͡ʃiá sneezing t͡saarit́ farting kunt͡ʃia ́kunt͡ʃia ́ noise of women or children talking taɨt́ taɨt́ noise of adult man yelling taɨt́ɨt taɨt́ɨt noise of adult men talking Animal and nature actions aḿpin butter/ies /ying kakia ́ sound of peccaries eating tańki tańki sound of river waves taḱia tańkia sound of snake slithering tɨtɨŕɨrɨr rocks falling from a cli$ ʊ́ʊ ʊ́ʊ sound/breeze preceding a rain wɨrɨ ́ buzzing (bee, wasp, hummingbird) Sounds of objects being manipulated, breaking, colliding pakɨt́ solid object colliding with another solid object parat́ object falling into the "re ʃakaú ʃakaú sound of water dripping, rocks, snails or rattle being moved, also sound of muzzle-loader ri/e tataŕtau di$erent objects crashing on each other t͡sɨút t͡sɨút sound of palm leaves being pulled out waá poking the "re Animal voices haii haii dog howling hau hau dog barking h̃mm jaguar’s voice (growling) hurur jaguar roaring pií ́ tapir’s voice 419 piʃ́ curasow’s calling saãs̃a ̃́ hoatzin’s calling tatat́at rooster crowing 8.13. Grammatical particles Grammatical particles are function words that work at the level of the clause and have grammatical meaning. The de"ning property of grammatical particles is that they are invariable and do not receive in/ection. There are two semi-copula particles in Wampis which I translate as ‘become’: has ‘become (+stative)’ and hak ‘become (- stative)’. They can be described as invariable forms that mark change of state. These types of particle are not reported for other Jivaroan languages. (136) wíi uún haś ui uunta has 1SG big become ‘I have become an adult (already).’ (137) húna urúk tukuiń haḱ hu=na uruka tuku-inu hak PROX=ACC how shoot.PFV-NMLZ become ‘How do I shoot this?’ Likely, both has and hak are grammaticalizations of two stems containing the root uaha ‘stand’. It is likely that has and hak are grammaticalized historically-complex 420 forms derived from the verb uaha ‘stand’ used in existential/copular constructions, as indicated in Table 8.22. With the attenuative aktionsart su#x -sa, uaha combines with a complement that has stative semantics, though the combinatorial meaning is inchoative (i.e. change of state). With the intensive aktionsart -ka, uaha yields more dynamic semantics. ‘>’ indicates hypothesized consecutive functional changes. Table 8.22. Probable grammaticalization path of has and hak Source form > Posture Verb > Existential/copula > Change of state uaha-sa ‘stand-att’ ‘be standing’ ‘exist’, ‘be’ has ‘become’ uaha-ka ‘stand- intens’ ‘stand up’ ‘exist’ hak ‘become’ The form has also functions as an semi-copular verb (i.e. receiving in/ection) with the meaning of ‘become’. Peña (Forthcoming) analyzes the grammaticalization of has in detail. 421 CHAPTER IX ADJECTIVES 9.1. Introduction This chapter analyzes the class of adjectives in Wampis. Adjectives constitute a “semi-closed” class in Wampis. Approximately 40 items constitute the class of adjectives in the language. Wampis possesses adjectivalizers, i.e. derivational morphology that derives adjectives. However, adjective derivation is no longer as productive as it seems to have been. In fact, one adjectivalizer is not very productive at all (the adjectivalizer - rama, see §9.8.3) and the other is historically based on nominalization (§9.8.5). There is one adjective-to-adjective derivational su#x that is productive but not very common (- taku, see §9.8.2). It seems that the distinction between nouns and adjectives used to be more prominent in past stages of the language, as the existence of adjective-derivational morphemes would suggest. The fact that there are not many adjectives and that adjective derivation is apparently no longer common does not mean that the language lacks ways to modify nouns. In fact, modern Wampis makes plentiful use of relativization to modify nouns, whereas the adjectives that do exist are more frequently use in predicative function. 422 Adjectives in Wampis are canonically post-nominal. Because of the cliticizing nature of most case and adposition forms in Wampis, adjectives can host case (but not all cases) and adpositions, which usually attach to the last element of the NP. The structure of this chapter is as follows: §9.2 provides the basic criteria to de"ne the adjective in Wampis, §9.3–§9.4 distinguish the category of adjectives from verbs and nouns. This is follow by an analysis of the morphological properties of adjectives in §9.5. Next, I describe the Adjective Phrase in §9.6, followed by the analysis of syntactic operations associated with adjectives in §9.7. Derivation of adjectives in dealt with in §9.8, whereas semantic and discoursive properties of adjectives are described in §9.9–§9.10. 9.2. De!ning the adjective in Wampis Adjectives are problematic to de"ne in many languages. There are languages that do not have a clear adjective class; what is more, it is well established that property concepts can be expressed by nouns, verbs or adjectives (Givón 1979: 14; Dixon 1982). However, I will propose that adjectives are a distinct class in Wampis. Croft (2001: 86 and $.) o$ers an analysis in terms of typological prototypes, discussing PROPERTIES as halfway between what he terms conceptual OBJECTS and ACTIONS in 423 terms of relationality, stativity, transitoriness and gradability. In Croft’s terms, a prototypical adjective is characterized by expressing a property and having a modifying function. In Croft’s view, an Adjective is a language-speci"c category that corresponds to the semantics of a PROPERTY, and modi"es an entity functioning as an OBJECT. Croft’s proposal for the semantic basis of parts of speech is reproduced in Table 9.1. Table 9.1. Semantic properties of prototypical parts of speech (from Croft 2001) Relationality Stativity Transitoriness Gradability Objects nonrelational state permanent nongradable Property relational state permanent gradable Actions relational process transitory nongradable For instance, a prototypical adjective in Wampis is pɨńkɨra ‘good’. It ful"lls all prototypical semantic properties of PROPERTIES: it is relational, that is, its de"nition involves a reference to another concept (a property of “goodness” cannot be established without something that is good and something that is bad). Pɨńkɨra ‘good’ expresses a state (as opposed to a process), it is prototypically permanent (that something is good is typically not seen as transitory) and it is gradable in a scalar dimension (for example, pɨńkɨra can be intesi"ed (§9.7.3.4)). 424 In addition, semantically, prototypical adjectives represent “single properties of prototype noun entities” (Givón 2002a: 53), encoding characteristics such as physical dimension, physical property, and so on. It will be seen that this can also be said about Wampis adjectives. Apart from their semantics, adjectives are also de"ned by their morphosyntactic and discursive features. Perhaps the most cross-linguistically common function of adjectives is that of attribution (Bolinger 1967) in which case adjectives usually occur as modi"ers of nouns in NPs. In Wampis, adjectives can also occur as modi"ers in NPs, however, they are not as commonly found as noun modi"ers as they are in predicative function with a copula. In terms of discourse, adjectives can serve to modify a referent already established in the discourse, or they can help introduced a new referent to the discourse (Thompson 1988). Morphosyntactically, in Wampis, adjectives head AdjPs, but they cannot head NPs or VPs. Unlike nouns, adjectives cannot be marked as possessors. In addition, the unmarked function of adjectives in Wampis is attribution,166 while the unmarked function of nouns is referential. The following sections will elaborate on the morphological, syntactic, semantic and discoursive distribution of 166. While adjectives more often occur in predicative function, they are usually marked with a copula construction in that function. 425 adjectives, but "rst, I discuss the basic distinctions between adjectives and verbs, as well as adjectives and nouns. 9.3. Adjectives vs verbs In Wampis, the distinction between adjectives and verbs is much more straightforward than the distinction between adjectives and nouns. Morphologically adjectives are never in/ected for any of the core categories associated with verbs, such as aspect, tense, person and mood, which de"ne the "nite verb in Wampis. In addition, adjectives do not receive any of the switch reference morphology associated with the subordinate verb in Wampis. In predicative functions, adjectives frequently occur with a copula, though it is possible that they occur alone in verbless constructions (see §9.7.2 for more details). 9.4. Adjectives vs nouns The distinction between adjectives and nouns is more subtle than the distinction between adjective and verbs just seen in the previous section. Table 9.2 summarizes the basic di$erences between adjectives and nouns in terms of their morphosyntax. 426 Table 9.2. Morphosyntactic properties of nouns and adjectives Adjective Noun Head of NP No Yes Modify head of NP Yes Yes Can be possessor No Yes Can be possessed Somea Yes Receive case and Adposition Only as phonological host Property of NPs Gradability Yes Nob Adjective-derivation with - rama and =hakinu No Yes Adjective-derivation with - taku Yes No a But notice that when adjectives are possessed, they are interpreted as nouns; e.g. ʃiirama-ri ̃‘his/ her beauty’ (ʃiirama ‘pretty, good’ + -ri ̃1PL/2PL/3.POSS). b Nominalizations in modifying function can be modi"ed with ʃiira ‘very’, but not with the intensi"ers ima, ti or mɨ (see §9.6). Nouns in referential function are not gradable. With regard to being the head of an NP, adjectives in Wampis cannot assume this function. For the related language Awajun, Overall (2007: 142) mentions that it is possible for adjectives to occur in headless NPs. The next example from Awajun illustrates this construction: (1) níŋka muúntan ʃináu nĩ-ka [muunta-na] ʃina-u 3SG-FOC [big-ACC] sing+IMPFV-REL ‘it (the kúgkup bird) sang with a big voice’ (Lit. it sang a big (one)) (Overall 2007: 142) 427 I have no attestation of that type of NP in the Wampis database. In fact, notice that a copula occurs with the adjective in Wampis: (2) ʃikaṕt͡ʃia húna amaśtahmɨ [ʃikapat͡ʃi=a hu=na] ama-sa-ta-hamɨ [small=COP PROX=ACC] give-ATT-IMM.FUT-1SG>2SG+DECL ‘I am going to give you this little one.’ (Lit.: ‘I am going to give you this [who] is little.’) The above example shows a relative clause: the copula is relativized with the proximal demonstrative. The proximal demonstrative, at the same time, functions as the head of the object NP. Concerning modi"cation of nouns, nouns can modify the head of an NP; however, in that case, word order is much looser. In a quick count of 40 NPs from texts where a noun to noun modi"cation occurs, 75% (=30) of cases observed the modi"er noun coming before the head. An example is presented in (3). (3) núu ʃuaŕ paḱi ah́aku tiḿaji nu ʃuara paki a-hak-u timaji NON.VIS person peccary exist-PT.HAB-NMLZ NARR ‘That person peccary used to exist.’ This is unlike modi"cation with adjectives, where there is a very strong preference for Noun-Adjective order. Notice that relative clauses also follow this pattern, occurring after the noun head. This seems to con"rm that the preferred position for the modi"cation of the head of the NP is post-nominal. To exemplify this 428 point, consider adjectives like uunta ‘big’. When uunta is used in a pre-nominal position, it acquires the meaning of ‘adult person, elder person, respected person’ (by extension, uunta is used as a noun when meaning ‘elder’ or ‘ancestor’): (4) úun pirút͡ʃ ʃuarań ma ̃h́akuiti uunta Pirut͡ʃi ʃuara=na ma=̃hak-u=iti elder Pirut͡ʃi person=ACC kill=HAB.PT-NMLZ=COP.3+DECL ‘The elder Piruch used to kill people.’ In that respect, uunta can act as a noun, heading an NP: (5) óo úun Limańmajakɛiti au uunta Lima=numa=ia=ka=iti MED adult.person Lima=LOC=ABL=Q=COP.3 ‘Is that guy (i.e. ‘adult person’) from Lima?’ But when uunta occurs post-nominally, it is clearly adjectival: (6) ou ʃuar úun hɨãj́i au ʃuar uunta hɨãj́i MED person big arrive+HIAF-3.PT+DECL ‘That big person just arrived.’ Interestingly, in lexicalized compound nominals that have an adjective incorporated, the adjective occurs "rst and there is no possibility at all that the Adjective-Noun compound can be interpreted as an adjective modifying a noun. For instance, uun_iauaa ̃is a compoun noun derived from iauaa ̃‘dog’ and uunta ‘big’. So, uun_iauaa ̃[uún ɲawaa]̃ ‘jaguar’ is never interpreted as ‘big dog’; the only way to say ‘big dog’ is by using the canonical Noun-Adjective order: 429 (7) ɲawaá ̃úun iauaa ̃ uunta dog big ‘big dog’ (8) úun ɲawaá ̃ uun_iauaa ̃ ‘jaguar’ With regard to possession, unlike nouns, adjectives cannot be possessors in Wampis. On the other hand, like nouns, adjectives can be possessed, but when they are possessed, their meaning and function is that of (i.e. they become) a possessed noun. Thus, when uunta ‘big’ is possessed, it acquires the meaning of ‘elder one’, ‘ancestor’ and can head NPs: (9) iíɲa úuntrika papaŋ́hɛɛ ̃júha hakaŕu iina uunta-ri=̃ka papanku=hai ̃ 1PL.GEN ancestor-1PL/2PL/3.POSS=FOC raft=COM iuha hak-ara-u walk.in.group HAB.PT-PL-NMLZ ‘Our ancestors used to travel by raft.’ Now compare the following sentences that contain the word ɨsarama ‘long’. In (10), ɨsarama functions as an adjective, modifying the noun head ‘snake’. Meanwhile, in (11) ɨsarama is possessed and is functioning as a noun in a genitive construction: 430 (10) naṕi ɨsaŕam tɨpaẃɛɛ napi ɨsarama tɨpa-ua-i snake long lie.down-3.SBJ-DECL ‘A long snake is lying down.’ (11) napiń ɨsaŕmariñ iista ́ napi ́ ɨsarama-ri=̃na ii-sa-ta ́ snake\GEN long-1PL/2PL/3.POSS=ACC see-ATT-IMP ‘Look at that snake’s length!’ The other points that appear in Table 9.2, such as the morphology associated with adjectives (including case and postpositions), gradability and derivation of adjectives, are discussed in detail in the next subsections. 9.5. Morphology associated with adjectives Adjectives have a unique su#x -taku that derives another adjective that acquires the meaning equivalent of ‘PROPERTY-ish’. This su#x is explained in §9.8.2. Adjectives can also receive a restricted number of morphemes that also occur on noun. Table 9.3 shows the su#xes that are shared by adjectives and nouns. Table 9.3. Nominal su#xes that can be received by adjectives Morpheme Gloss -t͡ʃi ‘Diminutive’ -t͡ʃau ‘Negative nominalizer’ Possessed noun markers (see §9.4) 431 9.5.1. Diminutive -t ͡ʃi with adjectives The diminutive -t͡ʃi can occur with adjectives as well as nouns. As it is typical of diminutives, -t͡ʃi does not only express smallness but it also can convey an idea of endearment. The marking of the diminutive in both the modi"er and head is optional, but it occurs in the majority of examples from texts. This suggests that the adjective tends to have some agreement with the noun head when marked with the diminutive. (12) nui ̃ ́ʃikaṕt͡ʃit͡ʃi jahaśmoot͡ʃ ɨŋkɨtún wɛiŋ́kʲaru tiḿaji nui ̃ ʃikapat͡ʃi-t͡ʃi iahasVmau-t͡ʃi inkɨtu-u=na there little-DIM insigni"cant.thing-DIM put.inside-NMLZ=ACC uaina-ka-ara-u timaji see-INTENS-PL-NMLZ NARR ‘They found there the little insigni"cant thing that was inside.’ There is a number of adjectives where the diminutive is lexicalized. Some examples are presented in (13). (13) aʃantat ͡ʃi ‘very old’ cf. aʃanta ‘elderly person’ ʃikapat ͡ʃi ‘little’, (ʃikapa is no longer found as a separate root or word) uhumat ͡ʃikɨ ‘brief’ cf. uhumakɨ ‘a little’ (uhuma ‘a little’+-kɨ ‘restr’) uuntat ͡ʃi ‘old’ cf. uunta ‘big’ ut͡ʃit ͡ʃi ‘young’ cf. ut͡ʃi ‘child’ 9.5.2. The negative nominalizer -t ͡ʃau and adjectives Some adjectives can be derived from adjectives, nouns and verbs using the negative nominalizer -t͡ʃau. One of the functions of verbal nominalization is to modify 432 nouns (i.e. via relative clauses), so it seems that through this function -t͡ʃau can derive items that become primarily modi"cational in function. Speci"cally, -tʃau derives opposites of certain adjectives, as many do not have a lexical antonym pair. -t͡ʃau is specially productive with evaluative adjectives. This is an interesting "nding, as it reveals a semantic tendency of Wampis lexicon related to adjectives: most adjectives expressing VALUE are positively oriented and do not have a lexical opposite. Examples are presented in (11) (14) pɨńkɨra ‘good’ — pɨńkɨra-t͡ʃau ‘bad’ t͡ʃit͡ʃirama ‘strong (of current)’ — t͡ʃit͡ʃirama-t͡ʃau ‘weak’ iman ‘superior (of quality or skill)’ — iman-t͡ʃau ‘unskilled’ iumintV167 ‘sweet’ — iumin-t͡ʃau ‘salty, sour’ iupia ‘be wild’ — iupi-t͡ʃau ‘docile’ 9.5.3. Adjectives and case and post-positional clitics Phonologically, adjectives can also host case and adpositional clitics (because of the very nature of cliticization in Wampis). However no adjective can stand alone and receive case or adpositional marking. However no adjective can stand alone and receive 167. This word almost always occurs as [jumiń], but it occurs as [jumińtri]̃ when it receives the 1PL/2PL/3 possessed marker -ri.̃ So, it seems that, synchronically, it is still possible to analyze this word as having a last /tV/ syllable. The deletion of the last syllable (/tV/) is the consequence of vowel elision: after V is elided, /t/ is deleted because /nt/ is an impossible coda. V is always deleted because it is in a position to be deleted. This a very regular Wampis pattern. 433 case or adpositional marking. Crucially, the most phonologically bound cases, the genitive and vocative (which are marked by high tone), cannot be phonologically borne by adjectives; they only occur on nouns. Case and adposition clitics that are attested in the data as being hosted by adjectives are given in Table 9.4 (instances of an adjective hosting the allative and the ablative were only obtained elicitation). Table 9.4. Case and adpositional clitics that can be received by adjectives Morpheme Gloss =na ‘Accusative’ =nVma ‘Locative’ =hai ̃ ‘Comitative’ =ni ‘Allative’ =ia ‘Ablative’ There are two ways in which case and adpositional clitics can be hosted by adjectives: • When adjectives occur at the end of a NP, the case or post-positional clitic occurs on the adjective. However, because they are case or adpositional markers, the scope of the clitic is never the AdjP, but the whole NP. In the following example, the adjective hosts the accusative =na: 434 (15) út͡ʃi ʃikaṕt͡ʃit͡ʃin súsaru [ut͡ʃi ʃikapit͡ʃi-t͡ʃi]=na su-sa-ara-u child little-DIM=ACC give-ATT-PL-NMLZ ‘They gave [him] the little child.’ However, in the next example, the accusative is hosted by the nominalized copula as it forms a relative clause: (16) númi pɨŋ́kɨr at́iɲan araḱmakuiti [numi pɨńkɨra a-tinu]=na araka-ma-ka-u=iti tree good COP-FUT.NMLZ=ACC plant-VBZ-INTENS-NMLZ=COP.3+DECL ‘She planted trees that are going to be good.’ Notice that if the adjective occurs in pre-nominal position, it does not carry the case marker either; rather, the accusative is hosted by the NP head: (17) pɨŋ́kɨr puhútnasha hukiɲ́ɛithʲɛɛ [pɨńkɨra puhu-ta]=na=ʃa hu-ki-inu=aita-ha-i good live-NMLZ=ACC=ADD take-WHILE.MOVING-NMLZ=COP-1SG.SBJ-DECL ‘I carried out a good life too.’ • The second case of adjectives receiving case or adpositional clitics is when a demonstrative occurs in the NP. Demonstratives have the peculiarity of triggering agreement between demonstratives, noun head and adjectives present in the NP, as the next examples show: (18) núna ʃuarań kakaŕman wɛińkʲar nu=na ʃuara=na kakarama=na waina-ka-ara NON.VIS=ACC enemy=ACC valiant=ACC see-INTENS-3PL.SS ‘[they] having seen that strong enemy . . .’ 435 (19) kartut͡ʃón susaŕmiahɛɛ ut͡ʃiń nat͡sań núna kartut͡ʃo=na su-sa-ara-mia-ha-i ut͡ʃi=na food=ACC give-ATT-PL-DIST.PT-1SG.SBJ-DECL168 child=ACC nat͡sa=na nu=na young=ACC NON.VIS=ACC ‘I gave cartridges to those young men (i.e. to hunt)’. In sum, at "rst look, adjectives may seem to receive some case and post- positional clitics. However, this is merely “accidental” because of either the phonological nature of most case and adposition markers, which are clitics that can attach to the last element of the NP; or because of a morphosyntactic property of demonstratives which trigger agreement with the head of the NP and adjectives. On the other hand, adjectives have at least one unique productive piece of morphology, the derivational su#x -taku, that does not occur with any other part of speech. 9.6. The adjective phrase (AdjP) Adjectives head adjectives phrases. The internal structure of adjective phrases in Wampis is clear: clearly because adjectives are gradable, only intensi"ers can modify adjectives. Intensi"ers occupy di$erent positions in the AdjP. Table 9.5 shows these 168. Notice that the plural -ara refers to the object. Third person objects are generally not marked on the verb, however a few examples of plural marking referring to a 3 person object occur in the data. It seems like the plural may be undergoing re-analysis to mark not only plurality associated with a 3 person subject, but also with a 3 person object. 436 positions. Table 9.5. Syntactic positions in an AdjP Modi"er + Adjective 1 2 3 ima ‘Intensi"er’ ʃiira ‘very’ ʃint͡ʃi ‘too, strongly’ ti ‘Intensi"er’ mɨ ‘Intensi"er’ Apparently, the intensi"ers ima, ti and mɨ are semantically equivalent, the meaning does not change when one or the other is used: (20) au ʃuaŕ mɨɨ́ ɨsaŕmajɛiti au ʃuara mɨ ɨsarama=aiti DIST person INTS tall=COP.3+DECL ‘That person is very tall.’ (21) núu ʃuaŕ ima ́sɨńt͡ʃi kakaŕam ah́akaruiti nu ʃuara ima sɨnt͡ʃi kakarama a=hak-ara-u=iti NON.VIS enemy INTS strongly valiant COP=HAB.PT-PL-NMLZ=COP.3+DECL ‘Those enemies used to be very valiant.’ (22) núu ʃuaŕ tií sɨńt͡ʃi kakaŕam ah́akaruiti nu ʃuara ima sɨnt͡ʃi kakarama a=hak-ara-u=iti NON.VIS enemy INTS strongly valiant COP=HAB.PT-PL-NMLZ=COP.3+DECL ‘Those enemies used to be very valiant.’ 437 The intensi"ers ima, ti and mɨ are in complementary distribution, they cannot co-occur together in the same phrase. (23) *nu ʃuara tií ima ́sɨnt͡ʃi kakaŕama ah́akaruiti *nu ʃuara ima ́tií sɨnt͡ʃi kakaŕama ah́akaruiti Adverbs ʃiira ‘very’ and ʃint͡ʃi ‘strongly’ can occur together, in which case the order is ʃiira sɨnt͡ʃi (the reverse order is prohibited): (24) húka ʃiír sɨńt͡ʃi úuntaiti hu=ka ʃiira sɨnt͡ʃi uunta=iti PROX=FOC very strongly big=COP.3+DECL ‘This is way too big.’ Canonically, AdjPs follow their noun head in the NP. The reverse order, N AdjP, is pragmatically marked—it usually highlights the adjective over the noun and is accompanied with a slightly forceful prosody. 9.7. Syntactic operations of adjectives In Wampis, adjectives can have an attributive function, in which case they modify the head of a NP (§9.7.1). Adjectives can also have a predicative function, occurring in copular and verbless clauses (§9.7.2). In addition, there is a range of constructions associated with the expression of comparison and superlative degrees (§9.7.3). 438 9.7.1. Adjectives as modi!ers of head of NP In Wampis, adjectives can also occur as modi"ers in NPs. The next examples illustrate adjectives modifying nouns: (25) ʃuaŕ naḱi núka takaś júat͡ʃu [ʃuara naki] nu=ka taka-sa ̃iu-a-t͡ʃau person lazy NON.VIS=FOC work-SUB\3.SS EAT-IPFV-NEG.NMLZ ‘A lazy person doesn’t eat working (i.e. it su$ers to eat).’ (Lit.: A lazy person, that [per- son], working, does not eat.’ (26) núwa aŋkań t͡sakakú mat͡sat́ɨa anú [nua ankanta] t͡saka-ka-u woman free grow-INTENS-NMLZ mat͡satɨ-a a=nu be.together.IPFV-IPFV COP=NON.VIS ‘The single women that were grown up’ 9.7.2. Adjectives in copular and non-verbal constructions In Wampis, adjectives are much more commonly found in predicative function as complements of a copula than as modi"ers to nouns. The following examples illustrate the occurrence of adjectives in predicate adjectives constructions: (27) ni ̃ŋ́kʲa jat͡ʃɛíti ni=̃ka iat͡ʃa=iti 3SG=FOC wise=COP.3+DECL ‘He is wise.’ 439 (28) najówɛithɛɛ naiau=aita-ha-i tall=COP-1SG.SBJ-DECL ‘I am tall.’169 Adjectives can also occur in non-verbal predicates without a copula: (29) wiḱʲa najóo ui=ka naiau 1SG=FOC tall ‘I am tall.’ Interestingly, other unrelated languages of the Amazon also show a preference for the predicative function of adjective. For instance, Ese Ejja (Takanan) apparently distinguishes two lexical classes: attributive adjectives and predicative adjectives, and predicative adjectives are far more frequent in that language (Vuillermet 2012). 9.7.3. Gradability Adjectives express properties that are gradable: they can be compared in terms of “degree” (less, equal, more) or occur in superlative constructions. The following sections describe di$erent constructions associated with the property of gradability where Wampis adjectives occur. 169. The word naiau seems to be an old nominalization of an old root naia ‘be.tall?, be.above?’, related to naia ‘height, tall, immense’, cf. naia-ma ‘make tall’ (where -ma is a verbalizer), naia- impi ‘sky’, naia-ntu ‘full moon’, naiant͡sa ‘ocean’ (< *naia-ɨnt͡sa, ‘immense-river’) and also cf. maybe related naia-pi ‘type of bird (and mythological hero)’. 440 9.7.3.1. Adjectives in comparative constructions There are two comparative constructions in Wampis. a. In the "rst construction, the comparand occurs with the focalizer =ka, the standard of comparison receives the ablative =ia and the adjective occurs with a copula. A demonstrative occurs before the copula, apparently introducing the standard of comparison (notice that de demonstrative is not the subject of the copular verb, which agrees with the entity being compared—this is clear in (31), where the subject of the copula is a 2SG person). Interestingly, several speakers translated the distal au as ‘more’ ( in Spanish) when I asked if there was any element in the sentence that indicated that meaning. (30) ni ̃ŋ́ka aḿɨa óo jat͡ʃɛiti ni=̃ka amɨ=ia au iat͡ʃa=iti 3SG=FOC 2SG=ABL DIST wise=COP.3+DECL ‘He is wiser than you.’ (31) aḿɨka wij́a óo ɨsaŕmɛitmɨ amɨ=ka ui=ia au ɨsarama=ita-mɨ 2SG=FOC 1SG=ABL DIST tall=COP-2SG.SBJ+DECL ‘You are taller than me.’ b. The second construction follows a subordinating strategy and involves the 441 verb apatu ‘put together’.170 In this construction, the standard of comparison appears with the comitative =hai1̃71 and the intensi"er ima is used to establish the meaning equivalent of ‘more’. The verb apatu is subordinated with the switch-reference marker - ma ‘Non-subject to subject’ and optionally followed by the conditional -ka. The adjective occurs in a predicate adjective structure heading the whole sentence. The whole construction can be roughly translate as ‘X with Y, if put together, X is more (Property).’ It is possible to consider that this construction has developed from a coordinated NP (coordinated with with the instrumental =hai)̃ in an appositive structure with a subordinated verb; i.e. ‘X and Y, if compared, . . .’. This analysis is consistent with the presence of the ‘Non-subject to subject’ switch reference marker -ma. Then, -ma would be referring to an elided object NP of ‘put together’ that is co-referent with the pre-posed coordinated NP. (32) ni ̃í aḿihʲɛɛ ̃apat́kamka namaḱ ma ̃átnumka ni ̃í ima ́jat͡ʃɛíti ni ̃ ami=hai ̃apatu-ka-ma-ka 3SG 2SG=COM put.together-INTENS-NON.SBJ>SBJ-COND 170. Also ‘copulate’. 171. Actually, it is better to consider that there is an Coordinate NP (coordinated with the instrumental =hai)̃. 442 namaka maã-ta=numa=ka ima iat͡ʃa=iti "sh kill-NMLZ=LOC=FOC ints wise=COP.3+DECL ‘He and you, if compared, he is better at "shing.’ Notice that in some cases the overt mention of the comparand can be dropped in the coordinated pre-posed NP. The comparand, however, is referred to in the main predication: (33) aḿihʲɛɛ ̃apat́kamka namaḱ ma ̃átnumka ni ̃í ima ́jat͡ʃɛíti ami=hai ̃ apatu-ka-ma-ka 2SG=COM put.together-INTENS-NON.SBJ>SBJ-COND namak maa-ta=numa=ka ni ̃ ima iat͡ʃa=iti "sh kill-NMLZ=LOC=FOC 3SG INTS wise=COP.3+DECL ‘If compared with you, he is better at "shing.’ The verb ii ‘see’ usually occurs with apatu ‘put together’, in which case ‘see’ takes the subordinating morphology: (34) wií aḿihʲɛɛ ̃apat́ka ijaḿka namaḱ maãt́numka ima ́jat͡ʃɛíthʲɛɛ ui ami=hai ̃ apatu-ka ii-a-ma-ka 1SG 2SG=COM put.together-INTENS see-IPFV-NON.SBJ>SBJ-COND namaka maã-ta=numa=ka ima iat͡ʃa=ita-ha-i "sh kill-NMLZ=LOC=FOC INTS wise=COP-1SG.SBJ-DECL ‘I with you, if compared, I am better at "shing.’ (Lit.: ‘I with you, if put together and seeing (i.e. considering), I am wiser at "shing.’) 9.7.3.2. Comparison of equality Comparison of equality is done with mɨtɨka ‘equal’ (cf. mɨtɨ ‘full’). The following 443 examples illustrate a comparison of equality. (35) wih́ʲɛɛ ̃mɨtɨḱɛitmɨ ui=hai ̃ mɨtɨka=ita-mɨ 1SG=COM equal=COP-2SG.SBJ+DECL ‘You are equal to me.’ (36) ni ̃í ʃuaŕ tiḱit͡ʃ mikúta núu nuú miḱuthɛɛ ̃mɛí mɨtɨḱʃa ni ̃ ʃuara tikit ͡ʃi Mikuta=a nu nu Mikuta=hai ̃ 3SG person other Mikut=COP NON.VIS NON.VIS Mikut=COM mai mɨtɨka=ʃa both equal=ADD ‘That person who was the other Mikut with the (original) Mikut, they were both equal too.’ 9.7.3.3. “Less” comparison In sentences that can be considered as equivalents of “less”-comparisons, the adjective occurs with the negative nominalizer -t͡ʃau and with the verb imatika (derived from the intensi"er ima). (37) amɨḱʲa wij́a namák mãátnumka imat́ikʲam jat͡ʃat́͡ʃuitmɨ amɨ=ka ui=ia namaka maã-ta=numa=ka 2SG=FOC 1SG=ABL "sh kill-NMLZ=LOC=FOC imatika-mɨ jat͡ʃa-t͡ʃau=ita-mɨ do.much.PFV=2SG.SS wise-nEG.NMLZ=COP-2SG.SBJ+DECL ‘You are less skillful than me at "shing (Lit.: ‘doing that much, you are not [so] wise’) Here is a similar construction with the adjective iman ‘superior’: 444 (38) iḿan húu jamɛí unúimamu tuiɲ́a iḿant͡ʃou iman hu jamai unuima-a-mau tu-ina imant͡ʃau superior PROX now study-IPFV-NMLZ say-PL.IPFV superior-NEG.NMLZ ‘[Those young men], skilled [as what] they call today’s professionals, they were not [as] skilled.’ 9.7.3.4. Adjectives in superlative constructions There is no superlative form in Wampis. Speakers use the intensi"ers ima or ti, or ʃiira ‘very’ and the adjective in an copular construction. There is apparently no di$erence in meaning when using either ima, ti or ʃiira, however ʃiira occupies a di$erent syntactic position than ima and ti (see §9.6). Notice that while this is a copular construction, it di$ers from a typical (typical in terms of frequency) attributive construction. In the latter, the subject usually occurs with the focus =ka and the adjective is marked with the copula or in a verbless predication. Example (39)–(40) illustrate this: (39) niŋ́kʲa ima ́ɨsaŕmajɛiti ni=̃ka ima ɨsarama=aiti 3SG=FOC INTS tall=COP.3+DECL ‘He is very tall’ (40) niŋ́kʲa ima ́ɨsaŕma ni=̃ka ima ɨsarama 3SG=FOC INTS tall ‘He is tall’ On the other hand, in the superlative construction, it is the adjective which 445 occurs focalized with =ka, the copula marks the subject. However, according to my teachers, this construction can also have an attributive (not necessarily superlative) meaning. The meaning of the intensi"er apparently is augmented in this construction, so I have tried to mimic this augmentation by translating it as “very, very”. (41) nijɛíti ʃiír ɨsaŕmaka ni=̃aiti ʃiira ɨsarama=ka 3SG=COP.3+DECL very tall=FOC ‘He is very, very tall.’ or ‘He is the tallest.’ (42) nij̃ɛíti ima ́ɨsarmaka ni=̃aiti ima ɨsarama=ka 3SG=COP.3+DECL INTS tall=FOC ‘He is very, very tall.’ or ‘He is the tallest.’ 9.8. Derivation of adjectives 9.8.1. Zero derivation There are examples of zero derivation between adjectives, nouns and verbs, but the direction of historical development is not clear in all cases: (43) sɨnt͡ʃi ‘strong’, cf. sɨnt͡ʃi ‘strengh’ (N) ( in the original Spanish). 176. I was asking about the /uorescent-like colors used in chicha or cumbia music posters in Peru. (Cumbia music is popular among the Wampis and I observed some posters in the villages advertising shows in the towns of Nieva and Bagua). Of course, the introduction of posters in the villages is a relatively recent phenomenon, so it is understandable that there is no term for the kind of colors used there. I was told that kapantutaku [kapańtaku] ‘reddish’ could be used but only “if we force the situation.” 448 an interesting semantic range of distinctions associated with color terms in Wampis that await more profound, culturally-appropriate analysis.177 9.8.3. Derivation with -rama The su#x -rama creates adjectives from adverbs, verbs and nouns. This su#x is apparently not very productive synchronically, but there are plenty of examples in the language where it can be clearly seen in use. A sample of words derived with -rama is given in Table 9.6. Table 9.6. Sample of adjectives derived with -rama Derived Adjective Probable Source asarama~ɨsarama ‘long, tall’ no clear source found, perhaps somewhat related to ɨsa~asa ‘burn’? kakarama ‘strong, valiant’ kaka ‘give strength’ kampũrama ‘thick, wide’ kampũa ‘type of tree (with thick buttress roots)’, ‘thick trunk’ kat͡surama ‘hard, solid’ kat͡su ‘harden’ kunturama ‘fat, fatty’ kuntu ‘be.fat’, cf. kuntu ‘arm’178 nuɨrama ‘fat’ nuɨ ‘get fat’ 177. For an analysis of color categories in Awajun, see Berlin and Berlin (1975). 178. Though notice that Jakway et al. (1987: 56) have an entry ‘greasy’ (Spanish ) so possibly from another unidenti"ed (but related?) lexical root kuuntu. 449 nuparama ‘dense (of vegetation), synchronically also ‘double’’ nupa ‘increase (weed)’ (V), cf. nupa ‘weed’ (N) uankarama ‘wide’ uanka ‘be ample’, cf. uanka ‘type of big pot’ iamarama ‘new’ iama ‘just now’, ‘newly’ (Adverb) iupirama ‘wild, sneaky, dour’ iupia ‘be wild’, cf. also the homophonous root iupia ‘be away’ 9.8.4. Unproductive adjectivalizer -patinu There is an old su#x -patinu that seems to have functioned as an adjectivilizer; however, it is no longer productive and is found only in a few items. This su#x can be roughly translated as ‘propensity to X’, ‘similar to X’, where X is the root with which it occurs. Apparently -patinu could derive an adjective from a verb, adjective, noun or even an ideophone (see ‘fragile’ in the example below). The su#x -patinu might be historically complex, probably composed of an old unidenti"ed formative *-pa and -tinu ‘Attributive’. Examples: (47) t͡ʃit͡ʃarpatinu ‘chatty, gossipy’ < *t͡ʃit͡ʃa-ri-̃patin ‘speak-1PL/2PL/3.POSS-ADJZ’ ʃakarpatinu ‘loose’ < *ʃaka-ri-̃patin ‘water.down-1PL/2PL/3.POSS-ADJZ’ pakɨrpatinu ‘fragile’ < *pakɨr-patinu ‘sound of object breaking-ADJZ’ 9.8.5. Augmentative =hakin The morpheme =hakin derives denominal adjectives that express the idea of 450 unusual largeness. By extension, =hakin sometimes creates new items with intensi"ed and negative connotations. The following examples illustrate the use of =hakin. (48) tantanɨ ‘belly’ → tantanɨ=hakin ‘pot-bellied’ uɨnu ‘mouth’ → uɨnu=hakin ‘big-mouthed’ numpa ‘blood’ → numpa=hakin ‘all bloodied’ hii ‘eye’ → hii=hakin ‘big-eyed’ Historically, =hakin likely comes from the nominalization of the stem uaha-ka ‘stand-INTENS’, which has been phonologically eroded, plus the agentive nominalizer -inu. So, apparently the old construction was a regular nominalization of ‘stand’ in its semi- copular function (in which it means ‘become’ §17.6). Synchronically, =hakin sometimes occurs as a separate word, occurring with the derived noun after a pause and bearing its own high tone. Some other times, it occurs phonologically bound to the noun it is deriving. 9.9. Semantic types of adjectives Dixon (1982) identi"es several types of adjectives in English, based primarily on their semantics: dimension, physical property, color, human propensity, age, value, speed. Not all of these types are found in Wampis, but most of them are. Dixon also indicates that if a language has a distinct class of adjectives, adjectives in that language at least have the types: age, dimension, value and color. Adjectives that express these 451 properties are found in Wampis. On the other hand, I have not found adjectives that can be associated to the category of ‘Speed’, most items related to this category are either adverbs or verbs. Table 9.7 summarizes the types of adjectives in Wampis based on their semantic properties. Examples are also provided. Table 9.7. Types of adjectives in Wampis according to their semantics Type Example AGE nat͡sa ‘young’, uuntat͡ʃi ‘old’, iamarama ‘new’ DIMENSION uunta ‘big’, ʃikapat͡ʃi ‘small, little’, asarama ‘long, tall’, sutara ‘short’ VALUE pɨńkɨra ‘good’, mɨtɨka ‘equal’ COLOR mukusa ‘black’, kapantu ‘red’, puhu ‘white’, iankú ‘yellow’ PHYSICAL PROPERTY hakɨ ‘warm’, samɨka ‘fresh’, karia ‘bitter’ SHAPE tutupini ‘straight’, tɨntɨ ‘round’ HUMAN PROPENSITY kakarama ‘strong, valiant’, iupirama ‘wild’ 9.10. Discoursive functions of adjectives Thompson (1988) identi"es two main discoursive functions of property concept words: a) to predicate a property of some stablished discourse referent, and b) to introduce a new discourse referent. Both functions are found in Wampis; however, predicating a property of some stablished discourse referent is the most frequent use of 452 adjective. This is not unexpected, though, as Thompson (1988) herself showed that the favored function of adjectives in English and Chinese was to predicate a property of a referent that is already introduced in the discourse.179 In addition, relative clauses play also similar roles of introducing a new referent in the discourse and indicating a property of an stablished referent. Thompson (1988) does not take into account relative clauses, but in a language like Wampis, where relative clauses play a substantial and most frequent role in modifying nouns, a more elaborated study looking at the di$erences or similitudes between the discoursive functions of adjectives and relative clauses is needed. For the moment, the following examples show the above mentioned discourse functions of adjectives in Wampis. Predication of property of an established referent: In (49), the speaker introduces the ‘cueva de los Tayos (Cave of Oilbirds)’ in the "rst lines (the name taiu ‘oilbird’ is used metonimically to refer to the cave where those birds live). Then in (49c) he uses an adjective to predicate a property of the already established referent: 179. However, the bigger part of the data on which I based my analysis are narratives, perhaps a bigger collection of conversational data may provide a di$erent perspective on the function of Wampis adjectives. 453 (49) a. . . . tajún pat͡ʃiśan óuhmatsattahɛɛ . . . taiu=na pat͡ʃi-sa-nu auhumatu-sa-tata-ha-i oilbird=acc mention-SUB-1SG.SS tell-ATT-DEF.FUT-1SG.SBJ-DECL ‘I am going to tell about the Cueva de los Tayos . . .’ b. . . . wiḱʲa tajúnka pɨŋ́kɨ nɨkat́͡ʃmiahɛɛ . . . ui=ka taiu=na=ka pɨnkɨ nɨka-t͡ʃa-mia-ha-i 1SG=FOC oilbird=ACC=FOC nothing know-NEG-DIST.PT-1SG.SBJ-DECL ‘I did not know anything about Cueva de los Tayos . . .’ c. . . . núka taj́u úuntɛiti nu=kataiu uunta=aiti NON.VIS oilbird big=COP.3+DECL ‘That Cueva de los Tayos is big.’ Introduction of new referent: This line introduces the protagonists of a Wampis myth: (50) joúnt͡ʃuk núwa himʲaŕ aŋkań puhú hakú tiḿaji iant͡ʃukɨ nua himaraankanta puhu hak-u timaji long.ago woman two single live PT.HAB-NMLZ NARR ‘Long ago, there used to live two single women.’ 454 CHAPTER X THE NOUN AND NOUN MORPHOLOGY 10.1. Introduction This chapter analyzes the morphological properties of nouns in Wampis and includes a description of noun classes. Section §10.2 discusses a working de"nition for Noun in Wampis. Next, §10.3 describes several nouns classes. Finally, §10.4 analyzes the noun morphology. 10.2. De!ning the noun in Wampis Nouns are one of the two major open word classes in Wampis, the other is the class of Verbs. Morphosyntactically, a noun in Wampis acts as the head of the NP, can be possessed (see §10.4.1) and possessor (see §10.4.4, §10.4.6.3). In addition, only nouns receive the vocative §10.4.6.4 and can be marked with the attributive -tinu §10.4.3, as well as modi"ed by uɨantu ‘group’ (see §11.3.5). Unlike adjectives, simple nouns are not gradable. Unlike verbs, simple nouns do not receive any in/ectional TAM morphology and cannot predicate on their own (they are marked with a copula when used in predicative function). Semantically, prototypical nouns in Wampis denote concrete objects or entities 455 and their unmarked use is referential. This coincides with Givón’s semantic criteria to de"ne nouns in terms of temporal stability (Givón 2002a) and Croft’s semantic properties of parts of speech (Croft 2001), respectively—cf. Chapter VII for a discussion. In discourse, nouns are used to introduce and manipulate referents in the scene (Hopper & Thompson 1984). For instance, in (1), two of the main characters of a story180 are introduced: in (1a) the uncle of the speaker—‘my mother’s brother’) and 1(b) the family’s hunting dog, Akum, who is hurt. The uncle then is only referred to by use of 3 person switch-reference morphemes in the next lines, whereas Akum, having been introduced in (1b), is anaphorically referred to in (1c) with the proximal demonstrative hu. (1) 1a. hintʲá nukurú úmaĩ maák wahóu hintá nuku-ru umaĩ trail\LOC mother-1SG\GEN sibling\1PL/2PL/3.POSS maa-kũ waha-u bathe+IPFV-SIM\3SG.SS stand+IPFV-NMLR ‘In the beach my mother’s brother was bathing’ 180. This story is authored and owned by one of my teachers, Dina Ananco Ahuananchi. She recorded it, wrote it down in Wampis and graciously shared the text with me for its use in this project. 456 1b. akum númpa hakín miɲóon wɛíɲa Akum numpa hakin wina-u=na waina Akum blood AUG come+IPFV-NMLR=ACC see+IPFV ‘having seeen Akum, who was coming all bloodstained’ 1c. íis húnaka ii-sã hu=na=ka see-SUB\3.SS PROX=ACC=FOC ‘having examining this (i.e. Akum). . .’ 10.3. Noun classes Noun classes distinguished in Wampis are Proper Nouns, Kinship terms and, based on their phonological and semantic properties, Compounds. I also refer brie/y to locational nouns. Nouns can also be distinguished for how they are marked when possessed: there are to types of nouns by this criteria (see §10.4.1). 10.3.1. Proper nouns Proper nouns are understood as names that denote unique entities frequently readily identi"able by the other people in the speech community. Traditional Wampis names can be common nouns in other parts of the grammar, for instance, they can be cultural artifacts, meteorological events, natural object or names of animals: Nantu ‘Moon’ (male), Tirinkasa ‘decorative ribbon’ (female), Hɨmpɨ ‘hummingbird’ (male), Yacuma ‘howler monkey’ (male), Ipaku ‘achiote’ (female), Nusɨ ‘peanut’ (female). Other names seem to be nominalizations of verbs: Nuninku ‘one who walks on wooden planks’ 457 (male, cf. nuninka-u ‘walk on wooden planks-NMLZ’), Antit͡ʃu ‘one who does not obey’ (male, cf. anti-t͡ʃau ‘listen+LOAF-NEG.NMLZ’). Traditional proper nouns nowadays are not completely in disused, but most Wampis people have adopted the Peruvian name style (two given names, paternal surname, maternal surname). The two given names are usually in Spanish, though in practice many Wampis use also familiar nicknames based on the traditional name system. The paternal and maternal surname are usually based on the traditional Wampis name system. There are some Spanish surnames that are very common, though, such as Navarro, Velazquez, Graña. Morphosyntactically, proper names form a small subclass because they can never be possessed. In addition, proper names can be modi"ed with a demonstrative in narratives (to specify the reference), but in conversations demonstratives are very rarely used with proper nouns. In addition, I have no examples of modi"cation of proper nouns with adjectives. 10.3.2. Kinship terms There are di$erent criteria for the semantic organization of kinship terms in Wampis. Particularly interesting are the terms for siblings, siblings-in-law and cousins 458 which are based on same or di$erent sex of female or male ego. Table 10.1 and Table 10.2 present these terms. Note that the Wampis traditional marriage system was between cross-cousins. Table 10.1. Kinship terms for siblings Lexeme Male ego Female ego iat͡su (1 and 2 person) [jat͡su] iat͡ʃi (3 person) [jat͡ʃi] same-sex sibling or parallel cousin ______ uma [uma] di$erent-sex sibling or parallel cousin di$erent-sex sibling or parallel cousin kai [kɛi] ______ same-sex sibling or parallel cousin Table 10.2. Kinship terms for siblings-in-law and cousin Lexeme Male ego Female ego sai [sɛi] same-sex sibling-in-law or parallel cousin ______ iua [jua] ______ same-sex sibling-in-law or parallel cousin uahɨ [wahɨ] di$erent-sex sibling-in-law di$erent-sex sibling-in-law unt͡su [unt͡su] di$erent-sex cross-cousin di$erent-sex cross-cousin 459 Other gendered kinship terms are presented in Table 10.3. The term ɨnkat͡ʃi denotes either ‘mother’ or ‘grandmother’. Table 10.3. Other gender-based kinship terms Concept Male Female Grandparent apat ͡ʃi nukut͡ʃi ɨnkat͡ʃi Parent or parent’s same-sex sibling apa nuku ɨnka Parent of di$erent-sex sibling iit͡ʃi t͡sat͡sa Parent-in-law uɨar t͡sat͡saa Child ut͡ʃi nauanta a The term t͡sat͡sa is modernly used as ‘mother-in-law’ because the term denotes the mother of a cross-cousin, whom a Wampis traditionally married. When referring to a group children, ut͡ʃi means ‘child’ in general. The word tiranki is not based on gender, it means ‘grandchild’. The Spanish words ‘brother-in-law’, ́ ‘father’, ́ ‘mama’́ and ‘uncle’ are also used as appropriate according to age and context. Kinship terms are the only class of nouns that can be marked for vocative and possession at the same time. 10.3.3. Location nouns Nouns that denote a location do not constitute a subclass per se. However, it is 460 important to mention that many can received a locative marker that consist in a high tone V́ on the last vowel, cf. ɨńt͡sa ‘river’ vs. ɨnt͡sa ́‘in the river’. Some of these nouns are grammaticalizing as location adverbs (see §8.6.3). 10.3.4. Compound nouns Compound nouns are common in Wampis, especially in the areas of plants and animal names. Many animal and sometimes plant name species or subspecies are based on names of animals that are very salient in Wampis culture; cf. panki ‘white-lipped peccary’ > iunki_paki ‘collared peccary’, iauaa ̃‘predatory carnivore’181 > suat͡ʃ_iaaua ̃ ‘species of black jaguar’ (cf. sua-t͡ʃi ‘huito182-DIM’). Morphosyntactically and semantically, compounds form a unit, in the sense that they receive nominal morphology only one time (at the end of the compound). Phonologically, compounds behave as two prosodic words, each part carrying their original high tone and undergoing other processes that occur at the level of the prosodic word, such as vowel elision. For instance, iakum_sampi [jakúm sampi]‘type of fruit’ is composed of iakuma [jakúm] ‘howler monkey’ and sampi [saḿpi] ‘shimbillo fruit (Inga 181. Used for ‘jaguar’ and ‘dog’. 182. Genipa americana, from which a black dye is obtained. 461 spp.)’. In the compound iakum_sampi, the word for iakuma undergoes vowel elision and surfaces with a high tone; sampi also surfaces with a high tone, but it does not undergo vowel elision because it only has two moras. Thus, prosodically both components of the compound behave like independent prosodic words. A good number of Wampis words have arisen from compounding but are synchronically one word (morphologically and prosodically); eg. naiant͡sa ‘sea, ocean’ (naia ‘big, immense’ + ɨnt͡sa ‘river’). Table 10.4 provides examples for the type of combination of roots or stems that are found in compound nouns in Wampis. Table 10.4. Compound word combinations Compound Gloss Components Type hapa iauaa ̃[haṕa ɲawaá]̃ ‘type of puma’ deer + dog N+N uun uuɨh [úun uwɨh́] ‘thumb "nger’ big+"nger Adj+N t͡sama mama [t͡sama ́mama]́ ‘manioc sp.’ become.brownish+manioc V+N ima napi [imʲa ́naṕi] ‘boa sp.’ intensi"er+boa Adv+N 10.4. Morphology The morphology of Wampis nouns is moderately complex. There are "ve slots, 462 1–4 are more derivational su#xes, and I have added a “position 5” for in/ectional morphology. However, this position 5, which designates the category of Case, is not strictly speaking one morphological “position”: the nominative is zero-marked, the genitive and vocative are marked with a grammatical high tone, and the accusative =na is a clitic that attaches to the last element of the NP, not just on the noun. With these observations, I describe Case in this chapter because case markers (with the exception of the accusative) work at the level of the noun. The structure of the noun morphology of Wampis is presented in Table 10.5. Table 10.5. Structure of the Wampis noun 0 1 2 3 4 5 Root Diminutive Possessive Attributive Benefactive Plural.SAP (Diminutive) Negative Case Generally, in Wampis a noun root can function as a stem without receiving derivation. However, a noun stem can also be complex; in actuality, a phonological word containing a noun in Wampis can be very complex. There are a number of discourse-oriented clitics that can be hosted by the noun, and copula clitics that can be 463 also hosted by a noun in a predicative function. The copula clitics are analyzed in detail in Chapter XVII. Discourse-oriented and modality clitics are analyzed in Chapter XVIII. 10.4.1. Marking on the possessed noun Nouns can be classi"ed into two classes (Type I and Type II), depending on how they are marked when they are possessed. There are no strict discernible semantic criteria corresponding to the two types of marking. In addition, a few nouns can be marked as either Type I or Type II. 10.4.1.1. Type I nouns Type I nouns are characterized by marking the possessed nominal (i.e. the head noun) with a su#x,183 which varies somewhat for the grammatical person of the possessor if the possessor is singular. Thus, as a system the su#xes can to some extent be viewed as a portmanteaux of person and number. Table 10.6 presents these su#xes. 183. This corresponds to Overall’s 2007 “su#xing class” in Awajun. 464 Table 10.6. Type I possessed noun marking Person\Number Singular Plural 1 -ru -ri ̃ 2 -rumɨ~rumi~-ramɨ -ri ̃ 3 -ri ̃ -ri ̃ As can be seen, for Type I nouns, there is a clear distinction between singular Speech Act Participant (SAP) possessors versus all the rest (including plural SAPs). This is interesting, as in other part of the grammar, plural speech act participants (1PL and 2PL) are marked di$erently than the rest, specially in the hierarchical system of argument indexation (cf. §14.3). The invariant su#x on the noun for 3rd person singular and all plural possessors might be viewed as just marking that the noun is “possessed”. 10.4.1.2. Type II nouns Type 2 nouns are characterized by a di$erent form of the possessed noun if the possessor is 2SG, 2PL, 1PL and 3 persons. For these persons and numbers, the last vowel of the root changes to either i or ɨ depending on morphophonological factors explained below. In addition, 3 and 1PL persons receive nasal prosody. For a second person 465 singular, the noun changes its vowel and adds the person marker -mɨ. For "rst person singular, the noun does not change its vowel, using instead -ru just as in the Type I paradigm. Type II possession marking is summarized in Table 10.7. Table 10.7. Type II possessed noun marking Person\Number Singular Plural 1 -ru Ṽ 2 V-mɨ (OR) V-mi Ṽ 3 Ṽ Ṽ As just noted, in Type II possessed nouns, the quality of the last vowel of the possessed noun stem changes between ɨ and i. The allomorphy depends on the quality of the preceding vowel of the noun stem (after glide derivation, cf. Chapter IV). Table 10.8 presents the criteria for vowel grade in the possessed form, and provides examples with the corresponding form for 1PL, 2PL and 3. It is important to note that the stem for 2SG also switches vowels, but in this case the vowel is not nasal. Nouns ending in i or ɨ never change their "nal vowels. The paradigms in Table 10.9 illustrate the application of possession marking Types I and II with the nouns equivalent to ‘daughter’ (Type I) and ‘dog’ (Type II). In 466 Table 10.9, optional independent possessor pronouns precede the possessed noun, and are in their genitive form. Table 10.8. Criteria for vowel-grade in Type II possessed noun marking Preceding vowel Last vowel of Noun (Nominative form) Last vowel of Noun (Possessed form) Nominative form Possessed form [-front, -low] a, u ɨ nuwa ‘woman’ numpa ‘blood’ muuka ‘head’ uwɨha ‘hand’ wɨnu ‘mouth’ nuwɨ ̃ numpɨ ̃ muukɨ ̃ uwɨhɨ ̃ wɨnɨ ̃ elsewhere a i jawaa ̃‘dog’ hinta ‘trail’ kata ‘penis’ jawai ̃ hinti ̃ kati ̃ NA i, ɨ i, ɨ suwɨ ‘throat’ japi 'face' suwɨ ̃ japi ̃ Table 10.9. Paradigms of Type I and Type II person/possessive marking Type I Example: nauantu ‘daughter’ Type II Example: muuka ‘head’ 1sg mina nauantu-ru mina muuka-ru 2sg amina nauantu-rumɨ amina muukɨ-mɨ 3sg nina nauantu-ri ̃ nina muukɨ ̃ 1pl iina nauantu-ri ̃ iina muukɨ ̃ 2pl atumí nauatu-ri ̃ atumí muukɨ ̃ 3pl nitá nauantu-ri ̃ nitá muukɨ ̃ 467 10.4.1.3. Possession marking of derived diminutive nouns Derived diminutive nouns default to Type I possession marking. There is no exception in the data. Consider the nouns iauaa ̃‘dog’, nanki ‘spear’, and uɨnu ‘mouth’, which in their underived form all receive Type II possessive marking. When su#xed with the diminutive -t͡ʃi, they switch to Type I. Table 10.10 illustrates this with a paradigm. Table 10.10. Possessive marking with nouns derived with the diminutive -t͡ʃi Root Type II marking (1pl/2pl/3) Diminutive stem: Type I marking (1pl/2pl/3) iauaa ̃'dog' iauai ̃ iawaã-t͡ʃi-ri ̃ nanki 'spear' nanki ̃ nanki-t ͡ʃi-ri ̃ uɨnu 'mouth' uɨnɨ ̃ uɨnu-t͡ʃi-ri ̃~ uɨnɨ-t͡ʃi-riã a I have both forms in my data. Apparently, the third person possessive form wɨnɨ has been reanalyzed as the basic form for 'mouth' by some speakers. 10.4.1.4. Possession marking of idiosyncratic nouns A few nouns show particular patterns when they appear in the possessed form. • With a second person possessor, the words nuku ‘mother’ and apa ‘father’ never take (of any type) of possessed marking; thus they remain unmarked in their second 468 person-possessed form. This may be related to the marking of 2 person, which elsewhere in the grammar also triggers peculiar patterns (cf. §14.3.3, §16.3). • The word iat͡su ‘brother of male’ occurs as iat͡ʃi ̃[jat́͡ʃi]̃ with the 3 person possessor. This is surely a phonological reduction of iat͡su-t͡ʃi ̃(brother-DIM\1PL/2PL/3.POSS). As an aside note, this word occurs often times with the Spanish diminutive -ito: [jat͡ʃito].) • t͡ʃit͡ʃama ‘speech’ has the possessed form t͡ʃit͡ʃamɨ even though by the rules given in Table 10.8 it would be expected to have the possessed form t͡ʃit͡ʃami. The word anɨnta ‘magical song’ also has a special possessed form anɨnti (not anɨntɨ, as would be expected from Table 10.8). 10.4.1.5. Arbitrariness of the Type I/Type II distinction In many languages, semantic categories associated with di$erences in nominal possession structures correspond to alienable/inalienable, inherent/non-inherent, or possessable/unpossessable features. In Wampis, kinship terms are obligatorily possessed, that is, they are grammatically inalienable. Virtually all nouns in the language, as far as I know, are grammatically possessable.184 184. With the exception of the terms for 'mother' nuku, and 'father' apa, which do not receive any marking of being possessed when the possessor is second person, as indicated above. 469 A survey of nouns indicates that there is no clear semantic criterion that allows us to predict which nouns are Type I and which are Type II. Some kinship terms are of Type I and others are of Type II, though kinship terms of Type I are more numerous. Some examples are given in Table 10.11. Those under the Type II label comprise an exhaustive list of Type II kinship terms in the data. Table 10.11. Type I and Type II kinship and a#ne terms Type I Type II aiʃu ‘husband’ apa ‘father’ iat͡su ‘brother’ iit͡ʃi ‘uncle’ nuku ‘mother’ ut͡ʃi ‘uncle’ amiku/kumpa ‘friend’ (from Spanish) auɨ ‘child of uma’ kai ‘sister (of female)’ pataa ‘relative’ sai ‘brother in law (of male)’ uma ‘sibling of opposite sex’ Body terms are more balanced in terms of number of members of one type or the other, though there are more body parts or a#nes that belong to Type II. Table 10.12 provides examples of Type I and Type II body part terms. 470 Table 10.12. Type I and Type II body part terms Type I Type II ampu ‘guts’ kuntu ‘arm’ munt͡su ‘nipple’ mɨɨ ‘kidney’ titi ‘Adam's apple’ ukunt͡ʃi ‘bone’ hii ‘eye’ iapi ‘face’ kanauɨ ‘branch’ kata ‘penis’ muuka ‘head’ suɨ ‘neck’ uɨnu ‘mouth’ uɨha ‘hand’ suat͡ʃi ‘lung’ Table 10.13 gives a few other nouns that belong to Type I versus Type II. Certain nouns that refer to culturally important objects or that can be seen as the extension of a person/property, like ‘dog’, ‘magical song’ and ‘spear’ are Type II. However, other nouns referring to salient cultural objects, such as tauasa 'feathered crown' (given only to people of utmost respect) and personal property like ‘boat’ or ‘pot’ are Type I. Table 10.13. Type I and Type II general nouns Type I Type II kanu ‘boat/canoe’ pininka ‘pot’ tauasa ‘feathered crown’ nunka ‘land’, ‘territory’ hinta ‘trail’ iauaã ‘dog’ anɨnta ‘magical song’ nanki ‘spear' 471 Finally, some nouns can take either Type I or Type II marking. Some examples are given in Table 10.14. Table 10.14. Nouns that can be either Type I or Type II Nouns haant͡ʃi ‘cloth’ hɨã ‘house’ iauaã ‘dog’ t͡ʃimpui ‘stool’ 10.4.2. Diminutive -t ͡ʃi As seen in Table 10.5, the diminutive -t͡ʃi occurs immediately after the noun root and creates a new noun that has the basic semantics of ‘smaller N’. The historical source of the diminutive clearly is the noun ut͡ʃi ‘child’. As a su#x -t͡ʃi clearly corresponds to the canonical [N Adj] word order in Jivaroan languages. Here are some examples of the diminutive in Wampis: (2) numi numi-t ͡ʃi ‘tree’ ‘little tree’ t͡ʃinki t͡ʃinki-t͡ʃi ‘bird’ ‘little bird’ munt͡su munt͡su-t͡ʃi ‘breast’ ‘little breast’ 472 iauaã iauaã-t͡ʃi ‘dog’ ‘little dog’ The diminutive morpheme has extended usages in which its meaning is not that of diminutive size or literally ‘little’. Rather, it shares some meanings with Spanish -ito which itself is not restricted only to indicating small size. For instance, one use of the Wampis diminutive that is similar to the Peruvian Spanish diminutive is its use to attenuate temporal or geographical distance. Here is one example: (3) iaki iaki-t͡ʃi ‘above’ ‘a bit further above’ Like in Peruvian Spanish, this does not necessarily mean that the actual distance is short, but it is used in a familiar/attenuative way to lower the sense of distance. The previous comparisons with Spanish does not necessarily mean that Wampis borrowed these uses from Spanish, as it is very common for diminutives to have this attenuative use (Payne 1997). Furthermore, the diminutive use in a familiar/attenuative so widespread in Peruvian Spanish have been clearly in/uence by languages like Quechua, where this use is very common. Notice that Jivaroan languages have been in contact with Quechua too. In addition, the diminutive is used to add tenderness or endearment: 473 (4) ut͡ ʃi ‘child’ ut͡ ʃi-t͡ ʃi ‘little child’ mama ‘mother’ mama-t͡ʃi ‘dear mother’ The Wampis diminutive is also used to add an a$ective meaning or to attenuate a proposal (i.e. make it seem like it’s not a “big deal”, or not very dramatic or extensive). The diminutive can also provide a sense of politeness to an expression. For instance, the next expression is commonly used to invite a bowl of manioc beer, the staple drink among the Wampis. (5) nihʲaḿt͡ʃit͡ʃʃa umarmi ́ nihamat̃͡ʃi-t͡ʃi=ʃa uma-ra-mi manioc.beer-DIM=ADD drink-DISTR-HORT ‘Let’s drink a little manioc beer.’ (Spanish: ‘vamos a tomar un masatito’) Though it is not really common in natural speech, the diminutive can be reduplicated. In such a case, the diminutive adds an intensive sense: (6) ut͡ʃi-t͡ʃi-t͡ʃi-t͡ʃi185 child-DIM-DIM-DIM ‘very very young’ 185. This is an elicited example. 474 (7) iaha-t͡ʃi-t͡ʃi-t͡ʃi far-DIM-DIM-DIM ‘very very far’ The diminutive can occur on nouns, pronouns, demonstratives, adjectives, and occasionally on verbs. In the next example, the diminutive attaches to ‘straight’ (an adjective in Wampis) and ‘good’ (also an adjective). In both cases, the diminutive derives a word with a intensive meaning: to paraphrase in English, the word tutupini-tʃi does not mean ‘a little straight’ but ‘quite straight’; and its use on pɨńkɨra ‘good’, interestingly, yields a more adverbial sense of ‘very’. (8) tutúpnit͡ʃ pɨŋ́kɨrat͡ʃ ɛɛ́ɲa núna huki ̃ ́ tutupini-t ͡ʃi pɨńkɨra-t͡ʃi a-ina nu=na hu-ki ̃ straight-DIM good-DIM COP-PL.IPFV ANA=ACC carry-DISTR\3SG.SS ‘Having carried the ones that are very straight . . .’ 10.4.3. Attributive -tinu The attributive attaches to a possessed noun stem and derives a new noun with the semantics of ‘owner of N’. The possessed noun stem may have either Type I or Type II possessive morphology. Examples (8)–(9) show instances where -tinu occurs with a Type I possessed stem are: 475 (9) t͡ʃit͡ʃámrintin t͡ʃit͡ʃama-ri-̃tinu186 speech-1PL/2PL/3.POSS-ATTRIB ‘with authority’ (Lit.: ‘owner of his speech’) (10) ɛɛʃ́rintin aiʃu-ri-̃tinu husband-1PL/2PL/3.POSS-ATTRIB ‘with husband’ (i.e. ‘married (woman)’) The next example is an instance of with Type II possessed noun morphology with the vowel ɨ ̃in the possessed form of ‘woman’ (the non-possessed root is nua). (11) nuwɨ ̃ńtin nuɨ-̃tinu woman\1PL/2PL/3.poss-attrib ‘With woman’ (i.e. ‘married (man)’) An attributive-marked noun usually su#ces to head a non-verbal clause that predicates ownership/possession, but the attributive-marked noun can also occur with a copula verb. Examples (12) and (13) demonstrate the attributive noun predicate and the form with a copula, respectively. 186. Notice that the possessed form of t͡ʃit͡ʃama ‘speech’ can also occur with Type II morphology, as we saw in §10.4.1.5. 476 (12) Oskar naártin Oscar naa-ru-tinu Oscar name-1SG-ATTRIB ‘My name is Oscar.’ (13) wikʲa apártiɲɛithɛɛ ui=ka apa-ru-tinu=aita-ha-i 1SG=FOC father-1SG-ATTRIB=COP-1SG.SBJ-DECL ‘I have a father.’ The attributive construction with -tinu is explained in more detailed in §17.4.1. 10.4.4. Benefactive -nau I gloss -nau as ‘benefactive’ to re/ect its primary use in texts, though it really seems to be a polysemic morpheme, capable of marking ideas of possession, benefactive, recipient, and source material. Its synchronic status as a case form is somewhat debatable, but it can clearly co-occur with other case markers, but its position is like that of other meaning-changing morphemes. Possessors inside of NPs can be marked with the genitive (cf. §10.4.6.3) or with the benefactive su#x -nau. The su#x -nau, thus, marks a noun as a possessor, but di$ers from the Genitive in both form and use. Morphosyntactically, the Genitive only occurs preceding the possessum; whereas -nau attaches to a possessor noun that occurs after the possessum, as in (14)–(15). 477 (14) t͡ʃimpui apa-ru-nau stool father-1SG-BEN ‘my father’s stool’ (15) papii haime-nau notebook Jaime-BEN ‘Jaime’s notebook’ Quite separately from possession, -nau is used to mark a benefactive participant. In fact, a survey of connected speech reveals that the primary use of -nau in Wampis is to mark a bene"ciary/recipient, as in examples (16)–(19). (16) húka apárnawɛiti hu=ka apa-ru-nau=aiti PROX=FOC father-1SG-BEN=COP.3+DECL ‘This is for my father.’ (17) súwa aaẃɛi tikit́͡ʃik nampɨtán eðérnoun Sua aa-ua-i tikít ͡ʃiki nampɨta=na Eder-nau=na Huito write+IPFV-3.SBJ-DECL one song=ACC Eder-BEN=ACC ‘Huito is writing a song for Eder.’ (18) wíi namakán ɨŋkat͡ʃírnaun itʲájɛɛ ui namaka=na ɨnkat͡ʃi-ru-nau=na ita-ha-i 1SG "sh=ACC mother-1SG-BEN=ACC bring-1SG.SBJ-DECL ‘I bring "sh for my mother.’ 478 (19) ámiɲu wíi wakɨŕamuka oúsati túsan wakɨŕahɛɛ ami-nau ui uakɨra-a-mau=ka aúsa-ti tu-sa-nu 2SG-BEN 1SG want-IPFV-NON.SBJ.NMLZ=FOC study-JUSS say-SUB-1SG.SS uakɨra-ha-i want+IPFV-1SG.SBJ-DECL ‘What I want for you is that you study.’ (lit. what I want for you, saying “that he stud- ies”, I want.’ ) Interestingly, when the benefactive -nau occurs alone (without the copula), it generally marks possession. But when it occurs with the copula, it tends to mark benefactive. In (20), the -nau-marked noun receives a copula and its interpretation is that of bene"ciary or recipient. On the other hand, in (21) -nau marks the noun as a possessor. (20) húka dínanuitʲi hu=ka Dina-nau=iti PROX=FOC Dina-BEN=COP.3+DECL ‘This is for Dina.’ (21) húka dínanu hu=ka Dina-nau PROX=FOC Dina-BEN ‘This is Dina’s.’ Another interesting use of -nau is to indicate the material or source from which something is made. 479 (22) miɲ́a híirka numíɲuiti mina hii-ru=ka numi-nau=iti 1SG.GEN"re.place-1SG=FOC "rewood-BEN=COP.3+DECL ‘My kitchen is made out of "rewood.’ Recall that, phonologically, sequences of the type Cau (C=consonant) can undergo internal vowel elision if the vowel is in a position to be deleted, a seen in Chapter VI. Consequently, when the underlying /a/ is elided from -nau, the morpheme surfaces as [nu], as in (22) where it is the third vowel from the left. One can speculate that, historically, -nau could be morphologically complex: it may be composed of an old form *na (perhaps related to the accusative =na?)187 and the Set II nominalizer -u (cf. §15.4.5). As we saw in Chapter VI (see also §10.4.5 below), the nominalizer -u triggers vowel-internal elision in Cau sequences. One di#culty with this hypothesis is that -u derives deverbal nouns elsewhere in the grammar, and is not otherwise attested in noun-to-noun derivation. 10.4.5. Negative nominalizer -t ͡ʃau The negative -t͡ʃau is historically composed of the verbal negative morpheme -t͡ʃa and the Set II nominalizer -u (§15.4.5). As the name indicates, -t͡ʃau negates or inverts 187. Notice that Wampis does not distinguish morphologically between direct/indirect object. Both Objects in ditransitive constructions are marked with the accusative =na. There conceivably could be a connection between the use of =na to mark benefactive/recipient referents in ditransitive constructions and the benefactive/recipient function of -nau. 480 the meaning of the word that receives it. Examples (23)–(25) illustrate its use with nouns. (23) ʃuara ‘man’ ʃuara-t͡ʃau ‘not a man’ (24) ikama ‘forest’ ikama-ri-̃t͡ʃau ‘unforested’ forest-1PL/2PL/3.POSS-NEG.NMLZ (25) aiʃu-ri-̃tinu ‘with husband’ aiʃu-ri-̃t͡ʃau ‘without husband’ (i.e. bachelor female) husband-1PL/2PL/3.POSS-NEG.NMLZ The negative nominalizer -t͡ʃau can also attach to verbs and adjectives, as seen in §15.4.7. The examples in (26) show -t͡ʃau with verbs, and examples in (27) illustrate it with adjectives. (26) iu ‘eat’ iu-á-t͡ʃau ‘one who did not eat’ (‘eat-HIAF-NEG.NMLZ’) t͡ʃit͡ʃa ‘speak’ t͡ʃit͡ʃa-t͡ʃau ‘one who does not speak’ (=‘quiet’) (speak+IPFV-NEG.NMLZ) (27) pɨnkɨra ‘good’ pɨnkɨra-t͡ʃau ‘bad’ uunta ‘big’ uunta-t͡ʃau ‘not big’ Similarly to -nau, -t͡ʃau can surface as [t͡ʃu] if the vowel /a/ is in a position to be deleted. For instance, the surface realization of t͡ʃit͡ʃa-t͡ʃau ‘quiet’, in (26) above, is [t͡ʃit́͡ʃat͡ʃu]. 10.4.6. Case Case marking here is viewed in a stricter sense of "the morphosyntactic 481 characterization of noun phrases that is imposed by the structure within which the noun phrase occurs" (Payne 1997: 100). True case markers in Wampis are the Nominative (which is unmarked), the Accusative, the Genitive and the Vocative. Case does not seem to be a coherent morphosyntactic category in Wampis, as some cases are marked by phonological means and therefore tightly related to the noun root or derived stem, and other cases, like the accusative, occur in the form of an enclitic that may or may not be borne by the head noun. 10.4.6.1. Nominative Wampis is a nominative-accusative language. Nominative case is zero-marked in Wampis. Intransitive and transitive lexical S and A argument NPs are treated the same way in the language in having no overt case marker, and in occurring generally as the "rst core argument in a sentence. (28) ɨt́ɨ wɨáwɛɛ [ɨtɨ] uɨ-́a-ua-i yellow.wasp go-IPFV-3.SBJ-DECL ‘The yellow wasp is going.’ (29) iḱam t͡sukaŋkán tukúmaji [Ikama] t͡sukanká=na tuku-ma-ji Ikam toucan=ACC shoot-REC.PT-3.PT+DECL ‘Ikam killed the toucan.’ 482 10.4.6.2. Accusative =na Noun phrases that are Objects of a verb are marked with the morpheme =na. In fact, =na marks a core argument of the verb that is “not-subject”, as it also marks the other argument of a ditransitive verb, as will be shown shortly. (30) napín karámrumhɛɛ [napi]=na karama-ru-ma-ha-i boa=ACC dream-APPL-REC.PT-1SG.SBJ-DECL ‘I dreamt of a boa (to my detriment).’ The enclitic =na usually occurs in the last element of an NP: (31) út͡ʃi ʃikaṕt͡ʃit͡ʃin súsaru [ut͡ʃi ʃikapit͡ʃi-t͡ʃi]=na su-sa-ara-u child little-DIM=ACC give-ATT-PL-NMLZ ‘They gave [him] the little child.’ (32) út͡ʃi kakaŕam at́iɲun wɛińkamiaji [ut͡ʃi kakarama a-tinu]=na uaina-ka-mia-ji child powerful.man COP-FUT.NMLZ=ACC see-INTENS-DIST.PT-3.PT+DECL ‘He saw a children that is will be a kakaram.’ An important note regarding the realization of the accusative =na is in order. Because =na frequently appears as the last morphological element of the noun word (or phrase—only discourse-level morphemes come after case clitics), its vowel usually undergoes apocope. This apparently has given rise to a further phonetic change in the accusative, whereby sometimes the nasal consonant /n/ is also lost if =na occupies the last position. In those cases, the only marking of the accusative is placement of high 483 tone—remember that =na induces shift of high tone one mora to the right, as described in Chapter VI. It will shown that this process is also the origin of the genitive marking in Wampis. For instance, a common collocation in Wampis is to use the verb hu ‘take’ with nanki [náŋki] ‘spear’ as its object. Often times, the noun nanki surfaces as [naŋkí]188 when it is the object NP, with the placement of high tone as the only evidence of accusative marking: (33) naŋkí húki ̃ nankí hu-ki ̃ spear\ACC take-WHILE.MOVING\3SG.SS ‘Having taken the spear…’ However, when more morphology is added to the nominal piece, the underlying /n/ surfaces. This is not optional; the /n/ always surfaces when another morpheme co- occurs after the accusative, as in (34). Further evidence that =na is present underlyingly is the surfacing of the entire form /na/ if the morphophonological environment is appropriate. In (35), the vowel of =na is not deleted because the (potentially) resulting the cluster [ɲk] in the coda is prohibited. Thus, consider: (34) naŋkínʃa húki ̃ nanki=na=ʃa hu-ki ̃ spear=ACC=ADD take-WHILE.MOVING\3SG.SS ‘Having taken the spear too…’ 188. But notice that this varies with [naŋkín], where the nasal consonant occurs at the end. 484 (35) nankíɲak hú-ki ̃ nanki=na=kɨ hu-ki ̃ spear=ACC=RESTR take-WHILE.MOVING\3SG.SS ‘Having taking the spear only…’ In terms of its distribution, the accusative can attach to demonstratives and personal pronouns. In this respect, an interesting distributional feature of =na is that when a demonstrative occurs in the noun phrase, the accusative =na attaches in the demonstrative, adjectives and relative clauses modifying the noun, as well as on the noun head. This is, by all accounts, a rare property of Wampis demonstratives, as they trigger agreement with certain categories, such as the accusative (see §8.3.1 for details). Example (35) shows the NP object ‘that person’ with =na on both elements of the phrase. This contrast with the distribution (37), where =na is on the last element only. (36) núna ʃuarán wɛińmiaji nu=na ʃuara=na waina-ma-ji ana=ACC person=ACC look-REC.PT-3.PT+DECL ‘He saw that person.’ (37) ʃuár sutarán wɛínmiaji ʃuara sutara=na waina-mia-ji person short=ACC look-dist.pt-3.PT+DECL ‘He saw a short person.’ 485 10.4.6.2.1. Accusative =na in ditransitive constructions The T (=Theme) and R (=Recipient) arguments of a ditransitive verb are both marked with =na. Bene"ciary/recipients can optionally be marked with -nau. In (38), the two objects of ‘give’, ‘beaded necklace’ and ‘T ͡sɨkt͡ʃɨku’, both receive =na. Likewise, in (39), the objects of ‘inform’, ‘that’ and ‘you’, also receive the accusative =na. (38) násɨ ̃t͡sɨkt͡sɨkun ʃuakán susámaji Nasɨ ̃ [T ͡sɨkt͡ʃɨku]=na [ʃuaka]=na su-sa-ma-ji Nasɨ ̃ T ͡sɨkt͡sɨku=ACC beaded.necklace=ACC give-ATT-REC.PT-3.PT ‘Nase gave T ͡sekt ͡seku a beaded necklace.’ (39) núna ámiɲa uháktathamɨ nu=na ami=na uha-ka-tata-hamɨ non.vis=ACC 2SG=ACC inform-INTENS-DEF.FUT-1SG>2SG+DECL ‘I am going to tell you that.’ 10.4.6.3. Genitive Genitive marking of a possessor noun (or noun phrase) is marked by cancelling apocope of the last vowel of the noun stem (cf. Table 10.5) and adding a high tone on the last vowel of the noun stem V́. Sometimes the genitive-marked NP surfaces with a "nal consonant n in addition to the high pitch. Thus the forms of the Genitive in Wampis are either V́ or V́n. The historical origin of the Genitive in relation with the 486 Accusative =na is discussed further below in this section. The most basic use of the Genitive in Wampis is to mark possession. In (40), the noun ‘our children’ is the possessor of ‘land’; it occurs with a high pitch, which marks the possessor. By contrast, in the Nominative of ‘our children’, the last vowel undergoes apocope and the high tone is placed on the second vowel, as shown in (41) . (40) ut͡ʃirí núŋka ut͡ʃi-ri ̃́ nunka child-1PL/2PL/3.POSS\GEN land ‘Our children’s land’ (41) ut͡ʃír ut͡ʃi-ri ̃ child-1PL/2PL/3.POSS ‘Our children’ In the following sentence, the word for ‘rain’ in the Genitive role has the high tone again in the last vowel,while in the Nominative form, the pronunciation is [júmi]. (42) jumí t͡ʃɨńt͡sak naŋkimaáta jumí t͡sɨnt͡saka nankima-a-́ta ́ rain\GEN dart throw-HIAF-IMP ‘Throw a rain’s dart!’ The next example shows the Genitive form V́n word-"nally; in the Nominative 487 form, the high pitch would be located in the second /a/: matáu. (43) matoún kat́i ̃ mataún kati ̃ palm\GEN shoot\1PL/2PL/3.POSS ‘The palm’s shoot’ 10.4.6.3.1. Possible historical development of the Genitive Overall has hypothesized the historical relationship between the Accusative and the Genitive in the related language Awajun in the following terms: “Thus historically, the accusative su#x may have been used to mark the possessor of a possessed noun, in a ‘benefactive’ sense. Then the dropping of "nal /n/ would have led to a reinterpretation, with the ‘possessor’ forms becoming a new genitive” (2007: 219). A similar development of the Accusative into a Genitive seems to have occurred in Wampis. As a "rst evidence of the development of the accusative into a genitive, we can observe the Genitive form of the personal pronouns (cf. §8.2.1.2), shown in Table 10.15. 488 Table 10.15. Nominative and Genitive forms of personal pronouns in Wampis Nominative Genitive 1SG ui mina ~ uina 2SG amɨ amina 3SG ni ̃ nina 1PL ii iina 2PL atumɨ atumí~atumina189 3PL nita nitá With the exception of 3PL, the Genitive pronouns are composed of the combining stems mi (1SG), ami (2SG), ni (3SG), atumi (2PL) plus the form na that, by all evidence, is none other than the accusative morpheme. In the case of the 3 person plural, the Genitive form does not have the element na, but the Genitive is marked with a high tone in the last vowel (the Nominative is pronounced [nítʲa] with high tone in the "rst vowel). The same occurs with the 2 plural variant atumí. Secondly, the accusative is one of the morphemes that induces shift in the placement of the high tone one mora rightward in two and three mora words (cf. §6.6). The vowel of the accusative normally undergoes apocope if it occurs word-"nally, 189. Actually, this form always surfaces as [atúmin] and the "nal /a/ is dropped following the rule of apocope in Wampis. By contrast, the 2 singular genitive form amina never deletes its last vowel, contravening the rule of apocope which deletes the last underlying vowel of the word. 489 which yields the form V́n, the same that is seen in the genitive. A further step in the re- analysis of the accusative would be the phonological erosion of the consonant /n/. Table 10.16 shows a comparison between nominative, accusative and genitive forms of nouns. Table 10.16. Nominative, acusative and genitive forms of numi 'tree' and naiapi ‘swallow-tailed kite’ Nominative Accusative Genitive numi [númi] numi=na [numiń] numi ́[numi]́ naiapi [najaṕ] naiapi=na [najapiń] naiapi ́[najapi]́ The fact that in Wampis the /n/ variably shows up in some realizations of the genitive form constitutes further evidence that the old accusative form is “still there” and its grammaticalization into marking possessor is relatively recent in the language.190 The third step would be the dropping of /n/, after which the only trace of the accusative is the high pitch, as in the word ‘bear’ or ‘woman’ in the table below. Further examples that link the genitive to the accusative form are provided in Table 10.17.191 190. In comparison, the reanalysis of the accusative into a genitive marker appears to have been faster in Awajun, where apparently the only marking of the genitive is -V́ (Overall 2007). 191. The "rst three examples in Table 10.15 are of words that were found in the database with the V́n genitive form, however notice that they can also occur simply as V́# in the genitive. 490 Table 10.17. Sample of nominative, accusative and genitive forms of nouns Nominative Accusative (=na)192 Genitive (V́n or V́) t͡súnki ‘water spirit’ t͡sunkína t͡sunkín matáu ‘palm’ mataúna mataún ʃuára ‘person’ ʃuarána ʃuarán t͡ʃái ‘bear’ t͡ʃaína t͡ʃaí núa ‘woman’ nuána nuá 10.4.6.4. Vocative The vocative identi"es a person being directly addressed during a communicative act. In Wampis, the vocative form of the noun is marked by canceling apocope and adding a high tone to the last vowel of the nominal word. The only other nominal morphemes compatible with the vocative in my data are the diminituve -t͡ʃi and the possessor person markers. Table 10.18 shows nominative and vocative forms of nouns with possessor person markers and the diminutive. 192. The underlined vowels in this column are dropped in the phonetic realization of the accusative form (thus making the Accusative forms basically equivalent to some of the Genitive forms). 491 Table 10.18. Vocative forms Nominative Vocative jat͡súr jatsu-ru brother-1SG nuwár nua-ru wife-1SG ut͡ʃít͡ʃ ut͡ʃi-t͡ʃi child-DIM nukút͡ʃ nuku-t͡ʃi mother-DIM nukút͡ʃru nuku-t͡ʃi-ru mother-DIM-1SG jatsurú jatsu-rú brother-1SG\VOC nuwarú nua-rú wife-1SG\VOC ut͡ʃit͡ʃí ut͡ʃi-t͡ʃí child-DIM\VOC nukut͡ʃí nuku-t͡ʃí mother-DIM\VOC nukut͡ʃrú nuku-t͡ʃi-rú mother-DIM-1SG\VOC 10.4.6.4.1. Vocative forms of kinship terms Some kinship terms have di$erent vocative forms than the one just described above, instead using the su#xes -ua ́and -ta.́ Both morphemes always receive high pitch and do not undergo apocope. Unlike the regular vocative form, these vocatives are apparently incompatible with person/possessive markers. Table 10.19 presents the 492 exceptional kinship forms. Table 10.19. Exceptional vocative kinship terms. Nominative Vocative apa ‘father’ apa-ua ́[apawa]́ nuku ‘moter’ nuku-ua ́[nukuwa]́ nauatu ‘daughter’ nauanta ́[nawanta]́ ut͡ʃi ‘child’ ut͡ʃitaá [ut͡ʃitʲa]́ nua ‘woman/wife’ nuataá [nuwata]́ iat͡su ‘brother of male’ iat͡sutaá [jat͡suta]́ a These forms alternate with a “regular” vocative form that is formed by the root plus a person/ possession marker; e.g. iat͡surú ‘my brother’ (vocative). 10.4.6.4.2. Vocative form of foreign names As many Wampis people have adopted Spanish names, the vocative forms of their names have been phonologically adjusted to "t into the Wampis structure. The basic rules of the vocative forms of Spanish names are the following: • Spanish names that end in a vowel cancel apocope and add a high tone like any normal Wampis noun: (44) Dina : Diná (Voc) Jaime : Jaimé (Voc) • Spanish names that end in a consonant (originally in Spanish) add an /a/ at the end if they do not end in /d/, /m/ or /n/ (I have no record of Spanish names ending in b or 493 p used by the Wampis as those are rare even in Spanish). Incidentally, this also shows that the vowel /a/ is part of the underlying form, i.e. one strategy to adapt foreign names in Wampis is to add paragogical vowel to maintain the word structure (no word in Wampis ends in a consonant underlyingly). (45) Abel : Abelá (Voc) Jesús : Jesusá (Voc) • If the Spanish name ends in /d/, /m/ or /n/, the Wampis add a syllable /ta/, /pa/ or /ka/, respectively. (46) Davíd : Davidtá (Voc) Simón : Simonká (Voc) Carmen : Carmenka ́(Voc) 10.4.6.4.3. Other cultural/familiar vocatives and terms to address people Apart from the vocative form of nouns, Wampis people use certain terms to address other people familiarly/colloquially or more respectfully: • Fathers use the words mat͡ʃu ‘boy’ (from Spanish ‘male’) or suki ‘testicle’ to address their sons in a familiar way. The term ʃuni ‘type of larva’ is also used to refer to/address a child. Parents and brothers also familiarly call a daughter/sister muhɨra ‘woman’ (from Spanish ). • In addition, the people who are considered uunta ‘elder’ (or ‘respected adult person’) are not called by their names, but are called uunta [úun] ‘elder’ or iit͡ʃi [iit́͡ʃ] ‘uncle’ if 494 there is a close personal relationship. Usually a possessive form marked with the vocative is used. • Women of status within the family or social group are called nuku ‘mother’+posessor marker, or nukut͡ʃi ‘grandmother’; also the Spanish ‘mother’ is used. • Close friends or relatives also use the form iat͡ʃito [jat͡ʃíto] (iat͡ʃi ‘brother+3’ plus the Spanish diminutive -ito) to address each other in informal contexts—this form is specially common among young male speakers.193 Otherwise, the more traditional forms iatsu-rú ‘brother-1SG\VOC, or iat͡suta ́(brother-VOC), which are still very common, are used. 193. The feminine form to refer to ‘brother’, uma, does not have a correspondent form with the Spanish diminutive. 495 CHAPTER XI THE NOUN PHRASE AND POSTPOSITIONS 11.1. Introduction Chapter XI explores the Noun Phrase (NP) in Wampis. The chapter begins with a de"nition of the NP in Wampis in §11.2. The discussion continues with a description of several construction involving a Noun Phrase in §11.3. After a brief note on case in §11.4, in §11.5 di$erent postpositional enclitics that form are discussed. 11.2. The noun phrase (NP) in Wampis In Wampis, the simplest noun phrase is headed by a noun or a pronominal form without modi"cation. (1) puhupat́ kutaŋkań nahańarmaji [puhupata] [kutanka]=na nahana-ra-ma-ji Puhupat bench=ACC make-DISTR-REC.PT-3.PT+DECL ‘Puhupat made a bench.’ The NP can be determined by a demonstrative, by tikit ͡ʃi ‘other’, by a personal pronoun acting as an de"nite article (infrequent in the data), by a numeral or non- numeral quanti"er—all of them typically precede the head (quanti"ers may appear after the noun). In addition, aia or aiatɨkɨ (both mean ‘only’) also can occur in the NP, typically preceding the head, but they can occur after the noun as well. In the data, all 496 of the above mentioned elements (demonstratives, tikit͡ʃi ‘other’, personal pronoun acting as articles, quanti"ers, aia and aiatɨkɨ ‘only’) do not co-occur.194 Modi"cation with uɨantu ‘group’, adjectives and relative clauses typically follow the head. More than one modi"er can occur in the NP, but text data shows that NPs in Wampis usually tend to be simple, consisting of a demonstrative (or other speci"er) and a noun head. Relative clauses are used more often than adjectives for noun modi"cation. More than one modi"er can occur in the NP, but only lexical noun heads can be modi"ed with adjectives. Figure 11.1 shows the elements of a NP. Note that RC=Relative clause. !! (Demonstrative) (tikitʃ͡i) N (uɨantu) (Adj) (Quantifier) (RC) (Personal Pronoun) (aia), (aiatɨk) NP Figure 11.1. Structure of Noun Phrase The structure provided in Figure 11.1 serves to introduce the NP constructions in the next sections. 194. A demonstrative and tikit ͡ʃi ‘other’ co-occur only when tikit ͡ʃi acts as the head of the NP. 497 11.3. NP constructions 11.3.1. The determined NP The second simplest NP (the "rst is just the noun occurring alone) is constituted by any of the elements that typically occur preceding the noun head; i.e. a demonstrative, tikit ͡ʃi ‘other’, quanti"er, personal pronoun and aia, aiatɨk ‘only’. These elements, in general, serve to specify the reference of the noun head, so it can be said that they function as determiners.195 I call this the “Determined NP”. Demonstratives have the distinctive property of triggering agreement with the noun head and its modi"ers, as explained previously in §8.3.1. (2) núna ʃuarań wɛińkamhɛɛ nu=na ʃuara=na pɨńkɨra=na uaina-ka-ma-ha-i non.vis=ACC person=ACC good=ACC see-intens-REC.PT-1sg.sbj-DECL ‘I saw a good person.’ 11.3.2. The possessive construction The simple possessive construction is done with a noun marked as possessed, the 195. It is not yet clear whether there is a Determiner Phrase in Wampis. Morphosyntactically, I have shown that all of the above mentioned elements that serve to determine an noun have very di$erent properties and there is no evidence that they form a class (see Chapter VIII). But the question of a Determiner Phrase merits further research—identifying tests such as coordination is di#cult as coordination is not very relevant in Wampis. Thus, more evidence, specially in terms of designing tests, are needed to consider a Determiner Phrase as a relevant structure in Wampis. 498 possessor may occur optionally, and is marked with the genitive. If a genitive noun occurs in the phrase, it always precedes the possessed noun and nothing stands between the two of them. (3) miɲa pataár mina pataa-ru 1SG.GEN relative-1SG ‘My relatives’ Apparently, the marking of possession is not obligatory when the possessor is not animate. Compare: (4) miɲ́a núŋkar mina nunka-ru 1SG.GEN land-1SG ‘my land’ (5) mit͡ʃa ́núŋka mit͡ʃa ́ nunka cold/GEN land ‘the land of cold’ There are cases of nesting with possessive NPs, as the following example shows. (6) miɲ́a nukút͡ʃru nukuri ́ mina nukut͡ʃru nuku-ri ̃ 1SG.GEN grandmother-1SG mother-1PL/2PL/3.POSS ‘my grandmother’s mother’ 11.3.3. The attributive construction In the attributive construction, Adjectives and Relative clauses modify the noun 499 head, attributing some property to the head noun. Any of the elements that can function as determiners may optionally occur. Modi"ers of nouns most frequently follow the noun head: (7) ʃuaŕ sutaŕ wɨáwɛɛ [ʃuara sutara] uɨ-a-ua-i person short go-ipfv-3.sbj-DECL ‘The short person is going.’ (8) núwa pɨŋ́kɨr taaḿaji nua pɨńkɨra ta-a-́ma-ji woman good arriv-hiaf-REC.PT-3.PT+DECL ‘A good woman arrived.’ Relative clauses are a function of nominalization in Wampis. There are two major relative clause constructions. One is done with nominalizers, most frequently with Set II nominalizer -u ‘subject nominalizer’ and -mau ‘non-subject nominalizer’ (cf. Chapter XV for more details on relativization): (9) núwa wɨńɨ ̃aparú [nua [uɨnɨ ̃ apa-ra-u]] womanmouth\1PL/2PL/3.POSS put.together-DISTR-NMLZ ‘the woman that sew her mouth’ (10) pińt͡ʃu nayaṕ tamóo [pint ͡ʃu [naiapi ta-mau]] raptor.bird swallow-tailed.kite say-NMLZ ‘The bird called Nayap’. 500 The other strategy of relativization is done by relativizing a copula verb with a demonstrative, as shown in (11) and (12). (11) núu ʃuaŕ mikút anú [nu ʃuara [Mikuta a=nu]] non.visperson Mikut cop=non.vis ‘that Person who was Mikut’ (12) ut͡ʃiri ́t͡sakɛíɲa ańuna [ut͡ʃi-ri ̃ [t͡saka-ina a=nu=]]na child-1pl/2pl/3.poss grow-pl.ipfv cop=dem=ACC ‘our children that are growing’ Relativization is described in detail in §20.2. 11.3.4. The plural construction Wampis does not mark plurality on the noun and plurality is not an obligatory category of nouns. However, plurality can be optionally marked at the level of the NP via the copula a plus the imperfective plural marker -ina. In this construction, the head always precedes the copula. The plural construction might have come historically from a relativization construction with a demonstrative. In fact, often times the demonstrative still occurs relativizing the copula, most frequently nu ‘non-visible’, but also au ‘distal’. I have no examples of the plural construction with the other demonstratives (although they can 501 relativize a copula for other relativization functions). (13) t͡ʃai ́t͡ʃit͡ʃaḿɨ ̃antúrak jawaá ɛíɲa núu t͡ʃai t͡ʃit͡ʃamɨ ̃ antu-ra-ka ̃ [iauaa a-ina nu] bear word\1PL/2PL/3.POSS obey-INTENS\3SG.SS jaguar COP-PL.IPFV NON.VIS ‘The jaguars listened to the bear’s words.’ (14) ʃuar pɨŋkɨrt͡ʃau ɛiɲa nuka [ʃuar pɨnkɨra-t͡ʃau a-ina nu]=ka person good-NEG.NMLZ COP-PL. IPFV NON.VIS=FOC ‘bad people’ There are plenty of examples that show that the structure [Noun a-ina] can occur alone, without the demonstrative relativizing the copula marked for plural. Note in particular the marking of the accusative in (15) over the whole plural construction (attaching after the copula), which is evidence that the plural construction is being analyzed as an equivalent of a NP. More evidence for this is given in (16), where all modi"ers of the noun receive the accusative =na, including the copula, in accordance with the rule of agreement marking triggered by a demonstrative; and in (17) where the locative is hosted by the copula. These examples suggest that the plural construction has been reanalyzed as a plural NP. (15) warit́͡ʃiri ̃ɛíɲan huúk [uari-t͡ʃi-ri ̃ a-ina]=na huu-ka ̃ thing-DIM-1PL/2PL/2.POSS COP-PL.IPFV=ACC gather-INTENS\3SG.SS ‘having gathered her little things’ 502 (16) núna ut͡ʃiń unúimaru ɛíɲan ɨaḱmiahɛɛ [nu=na ut͡ʃi=na [unuima-ra-u] a-ina]=na ɨa-ka-mia-ha-i NON.VIS child=ACC learn-DISTR-NMLZ COP-PL.IPFV=ACC look.for-INTENS-DIST.PT-1SG.SBJ-DECL ‘I looked for those educated children.’ (17) kuntiń ɛíɲanam [kuntina a-ina]=nama animal COP-PL.IPFV=LOC ‘in the animals’ Importantly, possession cannot be marked on the copula, only on the noun, as shown in (18). This corresponds with the idea that the plural constructions works at a level higher than the level of the noun word; i.e. at the level of the phrase. (18) pataár ɛíɲa pataa-ru a-ina relative-1SG COP-PL.IPFV ‘my relatives’ The plural construction is optional for the most part, but it is very frequent in enumeration of plural referents. 11.3.5. Construction with uɨantu ‘group’ The word uɨantu can modify a noun to provide the idea of ‘group’ or ‘class of N’. The minimal structure in which this construction occurs is: [N wɨantu] 503 In this construction, the noun has to be an overt expression in the NP and must precede the modi"er uɨantu. Interestingly, the copula plus plural marker (a-ina) never occurs when uɨantu occurs (i.e. they are in a complementary distribution). Semantically, the uɨantu-construction gives somewhat unspeci"c information in the sense that unlike typical class noun systems, in this cases no animacy/shape/number/gender or other distinction is made. The next examples illustrate the use of uɨantu ‘group’. The following phrase was translated as ‘sachamango and its varieties’: (19) apɛí wɨántu apai uɨantu sachamango group ‘sachamango (Gustavia superba) and its varieties’ In the next example the idea of “group” (that of disabled people) is clearly conveyed via the [N uɨantu] construction. The speaker is telling about a case of collective psychosis that occurred in her village. As a consequence of forceful physical acts (e.g. climbing on tress, jumping to the river from a high stand-point, and so on), a group of villagers became disabled. To refer to this group, the speaker uses uɨantu with the words meaning ‘crippled’ and ‘dislocated’. Note that the verb hasarmiaji ‘became’ is elided in the "rst line. Even though the English translation does not clearly transmit 504 that meaning, in Wampis the more literal meaning is ‘a group who are crippled.’ (20) a. nuí kuntúnam muúnkaru wɨántu nuĩ kuntu=nama muuna-ka-ara-u uɨantu there arm=LOC cripple-INTENS-PL-NMLZ group ‘they [became] crippled,’ (Lit.: ‘They became cripplers in their arms’) b. taŋkírin kuíkʲaru wɨántu hasármiaji tanki-rĩ kui-ka-ara-u uɨantu has-ara-mia-ji spine-1PL/2PL/3PL.POSS dislocate-INTENS-PL-NMLZ group become-PL-DIST.PT-3.PT+DECL ‘they became disable people’ (Lit.: ‘They became dislocaters of their spines’) 11.4. Case This is just a short note to remind the reader that two of the four cases of Wampis work at the level of the word: the genitive and the vocative, which are marked with a high tone. The nominative is zero-marked. The other remaining case, the accusative =na, is the only enclitic case and works at the level of the noun phrase. Nonetheless, for practical reasons, the accusative was described in the previous chapter, with the rest of case markers (cf. §10.4.6.2). 11.5. Postpositional clitics and the marking of oblique relations Postpositions mark oblique relations. Adpositions introduce a phrase that is typically an optional constituent in the sentence. In Wampis, all adpositions are enclitics with the exceptions of a high tone locative morpheme. 505 11.5.1. Locatives Wampis has three di$erent locatives, two involve consonant and vowel segments and the other is tonal: =nVma, =(n)i,̃ and high tone on the last surface vowel (symbolized here as V́), as indicated in Table 11.1. The "rst two are used for spatial and temporal location, whereas V́ is only used in a spatial sense and occurs only on a subset of nouns. As spatial locatives, all three di$er very little in their semantics. Table 11.1. Functions of locative markers Postpositional morphemes Spatial Temporal =nVma ✓ ✓ =(n)i ̃ ✓ ✓ V́ ✓ --- The main distinction between =nVma and =(n)i ̃is just morphological conditioning: =(n)i ̃only occurs with possessed noun stems and bare demonstratives, whereas =numa can occur with possessed or non-possessed nouns. All three locatives in Wampis convey “di$use” semantics; i.e. they do not specify the exact position of the "gure relative to a ground (‘inside’, ‘on’, ‘under’, etc). It will be seen in §17.5.3.3.2 on non-verbal predication that Wampis uses a set of verbal items to, 506 among other things, convey a more speci"c sense of position or location. The locatives also can be used for stative location with stative verbs or, with dynamic verbs, they indicate the direction of the movement. In (21)–(22) the locatives mark stative location: (21) hɨá̃ puháwɛi hɨã ́ puha-ua-i house\LOC live+IPFV-3.SBJ-DECL ‘He is in the house.’ (22) hɨá̃nam puháwɛi hɨã=nam puha-ua-i house=LOC live+IPFV-3.SBJ-DECL ‘He is in the house.’ In the next examples, the locative =numa marks direction with motion verbs: (23) hɨɰ̃ánum taáji hɨã=numa taa-á-ji house=LOC arrive-HIAF-3.PT+DECL ‘He arrived at the house.’ (24) hɨɰ̃ánum wɨḿaji hiã=numa wɨ-́ma-ji house=LOC go-REC.PT-3.PT+DECL ‘He went to the house.’ In the above examples, =numa is used; following are some examples with the other locatives with approximately the same propositional content as in (23)–(24) above: 507 (25) hɨá̃rui ̃taáji hɨã-ru=i ̃ ta-á-ji house-1SG=LOC arrive-HIAF-3.PT+DECL ‘He arrived at my house.’ (26) hɨá̃ taáji hɨá̃ ta-á-ji house\LOC arrive-HIAF-3.PT+DECL ‘He arrived at the house.’ 11.5.1.1. Locative =nVma The locative =nVma has two forms, =nama and =numa. The alternations between these two allomorphs is not predictable synchronically and they might be the remnants of two old competing forms. The morpheme =nVma is the most frequent of all three locatives in Wampis, and besides its adnominal function it is used to form relatives and adverbials. The locative =nVma usually attaches to the last element of the NP. (27) aʃí t͡ʃankín uúnnum t͡ʃimpíawɛɛ aʃi t͡ʃankina uunta=numa t͡ʃimpi-a-ua-i all basket big=LOC put.in-IPFV-3.SBJ-DECL ‘she is putting all in the big basket.’ Verbs of cognition and hitting/contact may have their arguments marked with the locative instead of the accusative. 508 (28) t͡sunkínmaʃa anɨńtaimtaiɲaiti t͡sunki=nVma=ʃa anɨntaima-tai=̃aiti water.spirit=LOC=ADD think-NMLZ=COP.3+DECL ‘In the water spirit also we (normally) think.’ Like in many languages, the locative =nVma sometimes shows metaphorically extended meanings, such as more abstract locative states. In the next example, the noun marked with =nVma can be more properly translated as ‘in the Wampis culture...’: (29) wampiśɲum núkap arútam aẃɛi uampisa=nVma nukapɨ arutama a-ua-i Wampis=LOC many power.vision exist-3.SBJ-DECL ‘In Wampis, there are many Arutam.’ Very commonly =nVma follows a nominalizer to create clauses functioning as relatives (of location) and adverbials (temporal/reason). The next two examples illustrate locative relative clauses: (30) urút͡ʃ kutámramu áunam [urut͡ʃi kutama-ra-mau a-u=nama] cotton thread-DISTR-NMLZ exist-NMLZ=LOC ‘Where there is a [ladder made of] threaded cotton’ (31) ɲawaa ̃ ́mat͡sámounam [iauãa mat͡sa-mau=nama] jaguar inhabit-NMLZ=LOC ‘[He brought it] to where there were jaguars.’ When a nominalized verb is imperfective or durative, the =nVma-marked clause indicates a simultaneous or durative action. The next two examples illustrate the use of 509 the locative in this temporal/adverbial construction: (32) mat͡sámsami támounam mat͡sama-sa-mi ta-mau=nama live.together-ATT-HORT say+IPFV-NMLZ=LOC ‘When we want to live’ (lit. ‘When we say “let’s live”’) (33) ɨnt͡sát͡ʃiɲam utúk wɨńamunam ʃuár pampá tímaji nuhíɲam ɨnt͡sa=t͡ʃi=nama utu-ku wɨ-ina-mau=nama river=DIM=LOC collect-SIM go-PL.IPFV-NMLZ=LOC ʃuara pampa timaji nuhi=nama person make.noise+IPFV NARR nose=LOC196 ‘While they were collecting [snails] in the little river, people were making noise…’ Likely through its use in temporal adverbials, =nVma has extended its uses to other the expression of other notions, such as reason or cause. Notice, however, that it is in perfective contexts where the locative provides the sense of reason or cause to the clause. Example (34) illustrates this use. (34) núu juwí ukámðramunam nuiñá núwɨk̃a waínkamuiti naháanaruiti tíɲu ármaji nu iuui ukama-ra-mau=nama nui=̃ia NON.VIS squash spill-DISTR-NMLZ=LOC there=ABL ‘Because that squash had been spilled, ever since then, nuɨ=̃ka uaina-ka-mau=iti nahana-ra-u=iti clay=FOC look-INTENS-NMLZ=COP.3+DECL make-DISTR-NMLZ=COP.3+DECL ‘It has been seen, it has transformed (the squash into clay),’ 196. The term nuhinama has the conventional meaning of ‘upstream’, or ‘above the river’s bank’ (the rivers are usually below the level where people actually live). 510 ti-inu a-ara-ma-ji say+LOAF-AGT.NMLZ COP-PL-REC.PT-3.PT ‘They said.’ In the next example, we see two consecutive clauses marked with =nVma (in the second, third and sixth lines of the example). In line 35.b, the clause was interpreted by my Wampis teachers with a temporal meaning. On the other hand, in line c, the interpretation given was an interpretation of reason. (35) a. jáunchukka huiñkʲa awarúnka at͡súhakuiti iaunt͡ʃukɨ=ka hui=̃ka auaruni=ka a-t͡su-hak-u=iti long.ago=FOC here=FOC Awajun=FOC exist-NEG-HAB.PT.NMLZ=COP.3+DECL ‘Long ago, there was no Awajun here’ b. húka jamɛɛ́ t͡ʃit͡ʃám naŋkáŋkamunam hu=ka jamai t͡ʃit͡ʃama nankana-ka-mau=nama NON.VIS=FOC now problem197 "nish-INTENS-NMLZ=LOC ‘Now when the problems have ended’ c. t͡ʃit͡ʃama iwáramunam t͡ʃit͡ʃama iuara-mau=nama problems "x-NMLZ=LOC ‘Because the problems were "xed’ d. awarúŋka kámɨ huniñkʲa ut͡saánawɛiti awarunka kamɨ hui-̃ni=ka ut͡saána-u=aiti Awajun INTERJ here-ALL=FOC enter-NMLZ=COP.3+DECL ‘The Awajun have entered over here’ 197. The word t͡ʃit͡ʃama means ‘speech, language, word’, and also ‘problem’. 511 e. huínkʲa kámɨ kahɨŕnajamu atsutí túsa hui=̃ka kamɨ kahɨ-ra-nai-a-mau a-tsu-tí tu-sã here=FOC INTERJ be.angry-DISTR-RECIP-IPFV-NMLZ exist-NEG-JUSS say-SUB\3SG.SS ‘having said, “Let it not exist hatred here,”’ f. maa túkɨ t͡ʃit͡ʃám umíkmounam iɲɛís̃amunam maa tukɨ t͡ʃit͡ʃama umi-ka-mau=nama198 inai-̃sa-mau=nama INTERJ INTERJ problem "nish-INTENS-NMLZ=LOC stop.doing-ATT-NMLZ=LOC ‘when they agreed and left them (the Awajun) alone.’ What is also very interesting is that in line 35.f of the above example, we "nd two locative-marked clauses that, semantically, have a temporal interpretation (i.e., to paraphrase, ‘when they agreed, when they left the Awajun alone, the Awajun entered over here’). 11.5.1.2. Locative =(n)i ̃ The locative =(n)i ̃occurs on demonstratives and nouns. The allomorph =ni ̃ occurs following another vowel i or u. Unlike =nVma, =(n)i ̃only attaches to possessed noun stems. Thus notice the following di$erent locative markings: (36) hɨã=nam ~ hɨa ̃́ ~ hɨa-ru=i ̃ house\loc house=loc house-1sg=loc ‘in the house’ ‘in the house’ ‘in my house’ 198. The phrase t͡ʃit͡ʃama umikamau has a conventionalized meaning of ‘agree’, ‘pact’ (‘"nish the problem’). 512 (37) míɲa piníŋrui ̃ mina pininka=ru-i ̃ 1sg.gen pot=1sg-loc ‘in my pot’ The locative =(n)i ̃probably has grammaticalized into a di$erent-subject marker in subordinated clauses (§19.4.2). It is also found grammaticalized in the Set I nominalizer -tai ̃(§15.4.4) that creates object or location nominalizations. 11.5.1.3. Locative V́ A subset of nouns can optionally shift their high tone to the last surface vowel to mark the locative, mostly nouns that are frequently used in locational predications. Table 11.2 presents a sample of words that can receive tonal locative marking in Wampis. A phonetic realization of the nominative and locative forms is given to illustrate the placement of high tone. Note that the examples in Table 11.2 are examples of nouns without other morphemes than the locative. The actual placement of the high pitch may vary depending on the morphological form of the complete noun (i.e. adding more morphology may lead to deletion of di$erent vowels, thus potentially leading to the high pitch to be located in a di$erent place). 513 Table 11.2. Sample of words that can receive the locative V́ Term Nominative Locative Gloss ikama [íkʲam] [ikʲám] ‘forest’ ɨnt͡sa [ɨńt͡sa] [ɨnt͡sá] ‘river’ hɨã [hɨ ̃á] ~ [hɨ ̃ɰ́a] [hɨá̃] ~ [hɨɰ̃á] ‘house’ numi [númi] [numí] ‘tree’ aha [áha] [ahá] ‘farm’ nunka [núŋka] [nuŋká] ‘land, earth, below (with loc)’ pininka [piɲíŋ] [piɲíŋ] ‘pot’ iaki [jáki] [jakí] ‘high place, above (with loc)’ nuhi [núhi] [nuhí] ‘upstream’ hinta [híntʲa] [hintʲá] ‘trail’ inita [iɲít] [iɲít] ‘inside, bottom’ From Table 11.2, some terms may seem ambiguous in their interpretation as their normal high tone may coincide with the locative high tone. For instance, inita ‘inside’ is usually realized as [iɲít] in the nominative and in the locative if there is no other morpheme following. However, when the restrictive =kɨ is added and the underlying last vowel surfaces, the high tone moves to the last vowel if it is marked for the locative. (38) aʃí aʃí aʃí iɲiták iɲiták it͡ʃíntukar aʃí aʃí aʃí inita=kɨ inita=kɨ i-t͡ʃina-tu-ka-ara all all all bottom=RESTR bottom=RESTR CAUS-dig-APPL-INTENS-3PL.SS ‘They dug all around underneath.’ 514 The following elicited examples further clarify the importance of distinguishing high tone placement for the locative. The verb uɨ ‘go’ is also used as a semi-copula meaning ‘become’ (see §17.6). In (39), where ‘Ikam’ (a proper noun) is the subject of ‘go’, the placement of high tone is on the "rst vowel. In this case, ‘go’ is interpreted in a semi-copular sense of ‘become’. On the other hand, in (40) ‘forest’ has the high pitch in its last surfacing vowel a, and the sentence is interpreted as a motion verb with the location being marked by the high tone. (39) íkʲam wɨáhɛɛ iḱama wɨ-́a-ha-i Ikam go-IPFV-1SG.SBJ-DECL ‘I am becoming Ikam.’ (40) ikʲám wɨáhɛɛ ikám wɨ-́a-ha-i forest\LOC go-IPFV-1SG.SBJ-DECL ‘I am going to the forest.’ All of the words in Table 11.2 can also occur with the other two locatives. There are no examples of the locative V́ where it is used with temporal semantics. All examples of this postposition are restricted to spatial uses. 515 11.5.2. Ablative =ia The ablative su#x marks a point of prior temporal or spatial location. The ablative can combine with nouns, pronouns, and demonstratives. Among its uses, it typically marks ‘origin’: (41) kanúsia Kanusa=ia Santiago River=ABL ‘from Kanús (Santiago River)’ (42) wabália Huabal=ia Huabal=ABL ‘from Huabal’ (43) tujá tu=ia where=ABL ‘from where?’ (44) kuámkajak ʃuarɛitmɨ? kuamaka=ia=ka ʃuara=ita-mɨ jungle=ABL=Q person=COP-2SG.SBJ+DECL ‘Are you a person from the jungle?’ When marking a spatial location, the ablative frequently occurs on a noun that is used with a locative sense, or else which bears the locative enclitic. In elicitation, the 516 locative-ablative structure was optional in most other cases. Thus, for instance, (45) and (46) bear exactly the same meaning. (45) wabáliajɛíthɛɛ Huabal=ia=aita-ha-i Huabal=ABL=COP-1SG.SBJ-DECL ‘I am from Huabal.’ (46) wabálnumiajɛíthɛɛ Huabal=numa=ia=aita-ha-i Huabal=LOC=ABL=COP-1SG.SBJ-DECL ‘I am from Huabal.’ A short non-exhaustive list of nouns that frequently occur with the locative when they carry the ablative is given in Table 11.3. Table 11.3. Words that occur with the locative when carrying the ablative Term Gloss ɨnt͡sa ‘river’ hɨã ‘house’ uaã ‘hole, cave’ numi ‘tree’ kut͡ʃa ‘puddle, lake’ nuhi ‘up stream, above stream’ (cf. nuhi ‘nose’) amai ‘other side of the river’ With verbs of motion, the ablative marks the starting point or source of the 517 movement: (47) iquítusiant͡ʃa itʲármiaji Iquitus=ia=na=ʃa ita-ara-mia-ji Iquitos=ABL=ACC=ADD bring-PL-DIST.PT-3.PT+DECL ‘They brought the one from Iquitos too.’ (48) puértormiɲiɲia wɨkáttahɛɛ nant͡ʃáhɛɛ ̃ puerto-rumi-ni=̃ia wɨka-tata-ha-i nant͡ʃa=hai ̃ dock-2SG-LOC=ABL travel-DEF.FUT-1SG.SBJ-DECL boat=COM ‘I am going to travel from your dock by boat.’ When indicating a point in past time, the ablative usually attaches directly to a noun or demonstrative. For instance, when the ablative follows nui ̃‘there’, either it indicates the starting point of a period of time (‘ever since, after’) or it works as a conjunction with the meaning of ‘moreover, also’. In (109), nui=̃ia establishes the starting point of a period: (49) nuiɲá ʃíir unúimaru ásar takaḱi wɨńawɛi nui=̃ia ʃiira unuima-ra-u a-sa-ri there=ABL well learn-DISTR-NMLZ COP-SUB-1PL.SS taka-ki wɨ-ina-ua-i work-WHILE.MOVING go-PL.IPFV-3.SBJ-DECL ‘Ever since then, having learnt well, they continue to work.’ And in (50), nui=̃ia (which appears in line b) means ‘also’: (50) a. núna t͡saníŋkʲan mamá kuntúriɲa óona nuiɲa paánmat͡ʃinʃa sɨámu tímayi nu=na tsaninka=na NON.VIS=ACC manioc.stem=ACC 518 mamá kuntu-ri=̃a au=na manioc\GEN arm-1PL/2PL/3.POSS=COP DIST=ACC ‘To the manioc stem, to the manioc’s branches [she pleaded]’ b. nuiɲa paánmat͡ʃinʃa sɨámu tímaji nui=̃ia paantama-t͡ʃi=na=ʃa sɨa-mau timaji there=ABL plantain-DIM=ACC=ADD plea-NMLZ NARR ‘also (Lit: ‘from there’) she pleaded with the little plantain [to get some food].’ The ablative is also used in one of the Wampis comparative constructions. The NP that serves as the standard of comparison bears the ablative. The next examples illustrate this point: (51) ámɨka wíja óu jat͡ʃáitmɨ amɨ=ka ui=ia au iat͡ʃa=ita-mɨ 2SG=FOC 1SG=ABL DIST wise=COP-2SG.SBJ+DECL ‘You are wiser than me.’ (52) ni ̃ŋ́kʲa wíja óu sutárt͡ʃitmɨ ni=̃ka ui=ia au sutara-t͡ʃi=ita-mɨ 3SG=FOC 1SG=ABL DIST short-DIM=COP-2SG.SBJ+DECL ‘He is shorter than you.’ It is interesting to note that some examples show that the ablative can co-occur with the accusative case. In (53), we see that in the NP [from the house] carries the ablative+accusative markers. Furthermore, in (54) the postpositional phrase ‘from Ecuador’ is treated as a core argument and receives also the object marker =na— interestingly, in this case the interpretation of the NP seems similar to a “headless” 519 relative clause (‘the brought [one] from Ecuador too’). (53) paánman hɨá̃nmajan juámaji paantama=na hɨã=nVma=ia=na iu-á-ma-ji banana=ACC house=LOC=ABL=ACC eat-HIAF-REC.PT-3.PT+DECL ‘He ate the banana from the house.’ (54) ecuaðórnumiant͡ʃa itʲaŕmaji Ecuador=numa=ia=na=ʃa ita-ara-ma-ji Ecuador=LOC=ABL=ACC=ADD bring-PL-REC.PT-3.PT+DECL ‘They brought [one] from Ecuador too.’ 11.5.3. Allative =ni The allative =ni indicates motion to a location. However, the allative is much less used than the locatives seen in 11.5.1 above which, as we saw, can also assume allative functions. The examples below illustrate the use of =ni. In both (55) and (56), the verbs predicate a movement from one place to another, thus =ni is used to indicate the location where the theme moves to. (55) nantuníʃa ʃuárka wakɨt́u hakú tímaji nantu=ni=ʃa ʃuara=ka wakɨtu hak-u ti-ma-ji moon=ALL=ADD person=FOC return HAB.PT-NMLZ say+LOAF-REC.PT-3.PT+DECL ‘To the Moon199 the people used to return, she said.’ 199. Here ‘Moon’ metonymically stands for the ‘space that mythical heroes inhabit (Moon, Stars, the Sun, etc.’. In the Wampis mythology, there used to be transit between di$erent worlds. 520 (56) amáiɲiɲ̃i katíŋkʲan wɨákun át͡ʃu ámaunam ímani ̃jahá wɨáhai amai=ni=̃ni katína-ka-nu wɨ-́a-ku-nu at͡ʃu a-mau=nama other.side=LOC=ALL cross-INTENS-1SG.SS go-IPFV-SIM-1SG.SS aguaje exist-NMLZ=LOC ‘To the other side of the river, having crossed, going where there is aguaje (Mauritia 0exuosa)’ With stative verbs, the allative can appear following the locative =nVma. In such cases, the allative =ni conveys a meaning of an unde"ned or unspeci"ed location, i.e. ‘to be somewhere around LOC’. In the next example, a stative verb ‘live/be’ occurs with the locative+allative construction: (57) aánman nápi áwai aa=numa=ni napi a-ua-i outside=LOC=ALL napi exist-3.SBJ-DECL ‘There is a snake (somewhere) outside.’ (58) ikámnuman puhawɛɛ́ ikama=numa=ni puha-ua-i forest=LOC=ALL live+IPFV-3.SBJ-DECL ‘he is (somewhere) in the forest.’ Notice that the ‘be somwhere around LOC’ meaning seems to be an extension proper of the allative itself, as demonstratives which carry the allative can have an allative interpretation. 11.5.4. Comitative =hai ̃ The comitative =hai ̃indicates company or instrumental use of an object. In the next examples, the NP that receives =hai ̃identi"es the one with whom the Agent is 521 carrying out the action; i.e. these are comitative uses of =hai.̃ (59) mankártuaka niŋkíkʲa ihámɨat͡ʃu nuwɨj̃ɛɛ ̃ihámu mã-karata-u=a=ka ni=̃kɨ=ka ihamu-a-t͡ʃau kill-1PL.OBJ-NMLZ=COP=FOC 3SG=RESTR=FOC jump-IPFV-NEG.NMLZ ‘The murderer, he did not jump alone,’ nuɨ=̃hai ̃ iha-mau wife\1PL/2PL/3.POSS=COM jump-NMLZ ‘he jumped with his wife.’ (60) nantú núwɨ ̃ásã kámɨ mat͡sátu tímaji nantújɛi ̃ Nantú nuɨ ̃ a-sã kamɨ Moon\GEN wife\1PL/2PL/3.POSS COP-SUB\3.SS INTERJ ‘Because she was Nantu’s wife,’ mat͡satu-u timaji Nantu=hai ̃ be.together-NMLZ NARR Moon=COM ‘She lived with Nantu.’ The next examples illustrate the instrumental use of =hai:̃ (61) ni ̃í wɛitín numíhɛɛ ̃awáʔtɨawɛɛ ni ̃ uaiti=na numi=hai ̃ auatu-a-ua-i 3SG door=ACC tree=COM hit-IPFV-3.SBJ-DECL ‘He hits that door with a stick.’ (62) númi naɨkhɛi ̃ hiŋkʲáamuiti numi naɨka=hai ̃ hinka-a-mau=iti stick rope=COM tie-IPFV-NMLZ=COP.3+DECL ‘The stick is tied with rope.’ The distinction between comitative and instrumental in Wampis has to do with animacy: with prototypical animate (especially human) entities, =hai ̃normally entails a 522 comitative use; with inanimate entities, the adposition =hai ̃normally conveys an instrumental function. Of course, it is possible to have an animate instrument. In (63), the comitative marks the means by which the theme (‘the document’) reaches its goal (Lima). By contrast, in (124) an inanimate entity (‘knife’) that can be loosely interpreted with the sense of ‘company’: (63) límanam papín akúpkarmiaji ámiɲa út͡ʃimhɛɛ ̃ lima=nama papi=na akupɨ-ka-ara-mia-ji amina ut͡ʃi-mɨ-hai ̃ Lima=LOC document=ACC send-INTENS-PL-DIST.PT-3.PT+DECL 2SG.GEN child-2SG-COM ‘They sent the document to Lima with your son.’ (64) kutʃírhɛɛ ̃wɨáhɛɛ kút͡ʃi-ru=hai ̃ wɨ-́a-ha-i knife-1SG=COM go-IPFV-1SG.SBJ-DECL ‘I am going with my knife.’ An NP marked with the comitative/instrumental also may function as a modi"er of a head noun. (65) ʃiámpin paántam hiámuhɛiñsha yuát͡ʃarmaʃi [ʃiampi=na [paantama hia-mau=hai]̃]=na=ʃa iu-á-t͡ʃa-ara-ma-ji hen=ACC plantain roast-NMLZ=COM=ACC=SPEC eat-HIAF-NEG-PL-REC.PT-3.SBJ+DECL ‘Maybe they ate hen with roasted plantains.’ In (66), the comitative attaches to the whole portion ‘that Mikut who had transformed at the beginning.’ In this case the action is more reciprocal. (66) inkúniakuiti tikíchik tsawántai nú míkut jáma naŋkámt͡ʃak míkut naháanaru anúhɛɛ ̃ inkunia-ka-u=iti tikit ͡ʃiki t͡sauantai nu Mikut meet.face.to.face-INTENS-NMLZ=COP.3SG+DECL one day ana Mikuta ‘One day that Mikut met’ 523 jáma naŋkámt͡ʃak míkut nahaanaru anuhɛɛ ̃ iama nankama-t͡ʃa-ka Mikutanahaana-ra-u a=nu=hai ̃ now happen-NEG-INTENS Mikut make-DISTR-NMLZ cop=NON.VIS=HAĨ ‘with that Mikut who had transformed at the beginning.’ The next examples are also reciprocal; notice the use of the reciprocal su#x on the verb: (67) wampíska awarúnhɛɛ ̃maánihakaruiti uampisa=ka auaruni=hai ̃ mãa-nai-hak-ara-u=iti Wampis=FOC Awajun=COM "ght-RECP-HAB.PT-PL-NMLZ=COP.3+DECL ‘The Wampis and Awajun used to "ght.’ (68) Ekuaðorhɛɛʃ̃a aʃí arántusar íiniamu awɛi Ecuador=hai=̃ʃa aʃí arantu-sa-ri ii-nai-a-mau a-ua-i Ecuador=COM=ADD all respect-SUB-1PL.SS look-RECP-IPFV-NMLZ exist-3.SBJ-DECL ‘With Ecuador too [we] all see each other with respect’ (lit. ‘With Ecuador there is the (reciprocal) look respecting’) The addressee in a communicative event predicated by the verb t͡ʃit͡ʃa200 ‘speak’ is marked with the comitative. Examples (69)–(70) illustrate this point. (69) polisíahɛiʃ̃a t͡ʃit͡ʃasmiaji polisia=hai=̃ʃa t͡ʃit͡ʃa-sa-mia-ji police=COM=ADD speak-ATT-REM.PT-1PL.SBJ+DECL ‘We had talked with the police.’ 200. When bearing the attenuative -sa, t͡ʃit͡sa acquires the sense of ‘talk, converse’. 524 (70) wíi miɲahɛɛ amihɛɛ ́t͡ʃit͡ʃástasan wi wina-ha-i ami=hai ̃ t͡ʃit͡ʃa-sa-tasa-nu 1SG come+IPFV-1SG.SBJ-DECL 2SG=COM speak-ATT-PURP-1SG.SS ‘I come to talk to you.’ Another context in which the comitative is used is for coordinating NPs. (71) míɲa papár uhátiɲu ajaji oujmátmoon pat͡ʃís tsɨrɨń iwʲáhɛɛ mina papa-ru uha-tu-inu a-ia-ji 1SG.GEN father-1SG tell-1SG.OBJ-AGT.NMLZ COP-REM.PT-3.PT ‘My father used to tell me’ auhumatu-mau=na pat͡ʃi-sa ̃ [Tsɨrɨ=na Iua]=hai ̃ inform-NMLZ=ACC mention-SUB\3.SS monkey.sp=ACC Iwa=COM ‘the story about Tsere and Iwa.’ In the previous example, it is interesting to note the position of the accusative =na, which usually occurs in the last position of the NP. The accusative occurs this time in the "rst element of the complex NP, but the semantic scope of =na is ‘Tsere and Iwa’, which constitute the object of ‘mention’. (72) wampískʲa awarunhɛinkʲa ʃíir ʃuár áhakuiti [uampisa=ka auaruni]=hai=̃ka ʃiira ʃuara a-hak-ar-u=iti Wampis=FOC Awajun=COM=FOC very enemy COP-HAB.PT-PLNMLZ=COP.3+DECL ‘The Wampis and Awajun were very enemies.’ 11.5.5. Comitative =tuk The morpheme =tuk may be a combination of a formative *tu plus kɨ ‘restrictive’. It always involves the idea of comitative, but unlike =hai,̃ it appears that =tuk does not have an instrumental use. 525 (73) aβraḿʃa nuwɨńtuk wɨḿiaji Abramka=ʃa nuɨ=̃tuk uɨ-mia-ji Abramka=ADD woman\1PL/2PL/3.POSS=COM go-DIST.PT-3.PT+DECL ‘Abraham also went with his wife.’ 11.5.6. Plural speech act participant =tí The Plural Speech Act participant enclitic =tí normally attaches to a noun phrase and marks it as a plural participant in the communicative event. The next examples illustrate the use of this morpheme. (74) ʃuarti ́ ʃuara=ti ́ person=SAP ‘we the people’ (75) wampisti ́ uampisa=ti ́ Wampis=SAP ‘we the Wampis’ (76) iíɲiati ii=ni=̃ia=ti ́ 1PL=LOC=ABL=SAP ‘one of us’ The next example shows that the speech act participant morpheme works at the level of the phrase. In (77) =ti ́attaches to the last element of the complex conjoined 526 NP. In the context of the text whence the example comes from, it has a scope over the entire phrase: it means “we the Awajun and the Wampis”. (77) awarún túra wampistí awaruni tura wampisa=tí Awajun and Wampisa=SAP ‘We the Awajun and the Wampis’ When =tí occurs after the ablative, it creates a plural ethnonym. An extended meaning of “among us” is also possible in this case: (78) perunmajatík Peru=nVma=ia=tí=kɨ Peru=LOC=ABL=SAP=RESTR ‘We the people of Peru’ or ‘Among us Peruvians’ The Plural Speech Act participant =tí can derive a second person plural participant in two ways in my data. First, it can attach to a noun followed by a second person plural marker: (79) ʃuartírmɨka ʃuara=tí-rumɨ=ka person=SAP-2PL=FOC ‘You (PL) the people’ The second way is by attaching directly to a copula that is marked with a plural person su#x: (80) ʃuár ɛɛ́ɲatiram [ʃuara a-ina]=tí-ramɨ person COP-PL.IPFV=SAP-2PL ‘You (PL) the people’ 527 This suggests that the construction [N cop-pl] used to pluralized nouns (see §11.3.4) is likely being reanalyzed as a noun phrase, rather than as a verbal clause, as =tí does not occur anywhere else in the data as a verbal enclitic or a deverbal derivational morpheme. In addition, the fact that =ti ́can receive person markers suggest a possible verbal origin for this enclitic (possibly an old copula?). The same reanalysis of the construction [N cop-pl] can also be observed when =tí also occurs marked on the copula to derive a 1st person plural: (81) [huti a-ina]=tí 1PL COP-PL.IPFV=SAP ‘We’ (lit. ‘we who are-us’) Sometimes =ti ́functions as a more inclusive pluralizer in combination with the additive =ʃa. In the next example, the speaker creates a semantically complex subject that includes elders and children. Thus, the sentence in (82) does not mean ‘we, we the elders, we the children…’ but ‘we—the elders and children altogether…’. (82) íiʃa uuntíʃa ut͡ʃitíʃa nakúnaku hákur ʃíir mat͡sámin ɛɛ́ɲahi ii=ʃa uunta=tí=ʃa ut͡ʃi=tí=ʃa 1PL=ADD elder=SAP=ADD child=SAP=ADD ‘We, the elder and children’ nakunaku hak-u-ri content become-NMLZ-1PL.SS ‘having become content,’ 528 ʃiira mat͡sama-inu a-ina-hi well inhabit.PL-NMLZ COP-PL.IPFV-1PL.SBJ+DECL ‘we live well.’ 11.5.7. First =á As the name indicates=á adds a meaning of ‘"rst’ or, by extension, ‘before’ to the element on which it occurs. (83) wampistijá uampisa=tí=á Wampis=SAP=FIRST ‘We the Wampis "rst’ First =á can provide a sentence with an idea of relative importance/priority: (84) ɨnt͡sanamá wɨḿi ɨnt͡sa=nama=á wɨ-mi river=LOC=FIRST go-HORT ‘Let’s go to the river "rst.’ The morpheme =á can also occur in pronouns. The semantic scope of =á in the examples is over the clause: (85) nijá wakóo ásã. . . ni=̃á ua-ka-u a-sã 3SG=FIRST climb.up-INTENS-NMLR COP-SUB.3SG.SS ‘(He) having climbed up "rst…’ 11.5.8. A&ective =a ́ The a$ective adds a meaning of sorrow for the entity (typically human) to which the noun or pronoun which carries -ia refers. 529 (86) wijá huní wɨkat́ɲɛitha ui=a ́ huni wɨka-tinu=ita-ha 1SG-AFF like.this walk-FUT.NMLZ=COP-1SG.SBJ+EXCL ‘Poor me, I’ll wander thus!’ (87) jat͡surua ́niŋ́ki liḿanam wɨáhɛi iat͡su-ru=a ́ ni=̃ki Lima=nama uɨ-́a-ha-i brother-1SG-AFF 3SG=RESTR Lima=LOC go-IPFV-1SG.SBJ-DECL ‘My poor brother, he is going to Lima alone.’ The origin of this morpheme may be purely sound symbolic, as it usually lengthens parts of the word in an iconic lamenting-way. It occurs frequently in nampɨta ‘drinking songs’. 530 CHAPTER XII INTRODUCTION TO THE VERB: CLASSES, VERB DERIVATION, STRUCTURE AND CONJUGATION PATTERNS 12.1. Introduction The goal of this chapter is to serve as an introduction to the study of the verb in Wampis. Because of the complexity of the Wampis verb, the present chapter and the next two chapters are dedicated to its study. This chapter presents the organizing principles of the verb structure, conjugation types, classes of lexical verbs (transitive, intransitive, labile, auxiliary verbs, among others), and also presents a discussion of verbalization. The structure of the chapter is as follows: in §12.2, I discuss what can de"ne the verb in Wampis; in §12.3–§12.5 I describe di$erent types of verbs having into account their transitivity properties as well as their lexical content or function (e.g. auxiliaries); in §12.6 I provide details about verbal derivation. This is followed by a discussion of the structure of the verb in §12.7, and then I describe verb conjugations in §12.8. 531 12.2. De!ning the verb in Wampis Morphosyntactically, the "nite verb in Wampis is de"ned by the presence of a root or an “inner stem” that functions as simple roots do (cf. §12.7 for a discussion of the verb structure, verbal root and what I call “inner stem”) plus a marking of aspect, tense, person and mood. The morphological properties of the verb are described in detail §12.7. First, however, I give a short overview of major lexical verb (root) types in §12.3 through §12.5. Semantically, prototypical verbs are the elements that exhibit temporal instability and serve the function of predication in the clause. Verbs prototypically codify changes in “either the state, condition or location of some noun-coded entity” (Givón 2002a: 52). Other semantic characteristics are also listed by Givón, such as temporal compactness, concreteness, complexity and spatial di$useness, and agentiveness and mental activity. Granted, the meanings of individual verbal lexemes can create a range of semantic variation around the above claimed prototype; thus, several semantic sub-classes of verbs can be distinguished in particular languages according to how well they "t in with the aforementioned prototype. In terms of semantic and propositional-act prototypes (Croft 2001), a prototypical verb is an 532 element whose unmarked function is predicative. In fact, Croft precisely de"nes the prototypical verb as the element that has a predicative function. 12.3. Valency and transitivity: verb classes It is well established that languages have di$erent morphosyntactic (including derivational) operations that allow them to adjust the relationship between the core semantic roles of a base verb root and the grammatical relations those core arguments hold in clauses. Usually, these operations are understood to be in the domains of valence and voice. The semantic valence of a verb is understood as the number of necessary participants in the scene predicated by the verb. In a broad sense, syntactic valence is the number of core arguments in a clause in which a verb is the main predicator.201 With regards to transitivity, there are four major classes of verbs in Wampis: intransitive, transitive, ditransitive, and labile. This classes are based on a the notion of transitivity restricted to a property of the root/stem (“lexical transitivity”) prior to the addition of any valence-changing morphology. Wampis can be characterized as a language in which transitivity is lexicalized at 201. Non-verbal clauses have syntactic valence, e.g. Mary is tall is a monovalent clause. The above notion of transitivity refers to lexical verbs. 533 the level of the root/stem for most verbs (Payne 2009: 13); i.e. there are not many lexically labile verbs. Table 12.1 provides examples of di$erent verbs according to their transitivity. Table 12.1. Common intransitive and transitive verbs Examples Transitivity kanu ‘sleep’, iukuma ‘swim’, ha ‘be sick’ Intransitive ma ̃‘kill’, uaina ‘see’, ɨsai ‘bite’, antu ‘hear’ Transitive su ‘give’, uha ‘tell, inform’, inii ‘ask’, nanki ‘throw’ Ditransitive amu ‘/ood’ or ‘plunge’, ɨsa ‘burn up/oneself’ or ‘bite’ Labile In terms of their ability to take valence morphemes, transitive verbs can take the re/exive and reciprocal morphemes. Wampis possesses detransitivizing morphemes, however they are not productive in the language and therefore their (restricted) use does not constitute a good practical criterion to distinguish transitive from intransitive verbs. Transitivity in Wampis is not sensitive to aspect or tense, but there are other mechanisms at the level of the clause that depend on the transitivity of the verb. For instance, a transitive verb can control a subordinated structure with the switch- 534 reference marker -tatamana (cf. §19.11.1), which indicates that the subject of the subordinate verb becomes the object of the main verb, as in (1). (1) wɨt́atman hɨɰat́tak wɨt́atman iṕati maãẃaru tiḿaji wɨ-tatamana hɨa-a-́tata-kũ wɨ-tatamana go-SBJ>OBJ arrive-HIAF-DEF.FUT-SIM\3SG.SS go-SBJ>OBJ iṕati ̃ ma-̃aú-ara-u timaji shoot+LOAF\3.SS kill-HIAF-PL-NMLZ NARR ‘He was going, when he was about to arrive, having shot him, they killed him.’ In (1), the subject of the subordinated verb (the person who was about to arrive) is coreferential with the object in the main clause (the one shot and killed). It can be seen that the lexical transitivity of the verb mã ‘kill’ allows for the use of the switch- reference marker -tatamana ‘subject to object’, even when this morpheme is used with an intransitive verb, such as ‘go’, in the subordinated clause. In contrast to transitive verbs, intransitive verbs in Wampis can be de"ned by the following principles: • With intransitive verbs, only one argument (S) may occur as a nominative noun phrase or pronoun in the same clause. Note that the nominative is zero-marked. (2) huwańka hakaḿaji huan=ka ha-ka-ma-ji Juan=FOC die-INTENS-DIST.PT-3.PT+DECL ‘Juan died.’ • Only one argument (S) is obligatorily marked in the verb. In (2), the 3 person is 535 marked with -ji. • Intransitive verb stems cannot take valency-decreasing morphemes, such as the re/exive -ma or the reciprocal -nai. (§13.2.1). As far as I can tell, there are no a-valent verbs in Wampis. Typical examples of avalent verbs in other languages may include meteorological verbs, but in Wampis they are always syntactically intransitive verbs that codify an S argument. All examples of meteorological verbs in the data have an overt NP subject, as in (3) and (4). (3) násɨ ampúa tiḿaji nasɨ ampu-a timaji wind blow-IPFV NARR ‘The wind was blowing . . .’ (4) júmi ʃíir jutúkti iumi ʃiira iutu-ka-ti rain very rain-INTENS-JUSS ‘Let it rain hard!’ (Lit.: ‘That the rain rains hard!’) 12.3.1. Labile verbs A few verbs roots/stems202 in Wampis are not speci"ed for transitivity, i.e. they have properties that can be associated with either transitive or intransitive verbs. For instance, the verb root amu has the (more general) semantics of ‘cover with water’, and 202. Stems that do not bear valence-changing su#xes. 536 it can be used with the sense of either ‘/ood’ or ‘plunge/ford’. In the "rst case, the verb behaves transitively; in the second, it behaves intransitively. For example, in (5) the subordinate clause ‘as though it was about to cover/reach us’ is transitive as can be seen by the presence of the 1 person plural object (P) as part of the verbal piece. In (7), by contrast, the clause is intransitive and has only one core argument (S). A locative adjunct can be present, but it is not required by the verb structure (6), hence amu in the sense of ‘plunge’ only requires one argument (marked on the verb and optionally as an nominative overt NP) and has the properties of an intransitive verb. (5) nuháŋka ʃíir sɨńt͡ʃi amútmastatak nuhankɨ=ka ʃiira sɨnt͡ʃi amu-tama-sa-tata-kũ river.rise=FOC very strong /ood-1PL.OBJ-ATT-DEF.FUT-SIM.3SG.SS ‘The swell of the river was very strong as though it was about to reach us [i.e. cover us].’ (6) óuʃa ʃuárʃa ɨnt͡sá amúawɛɛ au=ʃa ʃuara=ʃa ɨntsá amu-a-ua-i DIST=ADD person=ADD rive\LOC plunge-IPFV-3-DECL ‘That person is also plunging into the river.’ (7) óuʃa ʃuaŕʃa amúawɛɛ au=ʃa ʃuara=ʃa amu-a-ua-i DIST=ADD person=ADD plunge-IPFV-3-DECL ‘That person is also plunging.’ 537 12.4. Auxiliary verbs Auxiliary verbs in Wampis occur in mono-clausal constructions that consist of “a lexical verb element that contributes lexical content to the construction and an auxiliary verb element that contributes some grammatical or functional content to the construction” (Anderson 2006: 7). Except for the copulas, all other auxiliary verbs occur in progressive constructions and thus code aspectual and (to some extent) body posture. Table 12.2 present auxiliary verbs in Wampis. Table 12.2. Auxiliary verbs in Wampis Verb Original Meaning Auxiliary meaning a copula copula puhu live (grammaticalized as ‘be’) continuous/habitual mat͡satu inhabit, live together (plural) continuous wɨ go ‘gonna’/ ‘progressive’ waha stand stative/body posture tɨpɨ lie down stative/body posture ɨkɨma sit stative/body posture 12.5. Other lexical types of verbs 12.5.1. Lexically negative verbs Some verbs in Wampis are lexically inherently negative, i.e. they do not 538 necessitate any negative morpheme in order to change their polarity. Examples of this are tuhina ‘be unable, have di#culty’, nakita ‘not want’203 and kaninma204 ‘not "t (in a space)’. 12.5.2. Verbs that lexicalize a speci!c type of patient or location The lexical content of Wampis verb roots can have very rich/speci"c content and provide inherent information about the object or location that need not be mentioned—of course, many of these meanings are culturally-speci"c. Some verbs also include manner information in their lexical representation. To give a sense of what can be found in the Wampis lexicon (at the level of the root) here is a short list of verb roots that are di#cult to gloss with one or two words: (8) aɨpa ‘leave/put on /oor or bed’ apa ‘put a bite in mouth (also applies to putting bites of manioc during the process of preparing manioc beer)’ apuha ‘put on "nger’ at͡sɨnu ‘put on a garment for the head’ 203. The source of this verb is probably the adjective naki ‘lazy’. The portion -ta may be a variant of an old verbalizer -tu. 204. The underlying form of this verb likely is kaninVma, as the third vowel would be always dropped due to the vowel elision process explained in Chapter VI. Thus the root would be kaninV plus the re/exive -ma. Synchronically, I have not found an etymon for the (apparent) root kaninV. 539 ɨkɨna ‘put (something) on top of (something)’ kahi ‘die many "sh (specially after using barbasco)’ kauna ‘arrive several’ nami ‘chew food to give to animal’ napi ‘fold palm leaves (to make roof)’ nunku ‘put shirt, necklace, beaded wristband, shoes or watch on’ pampa ‘make noise’ panka ‘raise head’ patakɨ ‘put bracelet to someone else’ (from patakɨ ‘bracelet’) puku ‘plant manioc stems’ sɨpɨta ‘close a hole (in a wall)’ tɨa ‘pull out several (usually applied to plants)’ t͡sukapɨ ‘carry a light object in shoulder’ (from t͡sukapɨ ‘armpit’) 12.6. Verb derivation New verb stems can be derived from noun stems or roots, from demonstratives, from the intensi"er ima, or from other verb roots. 12.6.1. Zero derivation As the name indicates, zero derivation is regarded as the extension of an existing word to assume a new syntactic function without any perceivable change in form. Zero derivation is not productive, but there are examples of zero derivation in the language, some are presented in (9). (9) hɨã ‘house’ → hɨã ‘arrive’ iumi ‘water, rain (N)’ → iumi ‘rain’ nɨpt͡sɨpɨ ‘chest’ → nɨpt͡sɨpɨ ‘lie down with chest on ground’ nupa ‘weed’ → nupa ‘increase weed (in a farm plot)’ patakɨ ‘bracelet’ → patakɨ ‘put beaded bracelet to someone else’ 540 12.6.2. Denominal verbalizer -ma The su#x -ma creates a verb stem from a noun root. The created verb stem is typically transitive with the meaning of ‘make N’ or ‘create N’. The following examples illustrate derivations with -ma. (10) Nominal Root Derived Verb aaka ‘shack’ aaka-ma ‘build shack’ aha ‘farm’ aha-ma ‘make a farm’ amiku ‘friend’ () amiku-ma ‘make friend(s)’ anɨntai ̃‘heart’ anɨntai-̃ma ‘think, remember’ hɨã ‘house’ hɨã-ma ‘build house’ nant͡ʃiki ‘"ngernail’ nant͡ʃiki-ma ‘scratch’ nuku ‘cover roof with palm leaves’ nuku-ma ‘put leaves on (to protect oneself from rain)’ uasɨ ‘yarina palm leaves’ uasɨ-ma ‘put yarina threaded leaves on the roof’ ut͡ʃi-ri ̃‘his/her child’ (child-3) ut͡ʃiri-̃ma ‘give birth (to his/her child)’ uhaha ‘song of victory’205 uhaha-ma ‘sing uhaha’ uku ‘behind’ uku-ma ‘leave behind, depopulate’ The verbalizer -ma is used to render native forms of verbs that are borrowed from Spanish: 205. This is the song that were sung by women during the celebration of victory after a war party returned. The tsantsa ‘shrunken head’ were presented during this celebration. 541 (11) kanta-ma ‘sing’ (< Spanish ) kuita-ma ‘take care’ (< Spanish ) kusɨna-ma ‘cook’ (< Spanish ) 12.6.3. Other unproductive verbalizers There are several unproductive verbalizers in Wampis. Table 12.3 shows these unproductive verbalizers. They are explained in the discussion below. Table 12.3. Unproductive verbalizers in Wampis Morpheme Gloss -na Get noun -tu~-ta Meteorological -tu ~ -ta Action involve a sound The su#x -na creates a verb stem that is intransitive. This verbalizer attaches to a possessed noun stem, and acquires the meaning of ‘get noun’. (12) aiʃu-ri-̃na husband-1PL/2PL/3.POSS ‘get married (of a woman)’ nuɨ-̃na woman\1PL/2PL/3.POSS ‘get married (of a man)’ A homophonous su#x -na occurs in muku-na ‘smoke (V), suck’, from muku ‘smoke (N)’), but the stem is not intransitive and it is not possessed. 542 A su#x -tu~-ta is used to create verbal stems that indicate the action of a meteorological event: (13) ɨt͡sa ̃ ‘sun’ > ɨt͡sañtu ~ ɨt͡sañta ‘shine’ (usually of sun) pɨɨma ‘lightning’ > pɨɨtu ‘/ash lightning’ (no form pɨɨta was given for this form) nasɨ ̃ ‘wind’ > nasɨñtu ~ nasɨñta ‘blow (of wind)’ Though not meteorological verbs, the following derived verbs may be somehow related to the above examples: (14) maiai ‘air, breath’ > maia-ta ‘breathe’ mit͡ʃa ‘cold’ > mit͡ʃa-ta ‘refresh, calm fever’ Another homophonous su#x -tu~-ta seems related to the verb tu ‘say’, and derives a few verb stems indicating an action that involves a sound. (15) kuarVtu ‘snore’ uutu ‘cry’ haat͡ʃitu ~ haat͡ʃita ‘sneeze’ ũhutu ~ ũhuta ‘cough’ Though the -tu~ta that derives weather verb stems and the -tu~-ta that derives a stem that indicates predicates sound are similar phonologically, the semantic relationship seems unlikely. 12.6.4. Derivation with -tika and -ni: resumptive verbs The demonstratives and the intensi"er ima have a verbalized version and thus 543 form a small subset of verbs. The language makes use of the derivational su#xes -tika and -ni to derive the relevant verbs. The main function of these derived verbs is discursive: they are very often used in head-tail-like constructions (Thompson et al. 2007), that provide cohesion between one clause (typically but not always "nite) and the next. This construction is termed “bridging” in Overall’s 2007 grammar of Awajun. Table 12.4 lists the verbs derived with -ni and -tika. Table 12.4. Verbs derived with -tika and -ni Root Gloss Derived form with -ni Derived form with -tika Gloss nu Non-visible nuni nutika ‘do that’ hu Proximal huni hutika ‘do this’ aanu Medial aania aantika ‘do that’ aa Demonstrative adverb aini aitika~aatika ‘do like that’ ima Intensi"er imani imatika ‘do much’ a Speakers vary their pronunciation between [aanni] and [aani]. The di$erence between -ni/-tika derived verb forms is based on the following principles: • Forms derived with -ni always occur to indicate actions of a subject, typically agentive, that occurs in the next clause. • Forms derived with -tika typically occur when referring to a less agentive subject or 544 with prominent non-subjects (typically objects) of the next clause. The following examples help illustrate the use of these type of verbs. In (16), an agentive 1 plural subject prepares the manioc beer in line (a). Then the resumptive derived verb imani ‘do much’ occurs at the beginning of line (b) re-referencing the action of clause (a) and indicating continuity of the same 1 plural subject in clause (b) (which also continues as subject in line c). (16) a. […] nihʲamánt͡ʃ umíkʲar nihamãt ͡ʃi umi-ka-ri manioc.beer prepare-INTENS-1PL.SS ‘… having prepared manioc beed,’ b. ímanisar aʃí umíkʲar [imani-sa-ri] aʃi umi-ka-ri do.much-SUB-1PL.SS all prepare-INTENS-1PL.SS ‘doing that much, having prepared all’ c. íi juátɲuriʃa aʃí hukír wɨḿiahi ii iú-á-tinu-ri=ʃa aʃi hu-ki-ri 1PL eat-HIAF-FUT.NMLZ-1PL.SS=ADD all TAKE-WHILE.MOVING-1PL.SS wɨ-́mia-hi go-DIST.PT-1PL.SBJ+DECL ‘what we were going to eat also, having taken all, we went.’ 545 In the next example, the subject of the clause that immediately follows the resumptive verb nutika ‘do that’ is an (non-agentive) experiencer subject of the stative verb ‘be angry’: (17) a. […] míɲa núwar ʃuarhɛɛ ̃wɨḿatɛi ̃ mina nua-ru ʃuara-hai ̃ wɨ-́matai ̃ 1SG.GEN woman-1SG person-COM go-1SG/3.DS ‘. . . when my wife went away with another man,’ b. nútikʲan kahɨḱan [nutika-a-́nu] kahɨ-ka-nu do.that-HIAF-1SG.SS be.angry-INTENS-1SG.SS ‘having done that, having become angry,’ c. natɨmá umármiahɛi natɨmá uma-ra-mia-ha-i ayahuasca\ACC drink.PF-DISTR-DIST.PT-1SG.SBJ-DECL ‘I drank ayahuasca (i.e. to look for answers for the current bad situation).’ In the next example, two overt topical object NPs follow a verb derive with -tika in line (c): (23) a. taaḱum ikihmarta ́ ta-a-ku-mɨ ikihV-ma-ru-ta ́ come-IPFV-SIM-2SG.SS wash.hand-re/-APPL-imp ‘Coming (back from urinating), wash your hands!’ b. uwɨh́ɨm t͡suat́ɛiti tú hakúiti uɨhɨ-mɨ t͡suata=iti tu hak-u=iti hand.POSS-2SG dirty=COP.3+DECL say HAB.PT-NMLZ=COP.3+DECL ‘He used to say: “your hand is dirty”’ 546 c. aat́ikas ɛɛʃ́mankun t͡ʃit͡ʃaŕak nuwańt͡ʃa t͡ʃit͡ʃaŕak [aatika-sa]̃ aiʃumanku=na t͡ʃit͡ʃa-ru-ka ̃ do.like.that-SUB\3SG.SS man=ACC speak-APPL-INTENS\3SG.SS nua=na=ʃa t͡ʃit͡ʃa-ru-ka ̃ woman=ACC=ADD speak-APPL-INTENS\3SG.SS ‘doing like that, he advised the men, he also advised the women.’ 12.7. Basic structure of the Wampis verb In Wampis, an independent verb cannot consist of just a verb root. The root (or simple stem) must be obligatorily marked for aspect, tense, person and mood. The marking of Tense-Aspect-Mood (TAM) involves three kinds of devices: “derivational” perfective su#xes, more “in/ectional” tense su#xes, and auxiliary+nominalization constructions. The perfective aspect is obtained in the use of some derivational morphemes that, following (Overall 2007), are termed “aktionsart”. Past and future tenses are marked with su#xes, but actually the TAM system is greatly expanded with a series of copular constructions involving nominalizations that have been grammaticalized to express several TAM distinctions. Present tense is zero-marked. Beyond the obligatory morphemes, there are several morphological positions in the Wampis verb. A position class schema of the Wampis verb structure is presented in Table 12.5. 547 Table 12.5. Overall structure of the Wampis verb +Derivational ......................................................................................+In/ectional -1 0 1 2 3 4 5 6 7 8 Valence ROOT “Inner stem” Valence ‘Almost’ Object Aktionsart Imperfv. Potential Durative Pres. Hab. Neg. Number Tense (Mood) Person Mood Note: The morphological access to the Object and Person slots varies according to an uncommon hierarchical agreement that is explained in Chapter XIV. It is not always easy to distinguish between “derivational” and “in/ectional” morphology. Traditionally, in/ection is understood as a process that creates new forms of a same lexical item, a form that is required by the syntactic environment where the root appears. On the other hand, derivation is understood as the process of creating new in/ectable items, usually to indicate new concepts (Hockett 1958; Gleason 1961). However, in many cases it is better to speak of morphological formatives that are capable of both derivational and in/ectional functions (Bybee 1985; Payne 1985). Categories that are traditionally considered as in/ectional, such as “aspect”, “number” or “negation”, for instance, possess also derivational properties in Wampis. At least “aspect” is required in Wampis in all cases where a fully in/ected verb needs to be used, 548 and number and negation are of course required by certain syntactic constructions or the verb cannot be integrated into the discourse. But most of these su#xes have derivational properties. For instance, the root hapi ‘drag, pull’ usually forms a perfective stem with the aktionsart su#x -ka ‘intensive (action of A/S)’206 and a durative stem with the aspectual su#x -ma ‘durative’. This su#xes occupy the same position (cf. Table 12.5) and cannot co-occur together: (18) hapi-ka-ha-i hapi-ka-ta ́ hapi-ma-ta ́ drag-INTENS-1SG.SBJ-DECL drag-INTENS-IMP drag-DUR-IMP ‘I just dragged it.’ ‘Drag it!’ ‘Drag it for a little while!’ cf. *hapi-ka-ma-ta,́ *hapi-ma-ka-ta ́ However, the durative su#x -ma can derive a new stem from the root hapi ‘drag, pull’ with a new meaning: (19) hapi ‘drag’ > hapi-ma ‘sweep’ This new stem is treated just as a root, and can receive verbal morphology as any normal root would, including aktionsart su#xes (even though the durative is 206. The relation between perfective and aktionsart su#xes is examined in the next Chapter; cf. §13.3. 549 contained in the stem): (20) hapima-ka-ha-i sweep-INTENS-1SG.SBJ-DECL ‘I just swept.’ I call this kind of derived stem the “inner stem” (cf. Table 12.5). What I understand by “inner stem” is a verbal stem derived with any one of the su#xes that usually occupy positions 3-5 (which are su#xes that have derivational properties, beside the valence su#xes) and that form a stem that behaves like a “normal” root; that is, a stem that can receive again any piece of morphology from positions 1 to 8 depending on the morphosyntactic environment. I do not consider stems derived with valence su#xes as “inner stems” because they cannot carry another di$erent valence su#x when they are derived. From the above discussion, it can be said that a careful examination of some Wampis verbal morpheme properties shows that there is really no sharp boundaries between “in/ectional” and “derivational” morphology in the language. Thus, rather than making a strict divide between derivation and in/ection, I characterize the verbal morphology by placing the morphemes along a continuum between derivation and 550 in/ection. I describe the morphemes from -1 to 5 in Table 12.5 as “more derivational” and the morphemes from 6 to 8 as “more in/ectional”. The “more derivational” morphemes constitute the derivational level of the verb—they will be studied in Chapter XIII. I consider them “more derivational” in the sense that they most frequently create a stem that is in/ectable, i.e. a stem that can carry markings of person, tense, and mood. Aspectual morphemes are also obligatory to use a Wampis verb in discourse, but because of their position in the verbal piece and their derivational properties, they are considered in the derivational level. Thus, the root or the “inner stem” plus any one (or more, depending on the desired actual meaning) of the su#xes from -1 to 5 form what I call the level of the “stem”. On the other hand, the “more in/ectional” morphemes, those related to tense, subject and mood, are considered to form the in/ectional level— they are studied in Chapter XIV. Wampis is a language with a very interesting feature in the verb structure: most su#xes that are used for aspect are, at their very basic, meaning-changing su#xes. Despite being derivational, they are also obligatory, for aspect is an obligatory category of the independent verb. The imperfective is in a contrastive paradigmatic relation with the durative and the “aktionsart” su#xes, and both aktionsart and the durative su#xes 551 have clear meaning-changing properties. The relevance of such a language in relation to the larger theoretical body of literature discussing in/ection and derivation needs to be explored more in detailed and is one important topic for future research. Thus, for Wampis, one more or less clear divide between derivation and in/ection can be stablished after slot 5 in Table 12.5 above, whereas Aspect represents a cline between in/ection and derivation. Tense, subject/person and mood are in/ectional in that independent verb needs an indication of these categories in order to be used in the communicative event. The closest positions to the verb root are occupied by valency-changing morphemes, including the only pre"x in the language. There is also a su#x -nin ‘almost’ that is rather aspectual but occurs in position 1, close to the root. Positions 2 and 7 are generally related to the marking of object and subject, respectively. However, the morphological access to these positions varies as it is dependent upon a hierarchical agreement system. This is explained in Chapter XIV. Aktionsart/Aspectual morphemes and the potential morphemes occupy position 3. These morphemes create verb stems that can be in/ected. The negative su#x -t͡ʃa and the plural perfective su#x -ara occupy the next two slots. The plural -ara, though marking non-imperfective stems, does not 552 belong to the same paradigm of other aktionsart/aspectual morphemes. To summarize, all of the morphemes from positions 0-5 belong to a "rst level of morphology that is roughly related to “derivation” as most of them possess some degree of derivational properties. The second level of verbal morphology roughly correspond to “in/ection”. Morphemes in this level involve tense, person (mostly subject) and mood. 12.8. Verb conjugation patterns There are several conjugation patterns in Wampis that have to do with morphological changes in the verb stem in imperfective or perfective contexts. The simplest relevant verb stem may coincide with root, the “inner root”, or be a stem created with a root or “inner root” with any derivational su#x prior to position 3. The changes to the verb stem are not morphophonologically predictable. The imperfective and perfective stems are created with the imperfective and aktionsart su#xes, respectively. These su#xes occupy the same morphological slot that the durative, present habitual and potential (position 3 in Table 12.5). However, the durative, present habitual and the potential su#xes do not change the form of the verb stem. The su#xes that occupy position 3 in the verbal piece are listed in Table 12.6. The term “aktionsart” is explained in Chapter XIII. 553 Table 12.6. Su#xes that occupy position 3 in the Wampis verb structure Morpheme Gloss -a(́u) ‘High a$ectedness (change of state of Patient or of location of Agent) A k t i o n s a r t -i ‘Low a$ectedness (no salient change of state of Patient or Agent)’ -ka ‘Intensive (action of Agent)’ -ki(ni) ‘Do while moving (away)’ -ra ‘Distributed action’ -sa ‘Attenuative’ -ri ‘Do in proximity/while coming’ -u ‘Do in another location’ -a / -ina ‘Imperfective’/‘Plural imperfective’ -ma ‘Durative’ -na ‘Present habitual’ -mai ‘Potential’ 12.8.1. Imperfective conjugations Su#xation of the imperfective morpheme -a to the verb creates the most morphophonological changes in the verb stem. There are three general conjugation patterns when the imperfective is su#xed, all concerning changes to the last vowel of the verb stem. 12.8.1.1. Imperfective Conjugation I Imperfective Conjugation I is the easiest to understand. In Conjugation I, no changes are made to the verb stem when the imperfective is added. Most underlyingly 554 monomoraic verbs stems belong to this conjugation, as they do not lose their vowel. One exception, however, is tu ‘say’ which becomes ta in the imperfective. Stems that have more than one vowel and that have a "nal vowel /a/ either drop that last vowel with the imperfective -a (Conjugation II), or they drop the last vowel and insert a vowel ɨ (Conjugation III). As far as I can tell, there is no stem with more than two moras and a "nal vowel /a/ in Imperfective Conjugation I. The examples in (21) show verbs that belong to Imperfective Conjugation I. (21) Stem )nal a Stem Imperfective -a ha ‘die’ haa Stem )nal i or ii Stem Imperfective -a hapi ‘drag, pull’ hapia ii ‘see’ iia at͡ʃi ‘grab’ at͡ʃia Stem )nal ɨ or ɨɨ Stem Imperfective -a hatɨ ‘cut with axe’ hatɨa hurɨ ‘lay eggs’ hurɨa kɨɨ ‘burn’ kɨɨa 555 Stem )nal u or uu Stem Imperfective -a ihu ‘pierce, stab’ ihua iku ‘move away’ ikua pasu ‘be/put inside in untidy way’ pasua uu ‘hide’ uua207 12.8.1.2. Imperfective Conjugation II Imperfective Conjugation II is de"ned by replacement of the last vowel /a, u, ɨ/ or "nal /i/ (in a "nal /ii/ sequence) of the stem with the Imperfective -a. Within this conjugation pattern, the stem drops its last vowel when it receives the imperfective su#x -a. All examples that belong to conjugation II have more than two moras underlyingly, with the exception of tu ‘say’ as stated in §12.8.1.1. I have no examples of a last single vowel /i/ of the stem being dropped preceding the imperfective -a—i.e. all verb stems ending in a single vowel /i/ belong to the "rst conjugation, but there are examples of verbs ending in /ii/ which belong to Conjugation II. The following 207. The phonetic realization of this item is [uwa]. Under the current analysis, the second u of uu turns into a glide because it is between /u/ and /a/. This analysis is preferred for its simplicity, however it is possible to analyze [uwa] as uu dropping its last vowel u with the imperfective -a (following the pattern of Imperfective Conjugation II described below) and inserting a glide afterward. In general, there are not many verb roots with "nal uu and the analysis in such cases is always di#cult given the ambiguity generated by the phonological context. 556 examples illustrate the changes in the Imperfective Conjugation II. (22) Stem )nal a and aa Stem Imperfective akiina ‘give birth’ akiina haa ‘tear’ haa ipaa ‘invite’ ipaa taka ‘work’ taka Stem )nal ii Stem Imperfective hintii ‘"nd out’ hintia hii ‘take out’ hia inii ‘ask’ inia Stem )nal ɨ and ɨɨ Stem Imperfective tɨpɨ ‘lie down’ tɨpa anɨɨ ‘feel, be happy’ anɨa Stem )nal u and uu Stem Imperfective tu ‘say’ ta antu ‘listen’ anta huu ‘gather, harvest’ hua 557 12.8.1.3. Imperfective Conjugation III Stems belonging to Imperfective Conjugation III all end underlyingly in /a/. In this conjugation, the last vowel /a/ of the verb stem is dropped and an ɨ is inserted before the imperfective -a. If the verb stem ends in /i/ or /ɨ/, the stem is una$ected (i.e., no inserted vowel ɨ appears before /a/). Therefore, verbs ending with these vowels can not be said to belong to Conjugation III. An important note to take into account is that all stems belonging to Conjugation III have more than two moras. However, not all verbs with more than two vowels belong to this conjugation paradigm. For instance akiina ‘give birth’, has more than two moras, but it belongs to Conjugation II: its last vowel /a/ is dropped when preceding the imperfective -a: (23) akiina ‘give birth’ akiina ‘give birth (imperfective)’ The next examples show changes in the verb stems that belong to Imperfective Conjugation III. (24) Stem a Stem Imperfective uaura ‘despair, act in/become frenzy’ uaurɨa nankana ‘end’ nankanɨa ihaka ‘break, open hard-shelled fruits’ ihakɨa 558 If the verb stem has a high front vowel /i/ as the second-to-last vowel, the inserted vowel ɨ undergoes harmony and becomes i: (25) Stem Imperfective apiha ‘fold’ apihia 12.8.1.4. Derived verb conjugations Derived verb stems carrying valence, aktionsart, aspectual and/or object su#xes also belong to one of the Imperfective Conjugations I, II or III. The su#xes that have been identi"ed to trigger insertion of the vowel ɨ (i.e. which create stems belonging to Imperfective Conjugation III) are shown in Table 12.7. But not all verb stems derived with the su#xes in Table 12.7 belong to the imperfective conjugation III. In fact, the conjugation pattern they follow is entirely lexical. In other words, there are verbs stems derived with any of these su#xes that "t into the imperfective conjugation III and there are verb stems that are derived with the same su#xes that simply do not show the same pattern and belong in the Conjugation I or II patterns. 559 Table 12.7. Su#xes that insert a vowel ɨ in the imperfective stem Morpheme Gloss -tu ~ ru Applicative -tu ~ ru First singular objecta -ra Distributed Action -turama~tama ~ rama 1PL/2 Object -ma Re/exive -ma Verbalizer a The applicative and the "rst singular object share the same forms, but they are clearly distinct morphemes synchronically. 12.8.2. Perfective conjugations Three conjugations have been established for the perfective stem. For sake of comparison, examples of each conjugation in this section present the imperfective (just discussed in the preceding sections) and perfective forms of the verb. It is important to note that most verbs take an aktionsart su#x as a su#x with which they occur when they are used in perfective contexts (most verbs frequently appear with one aktionsart su#x from the paradigm given in Table 12.6). Thus, in addition, the perfective form is presented with the default aktionsart su#x taken by the verb. An analysis of the aktionsart su#xes and the perfective stem is given in Chapter XIII. 12.8.2.1. Perfective Conjugation I In the Perfective Conjugation I, the last vowel of the verb stem is not a$ected 560 when the aktionsart su#x is added. The examples in (26) illustrate the Perfective Conjugation I. (26) Stem )nal a (and aa) Stem Imperfective -a Perfective+Aktionsart naka ‘wait’ naka naka-ka (‘intensive’) aa ‘write’ aa aa-ra (‘distributed action’) Stem "nal i (and ii) Stem Imperfective Perfective+Aktionsart aki ‘pay’ akia aki-ka (‘intensive’) hii ‘take out’ hia hii-ra (‘Distributed action’) Stem "nal ɨ (and ɨɨ) Stem Imperfective Perfective+Aktionsart t͡sɨkɨ ‘run’ t͡sɨkɨa t͡sɨkɨ-i (low a$ectednes) hatɨ ‘cut with axe’ hatɨa hatɨ-a ́(‘high a$ectedness’) kɨɨ ‘burn’ kɨɨa kɨɨ-ka (‘intensive’) Stem "nal u (and uu) Stem Imperfective Perfective+Aktionsart puhu ‘live’ puha puhu-sa (‘attenuative’) umpu ‘blow’ umpua umpu-i (‘low a$ectedness’) kuarVtu kuarta kuartu-ka (‘intensive’) 561 12.8.2.2. Perfective Conjugation II In Perfective Conjugation II, a "nal high vowel /i/, /ɨ/ or /u/ of the stem is dropped and a vowel a is inserted before the aktionsart su#x. A Perfective Conjugation II stem (not including the aktionart su#x) thus in most cases becomes identical in form with its imperfective counterpart (but see ‘lock with bar’ below for a counter-example). However, in discourse these homophonous forms are not ambiguous, because the perfective stem adds an aktionsart su#x, which cannot co-occur with the imperfective. (27) Stem Imperfective Perfective+aktionsart inii ‘ask’ inia inia-sa (‘attenuative’) asakɨ ‘lock with bar’ asakɨa asaka-a ́(‘high a$ectedness’) kanu ‘sleep’ kana kana-ra (‘distributed action’) umu ‘drink’ uma uma-ra (‘distributed action’) Exempli"cation of the Perfective Conjugation II is given in 28 with the verb kanu ‘sleep’, which has the perfective stem form kana. The aktionsart su#x -ra occurs in this case (it is the most frequent aktionsart su#x selected by kanu in perfective contexts). Contrast (28) with (29), in the latter we can see the imperfective form. In (29), the las vowel /u/ of the verb kanu is dropped when preceding the imperfective -a, yielding kana. However, the imperfective form cannot carry the aktionsart su#x (-ra ‘Distributed action’ in (28)). 562 (28) kanármiaji kana-ra-mia-ji sleep.PFV-DISTR-DIST.PT-3.PT+DECL ‘He slept.’ (29) kanáwɛi kana-ua-i sleep+IPFV-3.SBJ-DECL ‘He sleeps/is sleeping.’ 12.8.2.3. Perfective Conjugation III In Perfective Conjugation III, nothing happens to the last vowel of the stem, but a consonant n is added following this last vowel. The nature of this consonant is unclear, as there is no trace of underlying nasalization to any of the composing morphemes (the n only appears in the perfective form). It is also unclear whether this nasal consonant belongs to a su#x; however, synchronically it does not seem to add any meaning to the verb. When a verb stem ends in a single-vowel quality multi-moraic sequence (e.g. katii ‘cross to other side of a river’), the last vocalic mora is replaced by n. Some verbs that belong in this conjugation are exempli"ed below. (30) Stem Imperfective Perfective anti ‘touch’ antia antin inku ‘"nd’ inkua inkun 563 katii ‘cross to other side of a river’ katia katin kauɨ ‘get together (many)’ kauɨa kauɨn mui ‘vomit, have nausea’ muia muin t͡sɨkɨ ‘run, jump’ t͡sɨkɨa t͡sɨkɨn Perfective Conjugation III is the least common conjugation for the perfective. Typically, verbs belonging to Conjugation III do not have a default aktionsart su#x with which they occur in the perfective form. Thus a possible analysis is that the nasal consonant n may be standing for the aktionsart su#x. It may be possible that the Conjugation III -n is related to the “placeholder” morpheme -na (§13.2.1.2). The “placeholder” su#x -na usually takes the morphological slot of a valence-related morpheme when they are not used (and it is replaced by them when they are used).208 However, that hypothesis is problematic given the following example in which the verb t͡sɨkɨ ‘run, jump’ shows that the nasal consonant occurs when the verb takes an aktionsart su#x. Therefore, the nasal consonant n is considered synchronically an integral part of the perfective form of the verb, rather a di$erent/separate su#x. 208. The placeholder su#x -na usually takes the morphological slot of a valence-related morpheme when the latter is not used (and it is replaced by them when they are used). 564 (31) t͡sɨḱɨŋki ̃iɲahkamá tuntari ̃ń kahínmatkiñ […] tsɨḱɨn-ki ̃ iia-hkama ̃ ́ run.PFV-WHILE.MOVING\3SG.SS fall-TERM tunta-ri=̃na kahinama-tu-ki ̃ quiver-1PL/2PL/3.POSS=ACC forget-APPL-WHILE.MOVING\3SG.SS ‘Having run, upon falling, having forgotten his quiver […]’ 565 CHAPTER XIII THE VERB I: "DERIVATIONAL" LEVEL 13.1. Introduction This chapter describes the "rst level of morphology in the Wampis verb structure, which roughly corresponds to “derivational” morphology, positions 1-5 in the verb structure (cf. §12.7). The categories to be studied in this chapter are the following: 1. Valence-adjusting morphemes, which are all su#xes, except for a causative that is the only pre"x in the language (§13.2). 2. Object/Person markers (§13.2.4). 3. Aktionsart, Imperfective, Durative, Present habitual and Potential a#xes (§13.3). 4. Negation (§13.4). 5. Non-imperfective plural (§13.5). 13.2. Valence operators The "rst layer of morphemes in the verbal piece, closest to the root, are valence operators. Languages almost always exhibit several morphemes that serve to adjust the 566 syntactic valence of clauses, increasing or decreasing it (Bybee 1985). In Wampis, a number of morphemes serve the purpose of increasing or decreasing valence. Valence- decreasing and valence-increasing operators are su#xal, with the exception of a pre"xal causative. Notice that there is no true passive in Wampis. 13.2.1. Valence-decreasing morphology Wampis possesses a subset of morphemes that decrease the valence of a verb, as shown in Table 13.1. Table 13.1. Valence-decreasing su#xes in Wampis Morpheme Gloss -na, -pa Detransitivizers -ma Re/exive -nai Reciprocal The detransitivizers are not very productive synchronically, especially -pa. The re/exive and the reciprocal commonly occur in natural speech. The fact that there are transitivizing and detransitivizing morphemes in Wampis apparently point to the importance of the transitive/intransitive distinction in past stages of the language, perhaps Jivaroan protolanguage more generally. Notice that some verbs have 567 detransitivizing su#xes as part of their lexicalized stem. 13.2.1.1. Detransitivizer -na The su#x -na turns a transitive verb into an intransitive one. Table 13.2 lists a sample of Wampis transitive verbs and their derived detransitivized counterparts. Table 13.2. Wampis derivations with detransitivizer -na Transitive root Detransitive -na eme ‘ruin’ eme-na ‘decrease’ hii ‘take out’ hii-na ‘exit’ kupi ‘break’ kupi-na ‘break one’s bones’ nankima ‘throw’a nankima-na ‘jump’ uka ‘spill’ uka-na ‘be spilt’ ut͡saa ‘throw away several objects’ ut͡saa-na ‘enter house (several people), fall in trap (several people/animals)’ a From nanki ‘spear’ and -ma ‘verbalizer’. The next examples illustrate the use of -na in natural speech. In (1) the transitive verb hii ‘take out’ is present. We can see that the verb, being transitive, receives a marking of 2 object. (1) nampɨḱ asámtɛɛ ̃hiírmakmiaji nampɨku a-sa-matai ̃ hii-rama-ka-ma-ji drunk COP-SUB-1SG/3.DS take.out-2.OBJ-INTENS-REC.PT-3.PT+DECL ‘Because you were drunk [in the soccer "eld], he took you out.’ 568 The following sentence shows the derived verb hii-na (take out+-DTR) i.e. ‘exit’. The verb stem takes only one S argument (‘you’): (2) Maʃiantá aḿɨ hiinkitʲá Maʃiantá amɨ hii-na-ki-tá Mashianta\VOC 2SG take.out-DTR-WHILE.MOVING-IMP ‘Mashianta, you go out!’ 13.2.1.2. “Placeholder” -na For Awajun, Overall (2007: 30) reports the existence of a morpheme -na that is homophonous with the detransitivizer -na seen in the preceding subsection. In contrast to the detransitivizer -na, this “placeholder” su#x -na does not change the valency of the verb and its function, synchronically, seems to stand there in the stem in place of a valency or object morpheme. This homophonous su#x also exists in Wampis. It is lexically restricted (appearing with only a few verbs) and seems to have the same behavior. If a valency or object morpheme is used, it replaces the non-valency changing -na. “Placeholder” -na occurs only with a small subset of transitive verbs, as listed in Table 13.3. 569 Table 13.3. Verbs where “placeholder” -na occurs Verb Gloss at͡ʃina ‘capture’ (from at͡ʃi ‘grab’) mukuna ‘suck, smoke’ (from muku ‘nipple’) ɨkɨna ‘put on (e.g. /oor, "re (to grill), bed, etc.) ’ ɨmɨtana ‘a"rm’ ɨpɨna209 ‘fence in’ uaina ‘see’ To illustrate the “placeholder” su#x -na, let us look at examples (3)–(5). In (3), the placeholder -na stands after the root ‘see’. Notice that this is not a detransitivizer, as the verb continues to be transitive. (3) wɛínkammɨ uai-na-ka-ma-mɨ see-NA-INTENS-REC.PT-2SG.SBJ+DECL ‘You saw him.’ In (4), the placeholder morpheme -na is replaced by the reciprocal -nai. (4) kaʃínkiʃa wɛínikami kaʃini=kɨ=ʃa uai-nai-ka-mi tomorrow=RESTR=ADD see-RECP-INTENS-HORT ‘Tomorrow we’ll see each other.’ 209. This example is particularly interesting. The root ɨkɨ means ‘put down’; with -na it means ‘put on "re, /oor (or some other surface)’. It appears like -na ‘placeholder’ may have had some derivational properties after all; unfortunately, there are not many examples like this in the lexicon. Notice that the same root ɨkɨ plus the re/exive -ma, i.e. ɨkɨma, means ‘sit’. In that case, - na ‘placeholder’ is “replaced” by the re/exive -ma: ɨkɨma ‘sit’. 570 And, in (5), it can be seen that the 1SG object marker -tu occurs in the position of the placeholder -na. (5) núwa wait́kʲarɛɛ túsa nua uai-tu-ka-ara-ai woman see-1SG.OBJ-INTENS-PL.PFV-APPR ‘May the women don’t see me.’ 13.2.1.3. Detransitivizer -pa The detransitivizer -pa is an unproductive su#x that can be found lexicalized in just a few verb roots. For instance, consider ʃiki ‘urine’, ʃiki-ki ‘take out liquids’ (with -ki ‘transitivizer’), ʃikitpa < *ʃiki-tu-pa ‘urinate’ (with -tu ‘applicative’ and -pa ‘detransitivizer’). There is also a more common version of ‘urinate’ with the re/exive - ma (§13.2.1.4): ʃikitma < *ʃiki-tu-ma. It is possible that the -pa is just an old variant of the re/exive -ma explained in the next section. The same [root-applicative-re/exive] structure is used in other lexicalized verb stems: iha-tu-ma ‘defecate’ (from iha ‘excrement’; at least two speakers told me that iha-tu-pa was also ‘correct’), t͡sapa-tu-ma ‘sprout’ (from t͡sapa ‘grow (of a plant)’), probably related to t͡sapa ‘gourd’). 13.2.1.4. Re2exive -ma(ma) The re/exive -ma(ma) reduces the valence of the clause by indicating that the agent and the patient of the verb are the same entity involved in the predicated event. 571 Semantically, prototypical re/exives mark an event in which the participant is the initiator and the end point of an action (Kemmer 1994). The typical used of the re/exive can be illustrated with the verb tɨmaʃi ‘comb’210. In (6), there is a re/exive action marked by the presence of -ma carried by the verb root. Thus, the agent and the patient of the verb are the same. By contrast, in (7), the action is between two di$erent participants, ‘you’ (agent) and ‘me’ (patient). Morphosyntactically, this is marked by the occurrence of a 2SG subject marker and a 1SG object on the verb. (6) intʲaʃiń tɨmaʃ́matɨamɨ intaʃi=na tɨmaʃi-ma-tu-a-mɨ hair=ACC comb-REFL-APPL-IPFV-2SG.SBJ+DECL ‘You comb your hair yourself.’ (7) tɨmaʃ́tamɨ tɨmaʃi-tu-a-mɨ comb-1SG.OBJ-IPFV-2SG.SBJ+DECL ‘You comb me.’ Further examples of derivations with the re/exive -ma are given in (8). Some examples reveal that the e$ect of the re/exive can be somewhat unpredictable 210. Notice the likely relation of this verb to the noun tɨma ‘louse’. 572 semantically. For example, though ‘sit’ can perhaps be interpreted as a hyponym of the re/exive concept ‘put oneself down’, ɨkɨ-ma does not have the broad meaning of ‘put one’s self down’, and it is lexicalized speci"cally as meaning ‘sit’; it cannot be extended, for instance, to ‘crouch’ or ‘lie down’ which are also hyponyms of ‘put oneself down’. In addition, t͡ʃit͡ʃa ‘speak’ with the re/exive plus the applicative -ru (§13.2.3) becomes ‘defend oneself’—this is likely related to the cultural importance of oratory speech in Wampis, but the meaning of the derived verb is not clearly obtained from the sum of its parts. (8) ati ‘untie’ → ati-ma ‘untie oneself’ ɨsa ‘burn’ → ɨsa-ma ‘burn oneself’ ɨnt͡sa ‘carry on the back’ → ɨntsa-ma ‘ride’ (‘put oneself on the back’) ɨkɨ ‘put down’ → ɨkɨ-ma ‘sit’ niha ‘wash’ → niha-ma ‘wash oneself’ t͡ʃit͡ʃamaru ‘defend’ → t͡ʃit͡ʃamaru-ma ‘defend onself’ Sometimes, the re/exive appears with the long form -mama. This is illustrated in (9) and (10). Note that the surface form with the ɨ [jumaḿɨahai] in (9) occurs because the stem (including the re/exive su#x) is conjugated following the imperfective conjugation III. Recall that in this conjugation type, a "nal vowel -a of the stem is replaced by a high central vowel ɨ when the stems receives the imperfective su#x -a (§12.8.1.4). 573 (9) jumaḿɨahɛɛ iú-mama-a-ha-i eat-REFL-IPFV-1SG.SBJ-DECL ‘I eat myself.’ (10) wiʃ́a wɨḿiahɛɛ konseho awaruna wamβisa tamaú aat́mammoo aḿaunam nui ̃ ́ ui=ʃa uɨ-́mia-ha-i konseho_awaruna_wamβisa ta-mau 1SG=ADD go-DIST.PT-1SG.SBJ-DECL Awajun-Wampis Council say+IPFV-NMLZ aa-tu-mama-mau a-mau=nama nui ̃ write-APPL-REFL-NMLZ COP-NMLZ=LOC there ‘I also went to the so-called Consejo Aguaruna-Huambisa, where they got registered themselves, there.’ Example (10) is interesting because the verb with the re/exive su#x is followed by the Set II non-subject nominalizer -mau. Thus the construal of the subject seems to be non-agentive (‘the ones who got registered themselves’).211 This appears to be a middle voice-like use of the re/exive in that the event is treated as a process in the sense of Chafe (1970); i.e. as a change of condition or state, rather than an action performed by an agent upon another participant or upon itself. Notice that when the verb is more stative/inchoative, as ‘be/get damaged’ in (11), the verb is nominalized with the subject nominalizer -u. 211. A subject nominalization with an active verb such as ‘write (register)’ would imply that the event is more prototypically transitive, with an agent acting upon a patient; i.e ‘the ones who register [someone]’. 574 (11) tunaámaru aśa ̃muút͡ʃrintin núka . . . naaŋ́ka tsunaŋ́ka tunaa-ma-ru-u a-sa ̃ get.damaged-REFL-APPL-NMLZ COP-SUB\3.Ss muút͡ʃi-ri-̃tinu nu=ka naa=na=ka t͡suna=na=ka conjunctivities-1PL/2PL/3-ATTRIB NON.VIS=FOC HESIT.PRO=ACC=FOC Tsuna=ACC=FOC ‘because he got damaged, with conjunctivitis, umm, that Tsuna [they saw].’ The connection between the functional domains of middle voice and re/exivity is well attested across languages (Shibatani 1985; Kemmer 1993; Givón 1994). Somewhat confusingly, the re/exive -ma is homophonous with the verbalizer su#x -ma (§12.6.2), the durative -ma (§13.3.4) and the recent past -ma (§14.2.2.2). Historically, it is uncertain at this point of study whether there is a connection between these functions, or not. However, the identity of all of these su#xes is clearly distinct synchronically. In the following example, we have the denominal verb ‘scratch’ derived with the verbalizer -ma, followed by the re/exive and the recent past morphemes, all occupying di$erent positions in the verb. (12) nant͡ʃiḱmamramhɛɛ nant͡ʃika-ma-ma-ra-ma-ha-i "ngernail-VBZ-REFL-DISTR-REC.PT-1SG.SBJ-DECL ‘I scratched myself.’ The elicited example in (13) shows an occurrence of the durative -ma with the verbalizer -ma. Notice that this cannot be an instance of the re/exive because the action 575 is not re/exive (it is an order addressed to a second person) (13) muúkrun nant͡ʃikmaamata ́ muuka-ru=na nantʃika-ma-ma-ta ́ head-1SG=ACC "ngernail-VBZ-DUR-IMP ‘Scratch my head (a little while more)!’ 13.2.1.5. Reciprocal -nai Similarly to the re/exive, the reciprocal -nai is a bound verbal morpheme in Wampis. The reciprocal indicates that the participants of the predication occupy both the role of agent and patient with respect to each other, i.e. they act equally upon each other. Examples of derivations with the reciprocal are given in (14). (14) Stem Reciprocal anɨɨ ‘think about’ anɨ-nai ‘love’ kahɨru ‘hate’ kahɨru-nai ‘be enemies/hate each other’ kumpa-ma ‘make friends’ friend-VBZ kumpama-nai ‘make friends (with each other)’ maã ́‘kill’ maã-́nai ‘"ght’ nanki ‘throw (something)’ nanki-nai ‘throw (something) to each other’ The next examples illustrate the use of reciprocal constructions. In (15), the comitative =hai ̃coordinates the complex NP that serves as the subject of the reciprocal verb (the "rst part of the complex subject is elided as it appeared a few lines before). 576 (15) awarúnhɛɛ ̃jaúnt͡ʃukka kahɨŕnɛi hakú ɛɛ́ɲawɛɛ awaruni=hai ̃ iaunt͡ʃukɨ kahɨru-nai hak-u a-ina-ua-i Awajun=COM long.ago hate-RECP HAB.PT-NMLZ COP-PL.IPFV-3.PT-DECL ‘[The Wampis] with the Awajun, long ago, hated each other.’ In (16), there is a plural subject of the reciprocal action. (16) iíʃa atúnikmi nɨkaś iíɲa núŋkɨ t͡ʃit͡ʃaḿruka puhústiɲa núu ii=ʃa atu-nai-ka-mi nɨkas iina nunkɨ ̃ 1PL=ADD help-RECP-INTENS-HORT truly 1PL.GEN land\1PL/2PL/3.POSS t͡ʃit͡ʃamaru-u=ka puhu-sa-tinu=a nu defend-NMLZ=FOC live-ATT-FUT.NMLZ=COP NON.VIS ‘We too let us help each other, truly, to defend our lands where we are going to live.’ The next example is also interesting. The reciprocal attaches to the main verb ‘fall’, so the interpretation is similar to a causative: ‘stumbling, they [made] fall each other.’ (17) tiḱit ͡ʃ ɛíɲaka tukúmrumak iɲańiarmiaji tikit ͡ʃi a-ina=ka tukuma-ru-ma-a-ku ina-nai-ara-mia-ji other COP-PL.IPFV=FOC stumble-APPL-REFL-IPFV-SIM fall-RECP-PL-DIST.PT-3.PT+DECL ‘Others, stumbling, they fell.’ (i.e. They made fall each other as they were stumbling.) Recall that the reciprocal -nai may undergo syncope as explained in Chapter VI, thus on the surface its realization may be [ni], as in (17) [iɲańiarmiaji]. 13.2.2. Valence-increasing operators 13.2.2.1. Causatives There are three morphological causatives in Wampis. Two of them are 577 commonly found in texts. There is one more that is unproductive and only occurs lexicalized in a few verbs. The pre"x V- and the su#x -mitika occur frequently in my data, whereas -ka is the unproductive one. Cross-linguistically, the connection between event integration and clause union in the expression of direct causation is very common, as direct causation is expressed with the most integrated morphosyntactic type (Haiman 1983; Givón 2002a; Shibatani & Pardeshi 2002). In Wampis, the pre"x V- and the su#x -mitika mark direct causation and the highest degree of control by the causer. When the causee shows some degree of volition, the causee is construed as an instrument manipulated by the causer, and the causative event does not necessarily imply success. In this kind of situation, Wampis speakers usually employ a quotative construction, showing less degree of integration. For instance, in (18) the causee (Sua) is sent to the forest in the context where the causer (Puhupat) needed someone to gather some plants to prepare medicine for a cold. In this case, we have indirect causation as Sua has some degree of independent control and the success of the caused event is less certain. Syntactically this is re/ected in a less integrated construction involving a quotative construction. (18) puhupát akúpkamaji ikámnum suwán wɨtá túsan Puhupata akupɨ-ka-ma-ji ikama=numa Sua=na Puhupata order-INTENS-REC.PT-3.PT+DECL forest=LOC Sua=ACC 578 wɨ-tá tu-sã go-IMP say-SUB\3SG.SS ‘Puhupat ordered Suwa to go to the forest.’ (Lit.: ‘Puhupat sent Suwa to the forest, say- ing: “go!”.’)212 Morphological causation in Wampis does not include the semantics of permission, unlike in some other languages. Permission is typically expressed also with a quotative construction, usually involving the jussive or the imperative. (19) aj́u wɨt́i túsan t͡sakat́kamiahɛi aiu wɨ-́ti tu-sa-nu t͡saka-tu-ka-mia-ha-i okay go-JUSS say-SUB-1SG.SS consent-APPL-INTENS-DT.PT-1SG.SBJ+DECL ‘I let them go.’ (Lit.: ‘Saying “okay, let them go”, I have consented’.) Causatives in Wampis can be applied to intransitive and transitive predicates. I have very few examples of causatives applied to ditransitive verbs. 13.2.2.1.1. Causative V- The causative V- is the only pre"x in Wampis. V- can adopt all vowel qualities present in the language (/i, ɨ, u, a/); however, the phonetic material that appears with the verb root is synchronically unpredictable. Thus, the form of this causative pre"x must be learnt in connection with the verb root and stored in the speakers’ memory. Examples of this causative pre"x are shown in Table 13.4. 212. Notice that the verb akupɨ also means ‘send’. 579 Table 13.4. Some verbs pre"xed with the causative V- Root V-+Root uaina ‘see’ i-uaina ‘show (cause to see)’ kari ‘ferment’ i-kari ‘make ferment’ t͡sɨkɨ ‘jump, run’ ɨ-t͡sɨkɨ ‘make jump, make run’ uɨka ‘walk’ ɨ-uɨka ‘guide (Lit.: ‘make walk’)’ iú ‘eat’ a-iú ‘feed’ uɨ ‘go’ a-uɨ ‘send’ ɨsa ‘bite’ a-ɨsa ‘cause to bite’ nuɨ ‘gain weight’ u-nuɨ ‘fatten up (animals)’ The next example illustrates the use of the causative pre"x carried by the verb roots uaina ‘see’ ( > i-uaina ‘show’) and ta ‘arrive’ ( > V-ta ‘bring’). Third person objects are not marked on the verb, note the applicative occurring to grammatically instantiate a reference to the theme (the ‘bird’) in (20). For comparison, sentences with underived (without causative) verbs are provided in (21) with a transitive uaina ‘see’, and (23) with intransitive ta ‘arrive’. Note that Wampis marks all NP objects in the same way, with the accusative =na. (20) t͡ʃinkiń ámʷɨ miɲ́a iwɛińturkatmɨ t͡ʃinki=́na amɨ mi=na i-uaina-tu-ru-ka-ta-mɨ bird=ACC 1SG=ACC 2SG=ACC CAUS-see-APPL-1SG.SBJ-IMM.FUT-2SG.SBJ+DECL ‘You are going to show me the bird.’ 580 (21) aḿɨ t͡ʃinkiń wɛińkatmɨ amɨ t͡ʃinki=na uaina-ka-ta-mɨ 2SG bird=ACC see-INTENS-IMM.FUT-2SG.SBJ+DECL ‘You are going to see the bird.’ (22) wií namakań itʲaáhɛɛ ui namaka=na i-ta-ha-i 1SG "sh=ACC CAUS-arrive+IPFV-1SG.SBJ-DECL ‘I am bringing "sh.’ (23) wií taáhɛɛ ui ta-a-ha-i 1SG arrive-IPFV-1SG.SBJ-DECL ‘I am arriving.’ The causative V- might have been used for denominal derivation in the past. There are a few instances of verbs with a pre"xed vowel and causative semantics that are derived from nouns (or at least for which no verbal base could be found synchronically). For instance, t͡sai is a species of tree whose wood is used to make houses and canoes. There is a verb u-t͡sai that has what appears to be the pre"xed causative and which means ‘build /oor’.213 To the best of my knowledge, there is no verb t͡sai from which the causativized u-t͡sai form may have been derived. And the verb u-munt͡su ‘breast-feed/suckle’ is more transparently related to munt͡su ‘nipple, breast’. 213. Though usually the /oor is build of pona (Irartea genus) palm wood. 581 The causative V- is more frequent in my data than the causative su#x -mitika214 that will be presented below. In addition, judging by its form and unpredictable phonetics, it seems that the pre"x belongs to a prior stage where there was a di$erent causation mechanism in the language. 13.2.2.1.2. Causative -mitika The su#x -mitika is another morpheme that serves the function of causation in Wampis. It attaches to transitive and intransitive stems. The next example has a transitive verb that becomes ditransitive when it receives the su#x -mitika. The objects are ‘him’ and ‘ground’. (24) nunkańt͡ʃa niɲ̃aḱ toómtikʲar [...] nunka=na=ʃa ni=̃na=kɨ tau-mitika-ra ̃ ground=ACC=ADD 3SG=ACC=RESTR dig-CAUS-DISTR\3SG.SS ‘[The soldiers] made him only dig the land [...]’ In (25) the transitive verb ii ‘see’ carries the causative -mitika. (25) tiḱit ͡ʃ hintʲańam iímtikʲas tikit ͡ʃi hinta=nama ii-mitika-sa ̃ other trial=LOC see-CAUS-SUB\3SG.SS ‘making him look to the other trail’ In the following example, the causative is used on the verb intransitive verb ‘act 214. Another causative su#x -ka is completely unproductive and found only in a handful of verbs. 582 crazy’ and yields the meaning of ‘make them act crazy’. The use of the re/exive is interesting. The re/exive is used on an intransitive verb (‘act crazy’) apparently functioning as a middle voice that serves to indicate the a$ectedness undergone by the experiencer (the people acting crazy), much like Spanish se is used in re/exive (golpear- se ‘hit oneself’) and middle voice (morir-se ‘die’) constructions. (26) ʃuaŕ uwiʃ́nuka puhaẃɛi húna ut͡ʃí hútika woúrmamtikouka túsar tiármiayi ʃuara uuiʃinu puha-ua-i person shaman live+IPFV-3.SBJ-DECL hu=na ut͡ʃi ́ hutika-a ̃́ uaura-ma-mitika-u=ka PROX=ACC child\ACC do.this-HIAF\3SG.SS act.crazy-REFL-CAUS-NMLZ=FOC ‘That person who is a shaman lives, having done that to the young ones, he is one to make them act crazy.’ 13.2.2.1.3. Causative -ka The su#x -ka is an unproductive morpheme that occurs only with a few verbs. The verbs that carry -ka in the database are listed in (27). (27) uaitu ‘su$er’ uaitu-ka ‘annoy’ hii ‘exit’ ii-ka ‘make exit/take out’ hɨa ‘arrive’ hɨa-ka ‘make arrive’ This su#x -ka may be a formative in the causative -mitika. Note that the above verbs cannot take the causative -mitika. 583 13.2.2.2. Transitivizer -ki There is a su#x -ki that turns an intransitive verb into a transitive one. It is an unproductive su#x, it only occurs lexicalized in a few verb stems. (28) uiʃi ‘laugh’ → uiʃi-ki ‘make fun of (i.e. ‘laugh at’)’ iha ‘defecate’ → iha-ki ‘defecate on someone/some place’ natuma ‘move onself away’ → natuma-ki ‘dodge, avoid’ Because of the semantics of the derived items, -ki potentially may have been a su#x with some form of locative or goal semantics. 13.2.3. Applicative -ru ~ -tu The applicative prototypically promotes a peripheral participant onto the “center stage” of a predicated event by making it core argument, in a Direct Object role. In addition, applicatives can rearrange the argument structure rather than necessarily increase the valence (Comrie 1985; Croft 1994; Peterson 2007) The general applicative in Wampis has two forms: -ru and -tu. They are in complementary distribution: the selection of one form or the other by a verb is purely lexical. Actually, verbs can be divided into two classes de"ned by the form of the applicative that they take. For the moment, this classes are referred to as “rV” and “tV” classes. In addition, there does not seem to be a semantic or phonological conditioning 584 for selecting one form or the other of the applicative. Examples of verbs belonging to “rV” and “tV” classes are given in Table 13.5. The distinction between “rV” and “tV” classes is maintained also through the object markers paradigm, as will be seen in §13.2.4.5. The applicative can be receive by transitive, intransitive and ditransitive verbs. In addition to the consonant di$erence, the applicative forms have allomorphs with a vowel a instead of u; i.e. -ta and -ra. They usually surface with vowel a in front of the agentive nominalizer -inu. A number verbs stems that are derived with the applicative (with either form -tu or -ru) also surface with the allomorphs -ta or -ra. In Wampis, the element promoted by the applicative is typically a location, a bene"ciary or male"ciary participant. In Wampis, the applicative does not necessarily increases the transitivity of the clause. The male"ciary and bene"ciary uses of the applicative, specially with transitive verbs, do not always increase the transitivity of the clause. This fact agrees with some claims in the literature that suggest that applicatives function re-arranging the argument structure rather than increasing the valence, including analyses of other languages of the Peruvian Amazon where applicatives are employed for several functions (Wise 2002; Valenzuela 2010). 585 Table 13.5. Sample of verbs that select forms -ru and -tu of applicative “-rV” Verbs “-tV” Verbs uakɨ ‘want’ transitive auɨma ‘send’ ditransitive t͡ʃit͡ʃa ‘speak’ transitive uha ‘tell’ transitive uɨ ‘go’ intransitive ma ̃‘kill’ transitive su ‘give’ ditransitive apa ‘put together, copulate’ transitive t͡supi ‘cortar’ transitive ɨnkɨma ‘enter’ intransitive nuhankɨ ‘/ood, drown’ trans/intrans ahɨka ‘throw in water’ transitive ta ‘come’ intransitive anuha ‘patch, glue’ transitive hu ‘take’ transitive pata ‘follow’ transitive antu ‘hear’ transitive hintina ‘teach’ transitive tɨpɨ ‘lie down’ intransitive ɨsa ‘bite’ transitive kanta ‘burp’ intransitive pɨɨ ‘be stuck’ intransitive naku ‘play’ intrans/transitive mat͡sa ‘inhabit, congregate’ intransitive taka ‘work’ intransitive ɨt͡sɨra ‘announce’ transitive kasa ‘steal’ transitive tɨmaʃi ‘comb’ transitive ahapɨ ‘throw away’ transitive With verbs of motion, the applicative may promote a location/goal to Object status. Intransitive verbs mark their locations with a locative adposition, as the next example illustrates. 586 (29) meśanam waḱaji mesa=nama ua-ka-ji table=LOC climb-INTENS-3.PT+DECL ‘I just climbed on the table.’ In the next example, the intransitive verb ‘climb’ receives the applicative and promotes a location/goal. In the passage, a woman tries not to move after she sees a Tijai (a human-like being in the Wampis folklore) from an elevated point when she was trying to hunt game. The Tijai realizes he is being watched and climbs to face the woman. The movement towards the woman is marked with the applicative. Notice that third person objects are zero-marked in the verb in Wampis. Note that the nominalized verb ‘stand’ (‘the one who was standing’—the location/goal), which is a fronted focused object, receives accusative marking and not a locative adposition (unlike the location in (29)). (30) wahaśunak tihɛíka nɨkaṕrin aśa ̃nɨkaṕaŕ warukú tiḿaji [uaha-sa-u]=na=kɨ tihai=ka nɨkapɨ-ra-inu a-sa ̃ stand-ATT-NMLZ=ACC=RESTR Tijai=FOC feel-DISTR-NMLZ COP-SUB\3.SS nɨkapɨ-ra ̃ ua-ru-ka-u timaji feel-DISTR\3SG.SS climb-APPL-INTENS-NMLZ NARR ‘To that who was standing, the tijai because he is one who feels, he having sensed [her], he climbed toward her’ The applicative occurs with transitive and intransitive verb roots in constructions that can be analyzed as external possession. In this construction, the 587 applicative is su#xed to the verb, increasing the valence of the clause allowing a semantic possessor to be promoted as an core argument directly dependent on the verb (Payne & Barshi 1999; Zavala 1999). In Wampis, this construction is de"ned by an overt NP optionally marked by a possession marker, and the verb marked by the applicative and a person marker that indicates the semantic possessor. In the next example, we can observe the promotion of a second participant (in this case, a ‘you plural’ participant) in a object syntactic role: the verb ‘set "re’ receives the applicative form -tu and the second plural object is marked on the in/ection of the verb. In the example, the 2PL participant are the owner of the house, thus the construction could be analyzed as an instance of external possession. (31) hɨãńʃa ɨkɨɨ́matkattahrumɨ hɨa=na=ʃa ɨkɨma-tu-ka-tata-harumɨ house=ACC=ADD set."re-APPL-INTENS-DEF.FUT-1SG>2PL+DECL ‘I am going to set "re to the house too (to your (PL) house).’ The next elicited examples show instances of an the applicative on an intransitive verb. Compare the use of the applicative-marked verb in (33) with the unmarked (for applicative) verb in (32). Note also that the nouns are marked by a possessor marker. 588 (32) nuwaŕ wɨkaśmaji nua-ru uɨka-sa-ma-ji woman-1SG walk-ATT-IMM.PT-3.PT+DECL ‘My wife took a walk’ (33) nuwaŕ wɨkat́rusmaji nua-ru uɨka-tu-ru-sa-ma-ji woman-1SG walk-APPL-1SG.OBJ-ATT-IMM.PT-3.PT+DECL ‘My wife took a walk (and I am somehow a$ected by it.215)’ Example (35) shows a malefactive use of the applicative with the verb ‘dream’. First, note that karama ‘dream’ is a transitive stem,216 as shown in (34). In (35), the applicative occurs on the verb, but clearly it cannot be being used for turning karama into a transitive verb, since it is transitive. Rather, the applicative is altering the argument structure of the clause to promote another argument, semantically indicating that the action of dreaming is to the detriment of the person who dreams (instantiated as the grammatical subject, indexed as -nu ‘1SG.SS’ on the copula and as -ha on the main verb). Note that, culturally, dreaming of a boa is seen as a bad thing by the Wampis. 215. Two examples of potential contexts where this phrase could be uttered: the wife of a person goes away without preparing food to carry, for example, when someone goes hunting; or when the husband is worried that he has not seen her wife around (Wampis villages are small so men and women usually know where the other person is). 216. From kara ‘dream (N)’ plus the verbalizer -ma. 589 (34) ʃuarań karaḿrahɛɛ ʃuara=na karama-ra-ha-i person=ACC karama-DISTR-1SG.SBJ-DECL ‘I dreamed of a person.’ (35) napiń karaḿru asań ʃiiŕt͡ʃoo nɨkaṕɨahɛɛ napi=na karama-ru-u a-sa-nu boa=ACC dream-APPL-NMLZ COP-SUB-1SG.SS ʃiira-t͡ʃau nɨkapɨ-a-ha-i good-NEG.NMLZ feel-IPFV-1SG.SBJ-DECL ‘Having dreamed of the boa, I feel bad.’ A benefactive use of the applicative is illustrated in (37) with the verb root ai ̃ ‘hang cloth’ (also ‘fold cloth, make the bed’). Example (36) illustrates the use of the same verb with a “neutral” interpretation. In comparison, in (37) the action is done for someone else, which is grammatically marked with the use of the applicative. (36) ɛi ̃ḱtatui ai-̃ka-tata-u-i hang.cloth-INTENS-DEF.FUT-3.SBJ-DECL ‘He is going to hang the cloths’ (37) ɛi ̃t́rukti ai-̃tu-ru-ka-ti hang.cloth-APPL-1SG.OBJ-INTENS-JUSS ‘Let him hang the cloths for me’ (Spanish: ‘Que me tienda la ropa.’) The bene"ciary of an action may be optionally marked with -nau ‘bene"ciary’ (cf. §10.4.4), followed by the accusative, as in the next example. 590 (38) mamań wiɲaún hiit́maji [mama=na] [ui-nau=na] hii-tu-ma-ji manioc=ACC 1SG-BEN=ACC pull.out.PFV-APPL-REC.PT-3.PT +DECL ‘He harvested manioc for me.’ Some cases of verbs that frequently occur with the applicative suggest an associative reading, or at least a mix of benefactive and associative meanings. For instance, when mat͡sa ‘inhabit, congregate’ carries the applicative, it means ‘be/live together’, or even ‘live in community’. In the following example, the part that contains the verb ‘inhabit’ receiving the applicative was translated into Spanish as se convive armoniosamente ‘we live together harmoniously’, i.e. with an interpretation of associative (‘live together’) and benefactive (‘harmoniusly’). (39) túmaiɲak ʃiír mat͡sat́ɨnawai húnaka aʃi ́ tií ukúkɲiti iiɲa uúntri ajakú=a=nu ɨt́͡sa naańtin tu-mai-na=kɨ ʃiira mat͡sa-tu-ina-ua-i say-POT-NMLZ=RESTR good inhabit-APPL-PL.IPFV-3.PT-DECL hu=na=ka aʃi ti uku-kini-u=iti PROX=ACC=FOC all INTENS leave-WHILE.MOVING-NMLZ=COP.3+DECL iina uunta-ri ̃ a-haku=a=nu 1PL.GEN elder-1PL/2PL/3.POSS COP-HAB.PT=COP=NON.VIS ɨtsa ̃ naa-̃tinu sun name\1PL/2PL/3.POSS-ATTRIB ‘It can be said that we live together harmoniously because of all [the teachings] that our ancestor called the Sun left.’ (Lit.: ‘It can be said that we live well [because of] all this left [by] that one that used to be our ancestor, [whose] name was Sun.’ 591 Many lexicalized verbs appear to be derived with either of the applicative forms -ru~-tu. A brief list of examples is given in (40). (40) Original verb Derived form ɨtsɨra ‘announce, inform’ ɨtsɨratu ‘accuse’ uka ‘pour liquid, spill’ ukatu ‘paint’ ɨnkɨ ‘put in’ ɨnkɨtu ‘save (something for oneself or for another person)’217 ɨnt͡sama ‘ride’ ɨnt͡samaru ‘climb on top of (animal or person’) anɨnta ‘magical song’ anɨntaru ‘sing anɨnta (i.e. for self or someone’s bene"t)’ 13.2.3.1. Homophony between the applicative forms and the 1SG object markers The applicative forms -tu/-ru coincide in form with the 1SG object marker forms - tu/-ru (§13.2.4.1). Thus when both the applicative and the object markers are used, the object marker switches its form to the alternate one, presumably to avoid haplology. Table 13.6 shows this alternations. 217. Synchronically also used for ‘save money’, including ‘save money in bank’. 592 Table 13.6. Combination of applicative and 1SG object su#xes Applicative 1SG Object Applicative+1SG Object -ru -ru -ru-tu -tu -tu -tu-ru The alternations in forms between the applicative and the object markers are described in §13.2.4.5. We will see that the applicative not only conditions the form of the 1SG objects markers, but also the forms of 1PL and 2 object markers. 13.2.3.2. Double marking of applicative? Sometimes, the applicative appears to be repeated after a 1SG object marker, as in the next example. (41) miɲ́aka jat͡súrnaka manturtuaẃarɛi titah́kama ́uútu mina=ka iat͡su-ru=na=ka ma-̃tu-ru-tu-aú-ara-i 1SG.GEN=FOC brother-1SG=ACC=FOC kill-APPL-1SG.OBJ-APPL?-HIAF-PL-3.PFV+DECL ti-tahkama ̃́ uu-tu-u say+LOAF-FRUST\3SG.SS cry-APPL-NMLZ ‘Trying to say “they just killed me my brother”, he cried.’218 The repetition of the applicative form is not very well understood, but it might 218. This example may be considered another example of external possession with the applicative extending the valency of the clause and the possessor marked on the verb and the object (‘my brother’). 593 be a case of adding intensity to a benefactive/malefactive action (I have no examples of applicative being double-marked when promoting a locative argument). Another possible analysis is that verbs that frequently occur with an applicative, like ‘kill’, are being reinterpreted. Some speakers may considered the form mantu (kill+applicative) as the root, to which they add the applicative form -ru and then 1SG object marker -tu. 13.2.4. Object marking on the verb The marking of object in Wampis constitutes a fascinating and complex area of the verbal morphology. As a reminder of the verb structure, Table 13.7 shows a simpli"ed version of the verb structure presented in Chapter XII. Table 13.7. Verb structure (simpli"ed) -1 0 1 2 3 4 5 6 7 8 ROOT Applicative (Valence) Object Person As seen in Table 13.7, Objects are for the most-part marked in slot 2 of the “derivational” morphological level. But there are some important caveats: • If the object is a 2 person, it is marked in position 7 of verbal piece, with portmanteau morphemes that index both subject and object. 594 • If a 3 person acts on a 2PL or 1PL object, then the grammatical person of the object is marked in slot 2 of “derivational morphology” (with an object marker) and in slot 7 of “in/ectional morphology” (with a subject marker; i.e. the verb agrees with the object). The pattern of verbal argument indexation is explained in detail in §14.3.3, after the subject a#xes are presented. For the present section, I will restrict myself to presenting the su#xes that serve to mark objects. Another interesting complication of Wampis verbal object morphology is that 1SG object markers and the applicative forms are identical, as noted earlier in §13.2.3.1. The presence of the applicative also determines allomorphic the form of the object markers for the "rst person plural objects and second person object. In addition, there are several su#xes used for the marking of singular and plural Speech Act participants. As in other areas of Wampis morphology, the selection of allomorphs of these di$erent object markers is entirely lexical and fairly untidy. Third person objects are not marked on the verb. 13.2.4.1. First person object First person object markers, singular and plural, are presented in Table 13.8. 595 Table 13.8. First person singular and plural object su#xes 1SG.OBJ 1PL.OBJ -ru -tu -rama -tama -karata~-karatu As can be seen, the same distinction of rV~tV forms that was previously distinguished for the applicative §13.2.3 is also found for the form of the object marker that the verb takes. Verbs that belong to the -rV class take -ru and -rama to mark 1SG or 1PL objects, respectively; verbs that belong to the -tV class take -tu for 1SG objects and the form -tama for 1PL objects. The forms -tama/-rama also mark 2 object (§13.2.4.2). The su#x -karatu~-karata is not a$ected by this allomorphic pattern and can occur with either “rV” or “tV” verbs. The su#x -karatu~-karata denotes an inde"nite "rst plural object. The following examples show the verb ‘inform’, which is a “tV” type verb. Thus, in (42), the verb takes the 1SG object marker form -tu. (42) óo uhat́kamaji au uha-tu-ka-ma-ji DIST inform-1SG.OBJ-INTENS-REC.PT-3.PT+DECL ‘He informed me.’ In (43), the verb takes the 1PL/2 object marker form -tama. Notice that when 596 third person subjects act upon plural Speech Act Participant objects, the verb agrees with the object, not with the subject. This pattern of hierarchical agreement is explained in §14.3.3. (43) óo uhat́makmahi au uha-tama-ka-ara-ma-hi DIST inform-1PL/2.OBJ-INTENS-PL-REC.PT-1PL+DECL ‘She/He informed us.’ The next two examples illustrate "rst person object marking with the “-rV” verbs ‘throw away’ and ‘cut’, respectively. In (44), the 1SG object marker form -ru occurs attached to the verb. (44) urúkunmak ahapruat́sa taa ́túsa ̃paŋkás uru-ka-u=numa-ki ahapɨ-ru-a-́tasa ̃ ta tu-sa ̃ how-Q-NMLZ=LOC-RHET throw.away-1SG.OBJ-HIAF-PURP\3.SS say-IPFV say-SUB\3.SS panka-sa ̃ raise.head-ATT\3SG.SS ‘Saying “how is it where [the bear] is going to throw me?”, he raised his head.’ In (45) the 1PL/2 object form -rama is present on the verb -rV verb ‘cut’. Notice that the use of the aktionsart su#x in this context (in this particular case, the ‘intensive’ su#x -ka) adds a perfective sense of an intensive ‘just done action’ to the verb. 597 (45) t͡supiŕmakhi tusa ́ t͡supi-rama-ka-hi tu-sa ̃ cut-1PL/2.OBJ-INTENS-1PL say-SUB\3SG.SS ‘Saying “he just cut us!”.’ As a "nal point, there is a 1PL object form -karatu~-karata that is usually translated as a 1PL object. (46) aḿɨka t͡sanúrkartamɨ amɨ=ka t͡sanu-ru-karatu-a-mɨ 2SG=FOC lie-APPL-1PL.OBJ-IPFV-2SG.SBJ+DECL ‘You are lying to us.’ This pronoun is sometimes used with a non-referential sense. In (47), a mestizo (non-Wampis person) reasons with Piruch, a Wampis leader that used to "ght soldiers (because of the abuses the latter commited). He tells him to stop killing because the soldiers may kill him (Piruch) as they had killed his brother. He uses the object form - karatu but not to mean a speci"c ‘us’ (the mestizo and Piruch were friends), but an non- referential (maybe impersonal?) ‘us’—these uses of -karatu~-karata need more examination. (47) Pirut͡ʃa ́jamɛíka mankartuaẃaipa Pirutʃa ́ iamai=ka ma-̃karatu-aú-ai-pa Pirucha\VOC now=FOC kill-1PL.OBJ-HIAF-APPR-PROH ‘Piruch, now do not kill anymore!’ Many derived generic nouns possessed this plural object marker: 598 (48) (a) hintiŋ́kartin hintina-karatu-inu teach-1PL.OBJ -NMLZ ‘teacher’ (‘one who teaches us’). (b) t͡ʃit͡ʃaḱartin t͡ʃit͡ʃa-karatu-inu speak-1PL.OBJ-NMLZ ‘adviser’ (‘one who speaks to us’) (c) sukaŕtin su-karata-inu give-1PL.OBJ-NMLZ ‘supplier’ (‘one who gives us’) The object marker -karatu~karata is also used when a second person agent (singular or plural) acts upon a "rst plural patient (49), however the other object markers are also available (50) in this case. (49) aḿɨja atupkratkata ́ amɨ=ʃa atupa-karatu-ka-ta ́ 2SG=add support-1PL.OBJ-INTENS-IMP ‘You too support us!’ (50) amɨ mantamaḿamɨ amɨ ma-̃tama-a-́ma-mɨ 2SG kill-1PL/2.OBJ-HIAF-REC.PT-2SG.SBJ+DECL ‘You killed us.’ 599 13.2.4.2. Second person object: 3>2 Second person singular and plural objects are marked in slot 2 only when a third person subject acts upon the second person object. Second person objects do not distinguish between plural and singular forms, but the structures are not ambiguous: • when the second person P argument is singular, the third person A argument is marked on the morphological position of subject (slot 7) of the verb. • when the second person P argument is plural, the third person A argument is not marked, and instead the verb agrees with the plural SAP participant in the morphological positions of object (slot 3) and subject (slot 7). See the examples below for an illustration of this pattern. Table 13.9 lists the 2 object markers. The forms -rama and -tama are shared with the forms for 1PL object and are taken by “rV” and “tV” verbs, respectively. The formative -ma present in -rama and -tama may also surface by itself marking a 2 person object. This indicates that there might be an old morpheme *rV ~ *tV as another formative in the -rama/-tama forms (probably related to the source of the applicative/ 1SG object forms?).219 219. Notice also that the plural form of the second subject marker is -rumɨ clearly historically decomposable into -ru (which, again, has the same form one of the forms used for the applicative 600 Table 13.9. Second person singular and plural objects indexes 2 object su#xes with a 3 person subject -rama -tama -ma Examples (51) and (52) have the verb ‘call’, which is a “rV” verb. Thus, the object form -rama is received by the verb. Notice that the indexation of the subject makes the sentence un-ambiguous with respect to the number of the object. In the case of 3>2SG (51), the verb agrees with the subject (with the su#x -ji). In the case of 3>2PL (52), the verb agrees with the object (with the su#x -rumɨ, which normally marks the 2PL subject). (51) nitʲa ́ɛíɲa aḿiɲa unt͡súrmakmarji nita a-ina unt͡su-rama-ka-ma-ji 3PL COP-PL.IPFV call-1PL/2.OBJ-INTENS-REC.PT-3.PT+DECL ‘They called you (SG).’ (52) nitʲa ́ɛíɲa unt͡súrmakmarmɨ nita a-ina unt͡su-rama-ka-ma-rumɨ 3PL COP-PL.IPFV call-1PL/2.OBJ-INTENS-REC.PT-2PL+DECL ‘They called you (PL).’ and 1SG object) and -mɨ (mɨ is directly related to the 2SG pronoun amɨ). 601 The next examples show instances of the form -tama with the “tV” verb ‘see’. (53) ni ̃í wɛitmakmaji ni ̃ uai-tama-ka-ma-ji 3SG see-1PL/2.OBJ-INTENS-REC.PT-3.PT+DECL ‘He saw you (SG).’ (54) ni ̃í wɛit́makmarmɨ ni ̃ uai-tama-ka-ma-rumɨ 3SG see-1PL/2.OBJ-INTENS-REC.PT-2PL+DECL ‘He saw you (PL).’ 13.2.4.3. Second person object: 1>2 Second person objects are marked in slot 7 (“Person”) when a "rst person agent acts on a second person patient. Table 13.10 lists the forms used. Table 13.10. Forms for marking 1A participants acting upon 2P participants A → P arguments 2SG P 2PL P 1SG A hamɨ himɨ 1PL A harumɨ hirumɨ The morphemes in Table 13.10 are actually portmanteau morphemes that mark both subjects and objects. They are easily decomposable: all of them are used to mark 1 and 2 person subject, singular and plural, as presented in Table 13.11 (see §14.3.3 for a 602 description of subject markers). Table 13.11. Subject markers, 1 and 2 persons Subject Marker Person -ha 1SG -hi 1PL -mɨ 2SG -rumɨ 2PL Examples (55)–(58) show the use of the 1 > 2 person markers. (55) tah́amɨ ta-hamɨ say+IPFV-1SG>2SG+DECL ‘I am telling you.’ (56) tah́armɨ ta-harumɨ say+IPFV-1SG>2PL+DECL ‘I am telling you (PL).’ (57) tah́imɨ ta-himɨ say+IPFV-1PL>2SG+DECL ‘We are telling you.’ 603 (58) tah́irmɨ ta-hi-rumɨ say+IPFV-1PL>2PL+DECL ‘We are telling you (PL).’ 13.2.4.4. A paradigm of object marking Table 13.12 o$ers a paradigm of object marking with the verb uaina ‘see’. The verb uaina ‘see’ is a -tV class verb; i.e. the object su#xes that it takes are -tu for 1sg objects and -tama for 1pl or 2pl objects. Recall that 3 person objects are zero-marked. Verbs that belong to the -rV class show exactly the same pattern, but switch their 1sg, 1pl and 2pl object su#x to the corresponding forms -ru and -rama. The "rst plural inde"nite object su#x -karatu occurs in the same position where -tama occurs in the Table; however, bear in mind the important observation that -karatu may occur with -tV or -rV verbs. The relevant object su#xes are in bold face type in the examples in Table 13.2. 604 Table 13.12. Paradigm of object marking on the verb 1SG → 2SG uaina-ka-hamɨ see-INTENS-1SG>2SG+DECL 1PL → 2SG uaina-ka-himɨ see-INTENS-1PL>2PL+DECL 1SG → 2PL uaina-ka-harumɨ see-INTENS-1SG>2PL+DECL 1PL → 2PL uaina-ka-hirumɨ see-INTENS-1PL>2PL+DECL 1SG → 3 uaina-ka-ha-i see-INTENS-1SG.SBJ-DECL 1PL → 3 uaina-ka-hi see-INTENS-1PL.SBJ-DECL 2SG → 1SG uaina-tu-ka-mɨ see-1SG.OBJ-INTENS-2SG+DECL 2PL → 1SG uaina-tu-ka-rumɨ see-1SG.OBJ-INTENS-2PL+DECL 2SG → 1PL uaina-tama-ka-mɨ see-1PL/2.OBJ-INTENS-2SG+DECL 2PL → 1PL uaina-tama-ka-rumɨ see-1PL/2.OBJ-INTENS-2PL.SBJ+DECL 2SG → 3 uaina-ka-mɨ see-INTENS-2SG.SBJ+DECL 2PL → 3 uaina-ka-rumɨ SEE-INTENS-2PL.SBJ+DECL 3 → 1SG uaina-tu-ka-i ̃ see-1SG.OBJ-INTENS-3.PFV+DECL 3 → 1PL uaina-tama-ka-hi see-1PL/2.OBJ-INTENS-1PL+DECL 3 → 2SG uaina-tama-ka-i ̃ see-1PL/2SG.OBJ-INTENS-3.PFV+DECL 3 → 2PL uaina-tama-ka-rumɨ see-1PL/2PL.OBJ-INTENS-2PL+DECL 3 → 3 uaina-ka-i ̃ see-INTENS-3.PFV+DECL 605 13.2.4.5. Applicative and objects Let us now turn the discussion toward an very interesting property of Wampis verbal morphology: when the applicative occurs with the object markers, the object markers switch from their “tV” forms to their “rV” forms and vice versa. Table 13.13 lists these forms for "rst person object markers. The forms -tuma and -turama are found very rarely in the data. They just seem like synchronically idiosyncratic formations associated to tu (perhaps the applicative?, or 1sg object?) +-ma and tu+-rama, respectively. Table 13.13. Combination of applicative and 1 person object forms Type\Su#xes Applicative 1SG Object 1PL Object “tV” verbs -tu -ru -rama -karatu~karata “rV” verbs -ru -tu -tama -tuma -turama -karatu~-karata Consider the verb ‘inform’, which is a “tV” verb; i.e. it selects the form -tu of the applicative and of the 1SG object. In (59), the verb receives the 1SG object marker -tu. 606 (59) óo miɲa uhat́kamaji au mi=na uha-tu-ka-ma-ji DIST 1SG=ACC inform-1SG.OBJ-INTENS-REC.PT-3.PT+DECL ‘He informed me.’ However, when the applicative is used (in this case, in a benefactive construction), the form of the object marker switches to -ru. (60) uhatrukata ́ uha-tu-ru-ka-ta ́ inform-APPL-1SG.OBJ-INTENS-IMP ‘Saying “tell her for me!”...’ Another example with a ‘rV’ verb, t͡ʃit͡ʃa ‘speak’, is shown in (61). (61) miɲ́a umaŕ t͡ʃit͡ʃaŕtak kakarta ́ɛíʃmankuitmɨ mina uma-ru t͡ʃit͡ʃa-ru-tu-a-kũ 1SG.GEN sister-1SG speak-APPL-1SG.OBJ-IPFV-SIM/3SG.SS kaka-ra-ta ́ aiʃumanku=ita-mɨ be.strong-DISTR-IMP man=COP-2SG.SBJ+DECL ‘While my sister was saying: “Be strong, you are a man!” The forms for the 1PL object are fairly messy when the applicative is present. Examples of di$erent forms like -turama and -tama are given in (62) and (63), respectively. (62) paańmarmin hurútramkimaji paantama-rumi=na hu-ru-turama-ki-ma-ji plantain-2.SG=ACC take-APPL-1PL/2.OBJ-WHILE.MOVING-REC.PT-3.PT+DECL ‘He took your banana (to your detriment).’ 607 (63) puhupat́ awat́rutmamhi Puhupata auata-ru-tama-ma-hi Puhupata hit-APPL-1PL/2.OBJ-REC.PT-1PL.SBJ+DECL ‘Puhupat hit us.’ Likewise, combinations of the applicative with the second object markers follow the same ‘switching’ pattern, as presented in Table 13.14. Table 13.14. Combinations of the applicative with 2 person object markers Type\Su#xes Applicative 2 Object “tV” verbs -tu -rama -ma -ruma “rV” verbs -ru -tama -turama -tuma 13.2.5. The su0x -nin ‘almost’ The su#x -nin creates a verb stems that adds the semantics of ‘almost’ to the verbal root. (64) junińrahɛɛ iu-nin-ra-ha-i eat-ALMOST-DISTR-1SG.SBJ-DECL ‘I almost ate.’ 608 (65) nahańaninsarji nahana-nin-sa-ara-ji transform-ALMOST-ATT-PL-3.PT+DECL ‘They almost transform.’ 13.3. Verbal slot 3: Aktionsart, Imperfective, Durative, Present Habitual and Potential su0xes In any given verb, only one su#x may occur in slot 3 of the verb structure, which I have labeled “Aspect” (cf. §12.7). That is, the relevant su#xes form a paradigm. Some members of the paradigm seem more derivational (or lexical), while others have a more in/ectional feel. In fact, with the exception of the -u ‘do away’, imperfective -a, the present habitual -na and the potential -mai, the rest of the su#xes in this paradigm have a verb to verb derivation property; i.e. they create what I have called an “inner stem” (cf. §12.7): a stem that can receive any piece of morphology from positions 1 to 8 in the verb (i.e. a stem that is treated as a root). Most of the su#xes in position 3 have more aspectual meaning, but at least one is more modal in meaning. However, they all comprise a single paradigm. 13.3.1. Overview Table 13.15 provides a list of the su#xes that can "ll the morphological position 3 in the verb structure. Following Overall (2007), I analyze a sub-group of su#xes 609 (discussed in §13.3.2) in Wampis as more ‘aktionsart’-like in nature. Table 13.15. Su#xes that "ll morphological position 3 in the verb structure (“Aspect”) -a(́u) ‘High a$ectedness’ -i ‘Low a$ectedness’ -ka ‘Intensive’ -ki(ni) ‘Do while going’ -ra ‘Distributed action’ -sa ‘Attenuative’ -ri ‘Do in proximity/while coming’ -u ‘Do away from location’ A k t i o n s a r t -a ‘Imperfective’, -ina ‘plural imperfective’ -ma ‘Durative’ -na ‘Present habitual’ -mai ‘Potential’ The term Aktionsart is understood to refer to lexical aspect. As ‘lexical’ aspect, these su#xes derive stems with sometimes subtle new meanings, but most verb roots also appear to have a “preferred” aktionsart su#x with which they appear. These “root + aktionsart” stem is obligatory for perfective contexts, most past tenses and future tenses and cannot co-occur with the imperfective, the durative, the present habitual and the potential, thus forming a paradigm. The selection of the “preferred” aktionsart su#x 610 by a root may not be semantically transparent synchronically; however, in terms of frequency it is fairly easy to predict which aktionsart will appear "lling position 3 when an aktionsart is needed for the verb to appear in certain morphosyntactic contexts (cf. §13.3.7). 13.3.2. Aktionsart It is di#cult to pinpoint the speci"c meaning for aktionsart su#xes in Wampis, as their semantics are rather vague and sometimes they just seem to mix more than one semantic property. These suggests that they are highly lexical in nature. There are several semantic domains grammaticalized via these morphemes, among them participant a$ectedness and associated movement, the label “Aktionsart” is more of a convenient term to try to classify these su#xes. Most aktionsart su#xes have a verb-to-verb derivational function. The verb t͡ʃit͡ʃa ‘speak’, for instance, can assume di$erent meanings according to which aktionsart it carries: With attenuative -sa: t͡ʃit͡ʃa-sa ‘converse’ (as in a quotidian, uneventful action) (66) nusɨ-hai ̃ t͡ʃit͡ʃa-sa-ma-ji nusɨ-COM speak-ATT-REC.PT-3.PT+DECL ‘He spoke (tranquilly) with Nuse.’ 611 With intensive -ka: t͡ʃit͡ʃa-ka ‘speak more forcefully’ (as in giving a speech— culturally, the Wampis speak forcefully when speaking in front of others, e.g. in an communal meeting). (67) unnta Alfonso Graña t͡ʃit͡ʃa-ka-ma-ji elder Alfonso Graña speak-INTENS-REC.PT-3.PT+DECL ‘The elder Alfonso Graña spoke (forcefully, showing his Arutam power, in a gathering front of the people)’220 With distributed action -ra: t͡ʃit͡ʃa-ra ‘speak to several people’ or ‘speak several times’. (68) puhupat t͡ʃit͡ʃa-ra-ma-ji Puhupata speak-DISTR-REC.PT-3.PT+DECL ‘Pujupat spoke (several times).’ Aktionsart su#xes are synchronically in complementary distribution with the imperfective, durative, present habitual and potential su#xes. However, aktionsart 220. This sentence was given as an example when I asked about di$erences in the use of aktionsart su#xes with the verb ‘speak’. The context I was given for this examples was (translated from my "eld notes): “for example, when there is a meeting of people discussing who to support [there were certain political problems in the villages]. For example, when uun Alfonso Graña speaks, everybody listens, his word is like...that is what you have to do, for example, there you can say: (t͡ʃit͡ʃa-ka ‘speak-intensive’), ‘the elder Alfonso Graña spoke’)”. Elders are very respected, and they manifest that they have a “vision” received from Arutam (see Chapter II for description of Wampis beliefs) through their oratory skills. 612 su#xes seem to have been more recursive in previous stages of the language. When an aktionsart su#x occurs lexicalized in a verb form, they can receive new aktionsart morphology—what I have called an “inner stem”. For instance, the verb uika ‘walk’ is a derivation of uɨ ‘go’ and the intensive aktionsart su#x -ka. The derived form uɨka receives the attenuative aktionsart su#x -sa, as in (69). (69) nuwaŕ wɨkat́ursaji nua-ru uɨka-tu-ru-sa-ji woman-1SG walk-APPL-1SG.OBJ-ATT-3.PT+DECL ‘My wife just took a walk (to my detriment; i.e. the speaker did not know she was going away).’ As far as I can tell, a verb can occur with di$erent aktionsart verbs depending on what the speaker is trying to communicate. However, not all aktionsart su#xes are available for use by any given verb—perhaps the restriction in the selection of an aktionsart su#x has to do with the lexical semantics of the verb itself (for instance, it is di#cult to think of a situation where a stative verb like puhu ‘live’ may use the associated movement -ki ‘do while moving’). That said, most verbs typically favor only one aktionsart su#x to occur in certain morphosyntactic environments, such as perfective, imperative and future, among others (see §13.3.7 for more on this topic). In the small Jivaroan tradition, other analysts have treated aktionsart su#xes as “perfective” (Larson 1963; Corbera Mori 1994; Gnerre 2010), though Overall (2007: 613 291) suggests that the “perfectiveness” may be a property of the root in Awajun. There are arguments in favor and against each position: • Verbs roots have an perfective form that sometimes di$ers from the bare root (§12.8), so it seems like perfectivy may be adjusted by the root itself. • Not all verbs need an aktionsart su#x to occur in perfective, just done-actions (§14.2.2.1) or with other past or future tenses (most verbs do, though). For instance uɨ ‘go’, ha ‘be sick’, tuku ‘shoot’ do not need an aktionsart su#x. In some cases, there may be an issue of avoiding ambiguity playing a role: uɨ without aktionsart su#xes means simply ‘go’, when derived with the intensive -ka it means ‘walk, wander in forest’; and ha ‘be sick’ with the intensive aktionsart su#x -ka means ‘die’. But “avoiding ambiguity” is not clearly a motivation in general, because one could ask why another di$erent aktionsart su#x (i.e. di$erent from -ka in the examples given) is not selected. This leads us to a semantic discussion of what each aktionsart su#x can or cannot derive, which is problematic given that synchronically, the selection of an aktionsart su#x seems highly lexical in many cases. • A few verbs do not need an aktionsart su#x, however most verbs do need an aktionsart stem for them to be able to be used in several speci"c contexts. For 614 instance, with the 3 person perfective, most verbs must obligatorily used an aktionsart (typically their “preferred” aktionsart su#xes, which are largely lexically determined) to yield the sense of a “just-completed” action: (70) kaŋkɨḱɛi ̃ kankɨ-ka-i ̃ roll-INTENS-3.PFV+DECL ‘It just rolled.’ cf *kankɨ-i ‘roll-3.PFV’ The aktionsart is incompatible with an imperfective-marked verb: (71) kaŋkɨáwɛi kankɨ-a-ua-i roll-IPFV-3.SBJ-DECL ‘It is rolling’ cf. *kankɨ-ka-a-ua-i ‘roll-INTENS-IPFV-3.SBJ-DECL’ • Related to the above examples, aktionsart su#xes cannot co-occur with any of the other su#xes in slot 3, imperfective, durative, present habitual and potential. (72) kankɨ-ka-i ̃ roll-INTENS-3.PFV+DECL ‘It just rolled.’ kankɨ-a-ua-i roll-IPFV-3.SBJ-DECL ‘It is rolling’ kankɨ-ma-ta ́ roll-DUR-IMP ‘roll it a little while’ 615 kankɨ-na-ua-i roll-PRES.HAB-3.SBJ-DECL ‘It usually rolls’ kankɨ-mai-inu=aiti roll-POT-NMLZ=COP.3+DECL ‘It can roll.’ • Aktionsart su#xes are required by most verbs in sequential subordination (§19.7), where the subordinated verb has a perfective interpretation. In (73), the verb nɨna ‘/y’ must occur with an aktionsart stem (its “preferred” aktionsart su#x is the intensive action -ka) to combine with the sequential di$erent subject -matai—̃note the verb tuku ‘shoot’ in the same example, which does not need an aktionsart: (73) t͡sukanka ́ nɨna-ka-matai ̃ iḱama tuku-ma-ji toucan /y-intens-1SG/3.DS Ikam shoot.PFV-REC.PT-3.PT+DECL ‘When the toucan /ew, Ikam shot it.’ Aktionsart su#xes are required by most verbs (i.e. they form an stem, which I will call the “aktionsart stem”) in the following morphosyntactic environments: - Just-done actions (§14.2.2.1) -Most past tenses (§14.2) -Future tenses (§14.2.3) -Most mood (§14.4) 616 -Sequential subordination (§19.7) From the above discussion, it seems that aktionsart su#xes have grammaticalized in the sense that they are required to "ll a morphological slot that is usually "lled by more aspectual-like su#xes (imperfective, durative) or the potential (which is not very aspectual-like but occupies the same position). Such grammaticalization process to occupy a morphological slot appears to be conditioned by the need of the language to distinguish between the aspectual-like nature of all the su#xes that occupy morphological position 3 in the verb structure. In turn, this process of grammaticalization has led to “semantic bleaching” of the aktionsart su#xes, which may explain why in current Wampis the “preferred” selection of an aktionsart by a certain verb root or “inner root” is not always semantically transparent. In fact, synchronically the selection of such or such aktionsart su#x by a verb may look rather ‘idiosyncratic’ in some cases. For instance, verbs of speech such as t͡ʃit͡ʃa ‘speak’, tu ‘say’ and uha ‘tell, inform’ have a di$erent “preferred” aktionsart su#x, as indicated in (74), but the semantic motivation for these di$erent selections is opaque: (74) Verb: “Preferred aktionsart”: t͡ʃit͡ʃa ‘speak’ -ka (Intensive) tu ‘say’ -i (Low a$ectedness of the Patient) inii ‘ask’ -sa (Attenuative) 617 Table 13.16 shows some examples of verbs with the “preferred” aktionsart su#xes with which they frequently co-occur. By “preferred” I mean aktionsart su#xes that have a very high frequency of co-occurrence with verbs in my text database. The aktionsart -ri (see Table 13.15) apparently is a recent con/ation of the applicative form -ru and the low a$ectedness aktionsart -i and it is not selected as ‘default’ by any verb in the database. The same happens with -u, which is unproductive synchronically: only a handful of examples occur in the data. Table 13.16. Sample of verb roots with their preferred aktionsart su#xes Verb Preferred Aktionsart ma ̃‘kill’, iu ‘eat’, t͡sanu ‘lie’, ta ‘arrive’, akinaa ‘be born’ -a(́u) ‘High a$ectedness’ tu ‘say’, ma ‘bathe’, ipanka ‘raise with hand’, ura ‘open (door, a book)’ -i ‘Low a$ectedness’ niha ‘wash, "sh with poison’, haa ‘tear’, hapi ‘drag, pull’, iikama ‘avenge’, paina ‘cook’ -ka ‘Intensive’ umu ‘drink’, karama ‘dream’, aiu ‘feed’, ikaa ‘make dry’ -ra ‘Distributed action’ hu ‘take’, mɨnanta ‘go away’, wakɨtu ‘return’, hina ‘exit’ -ki(ni) ‘Do while moving’ inii ‘ask’, auhumatu ‘tell’, ii ‘see’, tɨpɨ ‘lie down’, uara ‘enjoy’, aɨpa ‘put lie down sideways’ -sa ‘Attenuative’ 618 13.3.2.1. High a&ectedness aktionsart -a(́u) The high a$ectedness su#x -a(́u) indicates a high degree of a$ectedness (a change of state) of the patient of a verb or the change of location of the agent of an intransitive (typically motion) verb. The complete form -aú occurs in front of another vowel /a/. The verb tɨpɨ ‘lie down’ frequently appears with the attenuative -sa, where it conveys a more stative-like action: (75) nui ̃ ́tɨpɨśmaji nui ̃ tɨpɨ-sa-ma-ji there lie.down-ATT-REC.PT-3.PT+DECL ‘He lied down there.’ Compare with the next example, which includes a change of location. In this case, the high a$ectedness -a(́u) is used: (76) liḿanam tɨpɨaẃarmatɛi ̃ lima=nama tɨpɨ-aú-ara-matai ̃ Lima=LOC lie.down-HIAF-PL-1SG/3.DS ‘When we lay down on (i.e. arrived) Lima . . .’ The verb ‘kill’ uses the high a$ectedness su#x with great frequency, as in (77)– (78): (77) niɲa nuwɨñka mantuaẃaru nina nui=̃na=ka ma-̃tu-aú-ara-u 3.GEN woman\1PL/2PL/3.POSS=FOC kill-APPL-HIAF-PL-NMLZ ‘They were ones to kill his wife.’ 619 (78) jat͡ʃiń mantuaẃaruiti uun Piruchi ́jat͡ʃiń iat͡ʃi=na ma-̃tu-aú-ara-u=iti brother\1PL/2PL/3.POSS=ACC kill-APPL-HIAF-PL-NMLZ=COP.3+DECL uunta Pirut͡ʃi ́ jat͡ʃi=na elder Pirut͡ʃi\GEN brother.1PL/2PL/3.POSS=ACC ‘They killed his brother, elder Piruchi’s brother.’ The following example is one that shows the use of the high a$ectedness aktionsart su#x with the intransitive verb ta ‘arrived’. An aktionsart stem without any past tense marker assumes a just-completed action interpretation (§14.2.2.1). In this case, upon returning to his village, the speaker is telling about his experience in Pucallpa, a town that is very far from the Wampis villages (Pucallpa is located relatively close to the border with Brazil). The high a$ectedness is used to mark the change of location of the speaker (who has just returned from Pucallpa). (79) jaaḱtanam urúk puhúiñawa núna wɛińkʲan taah́ɛɛ hui ̃ ́ iaakata=nama uruka puhu-ina-u=a nu=na town=LOC how live-PL.IPFV-NMLZ=COP NON.VIS=ACC uaina-ka-nu ta-a-́ha-i hui ̃ see-INTENS-1SG.SS arrive-HIAF-1SG.SBJ+DECL here ‘Having seen how they live in the city, I just arrived here.’ 13.3.2.2. Low a&ectedness aktionsart -i The low a$ectedness su#x -i typically involves the expression of a sustained action where there is no salient change of state of the patient of a transitive verb or the 620 agent of an intransitive verb. In (80) there is a sustained action but no salient change of state. (80) hiiń umpui ̃ ́ hii=na umpu-i-i ̃ "re=ACC blow-LOAF-PFV.3 ‘He blew the "re.’ (i.e. to maintain the "re, so there is no change of state) Contrast the meaning of the low a$ectedness aktionsart su#x in the previous example with the next one. The verb root ‘blow’ can take the high a$ectedness aktionsart -a(́u), in which case it assumes the meaning of ‘play a wind instrument’. The derived stem does not take the low a$ectedness su#x, but the ‘distributed action’ -ra (81) (see below, §13.3.2.4), as playing an instrument typically refers a more complex action than just blowing the "re, as in the previous example. (81) puput͡ʃeń umpuaŕamahɛɛ puput͡ʃe=na umpua-́ra-ma-ha-i /ute=ACC blow.instrument-DISTR-REC.PT-2SG.SBJ-DECL ‘I played the /ute.’ By contrast, if we add the ‘low a$ectedness’ to the stem umpua ́the meaning is that someone blew a little air on the /ute, not that someone played it (this was an elicited example, the meaning resulting from adding the low a$ectedness su#x was 621 described as ‘odd’). (82) pupút͡ʃen umpuɛímhɛɛ puput͡ʃe=na umpua-́i-ma-ha-i /ute=acc blow.instrument-loaf-rec.pt-1sg.sbj-decl ‘I blew on the /ute.’ In the next example shows the low a$ectedness with the verb ‘bathe’, which is intransitive but in this case is causativized. ‘Bathe’ has the low a$ectedness su#x as its “preferred” aktionsart. (83) haḱɨ ɨt͡sɨśa imɛitʲaŕum hakɨ ɨt͡sɨ-sa i-ma-i-ta-́rumɨ warm.water heat-ATT CAUS-bathe-LOAF-IMP-2PL.SBJ ‘[With] warm water, cozy, give her a bath.’221 The low a$ectedness su#x may also be used to indicate the low topicality of a patient. For instance, as seen in the preceding subsection, the verb ma ̃‘kill’ frequently co-occurs with the high a$ectedness su#x -a(́u). However, in this text where one of my teachers explains how the Wampis go hunting, he uses the low a$ectedness su#x when the patient is ‘animals’. (By contrast, I have no examples of the low a$ectedness su#x when the patient role is human.) 221. The original Spanish translation of the verb ‘heat’ plus the attenuative is calientito, which I translate in the example as ‘cozy’. 622 (84) túra wií núnaka aʃiḱa mɛĩt́͡ʃuahɛɛ wií wariɲ́ak ma ́wakɨŕah núna tura ui nu=na=ka aʃi=ka ma-̃i-t͡ʃu-a-ha-i and 1SG NON.VIS=ACC=FOC all=FOC kill-LOAF-NEG.NMLZ-IPFV-1SG.SBJ-DECL ui wari=na=kɨ ma wakɨru-a-ha nu=na 1SG thing=ACC=RESTR INTERJ want-IPFV-1SG NON.VIS=ACC ‘And I do not kill all that [i.e. animals], I only [kill] things that I want.’ The previous example suggests that there might be more factors a$ecting the use of aktionsart morphology than just lexical aspect, such as contextual saliency, an animacy hierarchy or topicality. This issue is not fully explored yet. 13.3.2.3. Intensive aktionsart -ka Typically, the intensive -ka adds the idea of an action carried out with a relatively high degree of e$ort by the agent or, less commonly, an experiencer. The verb tukuma ‘kick’ when added the attenuative means that the kick is done with little e$ort. With the intensive -ka (which is the most frequent aktionsart “selected” by the verb tukuma), the verb acquires a sense of ‘kick "rmly’. This was explained to me in Spanish as the di$erence between a ‘little kick’ and a ‘strong kicking (as kicking the ball in soccer or someone in a "ght)’ (cf. the Spanish translation given to me: ‘patear con fuerza’). (85) puhupat́an tukúmsamji puhupata=na tukuma-sa-ma-ji Pujupat=ACC kick-ATT-REC.PT-3.PT+DECL ‘She gave Pujupat a little kick (i.e. joking or playing)’ 623 (86) puhupat́an tukúmkamji puhupata=na tukuma-ka-ma-ji Pujupat=ACC kick-INTENS-REC.PT-3.PT+DECL ‘She gave Pujupat a "rm kick.’ The verb ‘look for’ (also ‘look for game’, i.e. ‘hunt’) has the intensive -ka as its preferred aktionsart su#x: (87) waaḿak takaś nuŋkańʃa ɨaḱ nui ̃ ́kańamunmasha uaamakɨ taka-sa ̃ nunka=na=ʃa ɨa-ka ̃ quickly work-ATT\3.SS land=ACC=ADD look.for-INTENS\3.SS nui ̃ kańa-mau=numa=ʃa there sleep-NMLZ=LOC=ADD ‘He worked quickly, looking around for land, there, where to sleep.’ The verb root kahɨ ‘get/be angry’ usually takes the intensive -ka. If, instead, this verb occurs with the distributed action aktionsart -ra, it means ‘hate’. The “inner stem” kahɨra ‘hate’ can further take the intensive too: (88) ut͡ʃi ́kuitʲamkata ́kahɨŕkɛipʲarum ut͡ʃi ́ kuitama-ka-ta ́ kahɨra-ka-ai-pa-rumɨ child\ACC look.after-INTENS-IMP hate-INTENS-APPR-PROH-2PL.SBJ+DECL ‘Take care of that girl, do not hate her.’ 13.3.2.4. Distributed action aktionsart -ra The su#x -ra typically indicates an action that can be seen as complex, involving a process, a repetitive action or an action with more than one object 624 manipulated. The verb aɨpa means ‘put on the ground’ or ‘put on bed sideways’. It typically selects the attenuative -sa (89), but when it receives the distributed action su#x -ra, it means ‘put several (objects/people) on the ground or bed’. Example (90) was said by a speaker that was leaving his house (89) tikitt ͡ʃa aɨṕsamu jutúamu tikit ͡ʃi=ʃa aɨpa-sa-mau iutu-a-mau other=ADD put-ATT-NMLZ bury-IPFV-NMLZ ‘The other too, [he] put on the ground, buried.’ (90) aɨṕramoun ukúahai aɨpa-ra-mau=na uku-a-ha-i put-DISTR-NMLZ=ACC leave-IPFV-1SG.SBJ-DECL ‘I am leaving [the children]222 lying down.’ Other examples of the use of the su#x -ra are less transparent. For instance, unuima ‘learn’ typically takes the distributed action su#x -ra, as in (91), presumably because the action of learning is construed as a complex process (not punctual). (91) út͡ʃi unúimararti turaʃ́a nɨkaś kaḿɨ unúimarar kaḿɨ uúntriñ t͡ʃit͡ʃaḿrukti ut͡ʃi unuima-ra-ara-ti turaʃa nɨkas kamɨ unuima-ra-ara child learn-DISTR-PL-JUSS but truly INTERJ learn-DISTR-PL\3PL.SS 222. The only interpretation here is that there is more than one child, if the patient is singular, then the attenuative -sa is used. The speaker who said this utterance has three young children. 625 kamɨ uunta-ri=̃na t͡ʃit͡ʃamaru-ka-ti INTERJelder-1SG/3.poss=ACC defend-INTENS-JUSS ‘Let the children study, but, truly, having studied, that they defend their elders.’ Overall (2007) rightly notices that many actions involving liquids usually take the distributed action -ra (the Awajun cognate is -ha): umu ‘drink’, iutu ‘rain’, iniha ‘wet’. The same author mentions that since in Awajun there is a merger between Proto- Jivaroan *r and *h, it is probable that there were two aktionsart su#xes that have been neutralized (one for ‘liquid action’ and one for ‘complex actions/process’) (Overall 2007: 295). However, the Wampis data shows that there is only one su#x -ra for both ‘liquid action’ and ‘processes’. For instance, umu ‘drink’, iutu ‘rain’, ikihVma ‘wash one’s hands’ all select -ra as their default aktionsart su#x. There is still the possibility that there were two distinct su#xes (one for ‘liquid action’ and one for ‘complex actions/ process’), but it remains a possibility that is nearly impossible to prove, given that they would be homophonous. In any case, future comparative evidence may shed more light into this topic. 13.3.2.5. Attenuative aktionsart su0x -sa The attenuative denotes an action that is done without much e$ort or it decreases the ‘force’ of the verb, and in many cases it is the verbal equivalent to the diminutive that marks nouns and adjectives. The most common uses of the attenuative 626 is in imperative contexts, as has the pragmatic e!ect of reducing the force of a command, and with stative verbs (especially when the subject is an experiencer and with posture verbs). The next examples show the attenuative use in imperative clauses, where it serves to give a sense of a!ection223 to the command. Consider the di!erence between (92) and (93). In (93), the attenuative makes the command more subtle. (92) juat́a iu-a-́ta ́ eat-HIAF-IMP ‘eat!’ (93) jusat́a iu-sa-ta ́ eat-ATT-IMP ‘eat (please)!’ The attenuative -sa usually occurs with stative verbs. For instance, the stative verb puhu ‘live’ is frequently found carrying the attenuative as its “preferred” aktionsart, presumably because the action of living is construed as an action that is done without much e!ort or not involving salient changes on the participant: 223. Commonly translated as a command said ‘con cariño’ ‘with a!ection’, by my Wampis teachers. 627 (94) t͡ʃiŋkanaś nui ̃ ́puhusmiahɛɛ nukap t͡sawań t͡ʃiŋkanaśa nui ̃ puhu-sa-mia-ha-i nukapɨ t͡sauanta Chinganaza there live-ATT-DIST.PT-1SG.SBJ-DECL several day ‘Chinganaza, there I lived for a good time.’ Many verb roots that predicate a cause position (i.e. verbs of placement) usually take the attenuative -sa as their preferred aktionsart su#x, presumably because the result of the action is construed as stative: apuha-sa ‘place on surface’, aɨpɨ-sa ‘place on ground’ ‘place on bed sideways’, ɨkɨna-sa ‘place on "re (e.g. to cook)’, and so on. In addition, verbs that involve asking for something also frequently take the attenuative su#x -sa, presumably because they involve actions that pragmatically or socially may be construed as requiring a level of politeness: inii-sa ‘ask’, see also the periphrasis auakɨ ɨnɨntaima-sa (‘contrary think-Attenuative’)224 which means ‘di$er’, ‘think the contrary’. The attenuative is also used with other verbs where the “attenuative” sense is less transparent semantically. For instance, taka ‘work’ or auhu ‘study’ frequently “select” -sa. To study is maybe seen as a stative action, as opposed to ‘learning’ unuima which is more active and frequently uses distributive -ra. As was said above, in many cases the semantic ‘bleaching’ of the aktionsart su#xes makes their meanings di#cult to identify for all speci"c cases. 224. The verb auakɨ means ‘return’ literally. 628 13.3.2.6. Do action while moving -ki(ni) The su!x -ki(ni) belongs to a semantic domain known as ‘associated motion’ (Koch 1984; Guillaume 2006). -Ki indicates that the action is done while moving away from the point of reference. This is one of the most semantically coherent aktionsart su!xes, in the sense that it occurs with many verbs of motion and a!nes. By far, the most common occurrence of this su!x in narratives is with the verb hu ‘take’. (95) naŋki ́huki ́wɨḿiaji maat́sa ̃ nanki ́ hu-ki ̃ uɨ-mia-ji ma-̃a-́tasa ̃ spear\ACC take-WHILE.MOVING go-DIST.PT-3.PT+DECL kill-HIAF-purp\3.SS ‘Taking the spear, he went to kill.’ Notice the metaphorical use of the verb bearing -ki in the next example: (96) uunti ́nɨkat́ainkai núnia asań nunkaŋ́ka imatiḱan jukit́͡ʃmiahɛɛ uunta=ti ́ nɨka-tai=̃ka nuni-a a-sa-nu elder=SAP know-NMLZ=FOC do.that-IPFV COP-SUB-1SG.SS nunka=na=ka imatika-nu hu-ki-t͡ʃa-mia-ha-i land=ACC=FOC do.much-1SG.SS take-WHILE.MOVING-NEG-DIST.PT-1SG.SBJ-DECL ‘Because we the old people do not know, I did not take much land (i.e. ‘I didn’t buy much land’).’ The long form -kini surfaces before the 3 person perfective -i ̃(§14.3.2). 629 13.3.2.7. Do in proximity -ri The su#x -ri adds a sense of ‘do something in proximity to location at the point of reference’. This su#x seems to be a recent grammaticalization from the applicative form -ru plus the low a$ectedness -i. One of the functions of the applicative is to mark location or goal with intransitive verbs (§13.2.3). However, speakers seem to be re- analyzing -ri as distinct from the applicative, and they would often times correct me when I transcribed it as -ru-i (‘Applicative-Low a$ectedness’). The use of the low a$ectedness su#x in an associated motion su#x is a little unexpected, as low a$ectedness would presumably indicate rather that there is no change of state (i.e. no change of location); but this su#x is used with the idea of proximity to the point of reference, so one could claim that the change of location is not very salient. Example (97) shows an instance of the use of -ri with the verb hu ‘take’, which takes -ki ‘do while moving’ in the vast majority of cases, but here it receives -ri. Compare with (98), where -ki ‘do while moving’ implies that the action is moving away from the point of reference. (97) hurita ́ hu-ri-ta ́ take-DO.PROX-IMP ‘Come take him!’ (come to point of reference) 630 (98) hukita ́ hu-ki-ta ́ take-WHILE.MOVING-IMP ‘he took him (away from point of reference).’ 13.3.2.8. Do in another location -u The su#x -u is not very productive. It denotes an idea of action done somewhere else than the reference-point location, and occurs typically in the future. (99) wɨkóuthɛi wɨka-u-ta-ha-i hunt-DO.AWAY-IMM.FUT-1SG.SBJ-DECL ‘I will hunt (somewhere away from here).’225 13.3.3. Imperfective -a and plural imperfective -ina The imperfective -a is used to described an action with internal composition, i.e. viewing a situation as having “internal temporal structure” (Comrie 1976: 24). It usually marks progressive or habitual actions. Examples of the imperfective are presented in the next examples. (100) aḿiɲa jaaḱtarmiɲin puhaẃɛi iwiʃ́ɲuka amina jaakata-rumi=ni ̃ puha-ua-i iuiʃinu=ka 2SG.GEN town-2SG=LOC live+IPFV-3.SBJ+DECL shaman=FOC ‘The shaman lives in your village.’ 225. This example was uttered when the speaker was going o$ to hunt at Kampankis, a mountain range that is on the far east side of the Santiago river area, far from the village of Puerto Galilea where the example was collected 631 (101)amɨ júamɨ amɨ iu-a-mɨ 2SG eat-IPFV-2SG.SBJ+DECL ‘You are eating.’ There is a suppletive plural imperfective su#x -ina, illustrated in (102). (102) núka arútan iíkʲa tuiɲ́ahi nu=ka arutama=na ii=ka tu-ina-hi NON.VIS=FOC Arutama=ACC 1PL=FOC say-PL.IPFV-1PL.SBJ+DECL ‘That, we call Arutam.’226 The imperfective is used with the present tense, which is zero marked (recall that the other su#xes in position 3 cannot co-occur, so they cannot be used in present tense—except for the present habitual, which makes its own stem and does not co-occur with the imperfective). However, it is incompatible with future and past tense markers. Occasionally, the imperfective is used with a relative past tense interpretation that comes from the context. In the context of the next example, which comes from a mythological narrative, a woman arrives to the house of a man and stays there: (103)núka núwaka hɨã ̃ ́mat͡satɨáwai nu=ka nua=ka hɨã-a ̃́ matsatu-a-ua-i NON.VIS=FOC woman=FOC house-HIAF\3SS live.together-IPFV-3.SBJ-DECL ‘That woman arrived and was living there.’ 226. The pronunciation [arútan] is a haplology from arutama=na which is otherwise pronounced [arútman]. 632 13.3.4. The durative -ma The durative su#x -ma attaches to a verb root or “inner stem” and indicates an action of relative duration. The durative is more commonly used in imperative sentences. Notice that the durative usually lengthens the preceding vowel in the stem. A common example of the use of the durative is the typical Jivaroan formula to say good bye: (104) puhúumata puhu-ma-ta ́ live-DUR-IMP ‘Good-bye’ (Lit.: ‘Keep on living/being’.) Another example is given in (95): (105) muúkrun nant͡ʃiḱmaamata muuka-ru=na nantʃika-ma-ma-ta ́ head-1SG=ACC "ngernail-VBZ-DUR-IMP ‘Scratch my head (a little while more)!’ The durative, like the imperfective, may occasionally be used with a durative- past interpretation: (106) wií júmahɛi ui iu-ma-ha-i 1SG eat-dur-1SG.SBJ-DECL ‘I was eating for a while.’ Similarly to most su#xes in position 3 of the verb structure, the durative -ma has a verb-to-verb derivational property. Some examples that show this, for instance 633 consider: hapi ‘drag, pull’ → hapi-ma ‘sweep’, mit ͡suma ‘close one’s eyes’, mit͡suma-ma ‘be with eyes closed’, kahina ‘extinguish’, kahina-ma ‘forget’. There is a homophonous -ma recent past tense su#x (see §14.2.2.2). However, the durative and the recent past are morphosyntactically di$erent morphemes: the recent past, unlike the durative, can only occur on a stem already marked for aktionsart as in (107). The durative, as explained earlier, is also incompatible with aktionsart su#xes. (107) wií juaḿhɛi ui iu-a-́ma-ha-i 1SG eat-HIAF-REC.PT-1SG.SBJ-DECL ‘I ate (recently).’ 13.3.5. The present habitual -na The su#x -na provides a sense of habitualness in the present. It is not very productive as speakers seem to simply prefer to use the imperfective -a instead. The following examples illustrate the use of -na ‘present habitual’. (108) ahańamia peke-peke iíɲahɛi aha=nama=ia peque_peque ii-na-ha-i farm=LOC=ABL motor_boat see-PRES.HAB-1SG.SBJ-DECL ‘I usually see motor-boats from the farm.’227 227. The speaker’s farm plot is in an island in the middle of the Santiago river. 634 (109) ut͡ʃiń aɨṕnamɨ ut͡ʃi=na aɨpɨ-na-mɨ child=ACC lie.down-PRES.HAB-2SG.SBJ+DECL ‘You usually put the child on bed.’ 13.3.6. Potential -mai The potential morpheme -mai adds a sense of possibility, ability or capacity to the verb. In other languages, the semantic domain of possibility overlaps with the domain of permission (Bybee, Perkins, & Pagliuca 1994; Van der Auwera & Plungian 1998); however, I have no examples of permission with the potential in Wampis. Instead, for the expression of permission, the jussive is used (see §14.4.3.3). Examples of the potential morphemes are given in (110) and (111). (110) núka amɛít͡suhi nu=ka a-mai-t ͡su-hi that=FOC COP-POT-NEG-1PL.SBJ+DECL ‘We cannot be that.’ (111) turaʃ́a nui ̃ ́núwa nahaańaru tiíahu arúmmɛitsui turaʃa nui ̃ nua nahaana-ra-u but there woman make-DISTR-NMLZ tií=a=hu a-ru-ma-mai-t ͡su-i ̃ ints=COP=PROX exist-APPL-REFL-POT-NEG-1SG/3.ds ‘But, there, a woman transformed so big [i.e. transformed into a mountain] cannot exist.’ 635 Verbs carrying the potential su#x typically appear nominalized with the Set I nominalizer -inu or with nominalizer -na (nominalizers are explained in Chapter XV). The use of one or another su#x seems not to have any consequence for the interpretation of the verb marked for potential. Examples (112)–(113) illustrate this: (112) wari ́tumɛínt͡sukɛit wari ̃ tu-mai-na=t͡su-ka=iti what say-POT-NMLZ-INFER-Q=COP.3 ‘How can it be said?’ (113) inðih́enas timoutiʃ́a t͡ʃit͡ʃaḿrumakur wɨákur arańtukmoo amɛíɲaithi indihena ti-mau=ti=́ʃa t͡ʃit͡ʃamaru-ma-a-ku-ri indigenous say+LOAF-NMLZ=SAP=ADD defend-REFL-IPFV-SIM-1PL.SS wɨ-a-ku-ri arantu-ka-mau a-mai-na=ita-hi go-IPFV-SIM-1PL.SS respect-INTENS-NMLZ COP-POT-NMLZ=COP-1PL.SBJ+DECL ‘We so-called indigenous people, defending ourselves, we can be respected.’ 13.3.7. Verb stems: “Aspect” From the discussion in the preceding subsections, it can be established at this point that in Wampis there are up to "ve verbal stem types in terms of Aspect (and Mood with the potential). Those stems correspond to the verb marked with any of the su#xes listed in Table 13.17. 636 Table 13.17. Aspectual verb stems in Wampis “Aspectual” Stem Verb marked with Stem used with Aktionsart (Perfective) Aktionsart su#xes (most verbs—a few verbs do not select a “default” aktionsart su#x) Most past tenses and future tenses, imperative, jussive, hortative, apprehensive, prohibitive Imperfective -a Imperfective, present tense Durative -ma Durative, imperative Present Habitual -na Present habitual Potential -mai Potential 13.4. Negation There are two negative su#xes on the verb in Wampis: -t͡ʃa and -t͡su. They are in complementary distribution: a) -t͡su occurs in the following morphosyntactically conditioned contexts: • with the imperfective stem • with the existential verb a (§17.5) b) -t͡ʃa occurs everywhere else, including with the copular verb a (homophonous but analyzed as morphosyntactically distinct from the existential verb a, (§17.5.2.3). The examples below illustrate the di$erent contexts where the negative -t͡su 637 occurs. Examples (114)–(115) are instances of -t͡su with the imperfective228 and with the existential (116). (114) amɨḱa jukúmat nɨkat́͡smɨ amɨ=ka iukuma-ta nɨka-t͡su-mɨ 2SG=FOC swim-NMLZ know+IPFV-NEG-2SG.SBJ+DECL ‘You don’t know how to swim.’ (115)haańt͡ʃin nihat́suai haant͡ʃi=na niha-t ͡su-u=ai cloth=ACC wash+IPFV-NEG-NMLZ=COP.3+DECL ‘She/he doesn’t wash the clothes.’ (116)jaúnt͡ʃukka núwɨka atsú hakú iant͡ʃukɨ=ka nuɨka a-t͡su hak-u long.ago=FOC clay exist-NEG HAB.PT-NMLZ ‘Long ago, the clay did not exist.’ Examples of the use of the negative su#x -tʃa are given with non-imperfective stem (117) and the copula a (118). (117) awarún hat͡ʃaḿiaji Auaruna ha-t͡ʃa-mia-ji Awajun die-NEG-DIST.PT-3.PT+DECL ‘The Awajun did not die.’ 228. The present tense is zero-marked. 638 (118) wiʃiḱramu at͡ʃaḿi wiʃi-ki-ru-a-mau a-t͡ʃa-mi laugh-TR-APPL-IPFV-NMLZ COP-NEG-HORT ‘Let us not be laughed at.’ 13.5. Non-imperfective plural -ara Plural marking of 3rd person A/S arguments is indicated by the su#x -ara with non-imperfective stems. Notice that the morphological position of the non-imperfective plural -ara (in slot 5) di$ers from that of the plural imperfective -ina (in slot 3). Example (119) shows the use of the non-imperfective plural -ara with an aktionsart stem. Notice that the position of the aktionsart a#x is also the position where the imperfective -ina would occur in imperfective clauses, since they occupy the same morphological slot. This shows that the plural -ara occupies a di$erent position than -ina. (119) nit́a jawaáñ wɛiŋ́karmaji nita iauaa=̃na uaina-ka-ara-ma-ji 3PL dog=ACC see-INTENS-PL-REC.PT-3.PT+DECL ‘They saw a dog.’ There are some examples in the data, particularly with the hortative -mi, that show that the plural -ara can be optionally used with 1 plural persons. In this case, the su#x only marks ‘plural’, as hortative is always semantically understood as involving a 1 plural participant. It seems thus that -ara really only marks plural rather than 3 639 plural. Below, both (120) and (121) mean the same, but in (121) we have an example of the optional use of the plural -ara with the hortative. Notice that the vowel /u/ of the high a$ectedness aktionsart su#x -a(́u) surfaces when a vowel /a/ (in this case the /a/ of -ara) follows it. (120) juaḿi iú-a-́mi eat-HIAF-HORT ‘Let’s eat.’ (121) juaẃarmi iú-au-ara-mi eat-HIAF-PL-HORT ‘Let’s eat.’ Plurality for Speech Act Participants A/S arguments (with the exception just shown above) is marked with a portmanteau su#x in the person/subject morphological position (described in §14.3.3). 640 CHAPTER XIV THE VERB II: "INFLECTIONAL" LEVEL 14.1. Introduction As was seen in the previous chapters, there is what I call a derivational level and an in!ectional level in the verb structure. The "rst level, derivational, was described in Chapter XIII. The second level, in!ectional, is the topic of this chapter. The categories marked on the verb that I consider to be in the in!ectional level are: • A. Tense • B. Person • C. Mood Before going into the description of these categories, it is perhaps convenient to remember that the Wampis verb structure consists of several morphological slots, as presented previously in Chapter XII (cf. Table 12.5). What I refer to as a more “in!ectional level” are the categories marked in positions 6–8 of the verb structure in Table 12.5. Recall from the preceding chapters that, in general, verbal categories are not 641 neatly distributed in the Wampis verb structure. The category called “Potential” in Table 12.5, for instance, can be considered to be some kind of “mood”; however, it is marked with the su#x -mai in position 3 in the verb structure, the same position used for aspectual su#xes (cf. §12.7 for a description of the verb structure in Wampis). I describe in this chapter a hierarchical agreement system in Wampis where the morphological position 7, which is usually accessed to by the subject of the clause, is sometimes accessed by the object according to a hierarchy that is explained in §14.3.3. In addition, some mood markers occur at the end of the verb piece (for instance, the declarative), and some others occur before the person markers (for instance, the imperative). Thus, most of this chapter is functionally-driven according to the categories mentioned above (Tense, Person and Mood), as it would be very di#cult to follow a simple positional analysis for this verbal categories in a language like Wampis, which exhibits a particularly complex verbal morphology. The structure of this chapters is: §14.2 discusses the category of tense; §14.3 discusses the category that I call ‘Person’ and §14.4 discusses the category that I call ‘Mood’. 642 14.2. Tense The rich tense system of Wampis makes several distinctions in past and future; present tense is zero-marked. Table 14.1 shows the morphemes that mark di$erent tense distinctions in Wampis. Table 14.1. Wampis tense distinctions Morpheme Gloss -tata De"nite future -ta Immediate future ∅ Present -ma Recent past -ɨmia Intermediate past -mia Distant past -ia Remote past Aktionsart stem+person+mood Just-done actions =hak-u Habitual past Generally, tense markers are su#xal and occur in independent verbs. The past habitual =hak is morphologically di$erent as it can occur attached to the verb or not, forming its own prosodic word (§14.2.2.6), and it always occurs nominalized with Set II nominalizer -u (cf. §15.4.5). In addition, there is a perfective construction done with the 643 aktionsart stem: when the aktionsart stem is not marked by any tense, it assumes an interpretation of “just done” action. The “just-done actions” and past habitual constructions are also analyzed in this section, because, although they are morphologically di$erent, they are part of the tense system of the language. 14.2.1. Present tense Present tense is zero-marked on the verb. The use of the imperfective -a together with the present tense typically has a progressive reading, unless the context invokes a relative past tense interpretation of the imperfective. (1) wikʲa júahɛɛ ui=ka iu-a-ha-i 1SG=FOC eat-IPFV-1SG.SBJ-DECL ‘I am eating.’ Recall from §13.3.3 that there is a suppletive imperfective plural form -ina for 3 person: (2) nit́ʲa wɨńawɛɛ nita uɨ-́ina-ua-i 3PL go-PL.IPFV-3.PT-DECL ‘They are going.’ 14.2.2. Past tense There are several distinctions in the past in the Wampis language. Historically, it seems pretty straightforward that at the genesis of the current Wampis past tense 644 system there were two morphemes *ma and *ia. Synchronically, these forms correspond to the recent and remote past, respectively. They occur as formatives in the other past tense markers. The past tense system also has the habitual past =hak §14.2.2.6, but this morpheme shows di$erent morphosyntactic behavior and does not belong to the same paradigm of tense markers. The temporal interpretation of past tenses is summed up in Table 14.2. Table 14.2. Temporal interpretation of past tenses Morpheme Gloss Temporal interpretation with regard to moment of speech aktionsart stem+person+mood Just-done actions Immediately before -ma Recent past Hours to days/weeks -ɨmia Intermediate past Days to weeks -mia Distant past Months to years -ia Remote past Years =hak Habitual past (Typically) years Morphologically, most but not all past tense markers occur on an aktionsart stem. Table 14.3 summarizes the combinatorial properties of past tense su#xes with di$erent verb-stem types. 645 Table 14.3. Combinatorial properties of past tense su#xes Morpheme Gloss Received by ∅ Just-done actions aktionsart -ma Recent past aktionsart -ɨmia Intermediate past variesa -mia Distant past aktionsart -ia Remote past root =hak Habitual past root a The intermediate past may occur on bare roots as well as on aktionsart stems. 14.2.2.1. Just-done actions Actions that "nished just a moment prior to the moment of speech are zero- marked for tense; but unlike a present tense verb form, the verb must occur with an aktionsart su#x in a perfective function. In other words, the default interpretation of perfective verbs (when there is no overt tense marker) is that of ‘a-just-done’ action. For 3 person subjects (example (4)), there is a form that marks ‘3.PFV’ subject (cf. §14.3.2) (3) taah́ɛɛ ta-a-́ha-i arrive-HIAF-1SG.SBJ-DECL ‘I just arrived.’ (4) juaẃarai ̃ iu-aú-ara-i ̃ eat-HIAF-PL-3.PFV+DECL ‘They just ate.’ 646 14.2.2.2. Recent past The recent past -ma marks actions that have been done roughly between moments before the speech act and a few days. (5) kaʃiḱ ɛiʃ́ur miɲ́a ut͡ʃiŕ ɨɰaḱmaji kaʃi=kɨ aiʃi-ru mina ut͡ʃirún night=RESTR husband-1SG 1SG.GEN child-1SG=ACC ɨa-ka-ma-ji look.for-INTENS-REC.PT-3.PT+DECL ‘My husband looked for my son last night.’ The next example comes from a story in which a man wanted to know what was happening to the squash that he and his wife had planted, as every time he wanted to eat the squash he could not "nd a ripe one. He suspects his wife is eating the good squashes, so he goes to check his farm. After the man leaves, his wife speaks, using the recent past to refer to what had just happened (i.e. his husband leaving, saying that he was going to check the farm): (6) aúhuʃa ɛiʃ́ur wɨḿaji júwi kaḿɨ t͡samaḱmaji tiḿaji núna a. Aúhu=ʃa aiʃi-ru wɨ-ma-ji great.potoo=ADD husband-1SG go-REC.PT-3.PT+DECL b. iuui kamɨ t͡sama-ka-ma-ji squash INTERJ mature-INTENS-REC.PT-3.PT+DECL 647 c. ti-ma-ji nu=na say+LOAF-REC.PT-3.PT+DECL NON.VIS=ACC ‘Auju too: “my husband has gone,/ he said “the squash have become ripe”, that”.’ 14.2.2.3. Intermediate past The intermediate past -ɨmia indicates a time that goes farther in the past than the recent past. This su#x is not very commonly found in narratives and apparently is not productive in the language overall. In general, speakers prefer to use the recent past (instead of the intermediate past) for actions that are more relevant to the present and are not very distant in the past. Otherwise, the distant past is used. Most of the examples that I have of the intermediate past su#x come from elicitation tasks. An example from a conversation is given in (7), occurring in the second-to-last line. Notice that, in the example, the intermediate past is used for information that can be seen as “secondary” or anecdotal information, as speaker A is more interested in establishing what they will do when the person “At.” arrives: (7) A: wiḱʲa At. wakaḿtɛɛ ̃wɨt́ahɛɛ ui=ka At. ua-ka-matai ̃ ui-́ta-ha-i 1SG=FOC At. come-INTENS-1SG/3.DS go-IMM.FUT-1SG.SBJ-DECL ‘When At. comes, I’ll go.’ 648 B: núka At. miɲ́a unt͡súrnayi haʔá nu=ka At. mi=na unt͡su-ru-na-ji haʔa ́ NON.VIS=FOC At. 1SG=ACC call-1SG.OB-PRES.HAB-3.PT+DECL yes ‘That At., he always calls me, yes.’ A: aj́u aj́u okay, okay B: (laughs) At. nampɨḱu, M. wɛińaɨmiaji nampɨḱu A. nampɨku M. uaina-ɨmia-ji nampɨku A. drunk M. see-INTERM.PT-3.PT+DECL drunk ‘(laughs) At. was drunk, M. saw him.’ A: (laughs) wakaḿtɛɛ ̃warí miɲ́a uhaktá wa-ka-matai ̃ wari ́ mi=na uha-ka-ta ́ come-INTENS-1SG/3.DS quick 1SG=ACC inform-INTENS-IMP ‘(laughs) When he comes, tell me quickly.’ The su#x -ɨmia is actually decomposable into -ɨ and -mia:229 when the non- imperfective plural -ara is used for third person, the intermediate past tense su#x is broken into two discontinuous formatives by -ara: (8) nit́ʲa hapóunam wɨkaɨ́rmiaji nita hapau=nama wɨka-ɨ-ara-mia-ji 3PL cli$=LOC walk-INTERM.PT1-PL-INTERM.PT2-3.PT+DECL ‘They walked in the cli$ (not so long ago).’ For comparative purposes, it is interesting to note that in Awajun the distant past-amia is similar in form to the Wampis intermediate past -ɨmia. Overall (2007: 342) 229. And, in turn, -mia is historically analyzable as -ma (from the recent past) and -ia (from the remote past). 649 has hypothesized that the Awajun distant past -amia may be a grammaticalization of a complex construction involving the copula ‘be’ a marked with the past tense -mia. For Wampis, however, it is di#cult to relate the form ɨ to the copula a. 14.2.2.4. Distant past The distant past has a temporal interpretation that typically includes several months to years prior to the moment of speech. The distant past is very common in narratives, frequently used for backgrounded information, i.e. supportive information that do not advance the /ow of the discourse (Longacre 1976). In (9), a speaker tells a story that was similar to a case of “collective psychosis” which happened in a Wampis village several years ago (during the 2000s). Note the use of the distant past on the verb ‘go’, which in this case functions as an auxiliary: (9) ʃuara ́karaḿɛiɲak núniak woórkaru ɛɛ́ɲawɛɛ túmak jurúmkanʃa júu wɨt́͡ʃarmiaji ʃuara ́ karama-ina-kũ nuni-a-kũ person\ACC dream-PL.IPFV-SIM do.that-IPFV-SIM\3SG.SS uaura-ka-ara-u a-ina-ua-i tuma-a-kũ become.crazy-INTENS-PL-NMLZ COP-IPFV-3.SBJ-DECL do.thus-IPFV-SIM\3SG.SS iurumaka=na=ʃa iu wɨ-́t͡ʃa-ara-mia-ji food=ACC=ADd eat go-NEG-PL-DIST.PT-3.PT+DECL ‘They dreamed of the person, doing that, they became crazy, doing thus, they didn’t eat the food.’230 230. The phrase ‘they dreamed of the person’ is a literal translation. Culturally, when a shaman 650 14.2.2.5. Remote past The remote past tense -ia refers to a time that goes back many years ago, including the mythical past. The remote past typically attaches to the verb root. (10) aánmamtin míɲa nukút͡ʃruka tújaji turáʃa ʃiír anɨát͡shɛɛ aanu-mamtin mina nukut͡ʃi-ru=ka tu-ia-ji MED-SIMIL 1SG.GEN grandmother-1SG=FOC say-REM.PT-3.PT turaʃa ʃiira anɨ-a-tsu-ha-i but well remember-IPFV-NEG-1SG.SBJ-DECL ‘My grandmother said something like that, but I don’t remember well.’ The remote past tense of the 3 plural person is not done with the su#x -ia, but rather with a form whose apparent source is the recent past -ma plus the 3 person past marker -ji (§14.3.2): ma-ji. A paradigm with the remote past is given in Table 14.4 with the copula a. Table 14.4. Paradigm of conjugation with the remote past tense SG PL 1 a-ia-ha-i COP-REM.PT-1SG.SBJ-DECL a-ia-hi COP-REM.PT-1PL.SBJ-DECL 2 a-ia-mɨ COP-REM.PT-2SG.SBJ-DECL a-ia-rumɨ COP-REM.PT-2PL.SBJ+DECL 3 a-ia-ji COP-REM.PT-3.PT+DECL a-ara-maji COP-PL-3.REM.PT+DECL bewitches other people, he can do it through dreams. 651 14.2.2.6. Habitual past The habitual past =hak, as the name indicates, refers to habitual actions that are understood as occurring ordinarily in the past, typically many years ago from the moment of speech. The habitual past is most often used to set the scene and provide background information in narratives. For instance, the following example contains the "rst lines (after an introductory line) of the myth of Auju (Great Potoo) and Nantu (Moon), which explains how the Wampis obtained clay. We can see the used of the habitual past in the "rst and third lines: (11) jóunt͡ʃukka núwɨka atsú hakú tiḿaji hu óohmattɛinkʲa aúhu tútañɛiti jóunt͡ʃukka aúhu ʃuaŕ ahakú tiḿaji a. iaunt͡ʃukɨ=ka nuɨ=ka a-t͡su a=hak-u timaji long.ago=FOC clay=FOC COP-NEG exist=HAB.PT-NMLZ NARR ‘It is said that the clay did not exist long ago’ b. hu auhumatu-tai=̃ka auhu tu-tai=̃aiti PROX tell-NMLZ=FOC Common.Potoo say-NMLZ=COP.3+DECL ‘This story is called Auju (Common Potoo)’ c. iaunt͡ʃukɨ=ka auhu ʃuara a=hak-u timaji long.ago=FOC Common.Potoo person COP=HAB.PT-NMLZ NARR ‘It is said that Auju used to be a person long ago.’ Phonologically,=hak can occur attached to a verb root or it can occur forming its own prosodic word, separated from the verb. Morphologically, the verb that bears =hak is always nominalized with the subject nominalizer -u or the non-subject 652 nominalizer -mau (see §15.5, §15.6 respectively). Thus, the narrative modality marker timaji as in (11), or a copula as in (12) are usually needed to turn the nominalized form into a "nite expression. (12) miɲ́a nukút͡ʃruka túhakuiti arútmaka tikiʃ́kit ͡ʃuiti túhakuiti arútmaka núkap mina nukut͡ʃi-ru=ka tu-hak-u=iti 1SG.GEN grandmother-1SG=FOC say-HAB.PT-NMLZ=COP.3+DECL arutama=ka tikit ͡ʃiki-t͡ʃau=iti tu-hak-u=iti vision.power=FOC one-NEG.NMLZ=COP.3+DECL say-HAB.PT-NMLZ=COP.3+DECL arutama=ka nukapɨ vision.power=FOC many ‘My grandmother used to say “Arutam is not one”, she used to say,/ “there are several Arutam”.’ From the examples above, one can posit that the habitual past construction is a grammaticalization of a complex structure that involved a nominalization plus a copula. The lexical source of hak is most likely the posture verb uaha ‘stand’ plus the intensive aktionsart su#x -ka. Thus, the original construction can be hypothesize as [standverb- NMLZ COPULA] with the semantics of ‘one who stands doing X’. The fact that =hak occurs nominalized and also that it can still form its own prosodic word points to a relatively recent grammaticalization process, as it still does not completely behave like other verbal tense su#xes. Morphologically, unlike other past 653 markers, =hak attaches to the root or “inner root” (all other past tenses, except the remote past, attach to an aktionsart/perfective stem). 14.2.3. Future tense There are two morphological future tenses in Wampis: -ta ‘immediate future’ and -tata ‘de"nite future’. Verbs in any of the future tenses always occur with an aktionsart stem. 14.2.3.1. Immediate future The immediate future is marked with -ta. As the name suggests, -ta is used for actions that are to be carried out in the near future. The following examples show the use of this su#x. The passage in (13) comes from a narrative in which the person decides to sell her produce after seeing that others are doing well selling the fruits and plants that they have gathered: (13) wiʃ́a surúktahɛi ui=ʃa suru-ka-ta-ha-i 1SG=ADD sell-INTENS-IMM.FUT-1SG.SBJ-DECL ‘I am also going to sell (fruits).’ 654 And the following question implies an immediate answer: (14) wari ̃ ́titʲaḿɨk? wari ̃ ti-ta-mɨ-ka what say+LOAF-IMM.FUT-2SG.SBJ-Q ‘What are you going to say?’ The following sentence was said when the person who speaks was leaving immediately toward her farm to bring sugar cane: (15) paat́an huúkan utit́hʲɛi paata=na huu-ka-nu uti-ta-ha-i sugar.cane=ACC gather-INTENS-1SG.SS bring-IMM.FUT-1SG.SBJ-DECL ‘I’m going to bring sugar cane.’ (Lit.: ‘Having gathered sugar cane, I am going to bring.’) 14.2.3.2. De!nite future The de"nite future -tata conveys the idea that an action that will happen within or at a more or less speci"c time frame. There is typically an overt indication of when the event predicated will take place. For instance, many narratives start as in (16), with an adverbial indication of time (‘now’) and then the use of the de"nite future on the verb: (16) aj́u húuʔ t͡sawańtahui ̃óohmatsattahɛi ɨtsań pat͡ʃiśan aju hu t͡sauanta=a=hui ̃ auhumatu-sa-tata-ha-i okay PROX day=COP=here tell-ATT-DEF.FUT-1SG.SBJ-DECL ɨt͡sa=̃na pat͡ʃi-sa-nu Sun=ACC mention-ATT-1SG.SS ‘Okay, in this moment right here, I am going to tell about [the story of] Sun.’ (Lit.: ‘I am going to tell, mentioning Sun.)’ 655 In the next example, a war leader speaks to a man whose brother had been killed. In the Jivaroan culture, a man had to avenge the killing of a relative. Thus the use of the de"nite future is referring here to something that must happen: (17) aḿɨka uútipa uutɨámɨka iíkmakt͡ʃattamɨ jat͡súmi mantútramawaranuka amɨ=ka uuta-i-pa uutɨ-a-mɨ-ka 2SG=FOC cry-APPR-PROH cry.IPFV-IPFV-2SG.SBJ-COND iíka-ma-ka-t͡ʃa-tata-mɨ avenge-REFL-INTENS-NEG-DEF.FUT-2SG.SBJ+DECL iat͡su-mi ma-̃tu-turama-aú-ara=a nu=ka brother-2SG kill-APPL-1SG.OBJ-HIAF-PL=COP NON.VIS=FOC ‘Don’t cry, if you cry, you won’t take revenge on those who killed your brother.’ Another example is given in (18). Again, notice the overt temporal indication of when the action will happen (‘tomorrow’). (18) kaʃińkɨʃa wií t͡ʃit͡ʃaḿan umit́tahɛɛ kaʃini=kɨ=ʃa ui t͡ʃit͡ʃama=na umi-tata-ha-i tomorrow=RESTR=ADD problem=ACC "nish-DEF.FUT-1SG.SBJ-DECL ‘Tomorrow I am going to "x the problem.’ 14.3. Person The morphological position that I call ‘Person’ is where usually subjects are marked, however, it will seen that Wampis exhibits a rather complex way of indexing arguments on the verb structure. First, I list the su#xes used for 1, 2 and 3 persons S/A arguments in §14.3.1–§14.3.2. Discussion of the uncommon patterns of argument 656 indexation found in Wampis is provided in §14.3.3. 14.3.1. 1 and 2 person S/A S/A arguments for the 1 and 2 persons are marked with the su#xes listed in Table 14.5. In transitive clauses, these su#xes are used for 1 and 2 A argument actin onto a 3 P argument. Table 14.5. Marking of SAP S and SAP A acting on 3 person Singular Plural 1 -ha -hi 2 -mɨ -rumɨ Table 14.6 shows the person su#xes that index a 1 A argument and 2 P argument (i.e. when a 1 person acts on a 2 person). Table 14.6. Marking of 1A acting on 2P 2SG 2PL 1SG hamɨ harumɨ 1PL himɨ hirumɨ 657 In transitive clauses, the su#xes listed in Table 14.5 are used when a 1A or 2A argument acts upon a 3P argument as in (19)–(20), and when a 2A argument acts upon a 1P argument (21). Once again, remember that when a 1A argument acts upon a 2P argument, there is a portmanteau su#x that refers to both arguments of the transitive clause (22). (19) wií t͡sanúahai ui t͡sanu-a-ha-i 1SG lie-IPFV-1SG.SBJ-DECL ‘I am lying to her/him.’ (20) aḿɨ t͡sanúamɨ amɨ t͡sanu-a-mɨ 2SG lie-IPFV-2SG.SBJ+DECL ‘You are lying to her/him.’ (21) aḿɨ t͡sanúrɨamɨ amɨ t͡sanu-ru-a-mɨ 2SG lie-1SG.OBJ-IPFV-2SG.SBJ+DECL ‘You are lying to me.’ (22) tah́amɨ ta-hamɨ say+IPFV-1>2+DECL ‘I am telling you.’ From the examples above, it can be said the local domain (sap → sap) is split and shows di$erent marking depending on which grammatical person is the A 658 argument; whereas in the direct domain (sap → 3) the A argument is always marked in the same position (position 7 in the verb structure). 14.3.2. Third person S/A marking The marking of third person subject distinguishes between non-past and past, as Table 14.7 shows. These su#xes are used in declarative, interrogatives and exclamative sentences. The jussive -ti covers third person. The jussive is explained in §14.4.3.3. In contexts di$erent than declarative, interrogative, exclamative and jussive, third person is zero-marked. Table 14.7. 3 person subject su#xes Morpheme Occurs in -i ̃~ -ɨ ̃ ‘3 perfective’ Just-done actions -u(a) ‘3 (non-past)’ Future -u(a) ‘3 (non-past)’ Imperfective & Present tense -ji ‘3 past’ Past tense -ti ‘Jussive’ Jussive The non-past form -u(a) is used in the present (with the imperfective) and future tenses. In addition, I have some examples of a form -i~̃-ɨ ̃whose use seems to be 659 declining used for just-done actions (the 3 past su#x -ji can also occur in those instances). The actual pronunciation of -i~̃-ɨ ̃this form is quite variable, and most often surfaces as an oral vowel. As the su#x always occurs at the end of the word, it is di#cult to say whether it is underlyingly nasal or not (other nasal su#xes usually add a nasal consonant when other morphemes are added). Given that some speakers still pronounce it as a nasal, I assume that it was at least historically nasal, as some basic comparative evidence would suggest: in the related language Awajun, where it is apparently productive, the cognate su#x has the nasal form -ɨ ̃(Corbera Mori 1994; Overall 2007). The following examples illustrate the use of u(a). In (23), the su#x -u(a) attaches to the an imperfective stem. The same su#x occurs with the form -u in a future context with he semi-copula ‘become’ in (24). (23) miɲ́a ɛiʃ́ru ukúrɨawɛɛ mina aiʃi-ru uku-ru-a-ua-i 1SG.GEN husband-1SG leave-1SG.OBJ-IPFV-3.PT-DECL ‘My husband is leaving me.’ (24) húu unt͡súktatui aḿɨ wari ̃ ́wakɨŕam núnaka hu unt͡su-ka-tata-u-i DIST call-INTENS-DEF.FUT-3.PT-DECL 660 amɨ uari ̃ wakɨra-amɨ nu=na=ka 2SG what want+IPFV-2SG.SBJ+DECL NON.VIS=ACC=FOC ‘This (child) is going to call whatever you want, that.’ Contrast the previous examples with (25), in which the su#x -i ̃‘3.PFV’ is used for a perfective action in combination with the aktionsart stem (with the high a$ectedness su#x -a(́u)). (25) juaẃarai ̃ iu-aú-ara-i ̃ eat-HIAF-PL-3.PFV+DECL ‘They just ate.’ The other third person subject su#x, the su#x -ji, is used for all past tenses and fuses with the declarative su#x -i. Examples of the su#x -ji are given below with just- done actions (26), recent past (27) and distant past (28), respectively. (26) hakaj́i ha-ka-ji die-INTENS-3.PT+DECL ‘He just died.’ (27) ɨnt͡sa ́wɨḿaji ɨnt͡sa ́ wɨ-́ma-ji river\LOC go-REC.PT-3.PT+DECL ‘She went to the river (not long ago).’ (28) ajaḿrumaka pujústin t͡ʃit͡ʃaḿ aḿiayi lei ðe las natiβas inðihenas tamóo aiamaru-ma-ka puhu-sa-́tinu t͡ʃit͡ʃama a-mia-ji defend-REFL-INTENS live-ATT-FUT.NMLZ law exist-DIST.PT-3.PT+DECL 661 lei_de_las_comunidades_indiǵenas ta-mau Law_of_Indigenous_Communities say+IPFV-NMLZ ‘Living like that, there was a law to live defending ourselves, called Law of Indigenous Communities.’231 14.3.3. The typologically uncommon argument indexation pattern of Wampis Wampis has two morphological positions where arguments are indexed: position 2, ‘Object’, and position 7 ‘Person’ (cf. ?). Position 2 is where objects are usually indexed, and position 7 is where subjects are usually indexed. However, there are some patterns pertaining to the indexation of arguments on the verb that make the Wampis system a complex one. In transitive clauses, the Wampis system follows a typologically uncommon pattern of hierarchical indexation. In the Wampis transitive indexation pattern, in at least two cases, access to the morphological position of subject on the verb is determined not by the grammatical role of the arguments, but by their relative position on a person hierarchy. This is usually understood as a hierarchical agreement system (Silverstein 1976; DeLancey 1981; Zuñiga 2006; DeLancey Forthcoming), but the Wampis system is not a prototypical hierarchical agreement system: there is a split between singular and plural speech-act participants (SAPs), and only plural SAPs 231. The speaker is making reference to a Peruvian legislation commonly known as Ley de Communidades Nativas (Law of Indigenous Communities), which was passed in the 70’s. 662 objects access the position that is usually reserved for the subject, which is position 7. I call this position ‘Person’ precisely because even though it is the designated position of subjects, there are some exceptions whereby the A argument is not indexed—but there is always an indexation of an argument in this position. In an inverse domain, hierarchical agreement only occurs in a scenario where third person acts on a plural speech-act participant (3 → Plural.SAP):232 in this case, the plural SAP (the P argument) is indexed in the morphological position 7, designated by what I have called ‘Person’ (cf. Table 12.5). By contrast, in all other situations the verb always indexes the grammatical subject (the A argument) in the same ‘Person’ slot, including when there is a 3 → Singular.SAP situation. A/S arguments are marked by the same su#xes. The complications do not end in what argument has access to position 7, ‘Person’. Objects (P arguments) are also marked in a complex pattern: 3 person objects are never marked, but the marking of SAP arguments varies. A "rst person acting on a second person (1 → 2) indexes both arguments with portmanteau su#xes in position 7, ‘Person’. These portmanteau su#xes historically are transparently related to the "rst and second person 232. Except with the 1 plural object -karatu. With -karatu, the third person marker (A argument) occurs in position 7, and -karatu in position 2, ‘Object’. In that case, there is no agreement with P. 663 markers. When a third person acts upon a Singular.SAP (3 → Singular.SAP), the Singular.SAP (P participant) is indexed in position 2 of the verb, which I have called ‘Object’. Finally, S arguments are uncontroversial: they are always indexed in position 7. Table 14.8 summarizes the su#xes that mark the arguments of a verb. In the table, NA means ‘Not Applicable’. First, I include object markers (which are indexed in Position 2 of the verb structure), an ellipsis symbol “. . .” means that there are other morphological positions in the verb, and the su#xes that appear after “. . .” are the ones that occur in position 7 of the verb, ‘Person’ (which are reserved for A arguments in transitive clauses, with the exceptions discussed above). A symbol “---” means that nothing is marked in Position 2. The underlined elements are the patterns of indexation that are marked di$erently than the “expected” pattern (i.e. situations 1 → 2 and 3 → Plural.SAP, when objects are not marked (only) in position 3 but also in position 7). 664 Table 14.8. Indexation of subjects and objects on the verbTable 14.8. Indexation of subjects and objects on the verb A\P 1sg P 1pl P 2sg P 2pl P 3 P S 1sg A NA NA --- . . . -hamɨ --- . . . -harumɨ --- . . . -ha 1sg -ha 1pl A NA NA --- . . . -himɨ --- . . . -hirumɨ --- . . . -hi 1pl -hi 2sg A -tu/-ru . . . -mɨ -rama/-tama . . . -mɨ NA NA --- . . . -mɨ 2sg -mɨ 2pl A -tu/-ru . . . -rumɨ -rama/-tama . . . -rumɨ NA NA --- . . . -rumɨ 2pl -rumɨ 3A -tu/-ru . . .-ua, -ji or -i ̃ -rama/-tama . . .-hi -karatu . . . -ua, -ji or -i ̃ -rama/-tama . . . -ua, -ji or -i ̃ -rama/-tama . . . -rumɨ --- . . . -ua, -ji or -i ̃ 3 -ua, -ji or -i ̃ 665 14.3.3.1. Indexation of S arguments As Table 14.8 indicates, S arguments (the sole argument of an intransitive verb) are always marked in position 7 of the verb. The following examples illustrate S arguments (1, 2 and 3 persons) marked on the verb. (29) wɨáhɛɛ S: 1sg uɨ-a-ha-i go-IPFV-1SG.SBJ-DECL ‘I am going.’ (30) wɨáhi S:1pl uɨ-a-hi go-IPFV-1PL.SBJ+DECL ‘We are going.’ (31) aḿɨ wɨkaśmɨ S: 2sg amɨ wɨka-sa-mɨ 2SG walk-ATT-2SG.SBJ+DECL ‘You just went hunting.’ (32) at́um wɨkaśrumɨ S: 2pl atumɨ wɨka-sa-rumɨ 2PL walk-ATT-2PL.SBJ+DECL ‘You (PL) just walked.’ (33) Puhupat́ wiʃiŕmaj́i S: 3 (third person does not distinguish number) Puhupata uiʃi-ra-ma-ji Puhupata laugh-DISTR-IMM.PT-DECL ‘Puhupat laughed.’ 666 14.3.3.2. Indexations of arguments in transitive clauses The following examples illustrate the transitive indexation patterns presented in Table 14.8. In transitive clauses, the su#xes listed previously in Table 14.5 are used when a 1A or 2A argument acts upon a 3 person object as in (34)–(35), and when a 2A argument acts upon a 1 person object (36). Once again, remember that when a 1A argument acts upon a 2P argument, there is a portmanteau su#x that refers to both arguments of the transitive clause (37)–(38). (34) wií t͡sanúahɛɛ 1SG → 3 ui t͡sanu-a-ha-i 1SG lie-IPFV-1SG.SBJ-DECL ‘I am lying to her/him.’ (35) aḿɨ t͡sanúamɨ 2SG → 3 amɨ t͡sanu-a-mɨ 2SG lie-IPFV-2SG.SBJ+DECL ‘You are lying to her/him.’ (36) aḿɨ t͡sanúrɨamɨ 2SG → 1SG amɨ t͡sanu-ru-a-mɨ 2SG lie-1SG.OBJ-IPFV-2SG.SBJ+DECL ‘You are lying to me.’ (37) tah́amɨ 1SG → 2SG ta-hamɨ say+IPFV-1SG>2SG+DECL ‘I am telling you.’ 667 (38) tah́armɨ 1SG → 2PL ta-harumɨ say+IPFV-1SG>2PL+DECL ‘I am telling you (PL).’ From the examples above, it can be said that coding in the local domain (SAP → SAP) depends on which grammatical person is the A argument: if A is a 1 person, then both A and P are marked in position 7; wheras if A is 2 person, then only A is marked in that position (and P is marked in position 2). This suggests a hierarchy 2 > 1, as it seems that a 2P argument needs to be marked also in position 7. In the direct domain (SAP → 3), only the A argument is marked in the “subject” position. Now, let us consider the “inverse” domain (3 → SAP). (39) ɨsɛítmiaji 3 → 1 ɨsa-i-tu-mia-ji bite-LOAF-1SG.OBJ-DIST.PT-3.PT+DECL ‘She/he bit me.’ (40) ɨsat́mawɛi 3 → 2 ɨsa-tama-a-ua-i bite-2SG.OBJ-IPFV-3.SBJ-DECL ‘She/he is bitting you.’ (41) amútmaktathi 3 → 1PL amu-tama-ka-tata-hi "nish-1PL.OBJ-INTENS-DEF.FUT-1PL+DECL ‘He/they are going to kill us.’ 668 (42)amútmaktatrumɨ 3 → 2PL amu-tama-ka-tata-rumɨ "nish-2PL.OBJ-INTENS-DEF.FUT-2PL+DECL ‘He/they are going to kill you (PL).’ From the above examples, it can be seen that Plural SAP take preeminence over a 3 person, as they access position 7. To summarize, there are two morphological positions for indexing core arguments on the verb, SAP always access Position 2 (“Object”), SAP and 3 can access Position 7 as A, but as P arguments only Plural SAP can access position 7. Position 7 shows an interesting mismatch with regards to argument indexation: it is consistent in the forms for person, but not for role. Position 2 Position 7 1SG P 2SG P 1PL P 2PL P 3 A 3 A 1PL P (A form) 2PL P (A form) Overall, the hierarchy on which access to Position 7 depends as index of P can be summarized as: 2PL > 2SG > 1PL > 1SG/3. That is, 2 person outranks 1PL, and 1SG/3 are low in the hierarchy. The hierarchy on which access to Position 2 depends as index of P can be summarized as: 1PL/1SG > 2SG/2PL > 3. That is, in for position 2, 1 person outranks 669 2 person, and SAP person outrank 3 person. 14.4. Mood For purposes of this grammar, I use the term Mood for the grammatical category that encodes modality. Modality signals a speaker’s attitude toward a proposition, including the speaker’s belief in its reality, likelihood or relevance to her/himself (Bybee 1985; Timberlake 2007). In Wampis, the categories encoded under the label ‘Mood’ in this section are marked in the "nite verb, never in a subordinated verb. The Mood categories occupy di$erent positions in the verb structure: Mood Position Desiderative, Imperative, Jussive, Apprehensive, Hortative Position 6 Declarative, Exclamative Position 8 Narrative Marked with timaji 14.4.1. Declarative The declarative signals that the proposition expressed by a speaker’s utterance is o$ered as an unquali"ed statement of a fact. The declarative is marked with -i. (43) wií wɨáhɛɛ ui uɨ-a-ha-i 1SG go-IPFV-1SG.SBJ-DECL ‘I am going.’ 670 (44) tres t͡sawań puhústtahɛɛ tres t͡sauanta puhu-sa-tata-ha-i three day live/be-ATT-DEF.FUT-1SG.SBJ-DECL ‘I am going to be three days [hunting in the forest].’ 14.4.2. Exclamative The exclamative is zero-marked. Exclamatory sentences are frequently accompanied by an intonational raise. (45) hukitʲa ́tah́a hu-ki-ta ́ ta-ha-∅ take-WHILE.MOVING-IMP say+IPFV-1SG.SBJ-EXCLAM ‘Take it!, I say!’ (46) urutma ́úuntaita urutuma uunta=aita what big=COP.3.EXCLAM ‘It is big!’ An interesting property of exclamative (and also interrogative) sentences in Wampis is the addition of a negative morpheme that emphasizes the idea of the predication, rather than negates it: (47) óuʃa ɲawaánt͡ʃukɛit au=ʃa iawaa-̃t͡ʃau=ka=ita DIST=ADD jaguar-NEG.NMLZ=FOC=COP.3.EXCLAM ‘That is a jaguar!’ 671 14.4.3. Commands and manipulative mood: imperative, jussive, hortative and prohibitive Commands prototypically involve manipulation of other individuals. Imperatives necessarily take aktionsart stems; other stem types are incompatible with the imperative mood. The jussive, hortative and prohibitive are closely associated. The prohibitive is the negative counterpart of the imperative, whereas the jussive and the hortative express manipulative action towards di$erent persons (third person and "rst person plural, respectively). The following subsections describe these su#xes. 14.4.3.1. Imperative The imperative -ta ́addresses a second person singular by default and, in terms of polarity, is always a positive command (for negative commands, the prohibitive -pa is used (see §14.4.3.5). A 2 plural addressee is marked with the 2 plural subject su#x - rumɨ. The imperative is likely related to the the future marker -ta. Table 14.9 summarizes the imperative constructions with singular and plural forms. Table 14.9. Imperative constructions with 2 singular and plural persons 2sg 2pl Verb-Aktionsart-ta ́ Verb-Aktionsart-ta-́rumɨ 672 The next examples illustrate the imperative. Examples (48)–(49) are instance of commands where the addressee is singular. (48) núu úun mańt͡ʃu juhaẃaji túsa uhakta ́ nu uunta mant͡ʃu iuha-a-u a-ji NON.VIS adult mosquito walk.several-IPFV-NMLZ COP-3.PT+DECL tu-sa uha-ka-ta ́ say-SUB tell-INTENS-IMP ‘Tell that man that the mosquitos wander [around here]!’ (Lit.: ‘[To] that man, saying “the mosquitoes are walkers [around here]”, tell!’) (49) juat́a iu-a-́ta ́ eat-HIAF-IMP ‘Eat!’ In (50), the addressee is plural: (50) juat́armɨ iu-a-́ta-́rumɨ eat-HIAF-IMP-2PL.SBJ ‘(You PL) eat!’ I have found also some examples of the form -rama to mark the plural imperative. This su#x functions as a 1plural or 2 object in other parts of the grammar (cf. §13.2.4). The next example illustrate the use of -rama to form a plural imperative: (51) juat́arma iu-a-́ta-́rama eat-HIAF-IMP-2PL ‘(You PL) Eat!’ 673 Many commands make use of the attenuative -sa. In fact, a typicall use of the attenuative is to diminish the force of a command: (52) jusat́a iu-sa-ta ́ eat-ATT-IMP ‘eat (please)!’ 14.4.3.2. Familiar imperative There is a familiar imperative su#x -kia that is used in colloquial/familiar situations with close relatives and friends. For the most part, there is no apparent di$erence between -ta ́and -kia. I was told, however, that -kia is used para salir del caso ‘to get out of a (not very important) situation’. Such explanation is true for some but not all cases in the data. (53) iʃit́͡ʃik aśam wiḱia iʃit͡ʃikɨ a-sa-mɨ uɨ-kia little.bit COP-SUB-2SG.SS go-FAM.IMP ‘Go in a little while!’ (‘being a little [while], go!’)233 I was also told that a command given with ‘anger’ is expressed with -kia. Thus, a sentence like (54) may be said when a child does not want to eat and the father or 233. Notice that the surface realization of the verb ‘go’ (whose underlying form is /uɨ/) is [wi] and not [wɨ]. So, apparently -kia can establish backward harmony with a vowel /ɨ/ of the stem. Unfortunately, I have very few examples of -kia with a verb stem ending in /ɨ/, so for the moment I cannot make a more general statement on this. 674 mother becomes upset at the child’s behavior: (54) juaḱia iu-a-́kia eat-HIAF-FAM.IMP ‘Eat!’ (Order given with anger) However, -kia can also occur with the attenuative; thus the sentiment of ‘anger’ must be related to the familiar sense of this su#x, rather than being an intrinsic property of the su#x. Notice that with the attenuative, the familiar imperative surfaces with the reduced form [sia]. (55) iiśia huní wɨárɨ ̃ ii-sa-kia hu=ni wɨ-ara-ɨ ̃ see-ATT-FAM.IMP PROX=ALL go-PL-3.PFV ‘Look! They went over here.’ The plural for the familiar imperative is done by adding a form that surfaces as [rua]. This can be hypothesized as a phonological reduction of the plural -ara, the subject nominalizaer –u, and the copula a; i.e. a nominalization construction (‘be ones who [do] X’!). For the moment, I gloss -rua as a plural imperative form, as I have few examples of this enabling a more general statement. (56) juaḱiarua iu-a-́kia-rua eat-HIAF-FAM.IMP-PL:FAM.IMP ‘[You (PL)] eat!’ 675 14.4.3.3. Jussive The jussive involves the expression of positive commands, desires or permission directed toward a third person addressee. The jussive is marked with the su#x -ti. The jussive has probably arisen from the historical con/ation of the immediate future -ta and the third person marker -i ̃, although there is no trace of nasalization in the jussive (but recall that -i ̃surfaces very often as an oral vowel (cf. §14.3.2)). (57) jawaań huki ̃ ́wɨákʃa anɨńtroo kaḿɨ núŋkui surusti ́túsa ̃ iauaa ̃ hu-ki ̃ uɨ-a-kũ=ʃa anɨnta-tu-ra-u dog take-WHILE.MOVING\3.SS go-IPFV-SIM\3.SS=ADD sing.magical.song-APPL-DISTR-NMLZ kamɨ nunkui su-ru-sa-ti tu-sa ̃ INTERJearth.being give-1SG.OBJ-ATT-JUSS say-SUB\3.SS ‘[The man, when he goes to hunt...] when he is taking the dog too, he is one to sing anen (magical song) too, saying “Let Nunkui give me [good luck]”.’ In concordance with the aktionsart stem to which the jussive attaches, the plural is marked with the non-imperfective plural marker -ara, which, as was seen, is used for 3 persons (§5.4.2). (58) juaẃarti iu-aú-ara-ti eat-HIAF-PL-JUSS ‘Let them eat!’ The hortative also functions as a counterpart of the imperative in reported 676 speech. Thus, in the next example taken from a conversation, the command is understood as addressing the interlocutor: (59) kaʃ́i miɲit́i túsa turaḿɨnawɛɛ wilsonhɛi ̃ kaʃi uini-ti tu-sa tu-rama-ina-ua-i wilson-hai ̃ tomorrow come.PFV-TI say-SUB say-2.OBJ-PL.IPFV-3.SBJ-DECL Wilson-COM ‘They tell you to come with Wilson tomorrow.’ (Lit.: ‘Saying “Tomorrow that he comes”, they tell you, “with Wilson”.’ 14.4.3.4. Hortative The hortative -mi signals the speaker appeal or encouragement toward the addressee to bring about a future state of a$airs. The hortative is generally understood as having a ‘you+me’ subject . (60) umaŕmi uma-ra-mi drink.PFV-DISTR-HORT ‘Let’s drink.’ When there is more than one addressee, the plurality of addressees is marked with the non-imperfective plural -ara, which goes before the hortative su#x. (61) umaŕarmi uma-ra-ara-mi drink-DISTR-PL-HORT ‘Let’s drink.’ There is another su#x -tai that seems to also function as an hortative. In the data gathered for Wampis, it is mostly found mostly in formulaic phrases, as in (62). 677 (62) arúm wɛiɲ́iatɛi aruma uai-nai-a-tai later see-RECIP-IPFV-HORT ‘See you later’ (Lit.: ‘Let us see each other later.)’ Unlike all other su#xes related to the domain of command and manipulation, the su#x -tai can attach to an imperfective stem. This seems a little odd, as the morpheme -tai seems to be based on the basic immediate future -ta (which is likely also the source for the imperative). As stated before, the future always takes the aktionsart/ perfective stem, thus the use of an imperfective stem in this case, from a structural point of view, seems unexpected. 14.4.3.5. Prohibitive -pa The prohibitive -pa is used with negative commands. It addresses a 2 singular by default. If a 2 plural person is addressed, then the plural form -rupa is used. The su#x - rupa is historically composed of a plural formative *-ru and the prohibitive -pa. Optionally, speakers can use the su#x -pa followed by the 2 plural subject su#x -rumɨ. There is no di$erence between using -rupa or -pa-rumɨ. The next examples illustrate the use of the prohibitive. (63) naḱi ɛípʲa naki a-ai-pa lazy COP-APPR-PROH ‘Don’t be lazy!’ 678 (64) iʃaḿrukɛipʲa iʃama-ru-ka-ai-pa be.afraid-APPL-INTENS-APPR-PROH ‘Don’t be afraid!’ (65) tií sɨńt͡ʃi ɨt́͡sɨka imɛíɲirpa tii sɨnt͡ʃi ɨt͡sɨ-ka i-ma-i-ai-rupa INTS strongly heat-INTENS CAUS-bathe-LOAF-APPR-PL.PROH ‘[With] Very hot [water] do not bathe [her].’ Since the prohibitive typically refers to actions or events that are not desired to occur, it very frequently occurs following the apprehensive su#x, as can be appreciated in the above examples. 14.4.4. Apprehensive -ai The Wampis apprehensive modality is morphologically codi"ed via the use of the su#x -ai. The apprehensive serves to express that a potential event is seen as highly undesirable. Following Lichtenberk, it can be said that the Wampis apprehensive involves a miscellany of epistemic and attitudinal semantics related to the coding of “the speaker’s degree of certainty about the factual status of a proposition and with his or her attitude concerning the desirability of the situation encoded in the clause” (1995: 293). The apprehensive in Wampis covers the domains equivalents to what is typically 679 codi"ed with lest-clauses (Dixon 2002) and ev́itatifs (François 2003) in other languages. Unlike the morphemes seen in §14.4.3, the prohibitive occurs with all persons. With the second person, it occurs with the prohibitive -pa as was seen in the previous section. Third person does not distinguish between singular and plural. Table 14.10 shows a paradigm of person marking with the apprehensive. Table 14.10. Paradigm of person marking with apprehensive modality Person\Number SG PL 1 ai-ha ai-hi 2 ai-pa ai-rupa 3 aiã a The realization of this form is usually [in]. The apprehensive usually interacts with the prohibitive -pa, but it can also occur alone. In example (66), the apprehensive occurs along with the prohibitive in the verb ‘lie down’, but it occurs on its own in the main predication (the verb ‘ruin’). (66) t͡sawaŕa ɛɛ ̃ ́kanaḱum tɨpɨśaipʲa mɨsɨḿakɛim t͡saua-ra ai ̃ kana-ku-mɨ tɨpɨ-sa-ai-pa dawn- DISTR MED.LOC sleep+IPFV-SIM-2SG.SS lie.down-ATT-APPR-PROH mɨsɨ-ma-ka-ai-mɨ ruin-REFL-INTENS-APPR-2SG.SBJ ‘When it dawns, there, don’t keep sleeping, lest you dream your death!’234 234. The verb mɨsɨ ‘ruin’ also has the extended meaning of ‘dream with your own death’, which 680 Here are some examples of the use of the apprehensive, given with contextual information. (67) Context: A group of people is making t͡sant͡sa ‘head trophy’; the ritual leader gives them directions as the ritual was supposed to be very strict: ʃiiŕ t͡ʃinkaẃɛi túsa ̃ɨmɨśraih túsa ̃intʲaʃi ́aɨśawaih ʃiira t͡ʃinka-au-ai ̃ tu-sa ̃ ɨ-mɨsa-ra-ai-hi tu-sa ̃ well make.hole-HIAF-APPR say-SUB\3SG.SS CAUS-ruin-DISTR-APPR-1PL say-SUB\3SG.SS intaʃi ́ a-ɨsa-au-ai-hi hair\ACC CAUS-burn-HIAF-APPR-1PL ‘[He] saying “watch out for making holes”, [he] saying “may we not ruin it”, “may we not burn the hair [of the shrunken head]”.’ (68) Context: A Wampis woman marries an Awajun man. When his husband’s uncle dies, she is suspected of having poisoned him with poisonous manioc: turaʃ́a wií wɨkaś wɨḿiahɛɛ ɨt͡sɨŕtin ɛín turaʃa ui uɨka-sa uɨ-mia-ha-i ɨt͡sɨra-tu-inu a-i ̃ but 1SG walk-ATT go.PFV-DIST.PT-1SG.SBJ-DECL inform-APPL-NMLZ COP-APPR ‘But I left because I was afraid that they accuse me [of the killing].’235 14.4.5. Inferential -tai The inferential -tai marks a clause as being an inference or supposition on the is how the example was translated. 235. The verb ɨt͡sɨra ‘inform, give news’ plus the applicative acquires a sense of ‘accuse’. 681 part of the speaker. The following examples discuss the inferential modality. In (69), the story is about two sisters that meet a man, Nayap. The man tells them to go his house, as the women wanted to live with him. There are two trails, one goes to the man’s house and the other goes to another character, Tsuna. The sisters mistakenly go to Tsuna’s house. Nayap returns home and asks his mother for the sisters. Seeing that they are not there, he infers that they must be in Tsuna’s place: (69) a. at͡sá núwaka tɛíɲat͡sui at͡sa nua=ka ta-ina-t ͡su-u-i no woman=FOC arrive-PL.IPFV-NEG-3.SBJ-DECL ‘[His mother:] “No, the women do not arrive”, b. t͡sií nuiŋ̃kʲa saiŕ t͡sunáhɛɛ ̃hintińt͡suk ʃiakaratɛɛ t͡ʃii nui=̃ka sai-ru t͡suna=hai ̃ hintina=t͡su=kɨ okay there=FOC brother.in.law-1SG Tsuna=COM trail=INFER=RESTR ʃia-ka-ara-tai go.PL-INTENS-PL-INFER [Nayap:] “oh, then, with my brother-in-law Tsuna, they must have gone”.’ The next example, contains a description of how people talk about a man which possesses Arutam (a spiritual vision that gives a certain power to the person who sees it). In the Jivaroan tradition, people would know when a man possesses Arutam due to certain qualities that the man shows: leadership, oratory skills, "gthing skills, etc. In 682 this case, because the man did not die in wars, they infer that he must have the power of “Stone-Arutam”. People with the power of “Stone-Arutam” were supposed to be un- killable. (70) kaj́a wɛínkouskɛiti núnikoo asańt͡suk haat́͡sutɛɛ nunaʃ́a tiɲ́u aŕmaji kaia waina-ka-u=t͡su=ka=iti nuni-ka-u stone see-INTENS-NMLZ=INFER=FOC=COP.3+DECL do.that-INTENS-NMLZ a-sa=̃t͡su=kɨ ha-a-t͡su-tai COP-SUB\3.SS=INFER=RESTR die-IPFV-NEG-INFER nu=na=ʃa tu-inu a-ara-maji NON.VIS=ACC=ADD say-NMLZ COP-PL-3.REM.PT ‘Surely because he has seen stone-Arutam, surely he does not die, they said that too.’ 14.4.6. Narrative modality timaji Traditional narratives very frequently use the word timaji, which can be literally translated as ‘he/she said’ or ‘they said’ (< ti-ma-ji ‘say+LOAF-IMM.PT-3.PT+DECL’). This word is a marker of the narrative genre, and generally marks second-hand information. This relates to the fact that speakers frequently begin a narrative with something similar to ‘My grandparent/my elders/my parent used to tell me...’. Timaji does not occur in other types of genres with the second-hand information function. The source of narrative timaji is a complement construction of a speech report, in which timaji used to be a fully "nite verb and occurred with a nominalized verb as a 683 complement (this construction is still very frequent in narratives too). (71) iístahɛɛ wɨú tiḿaji ii-sa-ta-ha-i wɨ-u timaji see-ATT-IMM.FUT-1SG.SBJ-DECL go.PFV-NMLZ NARR ‘“I am going to see”—he went.’ Historically, an example like (71) clearly contained a "nite verb ‘say’ (timaji) and would translate as ‘“I am going to see”—he went—they said.’ However, crucially, in current Wampis timaji can also occur with verb forms that are not nominalized and are not marked for mood. This is evidence that timaji has been grammaticalized as a marker of narrative modality (72) hija ́tiḿaji aúhun hia timaji auhu=na reprimand+IPFV NARR Auju=ACC ‘He was reprimanding Auju.’ (73) nuni ́waá timaji nantuka nu=ni uaa timaji Moon=ka NON.VIS=ALL climb+IPFV NARR Moon=FOC ‘Over there Nantu climbed.’ (74) wɨɨ́ tiḿaji uɨ timaji go.PFV NARR ‘He went away.’ 684 (75) túu unt͡sua ́tiḿaji tu unt͡su-a ́ timaji like.that call-HIAF NARR ‘she called it (i.e. the manioc) like that.’ 685 CHAPTER XV NOMINALIZATION 15.1. Introduction This chapter discusses key aspects of the morphosyntactic properties of nominalization in Wampis. The analysis presented in the following sections also serves as an introduction to Chapter XX, where it will be seen that one of the salient syntactic features of Wampis is the use of nominalized clauses in a wide range of constructions that include relative clauses, complement clauses, and in copular clauses to create expressions that function at the level of the TAM system. Nominalization is understood here as the mechanism that derives a non-nominal element which then functions as nominals do in the language. Following the ideas discussed in Shibatani (2009), nominalization can be lexical or grammatical (cf. also Koptjevska-Tamm (1993), Gennetti et al. (2008)). The properties described here correspond to lexical nominalization, i.e. I will illustrate cases of class-changing derivations that become part of the lexicon of the language. After this introduction, section §15.2 discusses some aspects of the verbal morphology that are central to understand what type of verb stems carry nominalizers 686 in Wampis. This is followed by §15.3, where the nominalizers of Wampis and their morphosyntactic properties are presented. Finally, each nominalizer in Wampis is described in §15.4. 15.2. The nominalized verb To understand how nominalization works in Wampis, some morphosyntactic properties of the verb must be pointed out. Recall that one salient complication of the Wampis verb is that the distinction between “derivational” and “in/ectional” morphology is tenuous. The main "nite verb is obligatorily marked for aspect, tense, person and mood. However, aktionsart and aspectual su#xes of Wampis have clear derivational properties and can create a stem to be used in speech. Therefore I concluded that the terms “derivational” and “in/ectional” in this case are better regarded as a continuum as some formatives can assume functions related to both derivation and in/ection. The overall structure of the Wampis verb was presented in Table 12.5 (it was presented "rst in §12.7). The level of the “stem” that can enter into nominalization in Wampis roughly corresponds to the root plus “derivational” morphology (roughly the morphological positions from -1 to 4 shown in Table 12.5), incorporating valence, negative, number and aspectual information. That is, 687 nominalization occurs on a nominalizable stem, i.e. it can incorporate all the structural information up to the right border of position 4 in the verb structure (which roughly corresponds to what I call the “derivational” level). In general, nominalization in Wampis can retain a good portion of the verb structure. Depending on the speci"c nominalizing strategy, nominalized verbs in Wampis may preserve their valence, aspect and some argument (object) information. At least in one case, with the future nominalizer -tinu §15.4.2, the nominalized verb can also incorporate tense information. This is a very interesting property of Wampis (and Jivaroan languages):236 though cross-linguistically verbs tend to decategorize when they are nominalized, losing most of their verbal properties (Haiman 1983; Givón 2002b; Cristofaro 2003) including TAM and argument information, in Wampis the nominalized verb can carry some aspectual, argumental and, in one case, even tense information. As we saw in Chapter XIV, most verbs in Wampis select an aktionsart su#x with which they occur in perfective contexts (including most past tenses), as well as in future contexts. However, these su#xes do not occur in imperfective or potential contexts. This information is important because it will be shown that Wampis nominalizers di$er 236. Similar structural descriptions are observed for Awajun (Overall 2007), Shuar (Turner 1992; Saad 2014) and Shiwiar (Kohlberger 2014). 688 on whether they can nominalize an aspectualized stem or not. With regard to aspectual notions, recall from Chapter XIII (cf. §13.3) that the morphological position number 3 in Table 12.5 is shared by aktionsart, imperfective, durative, present habitual or potential morphemes, which thus enter in a system of oppositions. The occurrence of one or another su#x is conditioned morphosyntactically. Table 15.1 lists these su#xes. The stems that carry these su#xes are called “aspectual” stem. Table 15.1. Su#xes that occupy morphological position 3 in the verb piece -a(́u) ‘High a$ectedness’ -i ‘Low a$ectedness’ -ka ‘Intensive’ -ki(ni) ‘Do while going’ -ra ‘Distributed action’ -sa ‘Attenuative’ -ri ‘Do in proximity/while coming’ -u ‘Do away from location’ -a ‘Imperfective’ -ma ‘Durative’ -na ‘Present habitual’ -mai ‘Potential’ Examples (1)–(3) with the root ‘eat’ illustrate the di$erent morphosyntactic environments with aktionsart, imperfective, present habitual and potential verb stems. The verb ‘eat’ selects the high a$ectedness -a(́u) as its default aktionsart su#x, required in perfective contexts such as in (1). In imperfective (2), present habitual (4) and 689 potential (3) environments the aktionsart su#x does not occur. Recall that the high a$ectedness aktionsart su#x -a(́u) and the imperfective -a are distinguished by the ability of the aktionsart su#x to attract a high tone. In addition, notice that the potential stem usually occurs with the agentive nominalizer -inu and, interestingly, has ambiguous readings that can focus on the A/S or the P arguments of the nominalized potential stem. (1) juah́ɛɛ iu-a-́ha-i eat-HIAF-1SG.SBJ-DECL ‘I just ate.’ (2) júwahɛɛ iu-a-ha-i eat-IPFV-1SG.SBJ-DECL ‘I am eating.’ (3) mamań júnamɨ mama=na ju-na-mɨ manioc=ACC eat-PRES.HAB.2SG.SBJ+DECL ‘You usually eat manioc.’ (4) jumɛiń ju-mai-inu eat-POT-NMLZ ‘something that can be eaten (eat-able)’ or ‘one that can eat’ 690 Before proceeding further with the discussion of nominalizations, it needs to be pointed out that nominalizations are distinct from what, in a more restrictive way, I call “subordinate” verb forms in Chapter XIX. Unlike nominalizations, most subordinated verbs in Wampis can receive only valence and aspectual marking; in addition, unlike nominalizations, subordinate verbs: • are de"ned by an indication of switch-reference that is obligatory in those forms. • receive person/subject markers that occur in switch-reference contexts (these subject markers are not the same markers used for main clause verbs) • they do not receive nominative/accusative markers, unlike nominalizations, i.e. subordinate verb cannot syntactically "ll argument positions of another matrix verb (although functionally, some subordinated constructions are employ as semantic arguments of some verbs (see details in Chapter XX). • no verb form that carries a “subordinator” becomes a member of the lexicon of Wampis (i.e. “subordinators” do not derive lexical items), whereas nominalizers derive lexical items. Within this dissertation, thus, I de"ne the nominalized verb as a form of the nominalizable stem verb that receives any one of the nominalizers listed in §15.3. 691 Unlike what I de"ne as subordinate verbs, a nominalized verb never receives a switch- reference marker or person subject markers, and can function as syntactic arguments of predicates (i.e. nominalizations receive case marking). With this background on the nominalizable stem and the nominalized verb, we now turn to the discussion of the nominalizers themselves in Wampis. 15.3. Nominalizers in Wampis There are two sets of nominalizers in Wampis, which I call Set I and Set II. Both sets are used for lexical and grammatical nominalizations. Lexical nominalization “creates new lexical items belonging to the noun class of the language”, whereas grammatical nominalization “creates new referring expressions that have no lexical status” (Shibatani 2009: 187).237 There are two more nominalizers, the negative nominalizer -t͡ʃau (§10.4.5), and the nominalizer -na, which is not lexically productive and is mainly used in relative clauses and with the potential stem (§15.4.8). Set I and Set II nominalizers are presented in Table 15.2. 237. As I say in the introduction, the present chapter focuses in lexical nominalization. The functions of nominalization at the level of grammatical nominalization are described in Chapter XX. 692 Table 15.2. Nominalizing su#xes in Wampis Set I Set II -inu ‘Agentive nominalizer’ -u ‘Subject nominalizer’ -tinu ‘Future nominalizer’ -mau ‘Non-subject nominalizer’ -ta ‘Action nominalizer’ -tai ̃‘Patient/Location nominalizer’ In principle, a verb can take any nominalizer from Set I or Set II, depending on the morphosyntactic environment: this is not a distinction of verb classes, where one class takes one set of su#xes and another class takes the other set of su#xes. Rather, there are other important distinctions, if subtle, between Set I and Set II su#xes (cf. also the last part of the discussion in §15.4.5). The main morphosyntactic properties of Sets I and II are presented in Table 15.3. Table 15.3 also includes properties of underived nouns and adjectives so that nominalization properties can be compared in a larger picture of the language (see Chapter IX for a detailed analysis of adjectives). At "rst glance, from the semantics and morphosyntactic properties seen in Table 15.3, there seems to be some redundancy overlap in the nature of some of Set I and Set II nominalizing morphemes. However, we will see that they present particular morphosyntactic di$erences. 693 Table 15.3. Morphosyntactic properties of underived nouns, nominalizations and adjectives Properties Underived Nouns Set I Nominalization Set II Nominalization Adjectives Receive case and oblique clitics Yes Yes Yes Noa Can be possessor Yes Yesb Yesb No Can be possessed Yes Yes Yes Somec Can head an NP Yes Yes Yes No Can be pluralized with a-ina (COP- PL.IPFV) Yes Yes Yes No Gradable No In modi"cation function only In modi"cation function only Yes a Demonstratives trigger case agreement on certain elements of a noun phrase, including adjectives. Receiving case or oblique clitics is not considered a property of Adjectives (cf. §9.5). b Set I -inu and Set II -u cannot be possessed, since they do not mapped onto the semantics of a patient (a possessed thing is prototypically patient-like). c Notice that adjectives that are possessed usually function as nominals; e.g. ʃiirama ‘beautiful’ when possessed with -ri ̃‘1PL/2PL/3.POSS’ means ‘her/his beauty’ (i.e. when possessed, adjectives behave like possessed nouns—cf. §9.4). Morphologically, Set I and Set II di$er in one important property: with the exception of the future nominalizer -tinu, Set I nominalizing su#xes can occur on bare roots, but they do not attach to an aspectualized verb stem. In contrast, Set II morphemes can be attached to an aspectualized stem, as de"ned in §15.2. In addition, the Set II subject nominalizer -u, and Set I future nominalizer -tinu and patient/location 694 nominalizer -tai ̃can combine with the verbal negative su#x -tʃa. The distributional properties of Set I and Set II nominalizers are listed in Table 15.4. I also add information about the nominalizer -na, which combines with the potential stem marked by -mai ‘potential’, as well as the negative nominalizer -tʃau. Table 15.4. Distributional features of Wampis nominalizers Properties → Su#xes ↓ Can occur on root Can occur on aspectual stem Can occur on potential stem V-mai Can occur on negative stem V-tʃa -inu Yes No Yes No -tinu Yes Yes No Yes -tai ̃ Yes No No Yes -ta Yes No No No -u No Yes No Yes -mau No Yes No Yes -t͡ʃau Yes Yes No No -na Yes No Yes No 15.4. Derivational nominalization As lexical nominalizers, Set I and Set II su#xes in Wampis are productive and fully derivational. This section brie/y presents the basic derivational uses of the 695 nominalizers presented in §15.3. 15.4.1. Set I -inu, ‘agentive nominalizer’ The agentive nominalizer -inu creates a noun that denotes the subject of a verb and which is prototypically an agent and animate. The examples in (5) show instances of nouns derived with -inu. (5) arantu ‘respect’ → arantu-inu [arańtin] ‘a respectful one’ aúha ‘read, study’ → aúha-inu [óuhin] ‘student’ iuiʃi ‘heal’ → iuiʃi-inu [iwiʃiń] ‘shaman’ kutama ‘spin’ → kutama-inu [kutaḿin] ‘spinner (of cotton or chambira thread)’ iuara ‘make joke’ → iuara-inu [iwariń] ‘joker’ umu ‘drink’ → umu-inu [úmin] ‘drinker’ ma ̃ ‘kill’ → ma-̃inu [mɛín] ‘hunter’238 nakuma ‘draw’ → nakuma-inu [nakúmin] ‘drawer’ The agentive nominalizer -inu is very productive with the 1pl object marker - karatu. With -karatu, the nominalizer -inu creates generic nouns. (6) hintińkʲartin hintina-karatu-inu teach-1PL.OBJ-NMLZ ‘teacher’ (‘someone who teaches us’) 238. The form ma-̃inu ‘hunter’ is used in the conventionalized sense of ‘killer of animals’ and denotes a human. Animals that hunt are referred to as mañtinu [ma ̃ńtin] (see example (10)). The form for ‘killer’ in the sense of ‘assassin’ is mañkartin (see example (8)). 696 (7) mañkaŕtin ma-̃karatu-inu kill-1PL.OBJ-NMLZ ‘assassin’ (‘someone who kills us’) (8) nakúmkartin nakuma-karatu-inu draw-1PL.OBJ-INU ‘photographer’ (‘someone who draws us’) (9) amúkartin amu-karatu-inu exterminate-1PL.OBJ-NMLZ ‘exterminator’ (in the sense of ‘killer’, ‘someone who exterminates us’) At least one compound word involves a derivation with -inu: (10) ɲawaám̃ańtin iauaa_̃ma-̃tu-inu dog_kill-APPL-NMLZ ‘hunting dog’ (more literally: ‘a dog that kills for one (i.e. the owner)’) Interestingly, there are some lexicalized items that also carry the nominalizer su#x -inu, and in some of them the stem appears to have an aktionsart su#x (-i ‘low a$ectedness (of P or of location of A)’ in (11), and -ra ‘distributed action’ in (12). The 697 nominalizer -inu does not attach to a stem containing an aktionsart or any su#x that occurs in position 3 in Table 12.5 in current Wampis. However, it seems that the nominalizer -inu was able to attach to an aspectualized stem in past stages of the language. The following examples present historical analysis of words that are already lexicalized in Wampis and exhibit the su#x -inu: (11) uwɨḿtikartin u-uɨ-ma-tu-i-karatu-inu CAUS-go-REFL-APPL-LOAF-1PL.OBJ -NMLZ ‘savior’ (‘someone who let us go for our bene"t’)239 (12) t͡sɨŋ́kɨt͡sɨŋkɨŕin t͡sɨnkɨ-t͡sɨnkɨ-ra-inu branch.o$-branch.o$-DISTR-INU ‘tree that has many branches’ 15.4.2. Set I -tinu, ‘future nominalizer’ The su#x -tinu creates a structure that denotes either a future agent or a future patient of the verbal action or situation. For instance, the nominalization in (13) has two meanings, one focused on the agent of the verb and the other on the patient of the verb. 239. The verb uuɨma (< u-uɨ-ma as analyzed above) means ‘scape’ in current Wampis. 698 (13) óuhmatsatin auhumatu-sa-tinu tell-ATT-FUT.NMLZ ‘one who is going to tell’ or ‘what is going to be told’ More examples of the future nominalizer are presented in (14). (14) akina ‘be born’ → akina-tinu [akiɲ́atin] ‘one who is going to be born’ ɨakama‘hunt’ → ɨakama-tinu [ɨaḱmatin] ‘one who will hunt’ kaka ‘be.strong’ → kaka-ra-tinu [kakaŕtin] ‘one who will be strong’ (be.strong-DISTR-FUT.NMLZ) manka ‘get fat’ → manka-ra-tinu [maŋkaŕtin] ‘one who will get fat’ (get.fat- DISTR-FUT.NMLZ) From a historical perspective, the nominalizer -tinu is more than likely a con/ation of the immediate future marker -ta and the agentive/subject nominalizer -inu (cf. Overall (2007: 432). Because tense markers are formed on an aktionsart stem obligatorily, as seen in Chapter XII when we de"ned the verb, -tinu can be attached to an aspectualized stem (as in (13), where the stem has the aktionsart/perfective attenuative su#x -sa), a property that is not shared by -inu or the other su#xes of Set I. However, -tinu cannot occur with the imperfective, which separates -tinu from Set II nominalizers. The su#x -tinu is presented as a part of Set I nominalizers on the basis that it is derived from -inu. The future nominalizer -tinu can also derive an abstract action/stative noun that sometimes serves as a citation form of the verb. However, -tinu is not as commonly used 699 as a citation form su#x as the action nominalizer -ta (§15.4.3).240 The use of -tinu for citation forms probably derives from an eventive reading of patient nominalizations achieved with this morpheme. (15) iia ‘fall’ → iia-tinu [ij́atin] ‘to fall’ uɨ ‘go’ → uɨ-tinu [wɨt́in] ‘to go’ 15.4.3. Set I -ta, ‘action nominalizer’ The action nominalizer -ta creates an abstract noun that refers to the event/state expressed by the verb. (16) akina ‘be born’ → akina-ta [akiɲ́at] ‘birth’ ha ‘be sick’ → ha-ta [hat́a] ‘sickness’ mɨsɨ ‘ruin, die’ → mɨsɨ-ta [mɨsɨt́] ‘death, war’ puhu ‘live’ → puhu-ta [puhút] ‘life’ taka ‘work’ → taka-ta [takat́] ‘work (n)’ umu ‘drink’ → umu-ta [úmut] ‘to drink’ maã2́41 ‘kill’ → maãńaita [ma ̃ńit] ‘to "ght’ (maã-́nai-ta ‘kill-RECP-NMLZ’) As mentioned in the previous section, -ta nominalizations are commonly given 240. In general, because there is no conventionalized citation form in the language, Wampis speakers vary between -ta or -mau as choices for citation forms, besides -tinu. In other Jivaroan languages such as Achuar (Fast et al. 1996) and Shiwiar (Kohlberger 2014) the cognate of -tinu is used as a citation form. 241. There are two forms of the verb ‘kill’: ma ̃and maã.́ The second form is analyzable as ma-̃a ́ ‘kill-HIAF’, but it appears that for many speakers this form has been reanalyzed as the root. This is not unexpected, because aktionsart su#xes, like the high a$ectedness -a(́u) have derivational functions. 700 as the citation forms of verbs. 15.4.4. Set I nominalizer -tai,̃ ‘non-agentive nominalizer’ The su#x -tai ̃creates a noun that denotes a patient of a transitive verb, a location of transitive or intransitive verb, or an instrument; thus I call it ‘non-agentive’ to establish a contrast with the nominalizer -inu ‘agentive nominalizer’. Locations are usually (but not always, see examples such as aúhumatu-tai ̃or ɨkɨna-tai ̃in (17)–(18)) derived from intransitive verbs. The following examples show the most conventional meaning of the derived forms (a patient interpretation), but all of them can potentially denote a location of the verb too. (17) aki ̃ ‘put on ear’ → aki-̃tai ̃[aki ̃t́ʲɛɛ]̃ ‘earring’ (what is put on the ear’) kankɨ ‘roll’ → kankɨ-tai ̃[kaŋkɨt́ɛɛ]̃ ‘wheel’ (‘what is rolled’) umu ‘drink’ → umu-tai ̃ ‘drink (N)’ iu ‘eat’ → iu-tai ̃ ‘food’ (‘what is eaten’) iu ‘eat’ → iu-t͡ʃa-tai ̃ ‘inedible’ (‘what is not eaten’) (eat-NEG-NMLZ) maã ́ ‘kill’ → maã-́t͡ʃa-tai ̃[maãt́͡ʃtɛɛ]̃‘what is not killed’242 (kill-NEG-NMLZ) aúhumatu ‘tell’→ aúhumatu-tai ̃[óuhmattɛɛ]̃ ‘what is told’ or ‘place where something is told’ Examples of nominalized verbs that possess the su#x -tai ̃and have a 242. Specially to refer to animals that must not be killed for cultural reasons, such as dolphins or buzzards. 701 conventionalized meaning of location: (18) anuma ‘dock (V)’ → anuma-tai ̃[anumtɛɛ]̃ ‘dock (N)’ (‘where a canoe is docked’) aiama ‘rest’ → aiama-tai ̃[ajaḿtɛɛ]̃ ‘resting place’243 ihatuma‘defecate’ → ihatuma-tai ̃[ihʲatmatɛɛ]̃ ‘latrine’ ɨkɨma ‘sit’ → ɨkɨma-tai ̃[ɨkɨmtɛi]̃ ‘chair’ ɨkɨna ‘put on "re’ → ɨkɨna-tai ̃[ɨkɨntɛi]̃ ‘grill (N)’ ɨnkɨma ‘enter’ → ɨnkɨma-tai ̃[ɨŋkɨmatɛɛ]̃ ‘entrance’ There are some examples of words with -tai ̃which denote an instrument, rather than patient or location: (19) asakɨ ‘lock (V)’ → asakɨ-tai ̃[asaḱtɛɛ]̃ ‘key, lock (N)’ (‘what is used to lock’) ɨsɨka ‘wrap up warm’ → ɨsɨka-ma-tai ̃[ɨsɨkmatɛɛ]̃ ‘blanket’ (wrap.up.warm- REFL-NMLZ) maã ́ ‘kill’ → maã-́nai-tai ̃[maãńitʲɛɛ] ‘weapon’ (kill-RECP-NMLZ) nɨkɨ ‘grind’ → nɨkɨ-tai ̃[nɨkɨt́ɛi]̃ ‘grinder’244 The non-agentive nominalizer -tai ̃may have come from the combination of -ta ‘action nominalizer’ plus =(n)i ̃‘locative’ (but cf. Overall (2007: 436), who "nds not enough evidence for this hypothesis in a cognate form in Awajun). Synchronically, a nominalized verb with -tai ̃requires a locative marker: 243. Traditionally, aiamatai ̃are places near waterfalls where the Wampis go to rest after the ingesting of Ayahuasca. 244. Traditionally, this noun refers to a stone that is used to grind but is now extended to any kind of grinder. 702 (20) ajaḿtɛiɲ̃am wɨáhɛɛ aiama-tai=̃nama uɨ-a-ha-i rest-NMLZ=LOC go-IPFV-1SG.SBJ-DECL ‘I am going to the resting place’ Example (20) suggests that if there is a historical connection between the nominalizer -tai ̃and the locative -(n)i,̃ the time of development since the hypothesized grammaticalization of -tai ̃is relatively deep enough that speakers nowadays no longer see it as containing the locative—which leaves us with the need for more comparative data to attempt a connection between the nominalizer and the locative marker. On the other hand, the fact that the nominalizer -tai ̃can derive locations from a verb and the phonetic resemblance with the locative (which is also grammaticalized as a di$erent subject switch-reference (§19.5.2)) suggest that -tai ̃may after all be related to the locative. Corbera (1994: 149) has hypothesized that the Awajun cognate fomr -tai ̃is decomposable between the nominalizer -ta and an instrumental su#x -i which is found in Awajun,245 but not, as far as I know, in Wampis. It is interesting that both a locative and an instrumental interpretations are found in the nominalizer -tai ̃in Wampis. In the (yet not proven) case that there be a formative *-i ̃in -tai,̃ perhaps this is the only place 245. Overall (2007) lists this morpheme as -(a)i. He "nds not enough evidence for Corbera’s analysis. 703 in the grammar of Wampis where one can see an old Jivaroan instrumental, probably historically con/ated with the locative, in the nominalizer -tai.̃ This is an speculative suggestion that remains to be more thoroughly analyzed. 15.4.5. Set II nominalizer -u, ‘subject nominalizer’ The subject nominalizer -u encodes di$erent semantic roles—agent, experiencer/ undergoer, force—all of which map onto the A/S participant of the verb. Unlike -inu, which encodes prototypically animate agentive participants, -u can denote an animate or inanimate participant. For descriptive convenience, I use the term “subject nominalizer” to refer to the nominalizer -u. As mentioned in §15.3, an important feature of Set II nominalizers is that, unlike Set I nominalizers, they can nominalize verb stems which have aspectual information. (21) Examples with the imperfective -a arakama ‘plant’ → arakama-a-u [araḱmou] ‘planter’ ha ‘be sick’ → ha-a-u [haóo] ‘sick person’ nɨka ‘know’ → nɨka-a-u [nɨkóo] ‘wise’ t͡ʃit͡ʃa ‘speak’ → t͡ʃit͡ʃa-a-u [t͡ʃit͡ʃóo] ‘speaker’ Examples with aktionsart (perfective stem), -ra ‘distributed action, -ka ‘intensive’ auama ‘cut hair’ → auama-ra-u [awaḿaru] ‘one who cut his/her hair’ ha ‘be sick’ → ha-ka-u [hakóo] ‘dead, corpse’ unuima ‘learn’ → unuima-ra-u [unúimaru] ‘educated’ (‘one who learned’) ɨɨma ‘go forward’ → ɨɨma-ka-u [ɨɨḿkoo] ‘one who went forward’ 704 ɨsa ‘burn’ → ɨsa-ka-u [ɨśaku] ‘what/who burned’ manka ‘get fat’ → manka-ra-u [maŋkaŕu] ‘one who got fat’ mina ‘soften, melt’ → mina-ra-u [mińaru] ‘what became softened or melted’ kau ‘rot’ → kau-ra-u [kouróo] ‘rotten meat’ From the above examples, it can be seen that while there is some overlap between Set II -u and Set I -inu (§15.4.1) (for instance, the grammatical realization of the semantic role of the participants they denote is the subject of the nominalized verb), there are also some di$erences. The most notable di$erence is the ability of Set II -u to attach to aspectual stems, either imperfective or perfective. This allows Set II -u nominalizations to be able to actualize interpretations of the action of the nominalized verb in a scale of time (via the aspectual construal). Compare: (22) a. kana-u sleep+IPFV-NMLZ ‘one who sleeps’ or ‘one who is sleeping’ b. kana-ka-u sleep.PFV-INTENS-NMLZ ‘one who slept’ c. kanu-inu sleep-NMLZ ‘sleeper’ 705 In (a) and (b), we can obtain interpretations of the nominalized verb as being currently occurring or already occurred, respectively. By contrast, in (c) no such interpretation(s) is/are possible, because the action of the nominalized verb is now a property of the referent of the derived noun. This di$erence has important consequences in the grammar: in Chapter XX, we will see that Set II nominalizers are used primarily in modifying function (i.e. relative clauses), whereas Set I nominalizers are used primarily in predicative functions. 15.4.6. Set II nominalizer -mau, ‘non-subject nominalizer’ Three main derivational properties are related to the su#x -mau: a) it creates a noun that is understood as the patient or location of a verb; b) it creates a noun that is understood as an action nominalizer; c) -mau also creates a stem to create locative nominalizations, most frequently with the locative =nVma. To contrast -mau with the other member of Set II, -u ‘subject nominalizer’, I use the term ‘non-subject nominalizer’ to refer to -mau. Historically, -mau likely developed from an old su#x *-ma plus the nominalizer -u described in the preceding section. A likely source is the same morpheme that is the modern the non-canonical switch-reference marker -ma ‘subject to object’ (see §19.10.1). Similarly to -u, -mau can attach to an aspectualized stem. Examples of 706 derivation with -mau are presented in (23). (23) nakuma ‘draw’ → nakumɨ-a-mau ‘drawing (‘something that is being drawn’) (draw.IPFV-IPFV-NMLZ) t͡sua ‘heal’ → t͡sua-a-mau [t͡suwaḿu] ‘one that is healed’ (heal-IPFV-IPFV- NMLZ) amu ‘"nish’ → amu-ka-mau ‘"nished, exterminated’ ("nish-INTENS-IPFV- NMLZ) nakuma ‘draw’ → nakuma-ka-mau [nakúmkamu] ‘drawing (something that is already drawn’ (draw-INTENS-IPFV-NMLZ) The nominalizer -mau also appears to derive nouns whose interpretation seems to be that of an action nominalizer. (24) ha ‘be sick’ → ha-a-́mau [haaḿu] ‘death’ (be.sick-HIAF-NMZL) uutu ‘cry’ → uutu-mau [uútmoo] ‘cry (N)’ ɨmɨma ‘be proud’ → ɨmɨma-mau [ɨmɨmamu] ‘pride’ In relation to its ability to create action nominalizations, -mau is sometimes also used to give a citation form of verbs. The nominalizer -mau is also used to derived nouns that denote a location of the action of the verb, cf. akiina ‘be born’ → akiina-mau [akiíɲamu] ‘birth’ or ‘place where one is born’; arakama ‘plant (V)’ → arakama-a-mau [araḱmamu] (plant-IPFV-IPFV-NMLZ) ‘what is planted (i.e. seeds, etc.)’ or ‘farm (sown "eld)’. In most cases, however, -mau needs to carry a locative marker to derive location nominalizations, most frequently - 707 mau occurs with the locative =nVma, as in shown (25). (25) puha ‘live+IPFV’ → puha-mau=nama [puhaḿunam] ‘where we live’ amu ‘"nish’ → amu-a-mau=nama [amúamunam] ‘where something the ends’, ‘end’ ("nish-IPFV-NMLZ=LOC) The use of -mau in locative relativizations is described more in detail in §20.2.2 and §20.4 15.4.7. Negative nominalizer -t ͡ʃau The nominalizer -t͡ʃau can occurred attached to a root or aspectual stem. It negates the element to which it attaches. (26) t͡ʃit͡ʃa ‘speak’ → t͡ʃit͡ʃa-t͡ʃau [t͡ʃit́͡ʃat͡ʃu] ‘quiet person’, ‘mute’ kaka ‘be strong’ → kakait͡ʃau [kakɛít͡ʃu] ‘weak one’, ‘lazy one’ nɨka ‘know’ → nɨkat͡ʃau [nɨkat͡ʃau] ‘ignorant’ (‘one who does not know’) surima ‘be stingy’ → surima-t ͡ʃau [suriḿt͡ʃoo] ‘generous person’ The nominalizaer -t͡ʃau have developed from the negative verbal su#x -t͡ʃa and Set II -u ‘agent nominalizer’. The nominalizer -t͡ʃau can also combine with adjectives, other nouns and some adverbs, which indicates that formative -t͡ʃa in the nominalizer - t͡ʃau is no longer analyzed by speakers as a verbal-only morpheme, hence -t͡ʃau is analyzed as a synchronically unanalyzable morpheme. 15.4.8. Nominalizer -na The nominalizer -na does not have a productive derivational function, but for 708 sake of completeness it is described here. The nominalizer -na occurs with the potential stem (formed with -mai), as shown in (27), and relativizing verbs after the person marker, as shown in (28). (27) tumɛíɲa tu-mai-na say-POT-NMLZ ‘what can be said’ (28) ií puhah́ɲa húu kanús apat́͡ʃka rió santiáɣo tuiɲ́awɛɛ ii puha-hi-na hu kanusa 1PL live-1PL.SBJ-NMZL PROX Kanus apat ͡ʃa=ka rio santiago tu-ina-ua-i mestizo=FOC rio Santiago say-PL.IPFV-3.SBJ-DECL ‘Where we live, this Kanus river, the mestizo (=non-Wampis) call it Santiago river.’ 709 CHAPTER XVI SINGLE VERB CONSTRUCTIONS WITH LEXICAL VERBS, AND NOTES ON WORD ORDER AND ALIGNMENT 16.1. Introduction Chapter XVI explores the syntactic constructions which involve only one lexical verb as a main "nite verb. The discussion begins with some notes on word order in §16.2, followed by a description of the Wampis alignment pattern in §16.3, which constitutes a typologically uncommon syntactic pattern. Next, §16.4 describes the intransitive construction in Wampis. After a short note on copular constructions §16.5, §16.6 and §16.7 describe the transitive and ditransitive construction in Wampis, respectively. Finally, §16.8 and §16.9 describe the Wampis quotative and possession constructions, respectively. 16.2. Notes on constituent order In Wampis, the most basic verbal construction is a simple intransitive clause; that is, an intransitive verb marked for aspect, tense, person, and mood. The verb can stand alone to achieve a predication, not other constituent is needed. 710 (1) kana-ra-ma-ha-i sleep.PFV-DISTR-IMM.PT-1SG.SBJ-DECL ‘I slept.’ While a conjugated verb is enough to have a minimal verbal construction, a number of other elements can occur in a simple declarative clause. Apart from Subject and Object NPs, additional information can be expressed via postpositional phrases and adverbs of manner, time, location, demonstrative adverbs and intensi"ers. [(Adv) (Subj) (Obj) (Obl) (Adv) V (Obl) ] Figure 16.1. Wampis simple lexical verb word order The basic constituent order in Wampis is predicate "nal. In basic, pragmatically unmarked declarative clauses, the word order is A P V, where A= most Agent-like participant, P= most Patient-like participant, and V= Verb . However, the Wampis language allows for other order of subjects and objects. Adverbs and Oblique arguments are relatively free to occur before or after the predicate, but their preferred position in terms of frequency is pre-verbal. 16.3. Alignment and grammatical relations Grammatical relations concern the nature of the relationship between a 711 predicate or a construction and its core arguments: the single argument of intransitive constructions (S), the most agent-like argument (A) and the most patient-like arguments of transitive constructions (Comrie 1978). Alignment has to do with the patterns by which a language establishes the grammatical relations of those core arguments. Alignment may be manifested in a variety of coding structures including contituent order, participant-reference marking on verbs, case marking and various morphosyntactic combinations thereof. The following discussion concerns the marking of argument NPs in Wampis. The language exhibits also an uncommon pattern of argument indexation on the verb, which was described in Chapter XIV. Wampis exhibits a nominative-accusative alignment. S/A arguments are treated di$erently from P. Importantly, Wampis distinguishes between subjects and non- subjects, i.e. Wampis exhibits a symmetrical objects system (Bresnan & Moshi 1990). The coding properties and syntactic behavior of objects in Wampis (notional direct and indirect objects, objects of applicative) are identical. In Wampis S/A as a category is relatively straightforward. S/A Noun Phrases are treated in the same way at the syntax level. The marking of the subject (“Nominative” case) is zero. Aspect, de"niteness, deixis or other category do not play any relevant role 712 in the marking of A/S NPs. The following examples illustrate the marking of the single argument of an intransitive verb (S) for di$erent grammatical persons in (2)–(5), and the marking of the Agent-like argument of a transitive verb (A) for di$erent grammatical persons in (6)–(9). (2) S ui uɨ-a-ha-i 1SG go-IPFV-1SG.SBJ-DECL 'I am going' (3) S amɨ uɨka-sa-mɨ 2SG walk-ATT-2SG.SBJ+DECL 'You walked.' (4) S Puhupata uiʃi-a-ua-i Puhupata laugh-IPFV-3SG.SBJ-DECL 'Puhupat laughs.' (5) S Puhupata uiʃi-ra-ma-ji Puhupat laugh-DISTR-REC.PT-3.PT+DECL 'Puhupat laughed.' 713 (6) A P amɨ mi=na uaina-ka-ma-mɨ 2SG 1SG=ACC see-INTENS-REC.PT-2SG.SBJ+DECL ‘You saw me’ (7) A P amɨ mi=na uaina-a-ua-mɨ 2SG 1SG=ACC see-IPFV-2SG.SBJ+DECL ‘You are seeing me.’ (8) A P Puhupata nua=na tukuma-a-ua-i Pujupat woman=ACC kick-IPFV-3.SBJ+DECL ‘Pujupat is kicking the woman.’ (9) A P nua puhupata=na tukuma-ka-ma-ji woman Pujupat=ACC kick-INTENS-REC.PT-3.PT+DECL 'The woman kicked Pujupat' The notion of subject is de"ned by the following criteria: • Nominative (zero) case marking. • Marking on the verb with the regular patterns explained in §14.3.3.2. • Nominalizers that only codi"ed semantic notions that in Wampis are instantiated as 714 grammatical subject (Set I -inu and Set II -u). • Robust same and di$erent subject marking in subordinate clauses. • Controller of switch-reference -ma ‘non-subject to subject’ (cf. §19.10.1) Turning now to the discussion of the treatment of the most-patient like argument (P), this is another area where Wampis, as other Jivaroan languages (Corbera Mori 1994; Overall 2007; Gnerre 2010), exhibits a very unique pattern. The following principles apply to the marking of objects in Wampis: • First and second person NP objects always receive the accusative marker =na. • Third person NP objects receive the accusative marker =na only if the subject is 1sg, 3sg or 3pl person. • If the subject is 1pl, 2sg or 2pl, the third person object NP is left unmarked. Table 16.1 summarizes the marking of objects in Wampis. A zero symbol “∅” means that the object is not marked due to a pattern that will be discussed below. A straight line means “Not Applicable”. =na means that the object is marked with the accusative. 715 Table 16.1. Summary of object NP marking A↓ \ P→ 1sg P 1pl P 2sg P 2pl P 3 P 1sg A _______ =na =na =na =na 1pl A _______ _______ =na =na ∅ 2sg A =na =na _______ _______ ∅ 2pl A =na =na _______ _______ ∅ 3 A =na =na =na =na =na Examples (10)–(15) illustrate the marking of A and P with di$erent grammatical persons. The A arguments are 3, 1sg, 1pl, 2sg and 2pl, respectively. The P arguments are 1sg, 2sg, 2pl and 1pl, respectively. It can be seen that all object NPs (i.e. P) are marked when an A of any grammatical person acts upon a Speech Act Participant P. (10) A=3, P=1sg A P ni ̃ mi=na uaina-tu-ka-ma-ji 3SG 1SG=ACC see-1SG.OBJ-INTENS-REC.PT-3.PT+DECL ‘He saw me.’ (11) A=3, P=2sg A P ni ̃ ami=na uaina-tama-ka-ma-ji 3SG 2SG=ACC see-1PL/2.OBJ-INTENS-REC.PT-3PT+DECL ‘He saw you.’ 716 (12) A=1sg, P=2sg A P ui ami=na uaina-ka-ma-hamɨ 1SG 2SG=ACC see-INTENS-REC.PT-1SG>2SG+DECL ‘I saw you.’ (13) A= 1pl, P=2pl A P ii atumi=na uaina-ka-ma-hirumɨ 1PL 2PL=ACC see-INTENS-REC.PT-1PL>2PL+DECL ‘We saw you (pl).’ (14) A= 2sg, P=1sg A P amɨ mi=na uaina-tu-ka-ma-mɨ 2SG 1SG=ACC see-1SG.OBJ-INTENS-REC.PT-2SG+DECL ‘You saw me.’ (15) A=2pl, P=1pl atumi ii=na uaina-tama-ka-ma-rumɨ 2PL 1PL=ACC see-1PL/2.OBJ-INTENS-REC.PT-2PL.SBJ+DECL ‘You (pl) saw us’. Examples (16)–(19) illustrate the occurrence and non-occurrence of =na with NP objects. In (16) and (17), where the A is 3sg and 1sg, respectively, the NP object is marked with the accusative =na. On the other hand, in examples (18) and (19), the A is 2sg and 1pl, respectively, and the object NP does not receive accusative marking with =na. Thus, when an 1pl/2 A acts upon a Non-Speech Act participant P, P is left 717 unmark. (16) A=3sg, P=3 P A iauaã=na mã-á-ma-ji jaguar=ACC kill-HIAF-REC.PT-3.PT+DECL ‘He killed the jaguar.’ (17) A=1sg, P=3 P A iauaã=na mã-á-ma-ha-i jaguar=ACC kill-HIAF-REC.pt-1SG.SBJ-DECL ‘I killed a jaguar.’ (18) A=2sg, P=3 A P A amɨ iauaã mã-á-ma-mɨ 2SG jaguar kill-hiaf-REC.PT-2SG.sbj+DECL ‘You killed a jaguar.’ (19) A=1pl, P=3 P A iauaã mã-á-ma-hi jaguar kill-HIAF-REC.PT-1PL.SBJ+DECL ‘We killed a jaguar.’ From the above examples, it is evident that the marking of P in Wampis depends not only on the properties of the argument itself, but also depends on the nature of the other argument, A (most-agent-like argument). Two cross-linguistically uncommon 718 features of Wampis marking of NP-objects can be stated: • The structural marking itself is uncommon, since as shown in the above examples, it contains a “split” between 1SG and 1PL Speech Act participants (1pl/2 conditions the non-marking of the 3 NP object, whereas 1sg marks NP objects) . • This “split” is not based on the nature of the argument (i.e. speci"city, de"niteness, animacy, etc—categories usually understood as related to the notion of Di$erential Object Marking (Bossong 1983-1984; Aissen 2003)), but on the relationship with the other argument. • Interestingly, in Wampis the marking of P depends on a hierarchy in which what matters is “who acts upon whom”, and not what is the relationship of P to the verb. The uncommon syntactic case marking pattern of Wampis parallels the morphological patterns that are found in some Hierarchical agreement systems (DeLancey 1981; Siewierska 2003; Zuñiga 2006; DeLancey Forthcoming). Interestingly, Wampis also exhibits an uncommon type of Hierarchical agreement system that is di$erent from case marking (cf. §14.3.3). Thus, the following hierarchy appears to condition grammatical relations (and case-marking of P) in Wampis: 1PL/2 > 1SG > 3 719 1pl/2 are always marked as P, but as A do not trigger case marking of 3P. 1sg is also always marked as P, and as A always triggers case marking of P. 3 is sometimes marked as P, and as A always triggers case marking of P. There is one important coincidence between the hierarchical agreement system (as analyzed in §14.3.3) and case marking in Wampis: in the verbal hierarchical agreement, the same su#x -tama or -rama (see §13.2.4) is used for 1pl and 2sg, 2pl persons, which are the grammatical persons that trigger the “no-case marking” of 3 person Object NPs. So it seems like 1pl and 2 persons are (or were at one point in the past) construed as highly salient in discourse. Wampis seems to avoid marking situations where the Agent is somehow construed as related to the 2 person (singular or plural), i.e. the relationship between Speaker and the other Speech Act Participant appears to be avoided somewhat. This relates to other parts of the grammar of Wampis: for instance, we saw that the words for ‘mother’ and ‘father’ are never marked for second person possession (cf. §10.4.1.4), and patterning together of 1sg and 2 also occurs in other morphemes, such as the Plural Speech Act =ti ́(§11.5.6). 16.4. Intransitive constructions Intransitive constructions in Wampis have the valence requirement of one 720 argument, which is the grammatical subject. The intransitive constructions involve one participant and a "nal verb in a predication of an action, state or event. The single argument of an intransitive construction (S) receives the nominative, which is zero- marked (§10.4.6.1). An oblique argument may be present, as in (22) but is not obligatorily required by the construction. The overt NP may or may not occur in the construction, but the subject is always obligatorily marked on the verb. (20) atiĺio wɨḿaji Atilio uɨ-ma-ji Atilio go-REC.PT-3.PT+DECL ‘Atilio went.’ (21) ðińa úutɨawɛɛ Dina uutɨ-a-ua-i Dina cry.IPFV-IPFV-3.SBJ-DECL ‘Dina is crying.’ (22) ɨnt͡sańam jukúmaji ɨnt͡sa=nama jukuma-ji river=LOC swim+IPFV-3.PT+DECL ‘I am going to swim in the river.’ 16.4.1. Existential construction The existential is a lexical verb but because it is homophonous with the copula verb a and exhibits some important distributional properties relatable to the copula, it 721 is analyzed in detail in Chapter XVII, which is dedicated to possessive, existential, locational, attributive and equational constructions. 16.4.2. Weather constructions As far as I can tell there is no avalent verbs in Wampis. Typical examples of avalent verbs in other languages, such as meteorological verbs, are syntactically intransitive verbs that codify an S argument: (23) násɨ ampúawai […] nasɨ ampu-a-ua-i wind blow-IPFV-3.SBJ-DECL ‘The wind is blowing…’ (24) júmi ʃíir jutúkti iumi ʃiira iutu-ka-ti rain very rain-INTENS-JUSS ‘That the rain rains hard!’ Most verbs which can be used to communicate a weather event or state occur in intransitive constructions with a cogent subject. If a weather predication does not have an overt cogent subject , the interpretation of the grammatical subject may be problematic in some cases. For instance, if the verb umpu ‘blow’ in (23) occurs without an overt subject in the predication, it is not possible to know what or who blows. In those cases a weather event interpretation is problematic in Wampis. 722 16.5. Copular constructions Copular constructions are fairly used in the language. Copular constructions analyzed in great detail in Chapter XVII. 16.6. The transitive construction The Wampis transitive construction has a valence of two; that is, two arguments that are grammatically manifested as subject and object. The most frequent word order is Subject-Object-Verb (APV), although other orders are permitted in certain pragmatic environments. Objects marked for focus normally occur in initial position and the Subject occurs in post-verbal position (see (27) for an illustration of this pattern). The marking of overt noun phrases depends on the grammatical relation provided by the construction. Subject NPs are zero-marked, and all Speech Act Participant Object NPs are marked with the accusative =na. Third person Object NPs are not marked if the subject is a Plural Speech Act participant, according to the hierarchy observed in §16.3. (25) puhupat́ kaʃɛɛń maãḿiaji Puhupata kaʃai=na ma-̃a-́mia-ji Pujupat paca=ACC kill-HIAF-DIST.PT-3.PT+DECL ‘Pujupat killed a a paca’ 723 (26) paḱi ɨsɛińi ̃puhúpatan paki ɨsa-ini-i ̃ puhupata=na peccary bite-LOAF-3.PFV+DECL Pujupat=aCC ‘A peccary just bit Pujupat.’ (27) Suwaŋ́ka kuŋkuaśmaji Anðreś sua=na=ka kunkua-sa-ma-ji Andres Sua=ACC=KA kiss-ATT-IMM.PT-3.PT+DECL Andres ‘Andres kissed Sua.’ In text data, quotative constructions seem to favor post-verbal positioning of arguments. When there are two overt arguments in a quotative construction, the prefer order of the subject is post-verbal, and objects may occur at the end; i.e. they follow the order [Quotation V A P]. Quotations themselves are always direct quotations, they are never marked as arguments. See also §16.8. (28) jɛinkʲata ́tusa ̃t͡sɨŕɨ iwań iana-ka-ta ́ tu-sa ̃ t͡sɨrɨ iua=na help-INTENS-IMP say-SUB\3SG.SS Tsere Iwa=ACC ‘“Help me!” saying Tsere to Iwa.’ Semantic goals with predications of movement (‘go’, ‘come’, etc.) do not require an oblique argument. Constructions with emotion verbs do not “re-arrange” the mapping of semantic roles onto di$erent syntactic categories; that is, the location, experiencer or undergoer of an emotion predicate is instantiated as the grammatical 724 subject, and the theme or stimulus is instantiated as the grammatical object. 16.7. The ditransitive construction The ditransitive construction in Wampis can be de"ned as a transitive construction with a valence of three arguments. The three arguments are an Agent (A) codi"ed as the subject, a Theme (T) codi"ed as an object, and a Recipient (R) codi"ed as another object. There is no especial marking distinguishing T and R: both Theme and Recipient are treated identically: all objects in Wampis receive the accusative =na. Third person objects are zero marked on the verb. When there is a second person or "rst person object, it is marked on the verb, that includes second or "rst R arguments. The order of constituents in a ditransitive construction is A T V R. Other frequent order is A T R V, and A V T R is possible. On the other hand, the order R T V appears to be very disfavored. (29) A T R [Puhupat] [ukunt͡ʃi]=na su-sa-ji [iauaa]̃=na Puhupata bone=ACC give-ATT-3.PFV+DECL dog=ACC ‘Puhupat gave a bone to the dog.’ 725 (30) A T R [Puhupat] [ukunt͡ʃi]=na [iauaa]̃=na su-sa-ji Puhupata bone=ACC dog=ACC give-ATT-3.PFV+DECL ‘Pujupat gave a bone to the dog.’ The Recipient can receive the benefactive -nau optionally, but it still needs to be marked with the accusative. (31) A T R Pujupat ukunt ͡ʃi=na iauaa-̃nau=na su-sa-ji Pujupat bone=ACC dog-BEN=ACC give-ATT-3.PT+DECL ‘Pujupat gave a bone to the dog.’ 16.8. Quotative construction The quotative construction requires a speech verb and codi"es two core arguments, A and P, instantiated as the grammatical subject and object. The quotation itself is not mark as an argument grammatically and it is always a direct speech report. Quotative constructions are used in semantically equivalents of some complement constructions (see §20.3.3 for more details). (32) P ami=na sɨɨ ta-hamɨ 2SG=ACC thanks say-1SG>2SG+DECL ‘I tell you “thanks!” 726 16.9. The possession predication construction Possession predication construction can be done with the existential+applicative construction (see §17.5 for details), and in verbless clauses with the attributive -tinu (see §17.4.1 for details). There is one more lexical verb (the other is the existential a) that can be used to predicate possession: takaka ‘have’. This verb follows a simple transitive pattern. 727 CHAPTER XVII POSSESSIVE, EXISTENTIAL, LOCATIONAL, ATTRIBUTIVE AND EQUATIONAL CLAUSES 17.1. Introduction This chapter describes constructions that Wampis employs to express semantic functions such as possession, existence, location, attribution and equation. It includes an analysis of non-verbal clauses, as well as of clauses involving the copula verb, the copula clitics, the existential verb as well as other verbs that are used in copular and semicopular function. The structure of this chapter is: §17.2–§17.3 give a brief overview of the functions and the structure of non-verbal predication; §17.4 analyzes non-verbal (in the sense of truly verbless) clauses; §17.5 is dedicated to copular and existential predications; "nally, §17.6 mentions other verbs that have assume copular or semicopular functions. 17.2. Overview of functions of non-verbal predication Most languages have constructions that express proper inclusion, equation, attribution, location, existence, and possession. Following Payne (1997), proper 728 inclusion indicates that a particular entity belongs in a class of items speci"ed by the predicate (e.g. she is a teacher). Equative clauses serve to state that a speci"c entity is identical to the particular entity in the predicate (e.g. she is my mother). Attributive clauses predicate a property or attribute to the referent (e.g. she is tall). Locational clauses indicate a typically de"nite entity as being in a speci"c location—the location is the predicate (e.g. she is in the house). Existential clauses indicate the existence of usually an inde"nite entity (e.g. there is a football game) and may be accompanied by a temporal or locational reference. In discourse, existential clauses typically introduce a new participant. Wampis follows the locative/existential pattern to predicate possession and benefactive (e.g. this is for Shahar), a pattern that is not uncommonly heard across languages (Clark 1978; Stassen 2009). 17.3. Overview of the structure of non-verb lexical predications in Wampis A copular verb is a function word that links the subject with its complement. Copular verbs frequently do not possess much lexical content and much of the lexical semantics of the predicate typically resides in the complement of the copula. There are both copular and non-copular “nonverb” predicate constructions in Wampis. Firstly, non-copular constructions juxtapose a predicate constituent and a NP. 729 The predicate constituent can be either an AdjP or another NP. Secondly, non-verbal predicates in Wampis may occur with a copula a or copula clitics =aita~=ita (for speech act participants) and =aiti~=iti (for third person) which can be translated as ‘be’. It will be seen that the distinctions in the distribution of the copula verb a and the copula clitics are based on TAM restrictions and on whether the clause is subordinated or not (copula clitics never occur in subordinated clauses). In addition, Wampis distinguishes the copula a from an existential verb a. Both verbs are homophonous but morphosyntactically distinct (see §17.5.1). Thirdly, there are full lexical verbs that have been grammaticalized to assume copular functions. These verbs are used in existential/locative predicates; the most frequently used in this function are puhu ‘live’, iruna ‘crowd together, pile up, be dispersed’, and mat͡satu ‘live together’. Other full lexical verbs that are used in locative constructions are posture verbs such as tɨpɨ ‘lie down’ and uaha ‘stand’. Other verbs that are structurally used in copular constructions can be characterized as what have been termed semi-copulas (Hengeveld 1992). Wampis does not have a dedicated verb whose original lexical meaning is ‘become’. Instead, it uses full lexical verbs nahana ‘make’ and uɨ ‘go’, which in copular constructions assume the 730 meaning of ‘become’. 17.4. Non-verbal clauses 17.4.1. Juxtaposition constructions: equative, proper inclusion, attributive In general, a juxtaposition construction consists of a sequence of two elements without any relational or copular element between the two (a zero copula strategy). In Wampis, an NP and NP/AdjP can be juxtaposed to form a verbless clause. In the data used for this study, the "rst NP is always the subject and the second NP/AdjP is always the predicate. If the second constituent is an NP, it can be headed by a simple noun or a nominalized verb, or it can be marked with the attributive -tinu (see below). Juxtaposition constructions are used for equative (or identi"cation), proper inclusion and attributive predications. The following examples constitute instances of equative constructions. The NP subject can be a full NP (1), a demonstrative (2) or a pronoun (3): (1) núu úun míɲa papár [nu úunta] [mina papa-ru] NON.VIS elder 1SG.GEN father-1SG.POSS ‘That elder is my father.’ (2) hũṹ míɲa kumpár [hũ] [mi-na kumpa-ru] PROX 1SG.GEN friend-1SG.POSS ‘This is my friend.’ 731 (3) níŋkʲa míɲa kumpár [ni=̃ka] [mi-na kumpa-ru] 3SG=FOC 1SG.GEN friend-1SG.POSS ‘He is my friend.’ The following examples constitutes instances of proper inclusion using juxtaposition of NPs. In (4), the predicate of the clause is a nominalization. In this example, a speci"c entity (Nantu ‘Moon’) is being predicated to belong to a class (i.e. ‘hunter’). In (5), the subject identi"es himself as belonging to a class designated by the noun Wampisa. (4) Nántuka ɨaḱmou [Nantu=ka [ɨakama-u] Moon=FOC look.for.game-NMLZ ‘Nantu [was] a killer of animals.’ (i.e. ‘Moon hunted animals’) (5) wíkʲa ʃuár wampís [ui=ka] [ʃuara wampisa] 1SG=FOC person Wampis ‘I [am] a Wampis person.’ Attributive constructions can have an AdjP or an NP (including nominalizations) as their predicate. Examples (6) and (7) show a juxtaposition where the predicate is an AdjP, whereas (8) shows a juxtaposition structure where an NP constitutes the predicate 732 of the clause. (6) Kanús saár [Kanusa] [saara] Satiago River transparent ‘The Santiago river [is] transparent.’ (7) Wámpuka tíi sɨńt͡ʃi [Wampuka] [ti sɨnt͡ʃi] Wampuka INTS strong ‘The Wampuka246 [are] very strong.’ (8) núu úun umín [nu úunta] [umi-inu] NON.VIS elder "nish-nmlz ‘That elder [is] reliable.’ (Lit: ‘That elder [is] someone who "nishes his job/duty’) 17.4.2. Negation in juxtaposition constructions It is possible to negate juxtaposition constructions. However, there is a preference among Wampis speakers to use the copula a or copula clitics (depending on the morphosyntactic environment as described below) in negative clauses. It is more common to "nd negative juxtaposition constructions in attributive function than in equative and proper inclusion functions. This is probably due to the nature of a good 246. Native people of the forest who appear in folktales. The Wampis considered them to be un- contacted people who actually exist. The Wampis tell stories about seeing them while hunting or traveling in the forest. 733 portion of Wampis adjectives and modifying words that have their opposite terms simply coined by addition of the negative nominalizer -t͡ʃau (as explained in §9.5.2); e.g. kuitarit̃inu ‘rich’ vs. kuitari-̃t͡ʃau ‘poor’;247 pɨnkɨra ‘good’ vs pɨnkɨra-t͡ʃau ‘bad’. In any case, the main strategy of negation in juxtaposition constructions is the use of the negative nominalizer -t͡ʃau with the nominal or adjectival predicate. Thus, the negative construction of juxtaposing predicates is [NP NP/AdjP-NEG], as examples (9)– (10) show. (9) ɨmɨśtumamu pɨŋ́kɨrt͡ʃoo ɨ-mɨsa-tu-ma-mau pɨnkɨra-t͡ʃau CAUS-ruin-APPL-REFL-NMLZ good-NEG.NMLZ ‘Ruining (your own future) [is] bad.’ (10) núka aɨnt͡suka kakaḱt͡ʃoo nu=ka aɨnt͡su=ka kaka-ka-t͡ʃau NON.VIS =FOC man=FOC be.strong-INTENS-NEG.NMLZ 'That man is not strong.' As said above, equative and proper inclusion juxtaposition constructions are less commonly found in natural speech if negated. For instance, in example (11) below, a copula clitic is used. 247. The term kuitarit̃inu [kuitrintin] ‘wealthy’ (lit. ‘owner of his money’) is composed of kuita ‘money’ +-ri ̃1PL/2PL.1PL/2PL/3.POSS +-tinu ‘attributive’. The opposite term for ‘poor’ replaces the attributive -tinu with the negative -t͡ʃau. 734 (11) tihʲɛɛ́ ʃuárt͡ʃouwɛitʲi tihai ʃuara-t ͡ʃau=ait-mɨ tijai person-NEG.NMLZ=COP-2SG.SBJ+DECL ‘[…] you are not a Tijai248 person.’ 17.4.3. The construction NP + NP-tinu: possession When the attributive su#x -tinu attaches to the second NP, it indicates that it is possessed by the "rst NP. (12) ut͡ʃiŕtin ut͡ʃi-ri-̃tinu child-1PL/2PL/3.POSS-ATTRIB ‘with child’ (i.e. a father) In the possession construction with -tinu, a copula may optionally (but not necessarily) occur. Notice that -tinu does not derive a verb: the noun derived by -tinu never receives any type of verbal morphology and continues to be a noun. Thus, the construction with -tinu is a non-verbal clause type. Examples (12) and (13) illustrate non-verbal clauses where a -tinu-marked NP occurs. (13) kuɲáu Andrés ahártʲin kuɲau249 Andrés aha-ri-̃tinu brother-in-law Andres farm-1PL/2PL/3.POSS-ATTRIB ‘Cuñado Andrés has a farm.’250 248. In the Wampis folklore, human-like beings that inhabit the forests. 249. The term kuɲau is borrowed from Spanish (colloquialy pronounced [kuɲao] in Peruvian Spanish). In Peruvian Spanish, is used in the sense of ‘brother-in-law’ but colloquially is commonly used as ‘friend’. 250. This sentence was translated to Spanish as Cuñado Andrés es con su chacra ‘Andrés is with his 735 (14) óoka nuwɨńtin au=ka nuɨ-̃tinu DIST=FOC woman\1PL/2PL.3.POSS-ATTRIB ‘He is married.’ A common use of the attributive construction is to predicate people’s names, as illustrated in the next example: (15) ínt͡ʃis naártin Int͡ʃisu naa-ri-̃tinu Int͡ʃisu name-1PL/2PL.3.POSS-ATTRIB ‘Her name [was] Inchis,’ As mentioned earlier, a copula can occur after -tinu: (16) at͡ʃúrtiɲaithɛi at͡ʃu-ru-tinu=aita-ha-i aguaje-1SG.POSS-ATTRIB=COP-1SG.SBJ-DECL 'I have an aguaje (Mauritia 0exuosa) farm.' 17.4.4. Tense reference and juxtaposition Juxtaposed structures can have a past or present temporal grounding, as shown in previous examples. However, juxtaposition constructions are not attested as having a future interpretation in Wampis. A copula marked for future needs to be added in order to obtain a future tense interpretation (and therefore, in such case, the clause is not farm’ by my Wampis teachers (this translation is more faithful to the Wampis original). After I asked for clari"cation, my Wampis teachers opted to change the translation to tiene una chacra ‘has a farm’). 736 verbless). For instance, a possible future counterpart to (14) above is given in (18), where the copula a occurs and carries the immediate future marker -ta. (17) óoka nuwɨńtin at́awɛɛ [au=ka] [nuɨ-̃tinu a-ta-ua-i] DIST=FOC woman/1PL/2PL/3.POSS-ATTRIB COP-IMM.FUT-3.SBJ-DECL ‘He is going to be married.’ 17.5. Copular and existential predications 17.5.1. Brief overview of the copulas and the existential In general, copular verbs in many languages tend to be somewhat “irregular” in comparison with prototypical lexical verbs (verbs that receive regular conjugation and that usually do not have many morphosyntactic or semantic restrictions). Cross- linguistically, apart from their tendency to be irregular, copular verbs tend to belong in the semantic class of very stative verbs, and they tend to function as auxiliary verbs in other constructions (Payne 1997: 116). Copular clauses can predicate permanent or temporary states and their subjects typically have the semantic role of a patient. Syntactically, in Wampis, the subject of the copula construction is always nominative and the complement is either an AdjP or a nominative NP. Wampis possesses: 737 • A “be” copula a which occurs in non-present tense declarative contexts • Several copula clitics that are restricted to occur in present tense • A past tense copula clitic that occurs rarely in my data • An existential a (homophonous to the “be” copula a but morphosyntactically distinct). The historical relationship between the copula a and the existential a is quite possibly certain; however, synchronically they are completely di$erent verbs. Table 17.1 summarizes the principal di$erences found between the copula a, the copula clitics, and the existential a. Table 17.1. Distinctions between copula a, copula clitics and existential a Copula a Copula Clitics Existential a Function proper inclusion, equation, attribution, possession/benefactive (with possessive -nau) proper inclusion, equation, attribution location, existential, possession (with the applicative form -ru) Plural Takes plural -ina or - ara Does not take plural marking Takes plural -ia (except in 3rd person plural, which takes -ina) 738 TAM restrictions Does not occur in present declarative and polar/content interrogative Occur only in present tense declarative, polar/content interrogative and exclamative No restriction Occur in subordinating structures Yes, takes subordinating su#xes Main clause only Yes, but does not take subordinating su#xes itself Negation Marked with -t͡ʃa Not marked on the clitic (negative nominalizer -t͡ʃau marks the noun or adjective stem to which the copula clitic then attaches) Marked with -t͡su In the next sections, the morphosyntactic di$erences outlined above are discussed in order to better understand the behavior of the copula a, copula clitics and the existential. An analysis of their functions follows afterward. 17.5.2. Morphosyntactic distinctions 17.5.2.1. The expression of TAM categories With regard to TAM categories, the following principles distinguish the copula a, the copula clitics and the existential a in Wampis (they are explained in the subsequent discussion, below): • The copula a is used in non-present declarative moods (including past and future 739 tense references), tag questions and subordinated constructions. • In contrast, the copula clitics are used in present tense declarative, in content and polarity questions, and exclamative clauses. The exception to this copula clitic restriction is one copula clitic that is used in past tense: =ia. • The existential a has no major TAM restrictions. The following sentences are examples of di$erent non-present declarative moods where the copula a is used. They include instances of the copula carrying the distant past marker (18), a future nominalizer (19), the jussive su#x (20), and the hortative (21). (18) íiɲa úuntrika tíi sɨnt͡ʃi aŕmiayi iina uunta-ri=ka tii sɨnt͡ʃi a-ara-mia-ji 1PL.GEN elder-1PL/2PL/3.POSS=FOC INTENS strong COP-PL-DIST.PT-3.PT+DECL ‘Our ancestors were very strong.’ (19) ipʲákka átiŋkʲa […] nuŋká puhusú ipaku=ka a-tinu=ka nunká puhu-sa-u annatto=FOC COP-FUT.NMLZ land\LOC live-ATT-NMLZ ‘What was going to be Annatto tree […] was one to sit in the ground.’ (20) nú áti nu a-́ti NON.VIS COP-JUSS ‘Let that be so.’ 740 (21) nekáska núu ámi ipʲák nekas=ka nu a-mi ipaku truly=FOC NON.VIS COP-HORT annatto ‘Truly, let us be the annatto tree.’ (i.e. ‘Let’s transform into the annatto tree.’) In present tense declarative (22) and interrogatives (23), the copula clitics are used. Notice that the copula clitics do not occur in other moods, except in exclamative (24). (22) ɛɛ́ʃmaŋkuitmɨ áiʃmanku=ita-mɨ man=COP-2SG.SBJ+DECL ‘You are a man.’ (23) jɛɛ́tʲam ia=ita-mɨ who=COP-2SG.SBJ ‘Who are you?’ (24) mijá pɨŋ́kɨrɛitʲa! mia pɨńkɨra=ita INTS good=COP.EXCL ‘How beautiful [it] is!’ The copula verb can occur in questions only if they are not present tense declarative, or if there is no indication of temporal ground. In the following example, the speaker wonders why the hummingbird has not appeared and uses the copula a 741 subordinated with the su#x -sa:251 (25) hɨmpɨʃ́a urúka ásan naŋkámat͡ʃmia? hɨmpɨ=ʃa uru-ka a-sã nankama-t͡ʃa-mia? hummingbird=ADD how-Q COP-SUB/3SG.SS pass.through-NEG-DIST.PT ‘Why the hummingbird did not appeared?’ An interesting overlap in the use of the copula verb and copula clitics occurs with plural subjects. In non-attributive clauses, both the copula verb and the copula clitics can be used: (26) wabálnumiajɛithi Huabal=numa=ia=aita-hi Huabal=LOC=ABL=COP=1PL.SBJ+DECL ‘We are from Huabal.’ (27) wabálnumia ɛɛ́ʲɲahi Huabal=numa=ia a-ina-hi Huabal=LOC=ABL COP-PL.IPFV-1PL.SBJ ‘We are from Huabal.’ On a side (but interesting) note, the copula clitic can also be attached to Spanish words in code switching. The following example comes from a dialogue, with one of the speakers switching to Spanish occasionally. It can be observed that the copula clitic attaches to the Spanish past participle form for ‘paid’ (28) and ‘coordination’ (30). If we 251. Example (25) comes from a text relating the context in which many Wampis and Awajun people were protesting some laws passed by the government that could potentially a$ect their ecosystem. The non-appearance of hɨmpɨ ‘hummingbird’ is regarded as a bad sign. 742 consider the Spanish past participle is a nominalization (in pagado ‘paid), then the use of the copula is pretty coherent with its original Wampis use: (28) núka toðo a siðo así alimento hospeðaheʃa aʃí paɣaðowɛiti nu=ka todo ha sido así alimento hospedaje=ʃa NON.VIS =FOC everything has been thus food lodge=ADD aʃí pagado=aiti everything payed=COP3+DECL ‘That, everything has been thus, food and lodge, everything is paid.’ (29) núka una reunión preβia de coorðinasionkɛiti nu=ka una reunión previa de coordinación=ka=iti NON.VIS =FOC a meeting previous of coordination=FOC=COP.3+DECL ‘That is an early coordination meeting.’ As mentioned previously, the existential a does not have TAM restrictions and can occur in any tense reference and mood, including interrogative. The following examples show its occurrence with present and past temporal grounding, respectively (30) tikít ͡ʃik númi áwɛɛ tikit ͡ʃiki numi a-ua-i one tree exist-3SG.SBJ-DECL ‘There is one tree.’ (31) núkap kaáp ámaji nukapɨ kaapi a-ma-ji many tamshi.vine exist-REC.PT-3.PT+DECL ‘There were many tamshi vines.’ Additional examples of the existential a are given with the jussive (32) and 743 interrogative moods (33)–(34). (32) hui ̃ńkʲa kahɨŕnajamu atsutí hu-i=̃ka kahɨ-ra-nai-a-mau a-tsu-ti here=FOC be.angry-DISTR-RECP-IPFV-NMLZ exist-NEG-JUSS ‘Let it not exist hatred here.’ (33) awák a-ua-ka exist-3.SBJ-Q ‘Does it exist?’ (34) naráŋʃa arútramɨaka? naranka=ʃa a-ru-turama-a-ka orange=ADD exist-APPL-2.OBJ-IPFV-Q ‘Do you have oranges?’ 17.5.2.2. The marking of plurality in non-verbal predicates In copular and existential constructions, plural SAPs are marked by su#xes that indicate a combination of person, plural and subject argument information. The next examples show simple copular clauses with a "rst plural subject marked by -hi. (35) pɨŋ́kɨr ájahi pɨńkɨraa-ia-hi good COP-REM.PT-1PL.SBJ+DECL ‘We were good’ (36) pɨŋ́kɨrɛithi pɨńkɨra=ita-hi good=COP-1PL.SBJ+DECL ‘We are good.’ 744 As we saw previously in examples (26)–(27), there can be some overlap between the copula and copula clitics in the marking of plurality.252 Those examples are repeated below for ease of comparison. (37) wabálnumiajɛithi Huabal=numa=ia=aita-hi Huabal=LOC=ABL=COP=1PL.SBJ+DECL ‘We are from Huabal.’ (38) wabálnumia ɛɛ́ʲɲahi Huabal=numa=ia a-ina-hi Huabal=LOC=ABL COP-PL.IPFV-1PL.SBJ ‘We are from Huabal.’ For third person plurals, the situation becomes more irregular. The following points are to be noticed: • First, the copula a can take the plural -ina, which elsewhere in the grammar of Wampis occurs with plural imperfective stems (cf. §13.3.3), and -ara, which occurs with non-imperfective perfective stems (cf. §13.5). • Secondly, unlike the copula verb a, the copula clitics do not receive plural markers. 252. In addition, any verb that occurs with the distant past -mia takes the plural -ara regardless of whether there is an indication of plurality in the person marker attached to the verb: ii uɨka-inu a-ara-mia-hi 1PL walk-NMLZ COP-PL-DIST.PT-1PL.SBJ+DECL ‘We were travelers.’ 745 • Thirdly, the existential a has a suppletive plural form aia, which has the same cognate form in Awajun (Overall 2007).253 The following examples show simple clauses with the copula verb a and the existential a for third person plural referents. In this case, the existential occurs in the suppletive form aia (39), whereas the copula verb takes the plural su#x -ina in the imperfective (40). (39) ampúʃ numíɲam ájawɛɛ ampuʃa numi=nama aia-ua-i owl tree=LOC exist.PL-3.SBJ-DECL ‘There are owls in the tree.’ (40) ampúʃ ɛɛ́ɲawɛɛ ampuʃa a-ina-ua-i owl COP-PL.IPFV-3.SBJ-DECL ‘They are owls.’ Because third person plurals, unlike SAPs, do not have a dedicated person marker, in present declarative the notion of plurality is conveyed with a copula verb (a- ina), and then the predicate nominal ‘owl’ occurs with the copula clitic: 253. Overall analyzes this morpheme historically as a plural existential morpheme ai plus the imperfective -a (Overall 2007: 43). Phonetically, this is a most plausible analysis as it explains the surface form [aja] rather transparently (the vowel /i/ becomes an approximate between two vowels). 746 (41) óo ɛɛ́ɲa ampúʃɛiti au a-ina ampuʃa=iti DIST COP-PL.IPFV owl=COP.3+DECL ‘Those are owls’ 17.5.2.3. The marking of negation in verbal predicates Negation is also marked di$erently for the copula a, the copula clitics and the existential a. First, depending on the intending message and speci"c morphosyntactic environment, a negative clause with the copula a can be marked in two ways: a) the negative nominalizer -t͡ʃau marks the copula complement (42), or, b) the negative verbal morpheme -t͡ʃa is attached to the copula itself254 (43). (42) ʃuárt͡ʃou ájaji ʃuara-t͡ʃau á-ia-ji enemy-NEG.NMLZ COP-REM.PT-3.PT+DECL ‘They were not enemies’ (43) wiʃíkramu át͡ʃami tɨpɨŕkamu át͡ʃami uiʃi-ki-ra-mau a-t͡ʃa-mi laugh-TR-DISTR-NMLZ COP-NEG-HORT tɨpɨ-ru-ka-mau a-t͡ʃa-mi lie.down-APPL-INTENS-NMLZ COP-NEG-HORT ‘Let us not be deceived, let us not be dominated.’ (Lit. ‘Let us not be laughed at, let us not be laid down.’) On the other hand, unlike the copula verb a, copula clitics do not receive the 254. The su#x -t͡ʃa is a non-present negative marker. For present tense, the su#x -tsu is used. 747 verbal negative su#x -t͡ʃa. Instead, negation is marked in the clause by attaching the negative nominalizer -t͡ʃau to the nominal or adjectival complement. The copula clitics function basically as verbalizers, and thus follow the negative nominalizer -t͡ʃau. (44) ʃuárt͡ʃouwɛitʲi ʃuara-t͡ʃau=aiti person-NEG.NMLZ=COP.3+DECL ‘It is not a person.’ (45) ámɨka pɨŋ́kɨrt͡ʃouwɛitʲmɨ amɨ=ka pɨńkɨra-t͡ʃau=aita-mɨ 2SG=FOC good-NEG.NMLZ=COP-2SG.SBJ+DECL ‘You are bad.’ Finally, the existential receives the negative su#x -t͡su, which is only used in present tense (see §13.4 for details), as the following examples illustrate. (46) nihʲamañt͡ʃ at͡sáwɛi nihamãt ͡ʃi a-tsu-a-ua-i manioc.beer exist-NEG-IPFV-3SG-DECL ‘There is not manioc beer.’ (47) nui ̃ɲ́a naŋkamás jamɛíkʲa núu háta at͡sáwɛi nui-̃ia nankama-sa ̃ iamai=ka nu ja-ta a-t͡su-a-ua-i there-ABL happen-SUB\3SG.ss now=FOC NON.VIS be.sick-NMLZ exist-NEG-IPFV-3SG-DECL ‘Ever since it (i.e. the disease) happened, now that disease doesn’t exist.’ Importantly, the existential verb a never receives the non-present negative marker -t͡ʃa. Thus, if the temporal reference is not present declarative, the existential 748 still occurs with -t͡su but the non-present tense reference (in the example, a remote past) is conveyed by using a copula auxiliary, as non-present tenses are incompatible with - t͡su. Example (48) illustrate a construction with an existential negated with -t͡su followed by a copula in auxiliary function, marked for past tense. (48) hui ̃ŋ́ka awarúŋka at͡sú ájaji jóunt͡ʃukka hui=̃ka auaruni=ka a-t͡su a-ia-ji iaunt͡ʃukɨ=ka here=FOC Awajun=FOC exist-NEG COP-REM.PT-3.PT+DECL long.ago=FOC ‘Long ago, there was no Awajun here.’ 17.5.2.4. Copular and existential predicates in subordinate clauses The copula a can be subordinated with the subordinating morpheme -sa ‘Non- temporal manner’. The occurrence of a-sa (copula+subordinator) is frequent in texts and conversations. This structure is often used to establish meanings related to ‘reason’ or ‘cause’: (49) nítʲaʃa arát͡ʃou ásar kukút͡ʃ ɛíɲant͡ʃa aʃí sumɛíɲa asámtɛɛ ̃wíʃa surúktahɛɛ 49.a nita=ʃa ara-tʃau a-sa-ara 3PL=ADD sow-NEG.NMLZ COP-SUB-3PL.SS ‘Being that they did not grow [plants], 49.b kukut͡ʃa a-ina=na=ʃa aʃí suma-ina a-sa-matai ̃ cocona COP-PL=ACC=ADD all buy-PL.IPFV COP-SUB-1SG/3.DS ‘being that they buy the cocona fruits (Solanum sessili0orum), everything, 749 49.c wi=ʃa su-ru-ka-ta-ha-i 1SG=ADD give-APPL-INTENS-IMM.FUT-1SG.SBJ-DECL ‘I am also going to sell.’ (i.e. the speaker had grown her own produce to sell to those who did not grow theirs). In the data gathered for this research project, copula clitics do not occur at all in subordinating constructions (they only occur in the main clause). Thus, a potential subordinated clause as in (50) is not possible as far as I can tell; instead, only the subordinated verb occurs (51): (50) *(pɨnkɨra-tʃau=aiti a-sa-matai)̃ good-NEG.NMLZ=COP.3+DECL COP-SUB-1SG/3.DS (51) pɨŋ́kɨrt͡ʃau asámtɛɛ ̃. . . pɨnkɨra-tʃau a-sa-matai ̃ good-NEG.NMLZ COP-SUB-1SG/3.DS ‘being bad (people) […]’ In contrast, the existential verb a can occur within subordinate clauses, although in my data the existential itself never carries -sa. Example (52) shows the existential in a subordinate clause (second line in the analysis); notice however that it is the copula verb (not the existential) which carries the subordinating morpheme. (52) anɨńta ját͡ʃaka nuŋkúin pat͡ʃís anɨńtrua, nuŋkui ́ poderi óu asámtɛɛ,̃ núna anɨńtrua iɲɛíɲawai 750 52.a anɨnta jat͡ʃa=ka Nunkui=na pat͡ʃi-sã anɨnta-ru-a magic.song wise=FOC Nunkui=ACC mention-SUB\3SG.SS sing.anɨnta-APPL-IPFV ‘The [woman] who knows the anɨnta mentioning Nunkui, she sings, 52.b [Nunkui ́ power-ri ̃ a-u a-sa-matai]̃ Nunkui\GEN power-1PL/2PL/3.POSS exist-NMLZ COP-SUB-1SG/3.DS ‘being that Nunkui’s power exists,’ 52.c nu=na anɨnta-ru-ã inai-a-ua-i NON.VIS =ACC sing.anɨnta-APPL-IPFV stop.doing-ipfv--3.SBJ-DECL ‘she (the woman) stops what she is doing and sings to her (to Nunkui).’ Subordination in Wampis is described in detailed in Chapter XIX. 17.5.3. Functions of the copula a, copula clitics and the existential a The copula a and the copula clitics are all used for proper inclusion, equative and attributive predications. The main di$erences between the copula a and the copula clitics are not semantic, as evidenced from the previous discussion, but morphosyntactic. The existential is used to express locative predications, and to express possession. 17.5.3.1. The copula a The copula a is used in predicate nominal and predicate adjective constructions to encode proper inclusion, equative and attributive functions. Wampis does not 751 distinguish structurally between inclusion and equation, and the only structural di$erence of these with attribution is that in attributive clauses an AdjP can occur in the predicate. The basic construction with the copula is: NP X COP In the above construction, NP is the subject of the copula. A full NP, a pronoun, a determiner or a nominalized verb can occur as the copula subject. In addition, X above stands for the complement of the copula. X can be a NP, a nominalized verb or an AdjP. I have no examples of a pronoun or a determiner occupying the position of X. Sentences (53)–(54) constitute examples of proper inclusion and equative clauses with the copula. Sentence (53) has a nominalized verb as the complement of the copula. In (54), the copula complement is a noun (in this case the proper name Miik). (53) úun Tukúpʃa mankártiɲ ájaji [úunta Tukup=ʃa] [mã-karatu-inu] a-ia-ji elder Tukup=ADD kill-1PL.OBJ-NMLZ COP-REM.PT-3.PT ‘The elder Tukup also was a murderer.’ (54) míɲa nukút͡ʃruka Míik ámiaji [mina nukut͡ʃi-ru=ka] [Miika] a-mia-ji 1SG.GEN grandmother-1SG=FOC Miik COP-DIST.PT-3.PT ‘Miik was my grandmother.’ The next examples illustrate attributive clauses with the copula, where the 752 complements are ‘lazi’ and ‘short’, respectively. (55) nú apát͡ʃ náki áhaku tímʲaji [nu apat ͡ʃi] [naki] a-hak-u timaji NON.VIS mestizo lazi COP-HAB.PT-NMLZ NARR ‘That mestizo was lazy.’ (56) núka ʃuárka sútar ájaji [nu=ka ʃuara=ka] [sútara] a-ia-ji NON.VIS =FOC person=FOC short COP-REM.PT-3.PT ‘That person was short.’ 17.5.3.2. The copula clitics The copula clitics have allomorphs as summarized in Table 17.2, see §5.6.5 for a detailed explanation of morphophonological processes that condition the form of these allomorphs. Table 17.2. Wampis copula clitics. Type SAP Non-SAP Declarative =aita ~ =ita =aiti ~ =iti Interrogative =aita ~ =ita =aita ~ =ita Exclamative =aita ~ =ita =aita ~ =ita Preterit =ia =ia The copula clitics are used when the reference is present declarative, 753 interrogative and exclamative—in the latter, usually expressing surprise. The copula clitic =aita ~ =ita is used for SAP subjects, whereas the clitic =aiti ~ =iti is used for non-SAP, 3rd person subjects. There is a copula for past reference, =ia which occasionally occurs in the data. This preterit copula clitic seems to be declining in use, as for most referents in the past the copula verb, not the clitic, is used. Examples (57), (58) and (59) illustrate the use of the copula clitics in attributive, proper inclusion and equative clauses, respectively. (57) najóowɛɛthi naiau=aita-hi tall=COP-1PL.SBJ+DECL ‘We are tall.’ (58) ɛɛ́ʃmankuiti aíʃmaŋku=iti man=COP.3+DECL ‘He is a man.’ (59) aḿɨ jat͡surúitmɨ amɨ iat͡su-ru=ita-mɨ 2SG brother-1SG=COP-2SG.SBJ+DECL ‘You are my brother.’ An interesting note regarding the order of morphemes in interrogative clauses is that the question marker =ka comes before the copula clitic, hinting at a more recent 754 grammaticalization of the clitics: (60) warí tumɛínt͡sukɛit? uari ̃ tu-mai-inu-t ͡su=ka=aita what say-POT-NMLZ-INFER-Q=COP ‘How can it be said? The next example illustrates the use of the copula clitics in exclamative set- inclusion clauses. Notice that the question marker =ka can optionally occur in exclamative sentences and may be being reanalyzed as a non-declarative marker. If the question marker =ka occurs, the order of the morphemes is the same as in interrogatives, i.e. the question marker precedes the copula clitic. Note that the negative is used in this case to emphasize the idea of surprise of the expression, not in its literal ‘negation’ meaning. (61) auʃa ɲawánt͡ʃukɛit! au=ʃa iauã-́t͡ʃau=ka=aita DIST=ADD dog-NEG.NMLZ=Q=COP.EXCL ‘That is a jaguar!’ The sentence in (62) below contains an example of the use of the preterite copula clitic. I only have a handful of occurrences of this morpheme in my data, and it is apparently not very productive. For comparison, a question in present tense is also presented in (63). In the data for Wampis, the copula can refer to the remote or recent past. Notice that Overall (2007) reports a cognate remote past copula su#x -ya in 755 Awajun. (62) jája óuʃa ia=ia au=ʃa who=COP.PT DIST=ADD ‘Who was that?’ (63) jɛít óuʃa ia=ita au=ʃa who=COP DIST=ADD ‘Who is that?’ In addition, I was told by one of my teachers that a sentence like (64), which was pronounced with the clitic, should be transcribed as presented in (65); i.e. "corrected" with a fully conjugated copula verb. It is possible thus that the copula clitic =ia comes from the homophonous remote past tense marker -ia that is present in (65). The latter might have been reanalyzed after the remote past stem a-ia (Copula-Remote Past) was phonetically reduced. (64) huínka awáruŋka at͡sújaji jóunt͡ʃukka hui=̃ka auaruni=ka a-t͡su=ia-ji jauntʃukɨ=ka here=FOC Awajun=FOC exist-NEG=COP.PT-3.PT long.ago=FOC ‘Long ago, there was no Awajun here.’ (65) huiñkʲa awarúŋka at͡sú aj́aji jóunt͡ʃukka hui=̃ka awaruni=ka a-tsu a-ia-ji iaunt͡ʃukɨ=ka here=FOC Auajun=FOC exist-NEG COP-REM.PT-3.PT long.ago=FOC ‘Long ago, there was no Awajun here.’ 756 17.5.3.3. Existential clauses The existential construction in Wampis, as is to be expected, is based on the existential verb a. A location can (and usually) occur(s). The basic existential construction is: NP (LOC) Exist The LOC is typically either an NP, a locational word or an adverb and it is always marked with a locative marker. The NP subject is always nominative. The next examples illustrate the use of existential clauses. (66) ampúʃ numíɲam ájawɛɛ ampuʃa numi=nama aia-ua-i owl tree=LOC exist.PL-ua-i ‘There are owls in the tree.’ (67) namák junkúnamu áwɛi namaka junkuna-mau a-ua-i "sh grill-NMLZ exist-3.SBJ-DECL ‘There is "sh patarashca (=food grilled in leaves).’ (68) naɨḱ apíhmou piníŋnum áwɛɛ naɨka apiha-mau pininka=numa a-ua-i rope fold.pfv-NMLZ bowl.type=LOC exist-3.SBJ-DECL ‘There is a folded rope in the bowl.’ 757 Recall that the existential verb a has no TAM restrictions attested in the data. Morphologically, the existential verb is characterized by having a plural suppletive allomorph aia, and by receiving the negative marker -t͡su, as explained previously in §17.5.2. 17.5.3.3.1. Possession with existential a + applicative The existential verb a may be su#xed with an applicative to create a possessive stem; the possessive function might have originated in a benefactive use of the applicative. Unlike the original existential verb a, the possessive stem with the applicative is structurally transitive. In this possessive construction, the possessed thing appears as the subject of the verb and the possessor as the object. The possessor is indexed as an object on the verb if the possessor is a Speech Act Participant (SAP) as in (69); third persons do not have an overt verbal object marker and thus are left unmarked (70). Note that the possessed thing (the grammatical subject in the construction) can occur without being marked as possessed, as in (71), but semantically it is understood as possessed (hence the reason why this construction is considered as a possession construction). The possessor can occur in an overt NP marked with the accusative (71)—but two overt NP arguments of the verb are very rare in natural 758 speech, an overt NP possessor is rare. (69) arútam arútɨawɛɛ arutam a-ru-tu-a-ua-i power.vision exist-APPL-1SG.OBJ-IPFV-3.SBJ-DECL ‘I have an Arutam power’ (‘An Arutam power exists (to my bene"t)’) (70) ut͡ʃír araẃɛɛ ut͡ʃi-ri ̃ a-ru-a-ua-i son-1PL/2PL/3.POSS exist-APPL-IPFV-3.SBJ-DECL ‘He has a son.’ (‘His son exists (to his bene"t)’.) (71) arútam miɲ́a arútɨawɛɛ arutam mi=na a-ru-tu-a-ua-i power.vision 1SG=ACC exist-APPL-1SG.OBJ-IPFV-3.SBJ-DECL ‘I have an Arutam power’ The marking of Plural SAP’s possessor (the grammatical objects) on the verb show a di$erent (and uncommon) pattern, as Wampis argument indexation is morphologically and semantically complex (see §14.3.3 for details). For 1 and 2 plural possessors, the markings on the verb vary: the possessed thing no longer formally occurs as the subject of the verb. Instead both the object and subject markers, both referring to the possessor, occur at the same time. Thus, no indication of the possessed thing (a third person subject) overtly appears as the subject on the verb, as happens for the other grammatical persons/numbers. Notice, however, that this is actually the way 759 Wampis marks the situation of a 3 person subject acting on a 1 or 2 plural object (i.e. the 3 → PL.SAP scenario): the 1 or 2 plural argument is marked in both the object and subject verb slots. Examples (72) and (73) illustrate this pattern for 1pl and 2pl possessor: (72) arútam iíɲa arútramɨahi Arutam ii=na a-ru-turama-a-hi power.vision 1PL=ACC exist-APPL-1PL/2.OBJ-IPFV-1PL.SBJ+DECL ‘We have an Arutam power.’ (73) arútam aḿiɲa arútramɨarmɨ Arutam ami=na a-ru-turama-a-rumɨ power.vision 2SG=ACC exist-APPL-1PL/2.OBJ-IPFV-2PL.SBJ+DECL ‘You (PL) have an Arutam power.’ 17.5.3.3.2. Predicate locatives Locative constructions typically involve a construal in which a Figure (an NP that is being identi"ed) is located on a Ground (an NP that constitutes the referential location of the Figure) (Langacker 1987; Levinson & Wilkins 2006; Talmy 2007). Many languages employ copulas or existential verbs for the expression of spatial relations. Other employ posture verbs or positional verbs. For Wampis, the results of a study about the expression of spatial relations that I carried out were presented at the 2015 Society for the Study of Indigenous Languages of the Americas Annual Meeting (Peña 760 2015). Because of time and space constraints, only a summary is presented here. Of the types of spatial relations as de"ned in Ameka & Levinson (2007), Wampis actually employs most of them. Therefore, it is better to speak of di$erent strategies or constructions for Wampis, rather than trying to "t a strict typological parameters255 (Grinevald 2006). The copulas and the existential can be used in predications that express spatial notions, but a locative postposition must occur on the predicate NP, as can be observed in (74)–(75). (74) nu nanki mesa=nama a-ina-ua-i that spear table=LOC COP-PL.IPFV-3-DECL 'Those spears are on the table.' (75) nu nanki mesa=nama aia-ua-i that spear table=LOC exist.PL-3-DECL 'There are spears on the table.' Using the copula and the existential are by no means the only strategy that Wampis uses for locational predicates. Wampis possesses a fairly sophisticated lexical pool of verbs that are used in locatives constructions (especially for the subdomain of 255. Grinevald (2006), working on a varied sample of Amerindian languages, has proposed “intermediate” systems that allow for continuos, non-discrete, categories. 761 static and topological relations). In fact, the use of the copula or the existential to express static spatial/topological relations is infrequent in discourse. In Peña (2015), I presented an examination of Wampis expression of locational relations. Here I summarize the di$erent sets of verbs that can be used to express predicate locatives in the language. One "rst set of such verbs is composed of verbs that have a semantics similar of ‘live’ or ‘gather together’. Table 17.3 includes these verbs. Table 17.3. Set I lexical verbs used to code spatial notions. Verb Gloss Classi"catory Semantics mat͡sa ‘inhabit, be/get together (pl)’ +/- Animate mat͡sa-tu (mat͡sa-APPL) ‘be/get (pl.) together'256 + Animate irunu ‘reunite, crowd together, pile up' +/- Animate puhu 'live' + Animate Interestingly, Wampis is apparently forming specialized locational copulas from these verbs. In fact, to answer a basic locative question such as 'where is X?’, the typical answer involves the use of puhu ‘live’ if X is an animate singular or plural referent. If X 256. Also translated as ‘live in community’ by one of my teachers. 762 is plural and inanimate, either irunu ‘crowd together, reunite, pile up’ or mat͡sa ‘inhabit, be/get plural’ is used.257 If X is a singular inanimate entity, the answer usually involves a postural verb (see below). Di$erent examples of the use of these verbs are given in the following examples. Notice that, when used with the applicative, mat͡sa typically refers to human beings living/being together (thus notice the di$erence in the use of the applicative between (77) and (78)). (76) hɨɰã ́puháwɛɛ hɨã ́ puhu-a-ua-i house\LOC live-IPFV-3-DECL ‘She is in the house.’ (77) míik nuŋká mat͡sásmau miika nunká matsa-sa-mau bean gound\LOC get.together-ATT-3-NMLZ ‘The beans are on the /oor.’ (78) núka núwaka hɨɰã ́mat͡satɨna nu=ka nua=ka hɨã ́ mat͡sa-tu-ina that=FOC woman=FOC house\LOC inhabit-APPL-PL.IPFV ‘Those women were in that house.’ 257. But notice that irunu and mat͡sa are also used with animate plural referents. 763 (79) ampúʃ numíɲam irúnɨawɛɛ ampuʃ numi=nama irunu-a-wa-i owl tree=LOC crowd.together-IPFV-3-DECL ‘The owls are on the tree.’ A second set of verbs is composed of postural verbs. Wampis possesses a rather rich set of postural verbs, although not all of them are used with the same frequency. Other Amazonian languages exhibit very rich and complex sets of postural verbs that have grammaticalized as auxiliaries and aspectual markers (Queixalós 2009; Vuillermet 2012). Table 17.4 gives a sample of posture verbs that occur commonly in my data. Table 17.4. Sample posture verbs in Wampis. Verb Gloss uaha 'stand' a-uaha (cau-stand) 'be in a vertical position' tɨpɨ 'lie down' ɨkɨ 'sit' auankɨ 'be hanging' tɨã 'be reclined' tarima 'be standing "rm on two feet' atɨkVna 'be upside down' Sentences (80)–(84) provide examples of the use of di$erent verbs from Table 17.4 in locational constructions: 764 (80) Ampuʃa numi=nama ɨkɨ-tu-a-ua-i owl tree=LOC sit-APPL-IPFV-3-DECL ‘An owl is on the tree.’ (81) iauaã kut͡ʃa=nama uaha-sa-ma-ji dog puddle=LOC stand-ATT-REC.PT-3.PT ‘The dog was standing in the puddle.’ (82) ihu-a-ha-i namaka tepɨ-a-u=na uaina-ka-nu stab-IPFV-1SG.SBJ-DECL "sh lie.down-IPFV-NMLZ=ACC see-INTENS-1SG.SS ‘I stab, having seen "sh that are lying down.’ (83) pelota nunká tɨpɨ-a-ua-i ball ground lie.down-IPFV-3-DECL ‘The ball is lying on the ground.’ (84) muit͡sa numi_wahautin atɨkna-ka uku-ka-mau pot stump be.upsidedown-INTENS put-INTENS-NMLZ ‘The pot is upside-down on the stump.’ Finally, another set of verbs used in locative predicates is composed of positional/caused position verbs. Table 17.5 o$ers a sample of verbs that occur frequently in my data. 765 Table 17.5. Sample Wampis positional verbs Verb Gloss ɨnkɨ-tu (usually with applicative -tu) 'be inside' auanta 'cover' nɨna 'hang with rope' tɨntɨ 'be around, circle' nana 'be /oating//oat' tʃimpi 'put (plural) in recipient' atsi ̃ 'put on the head' atu 'be leaning' aɨpɨ 'be/put horizontal sideways' pasu 'be/put in an untidy way' ɨkɨna (from ɨkɨ ‘sit’) 'be on [place]’ aɨpa 'be/put on ground or bed' pɨɨ-tu (usually with applicative -tu) 'be stuck' Examples (85)–(87) illustrate the use of these verbs. (85) jurankɨ t͡sapa=nama ɨnkɨ-tu-a-ua-i fruit bowl=LOC be.inside-APPL-IPFV-3.SBJ-DECL ‘The fruit is inside the bowl.’ (86) nui ̃ ʃuara aepe-sa-ma-ji there person put.horizontal.sideways-ATT-IMM.PT-3.PT+DECL ‘There he had that person lying sideways.’ 766 (87) t͡sɨrɨ iakí pɨɨtu-a-ua-i spider above/LOC be.stuck-IPFV-3-DECL ‘The spider is stuck above (on the ceiling).’' 17.6. Other copulative verbs Apart from the copula verb a and the copula clitics, there are in Wampis three synchronically full lexical verbs that have also assumed copular functions. The verbs nahana ‘make’ and wɨ ‘go’ can be used in copular clauses (i.e. “linking” two nominative NPs) with the meaning of ‘become’. (88) Puhupata ampuʃa nahana-ra-ma-ji Puhupata owl make-DISTR-REC.PT-3.DECL 'Puhupat turned into an owl.' (89) amɨ=ka ʃiira ɨakama-u uɨ-a-mɨ 2=FOC good hunt-NMLZ go-IPFV-2SG.SBJ+DECL 'You are becoming a good hunter.' The other verb that is used as a semicopula meaning ‘become’ is hasa. This verb is likely a phonetic reduction of the verb uaha with the attenatuative aktionsart su#x - sa. With the attenuative, uaha has a more stative semantics of ‘be standing’. The verb uaha itself is used in predicative locative constructions: 767 (90) numi uaha-ina-ua-i tree stand-PL.IPFV-3SG.SBJ-DECL 'There are trees.' Probably through its function in existential constructions, uaha has developed other copular functions. The reduced form has functions as a semicopula in current Wampis: (91) ui=ka uunta has-ha-i 1SG=FOC big become-1SG.SBJ-DECL 'I have become an adult’ The same reduced form has has become an aspectual particle (i.e. an invariant element that does not receive any in/ection) that expresses a complete change of state: (92) t͡ʃit͡ʃama=ka uunta has problem=FOC big become 'the problem is big’ (it already became big) See (Peña Forthcoming) for a full analysis of the development of uaha into a semicopula. 768 CHAPTER XVIII DISCOURSE-ORIENTED AND MOOD PHRASE-LEVEL CLITICS 18.1. Introduction This chapter describes clitics whose functions are related to indicating discourse- oriented and mood information. These morphemes are best analyzed as clitics because they exhibit very few restrictions with respect to what can be their phonological hosts, and most of them lack a "xed position in the sentence. The discussion begins in §18.2 with an overview of the clitics that will be analyzed throughout the chapter. Each of the subsequent sections §18.3 through §18.10 is dedicated to describing each one of the clitics mentioned in §18.2. 18.2. Overview Table 18.1 lists the clitics that will studied in the next sections. With regard to their distributions, the restrictive =kɨ, additive =ʃa and focus =ka occur most frequently with NPs, but can also occur on verbs, adjectives and adverbs or marking several constituents in the sentence. Speculative =ʃa, inferential =t͡su, sudden realization and tag question =api and interrogative =ka can occur marking verbs and nouns. The restrictive =kɨ can co-occur with other clitics, and it is very productive co- 769 occurring with =ʃa. The additive =ʃa and the focus =ka cannot co-occur in the same word. Table 18.1. Mood and discourse-oriented clitics Morpheme Gloss =kɨ ‘Restrictive’ =ʃa ‘Additive’ =ka ‘Focus’ =ʃa ‘Speculative’ =t͡su ‘Inferential’ =ka ‘Interrogative’ =api ‘Sudden realization’, ‘Tag question’ =hama ‘Mirative’ 18.3. Restrictive =kɨ The clitic =kɨ restricts the reference of the element that receives it. The restrictive has an allophone =ki that occurs when the preceding vowel is /i/, following a vowel harmony process (cf. §5.5.1). The most basic semantics of the restrictive can be translated to English as ‘only’ or ‘alone’. (1) ɲawaãḱ iauaã=kɨ dog=RESTR ‘The dog only’ or ‘The dog alone’ 770 (2) niŋ́ki ni=̃kɨ 3SG=RESTR ‘He only’ or ‘He alone’ The restrictive occurs not only with nouns, but can also occur on pronouns, numerals and, occasionally, adverbs. The restrictive is also fossilized in words related to counting and to temporal or geographical distance. The following examples illustrate =kɨ with a pronoun (3) and a numeral (4). Example (5) shows =kɨ with the adverb ‘now’ and derives a word meaning ‘right now’. Example (8) shows =kɨ fossilized on number ‘one’. (3) ni=̃kɨ [níŋki] 3SG=RESTR ‘He only’ or ‘He alone’ (4) himara=kɨ [himárak] two=RESTR ‘Two only’ or ‘Two alone’ (5) iamai=kɨ [jamɛík] now=RESTR ‘right now’ 771 (6) tikit ͡ʃiki < *tikit͡ʃi-kɨ ‘one’ other-RESTR In discourse, the restrictive =kɨ usually serves to mark a stance. In the next example, the referents have been mentioned and clearly identi"ed because they are a speech act participant (‘you’) and a proper noun (‘Andrés’). In the conversation the statement ‘They are only two people’ occurs when the speaker is trying to convince his interlocutor to go to a meeting (he is trying to explain that two people have been chosen to go to this meeting—his interlocutor, referred by ‘you’, and ‘Andrés’, who is not present) and that everything is already set for his interlocutor and ‘Andrés’ to go. The strategy the speaker uses is to attenuate his statement with a diminutive and adding the restrictive =kɨ (in the line (c)): (7) 7.a nita=ka ii-sa-mau=aiti ʃuara 3PL=FOC see-ATT-NMLZ=COP.3+DECL person ‘They have chosen the persons,’ 7.b amɨ nu=i=̃ia Andrés 2SG NON.VIS=LOC=ABL Andres ‘You and Andrés […]’ 772 7.c ʃuara=t͡ʃi=kɨ=iti person=DIM=RESTR=COP.3+DECL ‘They are only two people.’ (Lit.: ‘They are two little people only.’) As the conversation goes on, the "rst speaker continues trying to convince his interlocutor to attend the meeting by stating ‘it is going to be only one day’—notice that "rst he says ‘one day only’ and then the second time he repeats ‘it is only one day’ (with a copula) to add illocutionary force to his statement: (8) un díak táwɛi un díakɨti un día=kɨ ta-ua-i un dia=kɨ=iti one day=RESTR say+IPFV-3.SBJ-DECL one day=RESTR=COP.3+DECL ‘One day only, he says, it is only one day.’ Another characteristic of the use of the restrictive in connected speech is its frequent co-occurrence with the additive =ʃa (see §18.5). One of the meanings of the “restrictive plus additive” structure usually involves a sense of relative immediacy (geographical or temporal). In (9), the =kɨ=ʃa structure provides a sense of geographic immediacy (being close in this case) and in (10) =kɨ=ʃa yields a sense of temporal immediacy (‘next morning’). (9) ikaḿkɨʃa wɨkaḱ aańkɨʃa wɨkaḱ ikama=kɨ=ʃa wɨka-kũ forest=RESTR=ADD walk+IPFV-SIM\3SG.SS 773 aa=ni=kɨ=ʃa wɨka-kũ out=ALL=RESTR=ADD walk+IPFV-SIM\3SG.SS ‘while walking in the forest (but not far away), while walking out there (but not far away).’ (10) hɨɨ ̃ń húu aŕmaji nútikʲa núu t͡sawaŋkɨʃa turaʃkuʃa kaʃiŋ́kɨʃa hɨɨ-̃ni hu-u a-ara-ma-ji house\1PL/2PL/3.POSS-ALL carry-NMLZ COP-PL-REC.PT-3.PT nuti-ka ̃ nu t͡sauanta=kɨ=ʃa do.that.PFV-INTENS\3SS NON.VIS day=RESTR=ADD turaʃkuʃa kaʃińi=kɨ=ʃa or.if.not next.morning=RESTR=ADD ‘[The women] carry [it] to their house, doing that, that same day or if not the next morning [they carry it].’ In the next example, the meaning of =kɨ plus =ʃa ‘additive’ is extended to ‘not even’. In the context where this utterance occurs, the speaker is introducing herself as belonging to the Wampis culture (I had asked her to tell me how she prepares manioc beer, hence she introduces herself as someone knowledgeable in Wampis culture, in which manioc beer—as in other cultures of the Amazon—is an important part of everyday life): 774 (11) wíkʲa apát͡ʃkɨʃa pat͡ʃímrat͡ʃmauwɛithʲɛɛ wíkʲa ʃíir wampiśɛithʲɛɛ 11.a ui=ka apat ͡ʃi=kɨ=ʃa 1SG=FOC mestizo=RESTR=ADD ‘I am not even a mestizo’258 11.b pat͡ʃima-ra-t͡ʃa-mau=aita-ha-i mix-DISTR-NEG-NMLZ=COP-1SG.SBJ-DECL ‘I am not mixed’ 11.c ui=ka ʃiira wampisa=ita-ha-i 1SG=FOC very Wampis=COP-1SG.SBJ-DECL ‘I am very Wampis.’ Notice that (11) does not mean “I am only a mestizo”, in fact, the speaker is, as she states, “very Wampis” (meaning in this case that both her parents were Wampis and she has lived her entire life in a Wampis village). Compare the preceding example with the next one, where =kɨ is received by the last element in the enumerative noun phrase. It also occurs with the additive =ʃa, but in this occasion, the meaning is restrictive: 258. The Spanish translation given for this clause was: ‘Yo ni siquiera soy mestiza.’ 775 (12) nui ̃ ́maátɛiɲɛiti tréskɨʃa ʃuʃuí káʃɛɛ hapákɨʃa nui ̃ mã-á-tai=̃aiti trés=kɨ=ʃa there kill-HIAF-NMLZ=COP.3+DECL three=RESTR=ADD ‘[At night] there, we (normally)259 hunt three [animals] only’ ʃuʃuí kaʃɛɛ hapa=kɨ=ʃa armadillo paca deer=RESTR=ADD ‘only armadillo, paca, deer.’ In terms of its distribution, =kɨ may actually vary its position: it can occur before (13) or after (14) the position occupied by the accusative =na: (13) ájatek aɨt́kɨnt͡ʃa wɛíɲah aiatɨkɨ aɨta=kɨ=na=ʃa waina-a-ha only green=RESTR=ACC=ADD look-IPFV-1SG.SBJ+EXCL ‘I "nd only the green (non-ripe) ones!’ (14) haŋkíŋkɨʃa mɨsɨḿain hanki=na=kɨ=ʃa mɨsɨ-mai ̃ thorn=ACC=RESTR=ADD ruin-POT\3.SS ‘Having stuck a thorn in too’ 18.4. Focus =ka The clitic =ka has varying functions in texts. =ka is most frequently attached to NPs and pronouns, but it can actually occur in almost any other element: verbs, adverbs, adjectives, numerals. The frequency of the focus marker with verbs and adverbs, in comparison with their occurrence on other word classes, is low. The term 259. The construction of the non-agentive nominalizer -tai ̃with a copula functions as a normative, which expresses how things are ‘normally’ done. 776 “focus” is more of a convenient term for a morpheme that achieves several important functions in the structuring of information in Wampis discourse. One of the most important uses of =ka is to introduce new participants in the discourse. The following example comes from a traditional narrative in which two women go looking for a husband. At some point they "nd Nayap, one of the major participants in the story. This is how the speaker introduces him in the text (note the use of =ka in several constituents): (15) núka úun ʃuaŕka túkɨ najaṕ tiḿaji kaḿɨ núna naaŕiñkʲa nu=ka uunta ʃuara=ka tukɨ naiapi timaji kamɨ NON.VIS=FOC adult person=FOC INTERJ Nayap NARR INTERJ nu=na naa-ri=̃ka NON.VIS=ACC name-1PL/2PL/3.POSS=FOC ‘That person, Nayap was his name.’ The focus =ka also serves to re-introduce participants in the discourse. That is, when a major participant is re-introduced in discourse, they usually occur as a full NP and marked with focus =ka. In the following passage, the narration is told from the perspective of the husband of the protagonist, who is planning to go see who is eating his squash. Then the perspective changes from the husband to his wife, the protagonist of the story (in 16.b), who is re-introduced in the discourse via the use of =ka. 777 (16) 16.a iístahɛɛ tiú timaji ii-sa-ta-ha-i ti-u timaji see-ATT-IMM.FUT-DECL say+LOAF-NMLZ NARR ‘He said “I am going to see (who is eating my squash)”,’ 16.b aúhuka ʃiír jurumiń asa ̃ auhu=ka ʃiira iurumin a-sa ̃ Auju=foc very eater cop-sub\3.ss ‘Auju being a big eater . . .’ The clitic =ka also is used for contrastive focus. In the next example, the speaker contrast how diabetes is said in Spanish (17.a) and in Wampis (17.b). (17) 17.a óuhmatsattahɛɛ wií jamɛí hatań pat͡ʃisan hat́a ðiaβetes tamóun auhumatu-sa-tata-ha-i ui iamai ha-ta=na tell-ATT-DEF.FUT-1SG.SBJ-DECL 1SG now be.sick-NMLZ=ACC pat͡ʃi-sa-nu ha-ta diabetes ta-mau=na mention-ATT-1SG.SS be.sick-NMLZ diabetes say+IPFV-NMLZ=ACC ‘Today, I am going to tell about a disease, mentioning what is called the disease diabetes’ 17.b turaʃ́a iíkʲa núka wat͡saḿat hat́a taj́i turaʃa ii=ka nu=ka uat͡sa-ma-ta ha-ta ta-hi but 1PL=FOC NON.VIS=FOC get.thinner-REFL-NMLZ be.sick-NMLZ SAY-1PL.SBJ+DECL ‘but we (i.e. the Wampis), that, we call it the disease that gets one thinner.’ The focus =ka is also used to express assertion. In the next example, a person 778 dreams that he is going to transform into a jaguar, but no one believes him. The character exclaims: (18) wiḱʲa kanaŕuithʲɛɛ ui=ka kana-ra-u=ita-ha-i 1SG=FOC dream.PFV-DISTR-NMLZ=COP-1SG.SBJ-DECL ‘I did dream (about that)’. In the next example, the narrator re-asserts a new information just provided. First he introduces the name of the character (“Inchis”) and then he re-asserts this new information: (19) 19.a ińt͡ʃis naaŕtin int͡ʃisu naa-ri-̃tinu Inchis name-1PL/2PL/3.POSS-ATTRIB ‘her name [was] Inchis’ 19.b naaŕiŋ̃ka núka núwaka ińt͡ʃis naa-ri=̃ka nu=ka nua=ka int͡ʃisu name=1PL/2PL/3.POSS=FOC NON.VIS=FOC woman=FOC Inchis ‘The name of that woman [was] Inchis.’ Another use of =ka is in parenthetical clauses. Parenthetical clauses are clauses that are not grammatically required; rather, they add information usually used to comment, qualify, clarify, or provide an afterthought to an statement. The following are the "nal lines of a text in which one of my Wampis teacher was teaching me how some 779 plants are used to cure wounds. After he "nishes (“those are [the plants that we use for healing]”) he adds an afterthought comment which is marked with =ka: (20) nu ɛínawɛi jatsúr haimito iíkʲa iḿanisrikʲa ðoktórnumka wɨát͡shi nu a-ina-ua-i iat͡su-ru haimito NON.VIS COP-PL.IPFV brother-1SG Jaime ii=ka iḿanis-ri=ka doktor=numa=ka wɨ-a-t͡su-hi 1PL=FOC INTS.ADV-1PL.SS=FOC doctor=LOC=FOC go-IPFV-NEG-1PL.SBJ-DECL ‘Those are [the plants we use for healing], brother Jaimito—we don’t go much to the doctor.’ Finally, I have found that =ka is used in cleft-like constructions, marking a relative clause in focus function (in this case, the structure in brackets [ ] is relativized by the non-visible demonstrative nu): (21) wiḱʲa nekaáti túsan wakɨŕah núka jũẃɛiti [wi=ka nɨka-a-ti tu-sa-nu uakɨra-ha nu=ka] 1SG=FOC know-IPFV-JUSS say-SUB-1SG.SS want+IPFV-1SG.SBJ NON.VIS=FOC hũ=aiti PROX=COP.3+DECL ‘What I want you to know is this.” (Lit.: ‘[What I want, saying: ‘let him know’], is this.’ The focus marker =ka and the conditional -ka (§19.11) are homophonous and likely related (cf. Overall (2007: 507) for an initial suggestion of this claim in Awajun, and Haiman (1978) for a cross-linguistic study on the development of conditionals). I have found some examples of what appears to be the focus =ka being used as a conditional. This is based on the fact that the conditional is only used in subordinate 780 verbs (cf. §19.11), but in the example below =ka occurs on pɨnkɨra ‘good’’, which is an adjective. (22) tikit́͡ʃik plańta aŕak haŋkiŕtin . . . pɨŋ́kɨrka waruri ̃ʃ́a aẃai one plant araka hanki-ri-̃tinu pɨńkɨra=ka one plant plant=FOC thorn-1PL/2PL/3.POSS-ATTRIB good=FOC waru-ri=̃ʃa a-ua-i worm-1PL/2PL/3.POSS=ADD exist-3.SBJ-DECL ‘[it is] a plant that has thorns. . . if it is good, there is waru too (i.e. there are a species of worms if the plant is good).’ 18.5. Additive =ʃa The additive =ʃa has a variety of functions. Its basic function is to serve to mark additive focus, i.e. =ʃa expresses “that the predication holds for at least one alternative of the expression in focus” ((Krifka 1998: 11)). A simple additive focus function is illustrated in the following example. (23) ɲawaá ̃arútmaʃa ah́aku tiḿaji jaguar arutama=ʃa a=hak-u timaji jaguar power.vision=ADD exist=PT.HAB-NMLZ NARR ‘There used to be also Arutam jaguar.’ The additive =ʃa is also used to mark the topic of a question: (24) nihʲama ̃ńt͡ʃiʃa aẃak nihamat̃͡ʃi=ʃa a-ua-ka manioc.beer=ADD exist-3.SBJ-Q ‘Is there manioc beer?’ The additive =ʃa has also other extended uses when it is used with the 781 restrictive =kɨ (cf. §18.3). 18.6. Speculative =ʃa The speculative =ʃa indicates that the speaker judges that the proposition expressed by his or her utterance is possibly true, but there is a possibility that it is not. This meaning is usually translated as “may be” (tal vez) by Wampis speakers. (25) ʃiampiń paańtam hiaḿuhɛiñʃa juaẃarmaji ʃiampi=na paatama hia-mau=hai=̃ʃa iu-aú-ara-ma-ji hen=ACC plantain cook.manioc-NMLZ=COM=SPEC eat-HIAF-PL-REC.PT-3.PT+DECL ‘Maybe they ate hen with cooked plantain.’ 18.7. Inferential =t ͡su The clitic =t͡su marks inference or supposition. It attaches mostly to nominals but can also be received by verbal elements. The following example comes from a ‘Pear "lm’ story (see Chafe (1980)). In this part, some children take a basket full of pears that a man has been collected. When the man climbs down the tree, he sees the children walking away and notices that one of his baskets is missing. From the evidence at hand, he infers that it has been the children who took the basket: (26) atákʃa út͡ʃi wakɨt́ki miníɲan íis nitʲát͡suk kasámturkaruiti túsa tú ɨnɨńtɛimʲar iɲɛísouwɛiti atakʃa ut͡ʃi wakɨ-tu-ki uini-ina-na ii-sã again child walk-APPL-WHILE.MOVING come-PL.IPFV-NMLZ see-SUB\3.SS 782 nita=t͡su=kɨ kasama-tu-ru-ka-ara-u=iti tu-sã 3PL=INFER=RESTR steal-APPL-1SG.OBJ-INTENS-PL-NMLZ=COP.3+DECL say-SUB\3.SS ‘Again, seeing that the children were walking, saying “They [the children] must have robbed me”.’ The clitic =t͡su can occur with a verbal su#x -tai that also marks inference §14.4.5. The di$erence between =t͡su and -tai is not well understood but there is apparently no semantic change in the inferential meaning added by these morphemes. In terms of their distribution, when -tai occurs on the verb, =t͡su can occur but only attached on a nominal constituent. The next example illustrates this. In the passage seen in the example, a group of women is making tuim /tuimpi/, a type of food that was served in some special occasions. In this case, the Wampis have just defeated a group of Awajun, and the warriors, following a ritual of celebration, were to eat tuim for the occasion. However, one of the women is half-Awajun and half-Wampis. The other Wampis women and the pamuk (a type of ceremonial leader) notice that the half- Awajun woman’s tuim smells of blood. (27) miɲ́a nukút͡ʃru nukuri ̃ ́ tuiḿriŋkʲa númpa númpa sɨɨ́r sɨɨ́r mɨhɨá tiḿaji núnitʲɛɛ ̃ paḿuk wa- hóo t͡ʃit́͡ʃak aḿɨka kaḿɨ awarún númpa takaḱɨamɨ [...] túkɨ patɛímit͡suk aíɲatɛɛ 27.a mina nukut͡ʃi-ru nuku-ri ̃ tuimpi-ri=̃ka 1SG.GEN grandmother-1SG mother-PL.SAP/3.POSS type.food-PL.SAP/3.POSS=FOC ‘My grandmother’s mother’s tuim’ 783 27.b numpa numpa sɨɨr_sɨɨr mɨhɨá nuni-tai ̃ blood blood IDEO:blood.smell stink+IPFV do.that-1SG/3.DS ‘[smelled like] blood, “it stinks like blood”, then,’ 27.c pamuka uaha-a-u t͡ʃit͡ʃa-ka ̃ leader stand-IPFV-NMLZ speak-INTENS/3.SS ‘the pamuk stood and said’ 27.d amɨ=ka kamɨ auaruna numpa 2SG=FOC INTERJ Awajun blood “you have Awajun blood” 27.e takaku-a-mɨ [...] patai-mi=t͡su=kɨ a-ina-tai have-IPFV-2SG.SBJ+DECL relative-2SG.POSS=INFER=RESTR cop-PL.IPFV-INFER “it must be your [dead] relatives (i.e. it must be your dead relatives who are causing the fod to smell bad)”. The inferential =t͡su might be related to the homophonous negative imperfective verbal su#x -t͡su (cf. §13.4). 18.8. Interrogative =ka The interrogative is marked with the question marker =ka, as in (28). (28) aj́atɨk awaḱɨamɨk aiatɨkɨ auakɨ-a-mɨ=ka only return-IPFV-2SG.SBJ=Q ‘Do you only rewind it [i.e. the cassette]?’ Though =ka occurs mostly on nouns and verbs, it actually can be received by 784 most parts of speech: pronouns, adjectives and adverbs. The following is an example in which the interrogative =ka attaches to a noun. (29) aḿɨ wari ́arútmak wɛińkʲatam amɨ wari arutama=ka waina-ka-ta-mɨ 2SG what power.vision=Q see-INTENS-IMM.FUT-2SG.SBJ ‘So you, which Arutam are you going to see?’ An important morphological property of the question clitic marker is that it occurs before the copula clitics: (30) pɨŋ́kɨrkɛita pɨnkɨra=ka=ita good=Q=COP.3 ‘Is it good? The question marker is historically related to the focus marker =ka (cf. §18.4). However, unlike the focus marker, which never undergoes apocope, the question marker does undergo apocope, as in (29) above. 18.9. Sudden realization =api and Tag question =api The clitic =api is primarily used when someone realizes that something has not gone as expected, with an outcome usually regarded as negative. Unlike the mirative, the event depicted by =api is typically directly associated to a volitive action done by the speaker and thus the event is not as strongly counter-expected as with the mirative. In the next example, a woman who is a big-eater eats up all the food without saving 785 some for her husband. So, while the realization that there is no more food is sudden, this is not totally unexpected information (as she herself has been eating it all up): (31) ɛiʃ́runka ampiŕat͡ʃhapi aiʃi-ru=na=ka ampi-ra-t ͡ʃa-ha=api husband-1SG=acc=FOC save.food-DISTR-NEG-1SG.SBJ=sud.realz ‘Oops, I didn’t save food for my husband!’ And in the following example, the speaker recounts how the villagers agreed in a community assembly to ask shamans from other communities for help, after they realize that the children in their village were behaving strangely due to witchcraft.260 She uses a construction with =api to talk about this situation: (32)iíkʲa húniakur puhut͡ʃuapitji wariń áa huiʃ̃a wariń hat́ak áa ii=ka huni-a-ku-ri puhu-t͡ʃu=api=iti-hi 1PL=FOC do.this-IPFV-SIM-1PL.SS live-NEG.NMLZ=SUD.REALZ=COP-1PL.SBJ+DECL wari ̃ a hui=̃ʃa wari ̃ hata=ka a what exist here=SPEC what sickness=Q exist ‘We do not live like that, what is there here, which sickness is there?’ In terms of the distribution of =api, it is interesting to notice that in the above example =api is received by a nominalized form, and in turn =api is followed by a copula clitic. So it seems that =api behaves distributionally in a similar fashion as the question marker =ka, which can be received by other non-verbal elements but works at 260. Notice, again, that this is—culturally—not totally unexpected information, as sickness and psychological problems are usually attributed to witchcraft in the Wampis worldview. 786 the level of the clause, and also precedes the copula clitic. The other use of =api in Wampis is in tag questions. In that function =api too attaches to a verb or nominal element and turns a declarative proposition into a question. Not surprisingly, most examples of =api in tag questions come from conversations, as in the following example. (33)t͡ʃií ɛítkas aań uhaḱtahɛɛ aḿɨkɨapi minit́tamɨ t͡ʃi aitika-sa aanu uha-ka-ta-ha-i yes do.like.that-SUB MED inform-INTENS-IMM.FUT-1SG.SBJ-DECL amɨ-kɨ=api uini-tata-mɨ 2SG-RESTR=TAG come-DEF.FUT-2SG.SBJ ‘Okay, I am going to inform like that, only you are going to come, aren’t you?’ 18.10. Mirative The category of Mirativity is related to the expectation of knowledge: “the term mirativity refers to the linguistic marking of an utterance which is new or unexpected to the speaker” (DeLancey 2001b: 370), information that is thus new and not yet part of the “speaker’s integrated picture of the world” (DeLancey 1997: 49). The mirative in Wampis is done with the morpheme =hama. This morpheme can occur su#xed or detached from the verb,261 forming its own prosodic word, thus I treated as a clitic-like 261. In some cases, while transcribing, speakers would correct me if I did not write hama as a separate word. 787 morpheme. The next example comes from a text in which a group of Wampis is besieged at their house by an enemy Awajun group. Having been attacked all night by the enemy, they are expecting the worse when they hear that more people come closer to their house: (34) 34.a huwańkoot͡ʃi huwańkoot͡ʃi wij́ɛíthɛɛ wij́ɛithɛɛ huanka-ut͡ʃi262 huanka-ut͡ʃi ui=aita-ha-i ui=aita-ha-i Juan-DIM Juan-DIM 1SG=COP-1SG.SBJ-DECL 1SG=COP-1SG.SBJ-DECL ‘[One of the people from outside spoke:] “Juancito, Juancito, It is me! It is me!’ 34.b uratritʲa ́wɛit́in tóo tiḿaji tútai ̃ ura-tu-ru-i-ta ́ waiti=na ta-u timaji tu-tai ̃ open-APPL-1SG.OBJ-LOAF-IMP door=ACC say+IPFV-NMLZ NARR say-1SG/3.DS ‘Open the door for me!’ he said; when he said that’ 34.c jaḱi núʃa núka ʃuaŕt͡ʃowahama ia-ki nu=ʃa nu=ka ʃuara-t ͡ʃau=a-hama who-INT.FOC NON.VIS=SPEC NON.VIS=ADD enemy-NEG.NMLZ=COP-MIR ‘[The ones inside the house said:] “who could that be?” “It is not the enemy!”’ 262. This is the only example that I have where the diminutive surfaces identical as its lexical source ut͡ʃi ‘child’. 788 34.d núka iíɲa patɛíɲahama nu=ka iina patai=̃a=hama NON.VIS=FOC 1PL.GEN relative\PL.SAP/3.POSS=COP-MIR ‘“It is our relative!”.’ In the example above, we can see that the new information acquired by the speakers is not expected, i.e. they did not expect that someone would come to help them as the enemy had surrounded their house. A more colorful example comes from a situation in which an clumsy linguist doing "eldwork used a blowgun that had a dart in it (no one knew that the blowgun was loaded with a dart) and accidentally shot someone in the leg, thankfully without hurting him. As the event was clearly not expected by anyone in the scene, the surprised linguist’s host exclaimed: (35) nui ̃ ́aḿa haḿa nui ̃ a-ma hama there exist-REC.PT MIR ‘There was a dart there!’ The next example is interesting and reveals the use of =hama for counter- expectations. I had asked a speaker to tell me about her village. When she starts describing her house, she marks the clause with =hama. 789 (36) waβaĺka miɲ́a hɨãŕka kaḿɨ ikaḿ húwa haḿa tikit́͡ʃ ʃuaŕ mat͡sat́ɨamunmaka nuiŋ́kʲa ihús húwatsui Wabal=ka mina hiã-ru=ka kamɨ ikaḿ hua-a hama Huabal=FOC 1SG.GEN house-1SG=FOC INTERJ forest\LOC be.located-IPFV MIR tikit ͡ʃi ʃuara mat͡satu-a-mau=nVma=ka nui=̃ka ihusa other person live-IPFV-NMLZ=LOC=FOC there=FOC near hua-a-t͡su-u-i be.located-IPFV-NEG-3.SBJ-DECL ‘In Huabal, my house, is located in the forest, it is not located near to where other peo- ple live.’ When I asked why she would use =hama in her description, she told me that she thought that in my knowledge of the world, I would think that all houses are part of a town. However, her house is relatively far from the village, as it was built following the old tradition where families used to live scattered, not forming villages or towns. So, to convey that unexpected information on the part of the interlocutor, she uses =hama. A similar example comes from a conversation. In this occasion, the son of one of my hosts in the community of Puerto Galilea had brought a DVD with videos and pictures of his son, my host’s grandson (who lives in the city, not in the village). Unbeknownst to my host’s son, their DVD player is broken: (37) A: βiðeo at͡ʃiaúk diβidi video at͡ʃi-a-́u=k dvd video grab-HIAF-NMLZ=Q dvd ‘Did you grab the DVD?’ 790 B: diβiði mɨsɨŕmahama mat́͡ʃu dvd mɨsɨ-ra-ma=hama mat͡ʃu dvd ruin-DISTR-REC.PT=MIR male ‘The DVD is broken, son.’ 791 CHAPTER XIX SWITCH-REFERENCE AND SUBORDINATION 19.1. Introduction In Wampis, what I call a subordinate clause is de"ned by the presence of a subordinate verb (see §19.2). A subordinate clause is subordinated relative to what I call a reference clause, i.e. a clause that functions as the main clause in the construction. Subordinate verbs in Wampis occur with su#xes whose main function is reference-tracking and topic continuity (Givón 1983; Watkins 1984; Mithun 1993). These su#xes indicate whether the dependent clause and the independent clause share the same subject (SS) or whether the dependent clause and the independent clause have a di$erent subject (DS). That is to say, Wampis exhibits what typologically is understood as a switch-reference system. This type of switch-reference where the subordinate clause is marked as sharing or not sharing the same subject with the reference clause is understood in the present work as canonical switch-reference (Haiman & Munro 1983). Wampis also possesses a non-canonical switch-reference system, which I de"ne as a reference-tracking device where at least one of the participants in the subordinate clause and in the reference clause is not a subject; i.e. 792 non-canonical switch-reference tracks continuity of some other participant in particular grammatical relations (Mithun 1993; Stirling 1993). Structurally, verbs that carry a canonical switch-reference marker in/ect for person, whereas verbs that carry a non- canonical switch-reference marker do not in/ect for person. In addition, Wampis forms subordinate clauses with subordinating su#xes accompanying the marking of switch reference on the verb. I call these su#xes “subordinators” in the restricted sense of su#xes that occur on a subordinate verb, as de"ned in §19.2. Most subordinators create a verbal stem that is marked for person, except in non-canonical switch-reference where person is not marked. In this way, I disntinguish between what I call “subordinate verbs” and nominalized verbs that occur in relative clauses and complement function. The distinction between what I de"ne as nominalized verb (as de"ned in Chapter XV) and what I de"ne as subordinate verb is made in §19.2. From the point of view of the functions a nominalized verb and a subordinate verb can have in Wampis, nominalized verbs are used for relativization. Complementation is accomplished using a range of means, including nominalization, subordination and quotative constructions.263 What I will call “subordination” (in the speci"c sense to this study of Wampis) is left to achieve 263. See Chapter XX for details on these di$erent constructions. 793 most other possible functional clause combinations in Wampis (i.e. expressions of cause, reason, simultaneity, sequentiality, frustration, and so on). This gives Wampis a highly hypotactic “character”, which is one of its salient features. This chapter follows this structure: §19.2 de"nes the subordinate verb and its morphosyntactic properties; §19.3 presents the subordinators in Wampis, §19.4 discusses person marking in subordinate verbs. This is followed by the description of non-temporal (§19.5), simultaneous (§19.6), sequential (§19.7) and imperfective DS subordinate clauses (§19.8). Section §19.9 describes di$erent subordinating su#xes that occur in SS clauses. Next, §19.10 discusses non-canonical subordination in Wampis, which is followed by an analysis of conditional (§19.11) and concessive (§19.12) constructions. The chapter closes with a short note on “clause chaining” §19.13. 19.2. The subordinate verb What I call “subordinate” verbs in Wampis cannot head independent clauses—an independent clause is characterized by having a main predication, prototypically this function in Wampis must be ful"lled by "nite main verbs, as de"ned in §12.2 and §12.7. Unlike nominalized verbs, subordinate verbs cannot modify nouns (i.e. they do not function as relative clauses), cannot be used as referring expressions, they are not 794 marked like arguments of the verb (crucially lacking properties of nominals such as case marking)—though they can function as arguments (i.e. in complementation), and have special morphology that is unique to subordination.264 The subordinate verb in Wampis is characterized by: • Presence of a switch-reference marker. • With regard to TAM marking, the subordinated verb is not marked for tense and most mood categories described in Chapter XIV. • Use of special forms to mark person, which are di$erent than the set of person markers used in "nite verbs. • Overt subordinating morpheme attached to the subordinated verb. For comparison, Table 19.1 shows the TAM properties associated with "nite independent verbs, subordinate verbs and nominalized verbs. The term “Aspectual Stem” in the table refers to whether the verb stem may contain a morpheme that can "ll the morphological position where aktionsart, imperfective, durative, present habitual or potential su#xes occur (see §12.7 for details). 264. Subordinators in Wampis do not possess the lexical derivation properties of nominalizers either—they do not create nouns that can be used at the lexical level. The scope of subordinators is always clausal. 795 Table 19.1. TAM properties of independent, subordinate and nominalized verb TAM\Verb type Independent Verb Subordinate Canonical Subordinate Non-Canonical Nominalized Verb Aspectual stem ✓ ✓ ✓ Somea Person ✓ ✓b --- Very marginal Tense ✓ --- --- ---c Mood ✓ Limited --- --- a Set II Nominalizers and Set I -tinu ‘Future Nominalizer’ can attach to an “Aspectual stem”. b Subordinated verbs use a paradigm of person markers that is distinct from the paradigm used by "nite verbs. c The future nominalizer -tinu is historically composed of the future marker -ta and the Set I agent nominalizer -inu. Synchronically, no tense marker can be received by the nominalized verb. Table 19.1 shows the “deranking” cline of subordinate and nominalized verbs relative to the "nite independent verb. Deranking is understood as the use of a verb form that is structurally di$erent than a verb form used in an independent clause; a deranked verb cannot be used in independent clauses (Stassen 1985: 77; Cristofaro 2003). Depending on the language, a deranked verb can occur with a special marker that signals that it is not an independent verb; however, depending on the language, it can also occur without certain verbal categories, such as TAM or agreement markers. Relative to subordinate verb forms in Wampis, it can be said that a deranked verb is 796 marked by reduced TAM morphology and by receiving a switch-reference morpheme. Unlike fully "nite verbs, subordinated verbs are not in/ected for tense265 and have very limited mood marking. It is possible to categorize verbs according to the place they occupy in the "nite/non-"nite continuum, as well as in terms of what type of clause type they can occur in. This is shown in Table 19.2. Note that quotative constructions serve as direct reported speech complements of a matrix speech verb, and are also used in subordinate clauses that are the functional equivalent of complements of cognition and desiderative verbs (see §20.3.2 and §20.3.3 for details). Table 19.2. Types of verbs and associated clause types Verb type Clause type Finite main verb Independent Dependent Subordinate Person-marking Non person-marking Main clauses “Coordinated” clauses, quotatives Canonical SS and DS subordinate clauses Non-canonical switch-reference subordinate clauses 265. An exception seems to con"rm the norm: a few examples in the database show that a verb marked with the de"nite future -tata can receive the simultaneous -ku (§19.7). In such occasions, the marked verb has an interpretation of ‘be about to Verb’ (which can still be seen as related to the simultaneous). 797 19.2.1. Morphological structure of the subordinate verb The structure of the subordinate verb is presented in Table 19.3 and Table 19.4. Table 19.3. Morphological structure of the canonical switch-reference subordinate verb 0 1 2 3 4 Verb Stem Subordinator Person Switch-reference Conditional Concessive Some Mooda a The only mood categories that can be marked in a subordinate verb are =ka 'Interrogative', =ʃa ‘Speculative’ and =t͡su ‘Inferential’. Table 19.4. Morphological structure of the non-canonical switch reference verb 0 1 2 Verb Stem Switch-reference Conditional Concessivea a The marking of mood categories with non-canonical switch-reference is not attested in the data. The person markers in SS clauses are actually portmanteau su#xes that mark person and identity of subjects. Di$erent-subject clauses add a DS marker -(n)i ̃which is historically related to the homophonous locative =(n)i.̃ On the other hand, clauses that take non-canonical switch-reference are not marked for person, as indicated in Table 19.4. Thus, subordinate verbs marked with canonical switch-reference show more complex structure than verbs marked with non-canonical switch-reference. 798 Morphologically, the subordinate verb di$ers from the main verb in the marking of certain categories. With regards to mood, only =ka ‘Interrogative’, =t͡su ‘Inferential’ and =ʃa ‘Speculative’ (these morphemes are described in Chapter XVIII) can be marked on the subordinate verb, but these are clause-level clitics that can be received by other constituents. Verb-only mood markers (i.e. all other mood su"xes in Wampis, cf. §14.4) do not occur with subordinated verbs. Subordinate verbs always have a morphological marking that indicates that the verb does not belong to the main clause. Subordinate verbs can receive some in/ection or not, depending on the type of switch-reference the subordinate verb co-occur with. Non-canonical switch-reference su#xes -tatamana ‘Subject to object’ and -ma ‘Non- subject to subject’ do not receive person marking, although they can occur with conditional and concessive clauses. When the subordinate verb co-occurs with a canonical switch-reference marker, it receives person marking. Aside from their verb forms, subordinate clauses essentially retain the same syntax as main independent clauses: case marking is the same; in general, relative order of noun phrases, obliques and adverbs is relatively free; dropping of pronouns and subject noun phrases is completely possible (actually, it is the most frequent pattern). 799 There is a strong tendency for the subordinate verb to occupy the "nal (right-most) position in the subordinate clause, though a number of examples show that it is possible that other constituents occupy that position, as in (1), where there is an NP (the name of the town) after the verb: (1) óuhmatsattahɛɛ wií miɲ́a jaaḱtarun pat͡ʃiśan Kanðúɣosan auhumatu-sa-tata-ha-i ui mina iaakta-ru=na pat͡ʃi-sa-nu tell-ATT-DEF.FUT-1SG-DECL 1SG 1SG.GEN town-1SG=ACC mention-SUB-1SG.SS Kandungosa=na Kandungos=ACC ‘I am going to tell mentioning my town Kandungos.’ 19.3. Subordinators The Wampis subordinating su#xes are presented in Table 19.5. I have divided the subordinators in four groups, marked by the heavy dotted lines. The symbol “---” means that the respective SS or DS category does not occur with the subordinating su#x. The "rst group is composed of subordinating su#xes that occur in SS and DS clauses. The non-temporal -sa belongs to this "rst group but it has a suppletive form -tai ̃ in DS clauses, which is a portmanteaux su#x that marks person and DS (see Table 19.7 below). The second group has only one type: the “zero form” subordinator occurs on an imperfective stem marked for person, and only occurs in DS clauses. The third group is composed of subordinators that only occur in SS clauses. Finally, the fourth group is 800 composed of the conditional and the concessive su#xes. Unlike the other subordinators, the conditional and the concessive occupy a di$erent morphological position (last— right most—position, cf. Table 19.3 and Table 19.4) in the subordinate verb. In addition, the conditional and the concessive are compatible with both canonical and non-canonical switch-reference marking. Table 19.5. Wampis subordinators Subordinator Meaning SS marking DS marking Type of switch- reference -sa/-tai ̃ Non-temporal ✓ ✓a canonical -ku (~ -ka) Simultaneous ✓ ✓ ∅ Sequential ✓ ✓ ∅ Imperfective --- ✓ -tahkama Frustrative ✓ --- -hkama Terminative ✓ --- -tasa Purpose ✓ --- -kaua Reduplicative ✓ --- -ka conditional ✓ ✓ canonical and non- canonical-ʃa concessive ✓ ✓ a The non-temporal subordinator -sa occurs only in SS clauses. In DS clauses, the suppletive form -tai ̃is a portmanteau of person and DS marking. All of the subordinators show a strong tendency to co-occur with a speci"c 801 aspectual stem.266 There are sporadic examples where the simultaneous action -ku, purpose -tasa and reduplicative -kaua su#xes occur with an aspectually-unmarked stem, but in general the subordinators co-occur with the stem types shown in Table 19.6. For greater descriptive accuracy, I have also added the non-canonical switch-reference markers -ma and -tatamana to the table, since the verb stems that carry these su#xes also vary according to the presence of any one of these switch-reference markers. Table 19.6. Aspectual stems associated with speci"c subordinators Type of stem Subordinator Meaning unmarked -sa non-temporal imperfective -ku simultaneous aktionsart/perfective ∅ sequential imperfective ∅ imperfective aktionsart/perfective -tahkama ́ frustrative aktionsart/perfective -tasa purpose imperfective -hkama ́ terminative imperfective -kaua reduplicative imperfective or aktionsart/perfective -ma non-subject to subject unmarked -tatamana subject to object imperfective or aktionsart/perfective -ka conditional imperfective or aktionsart/perfective -ʃa concessive 266. Recall that there are "ve stems related to aspect in Wampis: imperfective, perfective (done with aktionsart su#xes), potential, durative and habitual. See §13.3 for a description of these su#xes. 802 As can be seen, the conditional -ka and concessive -ʃa can occur with either perfective or imperfective stems; the non-subject to subject -ma can also occur with a perfective or imperfective stem. There is no attested occurrence of the durative (-ma), potential (-mai) or habitual (-na) stems in subordinate verbs (recall that the durative, potential and habitual occupy the same morphological position in the verb as imperfective and aktionsart su#xes, forming a paradigm, see §13.3). An auxiliary copula verb marked for subordination occurs in the case of the potential; or the potential-marked verb may occur nominalized as a complement of the subordinate verb as in (2), where the verb ‘feel’ carries the simultaneous and SS su#xes. (2) pakı́nkʲa ahápan ikúkmiahɛɛ humɛ́int͡ ʃu nɨkápɨakun paki=na=ka ahapa-a-nu iku-ki-mia-ha-i peccary=ACC=FOC abandon-IPFV-1.SS leave-WHILE.MOVING-DIST.PT-1SG.SBJ-DECL hu-mai-na-t͡ʃu nɨkapɨ-a-ku-nu take-POT-NMLZ-NEG.NMLZ feel-IPFV-SIM-1SG.SS ‘I left abandoning the peccary, feeling I couldn’t take it.’ The fact that the durative and the present habitual do not occur in subordinate clauses is likely due to the fact that subordinate clauses have stems marked for imperfective, simultaneous, or reduplicative action that generally overlap in the same functional domain as the durative and habitual (they all are somewhat associated to 803 imperfectiveness). 19.4. Person marking As de"ned in §19.1, canonical switch reference in Wampis is related to the to the marking of the subject in the dependent and reference clauses (canonical switch- reference). Canonical switch-reference verbs in Wampis mark the person-number of the subject. Table 19.7 shows the person markers used in canonical switch-reference (non- canonical switch reference, as de"ned in §19.1, does not in/ect for person—non- canonical switch-reference is covered in §19.10). “V” in Table 19.5 stands for stem "nal vowel. As can be seen in Table 19.7, the marking of person somewhat di$ers for SS and DS clauses. Non-temporal and sequential subordinate clauses use portmanteau su#xes that con/ate person and DS marking. In fact, in DS clauses, there is a striking neutralization in the marking of 1SG/3 persons, whereas 2 and 1PL remain distinctively marked. Note that, in the case of simultaneous and sequential DS clauses, the su#xes -i ̃ (which is the short form of the DS marker -(n)i—̃see §19.4.2) and -matai,̃ respectively, neutralize 1SG and 3 person.267 267. It is really striking how the distinction between 1SG/3 and 2/1PL in DS subordinate clauses is so neatly similar to distinctions in other parts of the grammar. In particular, the grouping of 804 Table 19.7. Person markers used in canonical switch-reference subordinate clauses Person Non-temporal Simultaneous Imperfective Sequential SS DS SS DS DS SS DS 1sg -nu -tai ̃ -nu -iã ∅ -nu -matai ̃ 2sg -mɨ --- -mɨ -mi -mɨ -mɨ -mi 3sg Ṽ -tai ̃ Ṽ -i ̃ Ṽ Ṽ -matai ̃ 1pl -ri -tai ̃ -ri -ri -ri -ri -ri 2pl -rumɨ --- -rumɨ rumi rumɨ -rumɨ -rumi 3pl -ara -tai ̃ -Ṽ -i ̃ -Ṽ -ara -matai ̃ a 1SG and 3 persons in simultaneous DS clauses are marked by attaching the DS marker -(n)i ̃ (§19.4.2) directly to the simultaneous subordinator -ku (§19.6). From Table 19.7, it can be seen that 1SG, 1PL and 3 grammatical person marking is neutralized with the non-temporal DS marker -tai.̃ We will see that the su#x -(n)i ̃is also used in sequential subordinate clauses with 2 and 1PL persons. Recall from §19.2.1 that the conditional and concessive subordinating su#xes occur after the person marking, in the right-most morphological position of the subordinate verb. The person- marking of subordinators that occur only in SS clauses (see Table 19.5) generally these categories is identical to how grammatical persons pattern together in the split marking system of participant arguments, as seen in §16.3. In that case, a distinction between these two groups, 1SG/3 and 2/1PL, is made: if a 1SG or 3 Agent-like participant acts upon a 3 Patient-like participant, the object noun phrase receives accusative marking. If a 2 or 1PL Agent-like participant acts on a 3 Patient-like participant, the object noun receives does not receive accusative marking. 805 patterns with the paradigm shown for simultaneous SS clauses in Table 19.7. 19.4.1. A note on the nasal marking of person in subordinate verbs Table 19.7 shows that nasalization occurs in the marking of third and 1 person in subordinate clauses. In Wampis, naasalization is realized in various surface ways, and sometimes actual nasalization is not produced (the realization can be oral, via high pitch—i.e. high tone; or by a consonant [n] on the margins of the syllable containing the nasal vowel). Where no other morpheme follows a morpheme marked for nasality, the realization is really variable and often times is not distinguishable, at "rst look, from an oral vowel. This phenomenon also occurs in subordination. However, examples like the following, where a nasal consonant surfaces in the verb [aḱuŋka], show that underlying nasalization is present to mark the third person: (3) iíɲa puhútnaʃa jat͡ʃaḿaru aḱuŋka núu ʃuaŕ núwɛiti pɨŋ́kɨr iina puhu-ta=na=ʃa jat͡ʃa-ma-ra-u a-kũ-ka 1PL.GEN live-NMLZ=ACC=ADD wise-VBZ-DISTR-NMLZ COP-SIM\3.SS-COND nu ʃuara nu=aiti pɨnkɨra NON.VIS person NON.VIS=COP.3+DECL good ‘if [a person] is one that has learned our way of living [Lit.: ‘our life’], that person is a good person.’ In (3), the simultaneous su#x -ku carried by the copula is marked for third person. This marking is via nazalization, as shown in Table 19.7. Contrast with (4), 806 where the simultaneous -ku is followed by the 2pl -rumɨ. As can be seen, no nasal surfaces in this case, as 2PL marking does not involve nasalization of its last vowel. (4) ʃuaŕ pɨŋ́kɨr aḱurmɨka tariḿat puhaŕum ʃuara pɨnkɨraa-ku-rumɨ-ka tarimata puha-rumɨ person good COP-SIM-2PL.SS-COND abundance live+ipfv-2PL.DECL ‘If you (PL) are a good person, you (PL) live in abundance.’ 19.4.2. Di&erent subject -(n)i ̃ The su#x -(n)i ̃occurs in simultaneous and sequential Di$erent Subject (DS) clauses only. It indicates that the subject of the dependent clause and the subject of the reference clause are di$erent. The su#x -(n)i ̃is distinct from the other DS markers, -tai ̃ and -matai,̃ because it occupies a morphological position after the person markers, whereas -tai ̃and -matai ̃occupy the position of person markers, as they con/ate person and DS marking (see §19.5.2 for a description of -taı ̃and §19.7.2 for a description of - matai)̃. In simultaneous clauses, -(n)i ̃co-occurs with all grammatical persons. In sequential clauses, -(n)i ̃is used with 2 and 1PL persons and the portmanteau -matai ̃is used with 1SG and 3 persons. Table 19.8 summarizes the forms of -(n)i ̃in simultaneous and sequential DS clauses, giving a paradigm with the verb puhu ‘live’. The imperfective stem—in this case, puha (puhu plus imperfective a)—is used for simultaneous clauses, and the root+aktionsart (in this case, attenuative -sa) are used for sequential clauses. 807 As we saw in §13.3.3 and §13.5, the imperfective plural -ina and the non-imperfective - ara are used in the verbal stem for 3PL persons. Table 19.8. Paradigm of -(n)i ̃ Person Simultaneous DS clausea Sequential DS clause 1sg puha-ku-i ̃ puhu-sa-matai ̃ 2sg puha-ku-mi-ni ̃ puhu-sa-mi-ni ̃ 3sg puha-ku-i ̃ puhu-sa-matai ̃ 1pl puha-ku-ri-ni ̃ puhu-sa-ri-ni ̃ 2pl puha-ku-rumi-ni ̃ puhu-sa-rumi-ni ̃ 3pl puhu-ina-ku-i ̃ puhu-sa-ara-matai ̃ a Imperfective DS clauses used the same paradigm of person and DS markers as Simultaneous, but do not include the simultaneous su#x -ku. See §19.8. The su#x -(n)i ̃has two forms. The form -i ̃occurs when the preceding vowel is not /i/ and the form -ni ̃occurs when there is a preceding /-i/. The marking of 1SG and 3 persons is neutralized in simultaneous DS clauses. Notice from Table 19.8 that the 2 person markers -mi change their usual last vowel /ɨ/ when they precede -(n)i.̃268 However, before -(n)i they occur as -mi and -rumi respectively. An alternative analysis would be that underlyingly the second person ends in /ɨ/ and this vowel undergoes 268. In other paradigms, including the person markers of "nite verbs, 2sg is -mɨ and 2pl is -rumɨ, both ending in vowel /ɨ/. 808 harmony with the vowel /-i/̃. This morphonological change would in turn trigger the occurrence of the consonant n in the form -ni ̃(likely as a means of dissimilation, as the form -ni ̃also occurs following the 1PL marker -ri, which also has an /i/). Based on homophony and morphophonological properties, the most obvious source of the DS -(n)i ̃is the locative =(n)i ̃(described in §11.5.1.2). The -i ̃of =(n)i ̃is probably also present in the other DS su#xes, -tai ̃and -matai,̃ though in those cases the development may have followed further steps (the portion /tai/̃ seems to correspond to the patient/location nominalizer -tai)̃. 19.5. Non-temporal subordination 19.5.1. Same subject -sa Non-temporal subordination is done with the su#x -sa in SS clauses. This su#x is used as a general SS manner subordinator. The su#x -sa attaches to an unmarked root. As the name indicates, non-temporal subordination indicates that the event of the subordinate clause is concomitant to that of the reference clause, without indication of a temporal relation between the dependent and the reference clause per se—however, a relative temporal meaning can be obtained depending on the context. For instance, in (5), there are two clauses marked with -sa that adopt a cause meaning. The "rst 809 subordinate clause marked with -sa was interpreted by my Wampis teacher as past tense and prior to the event of the main clause. The second subordinate clause marked with the su#x -sa was interpreted as present habitual269 whereas the main verb was interpreted as past with the intensive aktionsart su#x -ka giving the stem a perfective sense. (5) kuntiɲ́an taŋ́ku taŋkumat́sa ̃ wakɨŕa aśa ̃ naańt͡ʃa araḱnaʃa arakmat́sa ̃ wakɨŕiɲ aśa ̃ niŋ́ka t͡- sukiń huaḱuiti [kuntina=na tanku tanku-ma-a-́tasa ̃ wakɨra a-sa]̃ animal=ACC REDUP animal-VBZ-HIAF-PURP\3.SGSS want+IPFV COP-SUB\3SG.SS [naa=na=ʃa araka=na=ʃa araka-ma-a-́tasa ̃ HESIT.PRO=ACC=ADD plant=ACC=ADD plant-VBZ-HIAF-PURP\3SG.SS wakɨru-inu a-sa]̃ want-NMLZ COP-SUB\3SG.SS ni=̃ka t͡sukińta hua-ka-u=iti 3SG=FOC corner\LOC stay-INTENS-NMLZ=COP.3+DECL ‘[My mother] because she wanted to domesticate animals, umm, because she likes to grow plants too, she stayed [to live] in the corner270.’ 269. I asked the speaker if a past tense interpretation was possible and the answer was a#rmative. The point is that in a non-temporal construction, there is no intrinsically established temporal reference; any temporal interpretation comes from the context, not from the subordinator itself. 270. By ‘corner’ the speaker means on the outskirts of the village, where there is more free room for farming. 810 In the following example, the non-temporal subordinated clause acts as an adverbial of manner, indicating how the main action is done. Thus, the main action in the next example (‘living’) is done by ‘"shing’ and ‘eating’: (6) miɲ́a nuwaŕhɛinka maaśan júsan puhah́ɛɛ mina nua-ru=hai=̃ka maã-sa-nu iu-sa-nu puha-ha-i 1SG.GEN woman-1SG=FOC kill-SUB-1SG.SS eat-SUB-1SG.SS live+IPFV-1SG.SBJ-DECL ‘I live with my wife "shing and eating.’ Another example with the non-temporal subordinator -sa is in (7). The literal translation of this example shows that the semantics of the subordination is similar to the previous example, i.e. it behaves like a manner adverbial. Notice that the expression is very formulaic: this is the way how the general topic of most texts is stated in Wampis (‘I will say, mentioning X’)—the expression is usually translated as ‘I am going to talk about X’.271 Interestingly, the main verb ‘say’ is not in/ected for declarative, so it must analyzed as being inside the interrogative clause. (7) wii urúk t͡suaḿaru titʲah́ núna pat͡ʃiśan ui uruka t͡sua-ma-ra-u ti-ta-ha 1SG how heal-REFL-DISTR-NMLZ say+LOAF-IMM.FUT-1SG.SBJ 271. In the Wampis vocabulary (Jakway et al. 1987), the form is misinterpreted and glossed with the meaning of ‘about’ (‘acerca de’ in the Spanish original). However, I have no evidence that this form has developed into some sort of postposition or into an adverb. It behaves as a full verb in my data. 811 nu=na pat͡ʃi-sa-nu NON.VIS=ACC mention-SUB-1SG.SS ‘I am going to tell how I healed [from a wound in/ected by a peccary]’ (Lit.: ‘I am going to tell how I healed [from a wound...], mentioning that’.) 19.5.2. 1/3 Di&erent subject -tai ̃ In non-temporal DS clauses, -sa is not used. Instead, the su#x -tai ̃occurs. -Tai ̃is a portmanteau su#x that marks the clause as having a DS for 1 and 3 persons (see Table 19.7). The su#x is, as far as I know, not compatible with 2 person. This is very similar to how Overall (2007) describes the cognate DS marker in Awajun. As stated, - tai ̃indicates that the subjects of the dependent and reference clause are di$erent. In (8), the subject of the subordinated verb ‘go’ is a 3SG person, while the subject of the reference clause—which is a nominalization of the verb ‘follow’—is a third person plural.272 (8) naŋki ́at͡ʃiḱ ɨɨḿki wɨtɛɛ ̃nitʲaʃa pataátukaru nanki ́ at͡ʃi-ka ̃ ɨɨma-ki wɨ-tai ̃ spear\ACC grab-INTENS\3.SS move.forward-WHILE.MOVING go-1/3.DS nita=ʃa pataa-tu-ka-ara-u 3PL=ACC follow-APPL-INTENS-PL-NMLZ ‘[He] having grabbed the spear, [he] going forward, they too followed [him].’ 272. Notice that there is no overt 3PL subject in/ection on the verb (because it is nominalized), but there is a plural marking -ara that is used with third person. Hence, there is no problem in the interpretation of the argument of the nominalized verb. Recall that a nominalized verb does not take tense, person or mood su#xes. 812 In (9), the di$erent subject su#x -tai ̃occurs on the verb ‘say’ (actually, most examples of -tai ̃in the data occur with the verb ‘say’). In narratives, it is very common to use the verb ‘say’ with a SS or DS marker to recapitulate actions in between clauses. To simplify the example, a number of other subordinate clauses before the reference clause have been omitted in (9)—this is indicated by the ellipsis mark. (9) jumi ́ t͡sɨńt͡sak naŋkimaat́a ʃuaŕ jaha ́ wɨárɛi tuiɲ́a tiḿaji tútɛɛ ̃ maʃiánt͡ʃa aj́u taḱun t͡saaŋ́ anaŕmoo húakũ . . . jumi ́umpua ́tiḿaji 9.a iumi ́ t͡sɨńt͡saka nankima-a-́ta ́ ʃuara iaha ́ wɨ-ara-i rain\GEN dart throw-HIAF-IMP enemy far\LOC go.PFV-PL-APPR tu-ina timaji say-IPFV NARR ‘They were saying “Throw a rain’s dart!” “May the enemies not go away!”,’ 9.b [tu-tai]̃ say-1/3.ds ‘[when they said that],’ 9.c maʃianta=ʃa aiu ta-kũ t͡saanku ana-ra-mau hu-a-kũ Maʃianta=ADD okay say+IPFV-SIM\3SG.SS tobacco brown-DISTR-NMLZ take-IPFV-SIM\3.SS ‘Mashiań too, while saying “okay!” and taking browned tobacco leaves’ 9.d . . . jumi ́ umpu-a ́ timaji rain\ACC blow-HIAF NARR ‘. . . he blew (i.e. to invoke) the rain.’ 813 In the next example, -tai ̃attaches to the resumptive verb ‘do that’ and marks it as having a di$erent subject than the reference clause. The subject of the verb ‘do that’ is ‘Nantu’, whereas the subject of the reference clause is ‘Auhu’. (10) waá tiḿaji nańtu pitʲakri ́ húki ̃ túramunam núnitʲɛɛ ̃ naá aúhuʃa miɲ́a ɛɛ́ʃru ukúrɨawɛɛ tusa ̃. . . ni ̃íʃa waá tiḿaji 10.a ua-a timaji Nantu pitaka-ri ̃ hu-ki ̃ climb-IPFV NARR Moon traveling.basket-1PL/2PL/3.POSS take-WHILE.MOVING\3.SS ‘Having taken his traveling basket, Nantu was climbing [a ladder to the sky],’ 10.b tura-mau=nama [nuni-tai]̃ naa then-NMLZ=LOC do.that-1/3.DS HESIT ‘then, [when he did that],umm,’ 10.c Auhu=ʃa mina aíʃi-ru Common.potoo=ADD 1SG.GEN husband-1SG uku-ru-a-ua-i tu-sa ̃. . . ni=̃ʃa ua-a timaji leave.behind-1SG.OBJ-IPFV-3.SBJ-DECL say-SUB\3SG.SS 3SG=ADD climb-IPFV NARR ‘Auju too, saying: “my husband is leaving me!”, . . . she too was climbing.’ 19.6. Simultaneous subordination -ku The subordinating su#x -ku indicates that the action of the subordinate clause takes place simultaneously or overlapping with the action of the main verb. 814 The verb subordinated with -ku usually has in an imperfective stem.273 In simultaneous DS clauses, the DS su#x -(n)i ̃is added, as described in §19.4.2. Thus, the simultaneous subordinate verb has the following general structure: IMPERFECTIVE STEM-KU-PERSON-((n)i)̃ There are a few examples of the simultaneous -ku attaching to a verb marked with future. So perhaps the relation between the imperfective and the simultaneous -ku is epiphenomenal—after all, an ongoing action marked with imperfective is also very likely to be interpreted as simultaneous if it is related to another event. On the other hand, the construction of a verb marked with future+-ku has the semantics of ‘ABOUT TO VERB’. An example of the occurrence of -ku with a future-marked verb is given in (11). (11) iɲaŕtatak iɲaŕtatak wakóo iia-ra-tata-kũ iia-ra-tata-kũ ua-ka-u fall-DISTR-DEF.FUT-SIM\3.SS fall-DISTR-DEF.FUT-SIM\3.SS climb-INTENS-NMLZ ‘About to fall, he climb up [the tree].’ 19.6.1. Same-subject simultaneous subordinate clauses SS simultaneous clauses are marked by -ku plus the respective person marker as indicated in Table 19.7. Example (12) shows the use of -ku in a simultaneous SS clause. 273. There are occasional examples of -ku attaching to a stem unmarked for aspect. 815 (12) ɨmɨśtumarɛim iḿʲɛɛ ̃iḿʲɛɛ ̃wɨkaḱum ɨmɨsa-tu-ma-ra-i-mɨ imai ̃ imai ̃ wɨka-a-ku-mɨ ruin-APP-REFL-DISTR-APPR-2SG.SBJ over.there over.there walk-IPFV-SIM-2SG.SS ‘May you not ruin yourself [i.e. your own life] while you walk far over there.’ In (13), an interesting copy-verb strategy can be observed. The "rst instance of the verb ‘plead’ is subordinated with the simultaneous -ku, the second ‘plead’ is the main verb and occurs with the past habitual =hak (which in the example occurs as a separate phonological word). The passage tells how the Wampis elders who knew anɨnta ‘magical song’ pleaded with Nunkui, a being related to earth, agriculture and good hunting. (13) ut͡súmkaʃ puhústiɲa núna kaḿɨ arutti ́ túsar sɨɰɛíɲak núu úun jat́͡ʃa ɛɛ́ɲa núka sɨɰa ́ hakaŕuiti ut͡suma-ka-t͡ʃa puhu-sa-tinu=a nu=na kamɨ lack-INTENS-NEG live-ATT-FUT.NMLZ=COP NON.VIS=ACC INTERJ a-ru-tu-ti tu-sa-ara ̃ sɨa-ina-kũ exist-APPL-1SG.OBJ-JUSS say-SUB-3PL.SS plead-PL.IPFV-SIM\3.SS nu uunta iat͡ʃa a-ina nu=ka sɨa hak-ara-u=iti NON.VIS elder wise COP-PL.IPFV NON.VIS=FOC plead HAB.PT-PL-NMLZ=COP.3+DECL ‘To live without needing, [the elders] to that274 saying “to have (Lit.: ‘let it exist to my bene"t’)”, [the elders] while pleading, the wise elders used to plead.’ 274. The speaker is referring to Nunkui. 816 19.6.2. Di&erent subject in simultaneous clauses DS simultaneous subordinate clauses add the su#x -(n)i,̃ as stated in §19.4.2. Example (14) shows a simultaneous DS subordination. In the context from where the example comes, a young man talks to his father about becoming a peccary. The simultaneous marking of the subordinate gives an overlapping temporal sense between the actions (what the youth is saying and what his father is thinking while he hears it). (14) paḱi nahaańartasan wakɨŕahɛɛ túsa ̃ tóo tiḿaji taḱui ̃ apaŕiŋkʲa aańtrant͡suk tat́ɛɛ túsa ̃ . . . paki nahaana-ra-tasa-nu wakɨra-ha-i tu-sa ̃ peccary make-DISTR-PURP-1SG.SS want+IPFV-1SG.DECL say-SUB\3.SS ta-u timaji [ta-ku-i]̃ apa-ri=̃ka SAY+IPFV-NMLZ NARR say+IPFV-SIM-DS father-1PL/2PL/3.POSS=FOC aantura=̃t͡su=kɨ ta-tai tusa ̃ in.vain\3.SS=INFER=RESTR say+IPFV-INFER say-SUB\3.SS ‘[The youth] said: “I want to become a peccari”, while he was saying that, his father saying: “He must be lying” (Lit.: ‘He must be speaking in vain’) . . .’ There is an important alternation in the form of the simultaneous subordinator in DS clauses. The simultaneous -ku often times assumes the form -ka.275 This alternation of vowels is not well-understood at present, but it does not seem to be related to any 275. Actually, in most examples of simultaneous DS clauses the form of the simultaneous su#x is -ka. But notice that -ku also occurs, as in (13) above. 817 synchronic rule. What is more, the alternation in the vast majority of cases occurs in DS clauses (there are very few examples276 of the form -ka attested in SS clauses).277 For instance, in (15), the simultaneous+DS marker sequence is pronounced [kɛɛ]̃ and not [kui]̃. Thus, the underlying representation must be /kai/̃—recall that the sequence /ai/ in Wampis is pronounced either [ɛɛ] or [ɛi] (cf. §3.7.1).278 (15) kiaŕɛɛ ̃wɨkaiḿiahi kaḿɨ huaḱmiahi kintamɨakɛɛ ̃ kiarai ̃ wɨka-i-mia-hi kamɨ at.sunset walk-LOAF-DIST.PT-1PL.SBJ+DECL INTERJ hua-ka-mia-hi kintama-a-ka-i ̃ STAY-INTENS-DIST.PT-1PL.SBJ+DECL become.night-IPFV-SIM-DS ‘At sunset, we went, and we stayed when it was becoming dark.’ 19.7. Sequential subordinate clauses ∅ Sequential subordinate clauses present the events of two or more clauses in 276. Only three examples of the form -ka in SS clauses were found in a search in 40 Wampis texts and all of them occur with the resumptive verb nuni ‘do that’. The remaining occurrences of -ka are in DS clauses, occurring in 53 instances (95%) in the same texts. 277. Some other morphemes in Wampis also show this u~a alternation. For example, the applicative form -tu, sometimes surfaces as -ta. What remains a mystery to me, though, is why the simultaneous form -ka, as far as I know, occurs with such vast propensity in DS clauses. 278. Speakers also wrote instead of the expected -ku when I asked them to help me identify the su#x while translating texts (this is not unexpected, though, because the writing practice of Wampis speakers is not normalized yet; i.e. writing is based on how they hear the sounds). 818 succession or temporal sequence. In Wampis, subordinate verbs in sequential clauses are zero-marked. The subordinate verb must be formed with a perfective stem (with one of the available aktionsart su#xes)279 plus the respective person markers. The relationship between perfectivity and sequentiality has been argued for in the broader literature (Hopper 1979; Fleischman 1990); but see various studies (Payne & Shirtz 2015), which show it is not universal. The motivation to use the perfective is presumably to let the discourse move on by presenting discrete, successive actions. In SS sequential clauses the [perfective stem+person markers] structure is enough to identify the verb as sequential. DS sequential clauses are marked with a portmanteau -matai ̃in 1SG and 3 persons, and with a person marker plus the DS marker -(n)i ̃for 2 and 1PL persons. A paradigm of the sequential subordinate form with the verb puhu ‘live’ is presented in Table 19.9. The “preferred” aktionsart su#x that occurs with the verb puhu ‘live’ to form the aktonsart stem is the attenuative -sa. 279. But recall that not all verbs select an aktionsart su#x. 819 Table 19.9. Paradigm of verb puhu ‘live’ in SS and DS sequential clauses Person Sequential SS clause Sequential DS clause 1SG puhu-sa-nu puhu-sa-matai ̃ 2SG puhu-sa-mɨ puhu-sa-mi-ni ̃ 3SG puhu-sa ̃ puhu-sa-matai ̃ 1PL puhu-sa-ri puhu-sa-ri-ni ̃ 2PL puhu-sa-rumɨ puhu-sa-rumi-ni ̃ 3PL puhu-sa-ara puhu-sa-ara-matai ̃ 19.7.1. Same-subject sequential clauses SS sequential clauses have a zero form subordinator, and the subordinate verb is marked for person. In (16), the verb ‘make’ is marked as a SS sequential clause. The verb occurs with the high a$ectedness aktionsart su#x -a ́and it is marked with the 1PL person form the set of subordinate person markers (in this case -ri (cf. Table 19.9)). Notice that the reference clause contains a normative construction with a copula plus a nomizalition. One of the predicative functions of the patient/location nominalizer -tai ̃is to mark a normative tense with the meaning of “what we usually do”. Even though the copula is marked for 3 person, the normative is understood as 1PL, thus the SS person marking in the subordinated verb ‘make’ is 1PL. The verb ‘"x’ in the context of the example assumes 820 the meaning of ‘prepare’. (16) nihʲaḿt͡ʃikʲa tú iwʲartɛiɲti tú nahanaŕ umutɛíɲti wampistiḱʲa nihamat̃͡ʃi=ka tu iuara-tai=̃aiti manioc.beer=FOC thus "x=NMLZ=COP.3+DECL [tu nahana-a-́ri] umu-tai=̃aiti wampisa=ti=́ka thus make-HIAF-1PL.SS drink-NMLZ=COP.3+DECL Wampis=PL.SAP=FOC ‘The manioc beer thus we prepare, having prepare [it] thus, we the Wampis drink.’ The next example comes from a text in which the speaker explains how to make manioc beer. As can be observed, there is a string of sequential clauses, all marked as 1SG.SS. All dependent verbs are marked with an aktionsart su#x (in the example, the intensive -ka and the attenuative -sa occur on the subordinated verbs) and their interpretation is perfective. The dependent clauses are all subordinated to the clause containing the "nite verb ‘crush’, which appears at the end of the string. (17) hiiɲúm ɨkɨŋ́kan iɲaŕkan nui ̃ ́ɨnɨḱan apúhsan ihúahɛɛ [hii=numa ɨkɨna-ka-nu] [inara-ka-nu] "re=LOC set."re-INTENS-1SG.SS cook-INTENS-1SG.SS nui ̃ [ɨnɨ-ka-nu] [apuhu-sa-nu] ihu-a-ha-i there take.out.of."re-INTENS-1SG.SS put-ATT-1SG.SS crush-IPFV-1SG.SBJ-DECL ‘[I] having put it (i.e. the manioc) in the "re, cooked it, there, having taken it out of the "re and put it [on a surface], I crush it.’ In (18), there is another example with SS sequential clauses. The example comes from a text in which the speaker tells about the foundation of the Wampis town of Puerto Galilea (one of the founders is the elder mentioned in the example, Alfonso 821 Graña). The resumptive verb ‘do that’, plus ‘arrive’ and ‘speak’, are all ordered sequentially, and temporally the events in the subordinate clauses all occurred before the event of the verb ‘take’ in the reference clause (which appears at the end of the example). The clauses that contain a subordinate verb are in square brackets [ ]. (18) iriŋ́ku D. B. naańtin taaḿiaji nútikʲa taa ́ eskueĺa urɛími túsa ̃ t͡ʃit͡ʃaś huiɲa ́ túkɨ at́iɲa nú- na ut͡ʃiń hukiñ́ɛiti Alfonso Graɲa . . . Pukaĺpa hukiñ́ɛiti 18.a irinku D. B. naa-̃tinu ta-a-́mia-ji white.person D. B. name\1PL/2PL/3.POSS-ATTRIB arrive-HIAF-DIST.PT-3.PT+DECL ‘One white man whose name was D. B. arrived,’ 18.b [nutika-a ̃]́ [ta-a ̃]́ [[eskuela ura-i-mi do.that-HIAF\3.SS arrive-HIAF/3SG.SS school open-LOAF-HORT tu-sa]̃ t͡ʃit͡ʃa-sa]̃ say-SUB\3SG.SS speak-ATT/3SG.SS ‘having done that, having arrived, having said: “Let’s build a school”,’ 18.c hui=̃ia tukɨ a-tinu=a nu=na here=ABL INTERJ COP-FUT.NMLZ=COP NON.VIS=ACC ut͡ʃi=na hu-ki-inu=aiti Alfonso Graɲa child=ACC take-WHILE.MOVING-NMLZ=COP.3+DECL Alfonso Graña ‘he took that young man who was going to be from here (i.e. who would found the town), Alfonso Graña . . .’ 822 18.d Pukalpa hu-ki-inu=aiti Pucallpa take-while.moving-nmlz=cop.3+decl ‘he took him to [the city of] Pucallpa.’ 19.7.2. Di&erent subject sequential clauses As presented in Table 19.9, DS sequential clauses are marked with -matai ̃for 1sg and 3 persons, and with person+-(n)i ̃for 2 and 1PL persons. As the name indicates, DS sequential clauses describe events in temporal sequence, in which the subject of the event predicated by the dependent clause is di$erent from the subject of the reference clause. Instances of DS sequential clauses can be observed in (19)–(20). In (19), the di$erent subject is a 3 person (‘toucan’ in the subordinated clause, and ‘Ikam’ in the reference clause), so the su#x -matai ̃is used on the subordinate verb. (19) t͡sukaŋka ́nɨnaḱmatai ̃iḱʲam tukúmaji ́ [t͡sukanka ́ nɨna-ka-matai]̃ iḱama tuku-ma-ji toucan /y-INTENS-1SG/3.DS Forest shoot.PFV-REC.PT-3.PT+DECL ‘When the toucan /ew, Ikam shot it.’ On the other hand, in (20), the di$erent subjects are 2SG person in the subordinate clause and 1SG person in the reference clause. Thus, the 2SG DS of the subordinated clause is marked with a person marker plus the DS su#x -(n)i,̃ in accordance with the paradigm previously given in Table 19.9. 823 (20) umarú haíme wiʃ́a aḿɨ hɨɨ̃ ̃ḿin puhah́ɛɛ aḿɨ unt͡súrkaʔmin taah́ɛɛ uma-rú haime ui=ʃa amɨ hɨĩ-̃mi=ni ̃ brother-1SG\VOC Jaime 1SG=ADD 2SG house/POSS-2SG=LOC puha-ha-i live/be+IPFV-1SG.SBJ-DECL [amɨ unt͡su-ru-ka-mi-ni]̃ ta-a-́ha-i 2SG call-1SG.OBJ-INTENS-2SG-DS arrive-HIAF-1SG-DECL ‘My brother Jaime, I too am in your house, you having called me, I have come.’ 19.8. Imperfective di&erent subject subordination ∅ An imperfective stem with a “zero” subordinator can appear in switch reference constructions and its interpretation is somewhat similar to simultaneous clauses with - ku. However, unlike the subordinator -ku, this imperfective subordination only occurs in DS clauses, marked with the DS su#x -(n)i.̃ Thus, the structure of the imperfective DS subordinate verb is similar to the DS simultaneous subordinate verb (§§19.6.2), minus the subordinator -ku: IMPERFECTIVE STEM-PERSON-(n)i ̃ A paradigm with the verb puhu (the imperfective stem is puha from puhu ‘live’+-a ‘Imperfective’; for 3PL, -ina ‘Plural imperfective’ is used) is given in Table 19.10. 824 Table 19.10. Paradigm of puhu ‘live’ in Imperfective DS clauses Person DS marking 1SG puha-i ̃ 2SG puha-mi-ni ̃ 3SG puha-i ̃ 1PL puha-ri-ni ̃ 2PL puha-rumi-ni ̃ 3PL puha-ina-i ̃ Notice, again, that the marking of 1SG/3 persons is neutralized, whereas 2/1PL remain with their respective person markers. The next example illustrates the use of the DS marker -(n)i.̃ (21) tikit́͡ʃka paki ́mɛĩɲ́ɛi ̃niŋ́kʲa paki ́wɨámunam iḿani út͡ʃika pataátuki ̃paki ́wɨú tiḿaji [tikit ͡ʃi=ka paki ́ ma-̃ina-i]̃ other=FOC peccary\ACC kill-PL.IPFV-DS ni=̃ka paḱi wɨ-a-mau=nama 3SG=FOC peccary go-IPFV-NMLZ=LOC imani ut͡ʃi=ka pataa-tu-ki ̃ paki ́ wɨ-u timaji INTS child=FOC follow-APPL-WHILE.MOVING\3.SS peccary\ACC go-NMLZ NARR ‘Other [people] were killing peccaries, [but] the he (i.e. the young man), when the pec- caries were gone, the young man having followed the peccaries, he went away.’ 19.9. Same-subject only subordinators There are subordinators in Wampis that are only used in SS clauses. They receive the same person marking of the simultaneous SS clauses as described in Table 19.7. SS-only subordinators are: -tahkama ́Frustrative’, -hkama ‘Terminative’, -tasa 825 ‘Purpose’ and -kaua ‘Reduplicative’. 19.9.1. Frustrative -tahkama ́ The expression of frustration is a grammatical category in Wampis. The frustrative construction can be structurally de"ned as: [V-FRUST, VMAIN] The subordinate verb receives the frustrative su#x and an overt main reference clause states the source280 or the consequence of the frustration. Semantically, the frustrative codi"es an action that is done in vain, i.e. the expected outcome fails to be realized. Typologically, frustrative as a grammatically-marked category is not commonly found across-languages; however, it does seem to be a category employed by a fair number of languages of the greater Amazon region (Payne & Payne 1990; Dietrich 2006; Aikhenvald 2012: 183–185). The use of -tahkama ́is exempli"ed in (22). In this case, the frustrative su#x is received by a perfective stem of the verb ‘say’ (marked as perfective by the the low a$ectedness aktionsart su#x -i—in the example, ti is composed of tu ‘say’ plus -i ‘Low a$ectedness’): 280. Statements equivalent to being ‘unable’ or ‘have di#culty’ to carry out an action are also considered ‘source’ of frustration by this de"nition. 826 (22) miɲ́aka jat͡súrnaka manturtuaẃarɛɛ ̃titahkama ̃ ́úutu tiḿaji mina=ka iat͡su-ru=na=ka ma-̃tu-ru-tu-aú-ara-i2̃81 1SG.GEN=FOC brother-1SG=FOC kill-APPL-1SG.OBJ-APPL-HIAF-PL-3.PFV+DECL ti-tahkama ̃ ́ uu-tu-u timaji say+LOAF-FRUST\3.SS cry-APL-NMLZ NARR ‘[he] wanted to say “they killed my brother (to my detriment)”, but he cried.’ In the next example, because the character had been hiding the food from her husband, he got angry and tore her mouth, so she cannot pronounce words well. When her husband abandons her, she tries to say something but she is unable to: (23) auhuʃa nunik uutahkama ̃ ́waá ɛíʃrua titahkama ̃ ́waúhu waúhu tóo tiḿaji auhu=ʃa nuni-ka ̃ uuta-hkama ̃́ Common.Potoo=ADD do.that-INTENS\3SG.SS cry+IPFV-TERM/3SG.SS wa aiʃi-ru-a ti-tahkama ̃ ́ IDEO husband-1SG-AFF say+LOAF-FRUST/3SG.SS wauhu wauhu ta-u timaji IDEO IDEO say+IPFV-NMLZ NARR ‘Auhu, having done that, upon crying, she wanted to say “ay, my husband!”, but she said “wauhu wauhu”.’ Historically, the Wampis frustrative -tahkama ́may have developed from a periphrastic desiderative construction. First, for comparison consider Table 19.11, which shows the di&erent frustrative constructions in Jivaroan languages. 281. This is one of the examples where there seems to be a double marking of the applicative. This was previously commented in §13.2.3. 827 Table 19.11. Frustrative constructions in Jivaroan languagesa Shuar Achuar-Shiwiar Awajun Wampis -tah tukama ́ -tat kama ́ -takama -tahkama ́ a Data comes from: Turner (1992) for Shuar, Fast et al. (1996) for Achuar-Shiwiar and Overall (2007) for Awajun. Following arguments expressed in Overall (2007: 386), the frustrative su#x seems to have grammaticalized in Awajun and Wampis from a construction in which the subordinate verb received the desiderative -tah and was dependent on the verb tu ‘say’ marked with an (old?)282 terminative *-kama ́form. This can be more clearly observed with examples from Shuar:283 (24) defecate-DISTR-DES say-TERM\3.SS be.constipated-IPFV-3.SBJ-DECL ‘Though he wants to defecate, he is constipated’ (Spanish original: ‘Aunque quiera defe- car, esta ́estreñido’) (Pellizaro & Náwech 2005: 183) (25) work-ATT-DES say-TERM-1SG.SS complete-INTENS-NEG-PT-1SG.SBJ-DECL ‘Even though I worked, I did not complete it’ (Spanish original: ‘A pesar de haber traba- jado, (no cumpli)́’ (Turner 1992: 83) 282. Notice that the terminative su#x in Wampis is -hkama,́ whereas the cognate forms in other Jivaroan languages is -kama.́ See §19.9.2 for a description of the terminative in Wampis. 283. The examples are presented as they appear in the original sources. I retain the alphabetic conventions in those sources, thus = /h/ and = /t͡ʃ/. The glosses are mine—notice that Pellizaro and Naẃech do not gloss the example, and Turner glosses tukama as ‘even though’ (‘a pesar de’). 828 So the old complex construction had the semantics of ‘On saying I want to do X, (X failed to be realized)’. In Wampis, it seems that the verb ‘say’ that used to be in the original periphrastic construction has been phonologically eroded, and the remaining morphology collapsed onto the desiderative -tah. The desiderative -tah historically contained a future marker *-ta (which synchronically in Wampis is the ‘Immediate future’) plus the "rst singular person marker *-ha. This explains the “strange” modern form tahkama ́with a consonant /h/ as coda in the "rst syllable (as we saw in Chapter III, underlyingly, Wampis syllables always end in a vowel), as well as why the modern su#x tahkama ́does not delete any of its vowels (historically, the expression fossilized with an elided vowel). 19.9.2. Terminative -hkama ́ The terminative -hkama ́indicates the ending limit of the action of the subordinate clause relative to the (punctual) expression of the event of the reference clause. The terminative construction involves a subordinate verb with an imperfective stem marked with hkama ́plus a perfective-marked verb in the reference clause: [V-TERM, VMAIN] Semantically, the event of the subordinate clause marked with -hkama ́is 829 interpreted as an ongoing action that ends as the event of the reference clause is performed. Unlike simultaneous clauses, terminative clauses do not imply simultaneity in the events of the subordinate and reference clauses; and unlike sequential clauses, the event in the subordinate clause is seen as ongoing (whereas in sequential clauses the event is bound, perfective). This can be seen in the next example. A group of people goes searching for a young man that had left his family to live with peccaries. The action of searching concludes as they "nd the young man’s tracks: (26) paki ́pataátuk ijahkama ̃ ́nawɨ ́wɛińkʲar . . . paki ́ pataa-tu-ka ̃ ii-a-hkama ̃ ́ peccary\ACC follow-APPL-INTENS\3.SS see-IPFV-FRUST\3SG.SS nawɨ ̃ uaina-ka-ara foot\1PL/2PL/3.POSS see-INTENS-3PL.SS ‘When they were following the peccaries, they found his [the young man’s] foot prints . . .’ (Lit.: ‘Having followed the peccaries, upon looking around, having found the foot prints . . .’) In (27), the verb ‘thus happen’ receives the terminative subordinator: (27) miɲiɲ́a tiḿaji tumahkama ̃ ́wɛiŋ́kʲaru tiḿaji jaki ́ɨkɨḿas kɨnkúkanam at͡ʃiḿas uini-ina timaji tuma-a-hkama ̃ ́ uaina-ka-ara-u timaji come-PL.IPFV NARR thus.happen-IPFV-TERM\3SG.SS see-INTENS-PL-NMLZ NARR iaki ́ ɨkɨma-sa ̃ kɨnkuka=nama at͡ʃi-ma-sa ̃ above\LOC sit-SUB\3.SS bamboo=LOC grab-REFL-SUB\3SG.SS ‘They were coming, on happening thus, they saw [the girl] sitting above, grabbing her- self in the bamboo.’ The form of the Wampis terminative -hkama ́slightly di$ers from its cognates in 830 other Jivaroan languages. Awajun has the form -kama (Overall 2007: 384). There is no certain data from Achuar, Shiwiar and Shuar, but some examples found in Fast et al. (1996) and Pellizaro and Naẃech (2005) may suggest that there exists a form -kama ́ (perhaps not very productive anymore?)284 in Achuar-Shiwiar and Shuar. Without a context the clauses in the examples may be interpreted with a terminative meaning; nevertheless, the cited authors seem to translate them with a frustrative meaning (unfortunately there are no explanations provided for the examples). That said, notice that the structure of the frustrative construction presented by Fast et. al. (1996) for Achuar-Shiwiar and Turner (1992) for Shuar (as summarized previously in Table 19.11) is di$erent from what can be observed in (28)–(29): (28) (Achuar-Shiwiar) child food-ACC eat-TERM?/FRUST? choke-DISTR-3.PFV+DECL ‘The child choked trying to eat the food’ (Spanish original: ‘El niño se atoró tratando de tragar la comida’) (Fast et al. 1996: 153) (29) (Shuar) go.forward-TERM?/FRUST? die-INTENS-3.PFV+DECL ‘Because of going forward, he died.’ (Spanish original: ‘Por adelantarse se murió’) (Pellizaro & Náwech 2005: 158) Another important morphosyntactic di$erence with the Awajun terminative 284. Saad states that, in Shuar, -kama ́is only found fosilized in the frustrative (Saad 2014: 137). But see example (29) below for a possible example of -kama ́in Shuar. 831 construction is that in Awajun the form -kama attaches to an unmarked stem, whereas in Wampis it attaches to an imperfective stem. If it is assumed that the suffix -kama ́in the above examples from Achuar-Shiwiar and Shuar are cognates with the Wampis terminative suffix, then it can be seen that in those Jivaroan languages the cognate suffix -kama ́apparently also attaches to an unmarked stem. The difference in morphosyntactic behavior and phonological form seem to point to a different historical development for the Wampis terminative -hkama.́ Speculatively, it seems like in Wampis the suffix comes from an imperfect stem marked for person (a paralel with the frustrative construction suggests that this was a 1SG -ha)285 plus the old form *-kama.́ Eventually, the 1SG person marker+*-kama ́portion would have been reanalyzed into the modern terminative, whereas in the other languages *-kama ́continued to be the terminative. 19.9.3. Purpose -tasa The suffix -tasa introduces a purpose adverbial; i.e. a clause that exhibits an intention to motivate the realization of an event “which must be unrealized at the time of the main event” (Thompson, Longacre, & Hwang 2007: 250). The next examples 285. Another person marker that contains an /h/ is the 1PL -hi. 832 illustrate purpose clauses. (30) núna ʃuara ́wɛińkʲar maá t͡ʃit͡ʃaśtasa ipʲaŕmiaji nu=na ʃuara ́ uaina-ka-ara ma NON.VIS=ACC person\ACC see-INTENS-3PL.SS INTERJ t͡ʃit͡ʃa-sa-tasa ̃ ipa-ara-mia-ji speak-ATT-PURP\3.SS invite-PL-DIST.PT-3.PT+DECL ‘Having seen that person, they would invite him to speak.’ (31) húu t͡ʃit͡ʃaḿan uhaḱtasan tah́amɨ hu t͡ʃit͡ʃama=na uha-ka-tasa-nu ta-hamɨ PROX problem=ACC inform-INTENS-PURP-1SG.SS say+IPFV-1SG>2SG+DECL ‘I tell you to inform you of that problem.’ For Shuar, Turner (Turner 1992: 82-83) has suggested that -tasa may be a contraction of / which can be analyzed as: (Stem)-tah tu-sa (Stem)-DES say-SUB Following Turner’s analyzis, the modern Wampis purpose su#x -tasa likely developed from a similar construction, i.e. from a desiderative construction involving the verb ‘say’ marked with the non-temporal subordinator -sa. The phonological reduction may be explained by the intermediate form *tah-tsa (not present anymore in Wampis but reported by Turner for Shuar) with the vowel /u/ of -tu elided: *tah-tu- sa>tah-tsa>tasa. Additional comparative evidence comes from Awajun, where the fossilized verb tu ‘say’ can still be observed in the 3 person purpose marker -tatus <*-ta 833 ‘Desiderative future’, tu ‘say’ and *-sa ‘Subordinator’ (Overall 2007: 382). The purpose subordinator -tasa is used in complement functions with the desire verb uakɨru ‘want’ (§20.3.2). While semantically a complement, the structure involved in this type of complementation is the same as that used for switch-reference subordination. 19.9.4. Reduplicative -kaua The reduplicative -kaua indicates the reiterative nature of the action of the event described in the subordinate clause. The presence of the su#x -kaua (iconically) triggers the reduplication of all the phonetic material up until the second surface mora (i.e. second vowel from the left) of the marked verb. The reduplicated material always occurs preposed to the marked-verb. The reduplicative occurs on an imperfective verb stem. The vowel /u/ of the su#x -kaua allows for internal vowel elision of the preceding /a/ vowel in cases when this /a/ occupies a position in which it can be elided. This suggests that historically the su#x -kaua may be decomposable into the formatives *-ka and *-ua. The possible meanings of these hypothesized formatives are not known at present. The reduplicative construction can be summarized as: [REDUP Verb-REDUP] 834 As stated above, the event of the clause marked with -kaua presents a reiterative action that spans a certain amount of time. In (32), a 1SG participant comments about how he was "shing during a trip into the forest, and then how his group continues moving to "nd a place to camp: (32) putúŋka huwa ́ huwaḱuan ɛiḿkamiahɛɛ ójanam 0et́͡ʃahɛɛ ̃ ihʲún maá maákuan tikit́͡ʃ aʃi ́ ʃiír núkap maaḿiahɛɛ nutikan wɨá wɨákuar ɨt́sa kiakɛi ̃atakʃa nui ̃hintʲa ́huwaḱmiahi 32.a putu=na=ka huwa huu-a-kaua-nu carachama=ACC=FOC REDUP gather-IPFV-REDUP-1SG.SS aima-ka-mia-ha-i oja=nama "ll-INTENS-DIST.PT-1SG.SBJ-DECL pot=LOC ‘Gathering and gathering carachama "sh, I "lled the pot;’ 32.b /et͡ʃa=hai ̃ ihu-nu maa ma-̃a-kaua-nu harpoon=COM stab.PFV-1SG.SS REDUP kill-IPFV-REDUP-1SG.SS tikit ͡ʃi aʃi ʃiira nukapɨ ma-̃a-́mia-ha-i nutika-a-nu other all very many kill-HIAF-DIST.PT-1SG.SBJ-DECL do.that-IPFV-1SG.SS ‘having stabbed with the harpoon, killing and killing, many other ["sh] I "shed having done that;’ 32.c wɨa wɨ-a-kaua-ri ɨtsa ̃ kia-a-ka-i ̃ REDUP go-IPFV-REDUP-1PL.SS sun get.dark-IPFV-SIM-DS 835 atakʃa nui ̃ hintʲa ́ hua-́ka-mia-hi again there trail\LOC stay-INTENS-DIST.PT-1PL.SBJ+DECL ‘[we] going and going, when the sun was setting, there, again on the trail we stayed.’ In the next example, the reiterated action is ‘walk’: (33) waka ́wakaḱua ̃kaʃiki ́taa ́tiḿaji uaka uaka-a-kaua ̃ kaʃi=kɨ ́ ta-a ́ timaji REDUP walk-IPFV-REDUP\3.SS night=RESTR\LOC arrive-HIAF NARR ‘Walking and walking, he would arrived at night.’286 Apparently, an extended function of the reduplicative -kaua is to intensify the action of the marked verb. For instance, in (34), the reduplication emphasizes the action of crying, rather than indicating that the child cries several times: (34) awatiḿ kaḿɨ ut͡ʃi úutu timaji úu úutɨakua kamɨ ut͡ʃi jaki ́waḱoo timaji hɨɰ̃a ́nɛińtin auati-ma kamɨ ut͡ʃi uu-tu-u timaji hit+LOAF-NON.SBJ>SBJ INTERJ child cry-APPL-NMLZ NARR uu uu-tu-a-kaua kamɨ REDUP cry-APPL-IPFV-REDUP INTERJ ut͡ʃi iaki ́ ua-ka-u timaji hɨã ́ nainti=̃ni ̃ child above\LOC climb-INTENS-NMLZ NARR house\GEN top\1PL/2PL/3.POSS=LOC ‘When they hit her, the child cried, crying and crying, the child climbed up above to the roof of the house.’ 19.10. Non-canonical switch-reference Wampis possesses two switch-reference markers that do not behave like the 286. The verb ‘walk’ is uɨka. I was told that the form uaka, which occurs in this example, is a variant pronounced by some Wampis speakers. 836 canonical switch-reference system described in the preceding sections. In canonical switch-reference, subject continuity is tracked among di$erent clauses; however, in non-canonical switch-reference at least one of the participants in the subordinate clause and in the reference clause is not a subject (Mithun 1993; Stirling 1993). Wampis possesses two another reference tracking system that marks a transition between events where at least one co-referential participant is not a subject in one of the clauses of the subordinating construction: -ma ‘Non-subject to Subject” and -tatamana ‘Subject to Object’. In each case, the second-named role of the glosses is the role of the "reference" clause. Thus, the Wampis non-canonical switch-reference can be described as a device that tracks the continuity of a participant through a shift of scenes, regardless of whether it is a subject or not. Table 19.12 summarizes the marking of participants with -ma and -tatamana. Table 19.12. Non-canonical switch-reference markers in Wampis Stem Su#x Subordinate clause → Reference clause imperfective or aktionsart/perfective -ma Non-subject → Subject unmarked aspect -tatamana Subject → Object 837 A subordinate verb that carries -ma nor -tatamana cannot receive person marking. This suggests that a possible historical scenario from where these two switch- reference markers developed was a process of nominalization. Speci"cally, *-ma seems to have been an old nominalizer (this hypothesized morpheme seems to be a formative in Set II nominalizer -mau ‘Non-subject nominalizer’ too); the ‘subject to object’ switch reference -tatamana seems to be composed of a formative *tata, the hypothesized old nominalizer *-ma and *na, which is likely related to the current accusative =na in modern Wampis. A verb marked with non-canonical switch reference cannot take other subordinators (which in the verb marked with canonical switch-reference occupy the position that precedes the marking of person (cf. Table 19.3)). However, non-canonical switch reference is compatible with the conditional -ka (§19.11) and concessive -ʃa (§19.12) su#xes. Notably, these su#xes are historically related to the focus marker =ka (§18.4) and the additive focus marker =ʃa (§18.5), which are usually hosted by nouns and noun phrases. 19.10.1. Non-subject to subject -ma The su#x -ma attaches to an aktionsart/perfective stem and indicates that a non-subject participant (a P participant, an object of applicative or a Location) in the 838 subordinate clause becomes the subject (A/S participant) of the reference clause. If -ma occurs on transitive verbs, the resulting the sense of the verb is similar to a passive, but there is no true passive in Wampis. For instance, in (35) the subject ‘dream’ occurs overtly in the nominative case and is not demoted, whereas the object (‘him’) does not occur as an overt noun phrase (3 person objects are not overtly marked in the verb). (35) kaŕa maãḿ núʃa ɛíʃmaŋt͡ʃiʃa kanaŕui kara ma-̃a-́ma nu=ʃa aíʃimanku-t͡ʃi=ʃa kana-ra-u=i dream kill-HIAF-NON.SBJ>SBJ NON.VIS=ADD man-DIM=ADD sleep-DISTR-NMLZ=COP.3+DECL ‘When he was really sleepy (Lit.: ‘The dream having killed him’), that little man too fell asleep.’ In (36), the object of the applicative (‘her’) becomes the subject of the nominalized verb ‘melt’ in the main clause: (36) tií t͡sɨɨŕ ukat́ram tií t͡sɨɨŕ ukat́ram miɲaŕu tii t͡sɨɨra uka-tu-ra-ma tii t͡sɨɨra mina-ra-u INTS hot pour.liquid-APPL-NON.SBJ>SBJ INTS hot melt-DISTR-NMLZ ‘[Because] They poured very hot [water] [on to her], she was one to melt.’ The use of -ma to mark a Location-to-Subject interclausal relation is rare but entirely possible. In (37), the location within subordinate clause is co-referent with the subject of the reference clause. Here the location is the ‘road’ to the town of Bagua, on which a 1SG participant is traveling. 839 (37) túra wií wɨkaḿ Baɣ́wa hinti ̃ń kaḿɨ nui ̃ ́hińtʲa mɨt͡saŋ́krukamiaji tura ui wɨka-a-ma Bagua hinti=̃ni ̃ then 1SG walk-IPFV-NON.SBJ>SBJ Bagua trail\1PL/2PL/3.POSS=LOC kamɨ nui ̃ hinta mɨt͡sańkVru-ka-mia-ji INTERJ there trail collapse-INTENS-DIST.PT-3.PT+DECL ‘Then, when I traveled on the road to [the town of] Bagua, there, the road collapsed.’ 19.10.2. Subject to object -tatamana The su#x -tatamana indicates that the subject (an A/S participant) of the subordinate clause is co-referent with an object (a P participant) of the reference clause. (38) wɨt́atman hɨɰat́tak wɨt́atman iṕati maãẃaru tiḿaji wɨ-tatamana hɨa-a-́tata-kũ287 wɨ-tatamana go-SBJ>OBJ arrive-HIAF-DEF.FUT-SIM\3.SS go-SBJ>OBJ iṕati ̃ ma-̃aú-ara-u timaji shoot+LOAF\3.SS kill-HIAF-PL-NMLZ NARR ‘He was going, when he was about to arrive, having shooting him, they killed him.’ The following passage from a narrative nicely summarizes the complex conveyance of information that speakers can accomplish with non-canonical switch- reference -ma ‘Non-subject to subject’ and -tatamana ‘Subject to object’. The passage describes how a woman named Ipak turns into the achiote (Bixa orellana) tree, and why the red squirrel (who in the myth is a man named Kunam) is red-colored. To help guide 287. Notice here another example of the future+simultaneous construction previously mentioned in §19.6. 840 the reading of the passage, I include a parenthetical indication of the grammatical role of the co-referent participants (either Ipak or Kunam) on the right side of the gloss line. A “>” symbol means that the grammatical role of the co-referent participant changes in the next main clause. (39) ipʲaḱun nahaańartasa puhóon kaḿɨ nahaańar wahat́taman warúku tiḿaji naá kúnam nútikʲa waraḿ núka ipʲaḱka nahaańarka húu nahaańatɛik̃ nɨrɨḱ apih́ʲuma wahaśu tiḿa- ji nútika iḿani kúnam kaḿɨ watat́man tapit́ at͡ʃiḱ ʃaúkasa jaḱar akúpkau tiḿaji 39.a Ipaku=na nahaana-ra-tasa puha-u=na kamɨ Achiote=ACC make-DISTR-PURP live/be+IPFV-NMLZ=ACC INTERJ ‘That Ipak that was to transform... 39.b (Ipak: SBJ in this clause >OBJ in next clause) nahaana-ra uaha-tatamana make-DISTR stand-SBJ>OBJ ‘having transformed [into a tree]... 39.c (Ipak: OBJ) ua-ru-ka-u timaji naa Kunampɨ climb-APPL-INTENS-NMLZ NARR HESIT.PRO Red.squirrel ‘umm, Kunam climbed (i.e. Kunam climbed Achiote).’ 39.d (Ipak: OBJ in this clause > SBJ in next clause) nutika ̃ ua-ru-a-ma nu=ka ipaku=ka do.that+HIAF\3.SS climb-APPL-IPFV-NON.SBJ>SBJ NON.VIS=FOC Achiote=FOC nahanaa-ra=ka make-DISTR=FOC ‘Having done that, when he was climbing [Ipak], that Ipak having transformed...’ 841 39.e (Ipak: SBJ) hu nahaana-tai=̃kɨ nɨrɨ-ka ̃ PROX make-NMLZ=RESTR yield.fruit-INTENS\3.SS apihu-ma uaha-sa-u timaji bend-REFL stand-ATT-NMLZ NARR ‘when she was transformed, having yield a lot of fruits, she stood hanging low.’ 39.f (Kunam: SBJ in this clause > OBJ in next clause) nutika ̃ iḿani Kunampɨ kamɨ ua-tatamana do.that+HIAF\3.SS INTS Red.squirrel INTERJ climb-SBJ>OBJ ‘Having done that, when Kunam climbed far up... 39.g (Kunam: OBJ) tapit at͡ʃi-ka ̃ ʃouka-sa ̃ iaka-ra ̃ IDEO grab-INTENS\3SG.SS ornate-SUB\3SG.SS rub-DISTR\3SG.SS akupɨ-ka-u timaji let.go-INTENS-NMLZ NARR ‘[she] having whisked him away, having rubbed him painting him, she let go [of Kunam].’ 19.11. Conditional clauses with -ka Conditional constructions mark the protasis (which is the subordinated verb in Wampis) with -ka, whereas the apodosis contains the main 'nite verb. Unlike all other previously described subordinators, -ka occurs in the last position in the subordinate verb (fourth morphological slot, as seen previously in Table 19.3). The conditional su(x can combine with other subordinators that can appear in DS clauses (non- temporal, simultaneous, sequential and imperfective) and can occur on either an 842 imperfective or perfective stem. The structure of the conditional-marked verb can be summarized as: STEM-(SUBORD)-PERSON-((n)i)̃)-ka Person marking follows the Simultaneous SS and DS pattern (see Table 19.7), with one important di$erence: the 1PL can be optionally marked by the su#x -i and not by -ri. This 1PL su#x -i is unique to conditional clauses in the Wampis data. The person marking paradigm in conditional clauses is provided in Table 19.13. Table 19.13. Person marking in conditional clauses SS DS + Conditional 1sg -nu -i ̃ -ka 2sg -mɨ -mi-ni ̃ 3sg Ṽ i ̃ 1pl -i / -ri -i / -ri 2pl -rumɨ -rumi-ni ̃ 3pl -Ṽ -i ̃ The following examples illustrate the conditional. In (40), it can be observed that the conditional occurs on the verb ‘exist’ that contains the simultaneous su#x -ka plus the DS -ni:̃ 843 (40) kat́i ʃiír pɨŋ́kɨr kuiŕ aḱɛiŋkʲa mat͡ʃit́ri ̃húki ̃t͡supiḱ ahũaŕ kaḿɨ húu aŕmaji kati ̃ ʃiira pɨńkɨrakuira a-ka-i-̃ka shoot (plant)\1PL/2PL/3.POSS very well baby exist-SIM-DS-COND mat͡ʃita-ri ̃ hu-ki ̃ t͡supi-ka ̃ machete-1PL/2PL/3.POSS take-WHILE.MOVING\3.SS cut-INTENS\3.SS ahũa-́ra ̃ kamɨ huu-u a-ara-ma-ji knock.down-DISTR\3.SS INTERJ gather-NMLZ COP-PL-IMM.PT-3.PT ‘If there are very green palm shoots, [the women] having taken the machete, having cut and knocked [the palm shoots] down, they gather [the palm shoots].’ In (41), there are two conditional clauses subordinated to the verb umu-mai- na=iti ‘it can be drunk’. Notice that in the "rst verb (uma-ka-i-ka) marked with the conditional, the 1PL is marked as -i. The second verb (uma-ra-tah ta-ku-ri-ka) is a complex desiderative construction that literally means ‘if we say: let’s drink’ but is translated as ‘if we want to drink it’, the 1PL is marked with -ri. Apparently, there is no di$erence in marking the 1PL with -i or -ri. (41) kaʃiḱ t͡sawańtɛik̃ʃa umaḱɛika umúmɛiɲɛiti jumiń umaŕtah takúrka kaʃiki t͡sauanta-i=̃kɨ=ʃa uma-ka-i-ka early.morning day=LOC=REST=ADD drink+IPFV-SIM-1PL-COND umu-mai-na=iti drink-POT-NMLZ=COP.3+DECL iumin uma-ra-tah ta-ku-ri-ka sweet drink.PFV-DISTR-DES say+IPFV-SIM-1PL-COND ‘Early next day, if we drink [the plantain drink], it can be drunk (Lit: it is drinkable), if we want to drink it sweet.’ 844 In the next example there is a 2SG participant; the main verb is an imperative form: (42) aḿɨ ut͡súmakmɨka sɨata ́ amɨ ut͡suma-a-ka-mɨ-ka sɨa-ta ́ 2SG lack-IPFV-SIM-2SG-COND plead-IMP ‘If you need, request!’ Some examples in the data show that the conditional can occur with non- canonical switch-reference markers too. In (43), a speaker is narrating about the old quarrels between the Awajun and the Wampis, the su#x -ma ‘Non-subject to subject’ occurs with the conditional in the excerpt. Notice that there appears to be a small error of performance in the example—the speaker uses the 3SG pronoun instead of a 3PL. In Wampis, number is optional, so these “errors” are not usually salient as the referents are almost always clear from context.288 (43) maãḿka ʃuarań wampiśnaʃa kahɨŕak niĩʃa ma ̃ ́hakaŕuiti ma-̃a-́ma-ka ʃuara=na wampisa=na=ʃa kill-HIAF-NON.SBJ>SBJ-COND enemy=ACC Wampis=ACC=ADD 288. This was translated in Spanish as ‘El awajún también mataba’ taking the singular noun to represent the Awajun in general. Many varieties of Spanish allow for a singular noun to refer to a class of entities, though the agreement with the verb is always singular in that case—that is somewhat similar to what the Wampis speaker is doing, but there is an agreement mismatch in the example as there is a plural marking on the nominalized verb. 845 kahɨŕa-ka ̃ niĩ=ʃa ma ̃ hak-aŕa-u=iti be.angry-INTENS\3.SS 3SG=ADD kill PT.HAB-PL-NMLZ=COP.3+DECL ‘If [the Wampis] killed [the Awajun], [the Awajun] having gotten angry, they too used to kill the Wampis enemies.’ 19.12. Concessive clauses with -ʃa Concessive constructions indicate a relationship of contrast or counter- expectation of a main proposition (expressed in the reference clause) relative to a concession of another proposition (expressed in the subordinate clause). Concessive clauses in Wampis are marked by the su#x -ʃa. Like the conditional, the concessive occupies the last morphological position in the subordinate verb. Like conditionals, concessive clauses are compatible with canonical DS switch-reference as well as with non-canonical switch-reference. The following example is from a text where the speaker is telling how to make a plantain drink. The process consists of boiling the plantain longer than normally, until it is very soft, so that the plantain can be easily mashed and combine with water: (44) narúkmatɛiʃ̃a iɲɛísar ukuwa ́puhútɛiɲɛiti washukti ́túsar naru-ka-matai-̃ʃa inai-sa-ri uku-a cook-INTENS-1SG/3.DS-CONCESS leave-ATT-1PL.SS boil-IPFV puhu-tai=̃aiti uaʃu-ka-ti ́ tu-sa-ri live/be-NMLZ=COP.3+DECL become.brown-INTENS-JUSS say-SUB-1PL.SS ‘Even though [the plantain] is cooked, we leave it boiling so that it becomes brown (Lit.: ‘we leave it boiling, saying: “let it become brown”.’ 846 In the next example, a group of people is having trouble when their canoe is about to sink. In spite of asking some young men for help, they do not help: (45) tútaiʃ̃a út͡ʃikʲa ańtit ͡ʃarmiaji tu-tai-̃ʃa ut͡ʃi-ka anti-t ͡ʃa-ara-mia-ji say-1/3.DS-CONCESS child-FOC hear+LOAF-NEG-PL-DIST.PT-3.PT+DECL ‘Even though I told them [to help], the young men did not pay attention.’ In the context from where example (46) comes, the character feels ashamed because he is sickly. He does not want to be seen by a couple of women that were waiting for him: (46) wahóo tiḿaji hɨɰ̃a ́hɨa ̃ʃ́a ikʲaḿ kaʃ́i wɨḿatɛɛ ̃hɨãt́hɛɛ túsa ̃ uaha-u timaji hɨã ́ hɨa-̃ʃa stand+IPFV-NMLZ NARR house\LOC arrive.PFV-CONCESS ikaḿ kaʃi wɨ-matai ̃ hɨã-ta-ha-i tu-sa ̃ forest night come-1/3.ds arrive-IMM.FUT-1SG.SBJ-DECL say-SUB\3.SS ‘He was standing, although he had arrived, in the forest (i.e. outside of the house), say- ing “when the night has come, I will arrive.” The concessive is historically related to the additive =ʃa, which usually occurs with NP. According to Overall (2007), the additive in Awajun sometimes has concessive readings, as in the next example: (47) Awajun wɨ-mi dɨkas kaʃi-ʃa go:PFV-HORT really night-ADD ‘Let’s go, really, even though (it is) night.’ (Overall 2007: 508) I have not found examples like (47) in Wampis. What is very common in 847 Wampis is to have the sequence =kɨ=ʃa ‘=Restrictive=Additive’ assuming functions of scalar additives (“even”) (Krifka 1998; Gast & Van der Auwera 2011). An example of this function is given in (48). (48) wiḱʲa apat́͡ʃkɨʃa ui=ka apat ͡ʃi=kɨ=ʃa 1SG=FOC mestizo=RESTR=ADD ‘I am not even mestizo.’ Likely through this construction with a restricted focus, =ʃa developed further functions such as concessive marking. The relationship between additive and concessive appears to be very strong cross-linguistically (cf. König (1991), Gast and Van der Auwera (2011) or discussion in some of the papers in Haspelmath (2004)). 19.13. A brief note on “clause-chaining” Like many other predicate-"nal languages, Wampis is also a clause-chaining language, i.e. speakers can (and usually do) use long strings of subordinate clauses that are dependent on one reference (=main) clause. This pattern of clauses strung together constitute a important feature of the Wampis discourse. The following is an example from a Wampis narrative showing a long series of clauses strung together. The "rst clause (‘We ten people went’) is left in the example to provide a more complete context, as it connects with a previous string of clauses. 848 (49) 49.a [dies ʃuara=tí wɨ-mia-hi] ten person=SAP go-DIST.PT-1PL.SBJ+DECL ‘We ten people went,’ 49.b [taiu wɨ-ru-a-ku-ri] oilbird go-APPL-IPFV-SIM-1PL.SS ‘while going to Cueva de los Tayos,’289 49.c [wɨ-sa-ri] go-SUB-1PL.SS ‘going,’ 49.d [iamai=kɨ=ka iurumaka hu-ki-ri] now=RESTR=FOC food take-WHILE.MOVING-1PL.SS ‘at the beginning [we] having taken food’ 49.e [paantama nihamat̃͡ʃi umi-ka-ri] plantain manioc.beer prepare-INTENS-1PL.SS ‘plantain, having prepared the manioc beer’ 49.f [imani-sa-ri] do.much-SUB-1PL.SS ‘doing that much’ 289. The literally translation is ‘Oilbird’ but the speaker is metonymically referring to the caves where the oildbirds live, known with the Spanish name Cueva de los tayos. 849 49.g [aʃi umi-ka-ri] all prepare-INTENS-1PL.SS ‘having prepared all,’ 49.h [ii [iu-a-́tinu-ri=̃ʃa] [aʃi ́ hu-ki-ri] 1PL eat-HIAF-FUT.NMLZ-3/1PL/2PL.POSS=ADD all take-WHILE.MOVING-1PL.SS ‘what we were going to eat, having taken all,’ wɨ-mia-hi] go-dist.pt-1PL.sbj+decl ‘we went.’ In (49), all clauses except the "rst and last one are non-"nite: these are the only clauses which possess verbs fully marked for TAM, i.e. they are marked for aspect, tense, person and mood. All other clauses are both non-"nite and subordinated, as can be seen by the use of switch-reference su#xes and subordinators. This type of clause chain are very typical in Wampis narratives, the study of constructions related to chain clauses, including possible con"gurations, relations to aspectual distinctions and issues of word order remains as a point of research for future study. 850 CHAPTER XX RELATIVIZATION, COMPLEMENTATION AND OTHER COMPLEX CONSTRUCTIONS 20.1. Introduction This chapter describes the main strategies for relativization, complementation and other complex constructions. The chapter is functionally motivated, as relativization, complementation and the other complex constructions described in this chapter employ di$erent morphosyntactic strategies, specially nominalization (as de"ned in Chapter XV) and subordination (as de"ned in Chapter XIX). The structure of the chapter is as follows: §20.2–§20.3 provide notes on relativization and complementation in Wampis, respectively; §20.4 describes some adverbial constructions involving nominalization; "nally, §20.5 describes the Wampis equivalent of a tautological in"nitive construction. 20.2. Relativization A relative clause is one that functions as a modi"er of a noun. Relativization is a nominalization process in Wampis. Relative clauses in Wampis are typically postnominal, a syntactic property that they share with adjectives and appositional 851 phrases. There are two major relativization constructions in Wampis. The "rst involves Set II nominalizaers -u ‘Subject nominalizer’ and -mau ‘Non-subject nominalizer’ (see Chapter XV for a description of nominalizers). Both nominalizers can attach to an aspectualized stem. 20.2.1. Relative clauses with encliticized demonstrative The encliticized relative clause construction in Wampis consists of a verb that can carry aspect, tense and person, but not mood information. In terms of its internal structure, this construction preserves the argument structure of "nite clauses (i.e. the arguments of the relativized verb receive case marking) and the relativized verb always occurs in "nal position. In terms of its external structure, the relativized verb is in/ected for its syntactic role in the matrix clause (2). However, when relativized with a demonstrative, the copula a does not carry any verbal morphology (1). Relative clauses with encliticized demonstratives can be externally-headed (1) or internally-headed (3). (1) ʃuaŕ pɨŋ́kɨr anú taaj́i [NHead Modi"er]NP.subject ʃuar [pɨnkɨra a=nu] ta-a-́ji person good COP=NON.VIS arrive-HIAF-3.PT+DECL ‘The person that is good arrived.’ 852 (2) taẃa núna antúkhɛɛ [ta-ua nu]=na antu-ka-ha-i say-3.SG NON.VIS=ACC hear-INTENS-2SG.SBJ-DECL ‘I heard what he is saying’ (3) ni ̃í puhańui ̃wɨkaḱtahɛɛ [ni ̃ puhawa=nu]=i ̃ wɨka-ka-ta-ha-i 3SG live+IPFV-3.SBJ =NON.VIS=LOC travel-INTENS-IMM.FUT-1SG.SBJ-DECL ‘I am going to travel where he lives.’ 20.2.2. Relatives clauses with Set II nominalizers -u and -mau Relative clauses with Set II nominalizers -u and -mau can be headless, externally- headed or internally-head in Wampis (see discussion below for illustrations). In terms of the internal grammar of the relative clause, though relativized (i.e. nominalized) verbs retain certain verbal morphology, they are not fully in/ected as main, independent verbs are (see examples here, and Chapter XV for more details). In object relative clauses, the relativized verb in Wampis does not assign case to its object, the head of a relative clause. Thus, even when Wampis nominalizations allow for some verbal structure (they occur in aspectual stems, preferred order is verb-"nal), relative clauses in the language do not possess certain internal grammar characteristics that are related to a "nite verb, such as the marking of core-argument NPs and, more obviously, full "nite morphology. As for the external grammar exhibited by relative clauses in Wampis, it is basically that of a nominal, being able to receive most case/oblique and discourse- 853 related markers that a Wampis NP can usually take. 20.2.2.1. Externally and internally-headed relative clauses Externally-headed relative clauses have their head outside the relative clause. The most frequently in externally-headed relative clauses is Head-Modi"er. Internally- headed relative clauses have their head inside the relative clauses. When the common argument is the subject of the relative clause, the Set I nominalizer -u ‘Subject nominalizer’ is used. Example (4) illustrates the use of the nominalizer -u in an externally-head relative clause. The head noun is a core argument of the main verb, in this case the subject. There is a gap strategy employed by the construction, symbolized by ∅ in the morphemic analysis line. (4) túra tsɨrɨḱa nui ̃ ́puhaú hatʃań waińkauwaiti [ NHead NModi"er ]NP.subject tura tsere=ka [nui ̃ ∅ puja-u] and monkey.spec=FOC there live/be+IMPERF-NMLZ [N]NP.object V hacha=na waina-ka-u=ait-i axe=ACC see-INTENS-NMLZ=COP-DECL ‘And the monkey that was there looked for the axe.’ Relativization with -u in which the common argument is the object of the main verb is rare in the data gathered for Wampis. In (5) the accusative marker only occurs 854 attached to the last element of the object of the main verb, which happens to be the nominalized structure. (5) ʃuaŕ ni ̃h́ʲɛi ̃maãńiktasa miɲóunʃa nekah́akuiti [ʃuara [ni=̃hai ̃ ma-a-́nai-ka-tasa ̃ wina-u]]=na=ʃa person 3SG=COM kill-HIAF-RECP-INTENS-PURP\3SG.SS come+IPFV-NMLZ=ACC=ADD nɨka-hak-u-iti know-HAB.PT-NMLZ=COP.3+DECL ‘He knew the person that was coming to "ght him.’ In some other cases, such as in (6), the head and the relativized verb share the accusative marker.290 This is analyzed as a case of a nominalized clause (‘that was swimming’ in the example below) in an appositive construction with another noun (t͡sunki ‘water being’ in the example below). (6) t͡sunkiń jukumaún ɨntsańam waińkamiahɛi [t͡sunki]=na [juku-ma-u]=na ɨntsa=nama waina-ka-mia-ha-i water.being=ACC swim-DUR-NMLZ=ACC river=LOC see-INTENS-DIST.PT-1SG.SBJ-DECL ‘I saw a water-being that was swimming in the river.’ In the following example, the object relative clause shares the focus marker =ka with other members of the NP because of the presence of the demonstrative, which triggers agreement of =ka among members of an NP (cf. §8.3.1). The object 290. The only case where the accusative marker occurs in other elements of a NP is when a demonstrative is present. Demonstratives in Wampis make it obligatory that other nominal and adjectival/modifying elements in the NP occur with the accusative =na and the focus =ka. See example (7) below. 855 (‘Arutama’) of the relativized verb (‘see’) does not take accusative marker, neither it takes the focus =ka. This is an indication that, "rst, Arutama is internal to the relative clause (as it does not take =ka), and second, internally, the marked verb of the relative is treated as a nominal and not as a predicate (as it does not mark its object, Arutama, with the accusative). (7) núka ʃuaŕka arútam waińkouka nu=ka shuara=ka [arutama waina-ka-u]=ka that=FOC person=FOC spirit.power see-INTENS-NMLZ=FOC ‘that person who has seen Arutam. . .’ Compare the previous example with (8). In (8) there is a "nite verb with an accusative-marked object NP. (8) arútman waińkamiahɛɛ [arutama]=na waina-ka-mia-ha-i spirit.power=ACC see-INTENS-DIST.PT-1SG.SBJ-DECL ‘I saw an Arutam.’ I turn now to non-subject participants of the relativized verb. Such participants are relativized with the Set II nominalizer -mau ‘non-subject nominalizer’. The head noun is the object of the relativized verb. The example in (9) shows an object relativization, with a gap strategy. (9) iʃit́ʃik anɨákmaun aúhmatsahai [NHead NModi"er ]NP.object V [iʃitʃiki [∅ anɨa-ka-mau]]=na auhumatu-sa-ha-i little.bit remember-INTENS-NMLZ=ACC tell-ATT-1SG.SBJ-DECL ‘I’ve just told the little bit that I remember.’ 856 When the subject of an object relative clause is an overt nominal, the overt subject precedes both the object (which is the head of the relative clause) and the nominalized verb. That is, the structure as a whole is internally headed. This order shows a good degree of isomorphism between relative clause and main clause word orders, which are SOV. Notice that in internally headed relative clause structures, the head of the relativized verb does not carry the accusative marker, just as in example (9) above. The structural properties of this type of relative clause in Wampis is illustrated in (10). (10) wií arútam anɨákɨamuka kaḿɨ aẃai muukań [wi [arutama] [anɨaku-a-mau]]=ka kamɨ a-wa-i muuka=na 1SG spirit.power remember-IPFV-NMLZ=FOC INTERJ exist-3.SBJ-DECL head=ACC ‘Arutama that I remember...there is head-Arutam291.’ As we can see, in (10) the head of the relative clause, Arutam, is embedded as part of the relativized clause. Again, notice that the head is not marked with an accusative marker, unlike the object of the fully in/ected verb in (11). (11) muúk arútman anɨákɨahɛɛ muuka arutama=na anɨaku-a-ha-i head spirit.power=ACC remember-IPFV-1SG-DECL ‘I remember the head-Arutam.’ 291. The most important spiritual belief in the Wampis culture is that of Arutama, a spirit of power that manifests itself through visions. One of its manifestations is a monstrous head, muuka arutama ‘head-Arutam’. 857 Set II nominalizer -mau can also relativize an oblique (semantically, the location) argument of the verb. (12) núu núŋka wií akiíɲamu nu nunka [ui akiina-mau] NON.VIS land 1SG be.born.PFV-NMLZ ‘the land where I was born’ The locative relativization on a noun that functions as an object argument of the main clause usually involves a topic-comment strategy in the matrix clause. The noun phrase which contains the relative clause occurs initially, followed by a complete main clause which makes a predication about the initial NP. In (13) the main clause verb is hɨa-̃ma ‘build house’. Notice that in (13) there is no case marking on the fronted NP, but there is an obligatory co-referential (or resumptive) pronoun with the accusative case marker in the main clause (in (13), the resumptive pronoun is the non-visible anaphoric nu). (13) miɲ́a hɨ ̃ár puhaḿunam wiḱʲa núna hɨa ̃ḿramhɛɛ [mina hɨã-ru]j puha-mau=nam 1SG.GEN house-1SG live+IPFV-NMLZ=LOC wika [nu=na]j hɨa-ma-ra-ma-ha-i wi=FOC NON.VIS=ACC house-VBR-DISTR-REC.PT-1SG-DECL ‘My house where I live, I build it.’ 20.2.2.2. Headless relative clauses Headless relative clauses are those “which themselves refer to the noun that they 858 modify” (Payne 1997: 328). Headless relative clauses constitute a very common relativization strategy in Wampis. Headless relative clauses in Wampis take case marking. It is possible to have a headless relativization of an oblique argument with the Set II nominalizer -mau ‘Non-subject nominalizer’ (an example of an oblique relativization, which is semantically a location, is given in (17)), but speakers tend to favor the inclusion of an overt NP to disambiguate the reference of the subject relativized verb. The following examples illustrate headless relative clauses in di$erent argument positions as well as in nominal predicates. Headless relative functioning as subject: (14) aúhutumauka nuiń naŋkańɨawai [auhu-tu-mau]=ka nui ̃ nankana-a-wa-i tell-APPL-NMLZ=FOC there "nish-IPFV-3-DECL ‘What is told "nishes there.’ Headless relative functioning as object: (15) aúhmatuhakmaun anɨáku aśa [auhuma-tu=hak-mau]=na anɨa-kũ a-sa ̃ tell-APPL=PT.HAB-NMLZ=ACC remember+IPFV-SIM\3.SS COP-SUB\3.SS “Because she remembered what used to be told’ Headless relative functioning as complement in copular construction: (16) núwaiti kaḿɨ matsaḿsaruka kanúsa huińka nu=aiti kamɨ [matsama-sa-ara-u=ka kanusa hui=̃ka] NON.VIS=COP.3 INTERJ live-ATT-PL.PFV-NMLZ=FOC Kanusa here=FOC ‘those are the ones who live here in Kanus.’ 859 Headless relative clause functioning as oblique argument (location): (17) waβaĺkɛiti akiíɲarmu Huabalka=iti akiina-ara-mau Huabal=COP.3+DECL be.born.PFV-PL-NMLZ ‘Huabal is the place where they were born.’ 20.3. Complementation A complement clause is understood in this work as a clause that functions as an argument of some other (prototypically main) clause (Payne 1997; Noonan 2007). There is no simple complement construction in Wampis. Functional Wampis equivalents to what are complement clauses in other languages are realized by employing morphosyntactically distinct strategies: a) nominalization b) subordination (with switch reference)292 c) desiderative constructions d) quotative constructions In what follows I provide some brief notes on complement clauses that employ nominalizing and subordinating strategies in §20.3.1. Desiderative and quotative constructions are described in §20.3.2 and §20.3.3, respectively. 292. I am restricting the meaning of ‘subordination’ as this concept is de"ned in Chapter XIX. 860 20.3.1. Complementation using nominalization and subordination There is no unique pattern of complementation for verbs in Wampis; that is, what type of strategy is used depends on the matrix verb. Generally, it can be said that nominalization is the favored strategy when the subject of the complement clause and the subject of the matrix are co-referent. In turn, subordinating su#xes are used when the subject of the complement clause and the subject of the matrix clause are di$erent. Some verbs of perception like antu ‘hear’ select a complement clause that refers to an event, in which case the non-subject nominalizer -mau is used. This is illustrated in the next examples. The nominalized clause, functioning as an object of the verb, receives the accusative =na. (18) utʃiri ̃ ́uútmaun antúk [utʃi-ri ̃ ́ uu-tu-mau]=na antu-ka ̃ child-1PL/2PL/POSS\GENcry-APPL-NMLZ=ACC hear-INTENS\3SG.SS ‘The bear, having heard the cub’s crying.’ (19) miɲ́a hintaŕka jusá uhukɨ ̃ ́wahaúwayi tamaún antúk [mina hinta-ru=ka iusa ́ uhukɨ ̃ waha-u=a-ji 1SG.GEN trail-1SG=FOC macaw.sp\GEN tail\3.POSS stand+IPFV-NMLZ=COP-3.PT+DECL ta-mau]=na antu-ka ̃ say+IPFV-NMLZ=ACC hear-INTENS\3.SS ‘She heard that he was saying: “the macaws’s tail is the one hanging in my trail”.’ In the above examples, the subject of the complement clause and the subject of 861 the main clause are di$erent. Another strategy when the subject of the complement verb is di$erent than the subject of the matrix verb involves a subordinate construction that consists of the subordinator -ku ‘simultaneous’ plus a switch-reference person marker (in this case, the di$erent subject marker -ni)̃. (20) kaʃit́in ikaḿnumia taákmin antúkmahai kaʃi-tin ikama=numa=ia ta-a-ku-mi-ni ̃ antu-ka-ma-ha-i night-TIME forest=LOC=ABL arrive-IPFV-SIM-2SG-DS hear-INTENS-REC.PT-1SG-DECL ‘Last night I heard you arriving from the forest.’ (lit. ‘At night, you arriving from the forest, I heard.’) In (20), the subordinated verb ‘arrive’ receives the ‘simultaneous action’ subordinator -ku, a person marker -mi and an di$erent subject marker -ni.̃ Rather than a nominalized complement clause, what we have is a typical clause combining construction in Wampis (cf. Chapter XIX). Other verbs of perception such as uaina ‘see’ select a participant nominalization with the subject nominalizer -u in their complement structure, as illustrated in (21). (21) eskúβi múnt͡su umóun waińkamhi [Sccoby munt͡su uma-u]=na uaina-ka-ma-hi Scooby breast drink+IPFV-NMLZ=ACC see-INTENS-REC.PT-1PL+DECL ‘We saw that Scooby drank milk.’ Another verb of cognition such as nɨka ‘know’ typically takes a complement with Set I action nominalizer -ta if the subject of the main and the dependent clause are the same, as in (22). Note that the nominalized verb does not receive case marking because 862 of the case-marking pattern of Wampis (if 2 A acts on a 3 P, P does not receives accusative marking, as explained in §16.3). (22) aḿɨka jukúmat nɨkaḿɨ amɨ=ka iukuma-ta nɨka-a-mɨ 2SG=FOC swim-NMLZ know-IPFV-2SG+DECL ‘You know how to swim.’ With the same verb, nɨka ‘know’, when the complement clause involves a di$erent subject, there is a “relativization” strategy. The dependent verb is nominalized with the subject nominalizer -u from set II. (23) ʃuaŕ naŋkí huki ̃ ́miɲaúnt͡ʃa nɨkah́akuiti [ʃuar nanki ́ hu-ki ̃ wina-u]=na=ʃa person spear\ACC take-WHILE.MOVING\3SG.SS come+IPFV-NMLZ=ACC=ADD nɨka=hak-u=iti know-PT.HAB-NMLZ=COP.3+DECL ‘He also knew the person that come carrying a spear.’ A nominalized complement clause with Set I -ta ‘action nominalizer’ functions as the object of the inchoative complement-taking verb ‘begin’. Notice that, in this case, the nominalized verb receives accusative marking. (24) nui ̃ ́tʃitʃatań naŋkaḿawaru nui ̃ [tʃitʃa-ta]=na nankama-aú-ara-u there speak-NMLZ=ACC begin-HIAF-PL.PFV-NMLZ ‘There, they started to speak.’ 20.3.2. Desiderative constructions We have seen some complementation uses of nominalizations in the previous 863 examples. However, not all verbs use nominalizers for complement clauses. Modality verbs such as wakɨru ‘want’ express their complement with an purpose construction with the subordinator -tasa ‘purpose’ (25) (see §19.9.3 for a description of -tasa) or they use a desiderative construction with -tah ‘desiderative’ (26) (see discussion below for a description of the desiderative construction in Wampis). When subordinators are used, the clauses they subordinate do not receive argument marking; i.e. they are not treated as nominals, unlike nominalized complement clauses. (25) wií sɨɨ́ tit́asan wakɨŕahai wi sɨɨ ti-tasa-nu wakɨru-a-ha-i 1SG thanks say+LOAF-PURP-1SG.SS want-IPFV-1SG-DECL ‘I want to say “thanks”.’ The desiderative construction consists of the desiderative su#x -tah, which attaches to an aktionsart verbal stem, plus the verb ‘say’. (26) wií sɨɨ́ titah́ tah́ai wi sɨɨ ti-tah ta-ha-i 1SG thanks say+LOAF-DES say+IPFV-1SG-DECL ‘I want to say “thanks”.’ (Lit: ‘I say, wanting to say thanks!’) For di$erent subjects with ‘want’, speakers use a quotative construction with the subordinator -sa, as shown in the next example. (27) tat́i túsan wakɨruthɛi ta-ti tu-sa-nu wakɨru-tu-ha-i arrive-JUSS say-SUB-1SG.SS want-APPL-1SG-DECL ‘I want you to come.’ (Lit.: ‘I want, saying: “may you come”.) 864 20.3.3. Quotative construction Manipulative verbs in Wampis also typically coach their “complements” in quotative constructions, as shown in (28). (28) Puhupat́ akúpkamaji ikaḿnum Suwań wɨta ́túsan Puhupata akupɨ-ka-ma-ji ikama=numa Sua=na uɨ-ta ́ tu-sa ̃ Puhupata send-INTENS-REC.PT-3.PT forest=LOC Sua=ACC go-IMP say-SUB\3SG.SS ‘Puhupat ordered Suwa to go to the forest.’ (Lit.: ‘Puhupat sent Suwa to the forest, say- ing: “go!”.’) The verb of perception ii ‘see’ usually occurs in quotative constructions with the extended meaning of ‘think’ (i.e. the quotative construction with a perception verb is used to express a cognitive process), as in the next example. (29) aímkamapha túsa ijahkamá tikít͡ʃik t͡ʃankín atsóon wɛíɲak aima-ka-ma-api-ha tu-sã ii-a-hkamã ́ "ll.up-INTENS-REC.PT-SUD.REAL-1SG.SBJ+EXCL say-SUB\3SG.SS see-IPFV-TERM\3.SG.SS tikit ͡ʃiki t͡ʃankina a-tsa-u=na uaina-ka one basket exist-NEG-NMLZ=ACC see-INTENS ‘upon thinking, saying: “I "lled it (i.e. the basket) up!”, having seen that there was one basket missing. . .’ (Lit. ‘Upon seeing, saying: “I "lled the basket up”. . .) 20.4. Other adverbial constructions involving nominalization Nominalized constructions with -mau plus a locative marker can be used for adverbial functions, typically temporal. Nominalized temporal constructions are illustrated in the next examples, where they involve event nominalizations. 865 (30) kaʃ́i haśmaunam hɨɰaú [kaʃi has-mau=nama] hɨa-u nightbecome-NMLZ=LOC arrive-NMLZ ‘When it became night, he was one to arrived.’ (31) jamaí puhúsa wɨámunam ma ́jaaḱat úun hase ́túsan taj́ame [jamai puhu-sa-nu wɨ-́a-mau=nam] ma now live-SUB-1SG.SS go-IPFV-NMLZ=LOC INTERJ jaakatauunta has-ɨ tu-sa-nu ta-hamɨ town big become-3.PFV+DECL say-SUB-1SG.SS say-1SG>2SG+DECL ‘Now while I go on living, the town has become big, I tell you.’ The next example was translated with an interpretation of reason, rather than with a temporal interpretation. The construction is basically the same as the previous examples: (32) tʃitʃaḿ iwaŕumunam awaŕunka kaḿɨ huɲińka utsaánawaiti [tʃitʃama iwaŕa-mau=nam] awaruni=ka kamɨ problem "x.PFV-NMLZ=LOC Awajun=FOC hesit hui=̃ni=ka utsaana-u=aiti here=ALL=FOC enter.PFV-NMLZ=COP.3+DECL ‘Because the problem [i.e. the war] was "xed, the Awajun entered over here.’ 20.5. “Tautological in!nitive” construction with -tai ̃ One stylistic construction found in Wampis narratives is similar to what have been called “tautological in"nitive” or “copy-verb” constructions in other languages (Goldenberg 1998). In this construction, a nominalized form of the same verb occurs before a semantically main verb. 866 In the Wampis data collected, the semantically main verb (i.e. the one with assertive force) can occur in its bare root form or as a nominalized form.293 The verb in initial adverbial clause position, takes the Set I nominalizer -tai ̃(§15.4.4) and then adds the restrictive =kɨ and the additive =ʃa morphemes. (33) tukutɛik̃ɨʃa tukútmaji tuku-tai=̃kɨ=ʃa tuku-tu-ma-ji shoot-NMLZ=RESTR=ADD shoot-APPL-REC.PT-3.PT+DECL ‘Shooting him, he shoot him.’ The combination of the restrictive plus the additive clitics usually indicates temporal or locational continuity (see §18.3). This semantic property (presumably time continuity) is likely a motivating factor for the use of =kɨ=ʃa in this construction, because the initial nominalized verb indicates what is to happen immediately afterwards in the main predication. The construction serves to intensify an action that has been recently introduced in the discourse. In fact, while it is not as commonas nominalized complement, relative and adverbial clauses, the “tautological in"nitive” construction often appears at key moments of the narratives to highlight an important action. The next examples illustrate this construction. Example (34) comes from a text that tells the story of how a group of Awajun, 293. This is interesting from a comparative perspective, as most “tautological” in"nitive constructions I am aware of have a "nite main verb. 867 who used to "ght intertribal wars with the Wampis wanting revenge for one of their dead relatives, come to kill the protagonist of the story. The passage in (34) narrates the moment where the Awajun come looking for the main character, laying siege to his house. Notice that third person objects are not marked on the verb, and the second verb (semantically main verb) is formally nominalized. (34) ʃuaŕ túkɨ awarún aíɲa ɨɰaḱaru tiḿaji matsat́un nɨkaẃaru nútika nekás naan kaʃi tariar ɨpɨńtaik̃ʃa ɨpɨɲ́aru hɨɨ ̃ń 34.a ʃuara tukɨ awaruna a-ina ɨa-ka-ara-u ti-ma-ji person INTERJ Awajun COP-PL.IPFV look.for-INTENS-PL.PFV-NMLZ say-REC.PT-3.PT+DECL ‘Those persons...the Awajun looked [for him], 34.b matsa-tu-u=na nɨka-aú-ara-u nutika-a nɨkas naa=na live-APPL-NMLZ=ACC know-HIAF-PL.PFV-NMLZ do.that-HIAF truly HESIT.PRO=ACC ‘they found out that he was one who lived; doing that, truly,’ 34.c ɨpɨna-tai=̃kɨ=ʃa ɨpɨna-ara-u hɨɨ̃=̃ni ̃ fence.in-nmlz=restr=add fence.in-pl.pfv-nmlz house\1pl/3=loc ‘as if294 shutting him in, they shut him in in his house.’ 294. The translation given in Spanish for this construction is como si... ‘as if...’ or casi como... ‘almost like [doing verb...]’. 868 The next example comes from a text in which a man finds a bear’s cub and tries to take it home to domesticate it. In the passage shown in (35), the bear’s mother catches them up and starts beating the man. (35) awa ́awat́kawa kaḿɨ maã/́ maãt́aik̃iʃa maã ́[...] awa awata-kawa ̃ kamɨ ma-̃a ̃́ redup hit+IPFV-REDUP\3SG.SS INTERJ kill-HIAF\3SG.SS [ma-̃a-́tai=̃kɨ=ʃa ma-̃a ̃]́ kill-HIAF-NMLZ=RESTR=ADD kill-HIAF\3SG.SS ‘hitting and hitting, she [the bear] killed him, as if killing him, she killed him [...]’ 869 CHAPTER XXI THIS GRAMMAR IN BROADER CONTEXT This grammar has provided detailed documentation and analysis of the structures of Wampis, as well as of the functions those structures serve. The main goal has been to provide an accurate description of the language, covering sound patterns, prosody, morphological processes, word classes, simplex and complex constructions, and selected notes on discourse. I have provided hypotheses to try to understand the synchronic phenomena found in the Wampis language, either by relating them to processes of grammaticalization or to greater cognitive pressures, as well as by relating my "ndings to typological claims in the broader literature. In addition, the present work has also provided basic ethnohistorical and ethnographic notes about the Wampis people. Any language that is spoken is in constant movement, i.e., it is shaped by the constant activity of its speakers. A grammar is never "nished. I have no illusions of having captured everything that can be found in a language like Wampis, or that the hypotheses presented in this work will remain unchallenged. Much investigation remains to be done for Wampis, especially in the "elds of prosody (including 870 nasalization), discourse structure, comparative reconstruction (with other Jivaroan languages), and language contact. For the most part, this grammar has been written employing linguist’s jargon and in a speci"c (academic) writing tradition. It is thus my hope that this grammar sparks discussion amongst specialists looking at patterns of the Wampis language, as well as at how those patterns may inform current theories of language, regardless of the theoretical perspective one takes. As we approach the end of this work, I would like to dedicate a few words to discuss, if brie/y, one "nal topic: the impact of this work for the greater scienti"c community, as well as for interested members of the public, including the community of Wampis speakers. I have tried to be careful in presenting de"nitions as clearly as possible to make the grammar more accessible to non-linguists. Some parts of this work may constitute a "rst introduction to the Wampis world for scholars from other "elds, for instance, anthropologists, ethnographers, educators or historians. While this is a grammar written by someone who views language through analytical tools provided by linguistic theories, I would be happy if part of the knowledge transmitted here, which has been 871 shared with me by my Wampis teachers and collaborators, provokes the reader to become interested in learning more about the Wampis language and the Wampis people. The Introduction in Chapter I stated that the Wampis are very conscious of their language and other cultural aspects that come with language. They want to do things with their language, and they have expressed needs related to language such as materials for intercultural education, vocabulary, development of jargon for speci"c "elds, and translation of Peruvian laws and policies into Wampis for future consultation of new legislation with the Wampis population, especially in regard to land prospection and land rights. One of the issues of writing this dissertation in English is a basic problem of accessibility: many Wampis and Peruvians will have to work harder to read and understand this grammar, as English is not their "rst language—that is very unfair. One "rst step to remedy this problem is to make the data collected for the research presented here available to the public, especially to the Wampis. A second step is to try to make a bilingual pedagogical grammar and a dictionary (Wampis/Spanish), taking the present work as a basis to achieve those goals. Such products will help produce materials for education as well as help in translation of laws, for instance. In general, 872 the ultimate goal is to facilitate that the Wampis people themselves be able to use these recourses as they see "t, and to facilitate resource development with the ideal goal that, at some point in the future, the Wampis people can produce materials of diverse types in their own language. That is a di#cult goal to achieve, and perhaps it will be some time before it can be feasible, but it is not a crazy idea. Though endangered, the Wampis language is still being transmitted to future generations, and younger generations are able to write in Wampis. Though writing has not been standardized, there are interesting writing practices using Wampis, both inside the community—many Wampis write poems and songs—and outside the villages—such as using technologies (chat and other social media) to communicate over distances, for instance. But there are still many complex challenges. Consider the following example: Just three weeks before I submitted my dissertation to my committee, I received an email from a Wampis friend who is now in Lima. She has received higher education at university level in Lima. She is one of the most intelligent and caring people I have met, and is doing some extraordinary things; for instance, she is writing literature in Wampis. In Peru, especially in the area of the Amazon, that is truly unique. I translate speci"c parts of her message with her permission (the email was originally written in Spanish; I have 873 anonymized personal references and the reason for underlining certain portions of the translation will be explained shortly): I write to you hoping that you can read this message. I really do not have many linguists’ contacts, especially those ones who study Wampis, so every time I have di#culties to understand my own language, strange as it may seem, I have hopes of being able to read your work . . . Not long ago I had a meeting with someone who works in DIGEIBIRA at the Ministry of Education295 and he recommended that "rst I must learn Wampis if I want to write. It really left me thinking. I have many doubts that are not resolved . . . To give you an example, according to the special- ist at DIGEIBIRA, the accent marks that I have used in my short stories should not be there. My short stories are now without accent marks, the worst part is that they do not have a good argument [to tell me why I should not use accent marks], as for me, now I cannot explain why I put accent marks . . .” Some of the passages of this email, especially those that are underlined, reveal several complex issues (some of them would probably need another dissertation to explain!), so I o$er here just brief comments. On one hand, this email reveals how important language is for the Wampis: my friend is truly concerned about whether she is representing her own native language, Wampis, in a fair way. On the other hand, my friend has a concern about being able to access this present work, but more importantly, whether she will be able to read and understand it. As I said, I am conscious that having written this grammar in English and in a speci"c academic 295. DIGEIBIRA=Dirección General de Educación Intercultural, Bilingüe y Rural (General O#ce of Intercultural, Bilingual and Rural Education). 874 tradition may be unfair to those most interested in reading it but who do not read English; so it is my task to make this work more accessible to non-English readers, specially to the Wampis. My friend’s message also reveals some pre-conceptions of people about language. That one person suggests to another person that she has to learn her native language (which she already orally speaks, and writes) is striking, and sad, to me. In the best case scenario, the person at the Ministry of Education really wants to help but has no clue about how to do it (note also that he may be confusing oral with written language). The ideas 1) that one has to (re-)learn their own native language which they already speak (in the case of my friend, Wampis); 2) that someone discourages the use of my friend’s native language to express herself (in the case of my friend, via writing literature in Wampis) left me very perplexed. There is much work to do in the "eld of intercultural education in Peru. The Wampis are no exotic people. I consider it to be demagogic and very un- critical to treat them as victims or “subaltern subjects”. It is true that, through colonial and independent times, the Wampis have su$ered many abuses, but they have defended themselves too. Unlike many other native groups in the Americas, the Wampis culture 875 was never suppressed by the dominant colonial power, or by republican powers in independent times—that combative spirit and sense of freedom has likely prevented the extinction or total assimilation not only of the Wampis, but of the Jivaroan peoples in general. The Wampis identity is centuries old, and dates back to pre-national Peruvian times. Just like their language, the Wampis consider themselves subjects in a modern culture, the Wampis culture, within the greater Peruvian national society. The Wampis culture is not an immobile atavistic culture. It is very modern on its own way, and just like the language, the Wampis culture is in constant adaptation. I "nished Chapter II with a quote by one of my Wampis teachers: “we no longer defend ourselves with weapons, we defend ourselves with words”. It is perhaps no coincidence that one word for ‘to defend’ in Wampis is based on the word t͡ʃit͡ʃama ‘word, speech’.296 Language and communication are at the center of Wampis culture, and much of what I found in Wampis discourse about how they conceive of themselves and their problems in current times centers around being able to communicate their needs to others. As Óscar Jimpikit, secretary of the Community of Candungos, puts it:297 296. The reader will "nd several instances of this word written as a stem in the examples below. 297. In the following examples, my focus lies beyond grammatical explanations, so I just provide 876 yatsuru tajame wait anesam chichamkartukta wari uun sunaiyachkursha chichamjai papijai ‘my brother, I tell you, please support us, we are not asking for a big thing (i.e. money or other material things), but [help us express our needs] with words, with documents’ ameja atupkratkata tusa chicham etserkata iina yaaktari uunta jui ‘saying help us, inform with words here in our big city (i.e. Lima, capital of Peru)’ nunka Peru tamau nankata jui ‘here at the end of what is called Peruvian land’ (i.e. inform others of the situation here where Peru ends=the Wampis territory)’298 The Wampis people, just like any other group, want to be respected and be able to negotiate policies and laws that may a$ect them. They do not want someone else to take decisions in their name. I asked a Wampis leader, Juan Luis Nuningo, Apu (local Chief) of the Community of Puerto Galilea, to record a text about current problems of the Wampis, what they would like for future generations from his perspective. He immediately construed a very interesting and wise speech, in which he considered a good person (from a Wampis perspective) to be someone who knows the Wampis language and culture as well as the national, mestizo culture (including Spanish, the national language of Peru)—that is, the Wampis do not consider themselves an isolated direct translations of the examples (and transcribe using the Wampis alphabet, not IPA symbols). 298. The Wampis territory is located in the border region with Ecuador. 877 group, but they really want to be part of a national group while maintaining their own identity: ii wampisti unuimakrika kame iiña pujuti kajinmatkimuka amaichuwaiti turamtai ‘we the Wampis, if we learned (i.e. if we receive education), we should not forget our own culture’ apatka mai apatka unuimamaiñaiti apach pujut apach chichama nuu ‘together both we must learn the mestizo culture, the mestizo language [and the Wampis culture/language]’ maa untsu nuniaku tumainaitji maa wiyaitjai wampisan ‘then, doing that, we can say "I am Wampis"’ kame unuimarunka tumaiñaitji turamtai antsuk iiña pujuti nekachu ‘we can call [the person who has been educated] “one who has learned”, and without doing that, then she/he doesn't know our culture’ kame aya apachi pujutiña nuke nekakrika ‘if we only learn the mestizo culture’ apachnumani yunutkauwa anin wantiniaji nuka kame ‘it seems like we have included ourselves in the mestizo['s society]’ (i.e. ‘it seems like we have been assimilated to the national culture’) nunia nuka ayatek menká ‘having done that, that one is lost’ Juan Luis Nuningo considers access to education, respect and communication at the center of Wampis e$orts for future generations: indigenas timautisha chichamrumakur weakur arantukmau amainatji ‘we the so-called indigenous people, if we go around defending ourselves we can be respected’ 878 unuimarti tusa wakeraj nuka wakerajai uchi ‘we want that our children learn (=receive education)’ imainsha nekas ekemas niisha ma ‘far over there too [=in Lima], truly, sitting as authorities’299 tuke ma uun matsatea imai ‘far there where the big ones (=authorities) are’ . . . nui chichak kame unuimara asa nui chichaiñakai nekas nuka nuni ‘there, speaking, having learned, when they speak, truly, in that way’ yachamaku chichaiñakai iincha aranturmaktiñaiti gobiernoka ‘when those [of us] who have learned speak, the government can respect us too’ To conclude, consider the following words also from Juan Luis Nuningo, Apu of Puerto Galilea: chicham weawai jutika iiña nunke surimkami iña nunke ayamrukmi ‘the word goes around like that, let us take care of our land, let us defend our land’ turamtai iiña pujutiña nu nekas kame kakaram asa iisha kame arantukmau atai ‘thus happening, because we truly strengthen our lives, we too are respected,’ iruntra takakmakur ‘while we work together,’ 299. The word (/ɨkɨma/) means ‘sit’, but in this context it means ‘sit as authority’. In Wampsis culture, the respected adults and elders have a seat (/t͡ʃimpuiì/), which represents their authority, where they sit to discuss di$erent matters with other people. 879 kame matsamsami nutikakur iisha tuke kishmataiksha kishmakar matsamsami kame iiña waitkarmaiña nuu ‘let us live as if we shoo away, let us live shooing away those who bother us’ wishikramu achami teperkamu achami ‘let us not be laughed at, let us not be dominated’ kame nuu chicham weawai kame nuwaiti ii anentaimsar ‘that word goes around, that is what we think’ unuimararti nekas uchi tusa wakeraj nuka ‘saying, truly, “let the children get educated”, that is what we want’ nuni ayamruniakar nuni chichamruniakar kame ii iruntrar matsamsar kame iisha kakaram asarmi tusar ‘having done that, having defended ourselves, having done that, having defended ourselves with words, we, living altogether, let us be strong’ uchika chicharnawai kame nuwaiti juti anentainka ‘we advise our children, that is our way of thinking.’ It is my hope that this work contributes not only to the knowledge of the Wampis language, but also to a better understanding of the Wampis culture—of which language is a central part. Moreover, it is my hope that the present work and its derived future works contribute to a more profound knowledge of the speakers, and facilitates a fair intercultural communication (cf. the allusions above to talk with the government, or to inform others in the “big city” (Lima) of the problems the Wampis’ are facing, etc.) 880 This communication must start with the realization that the Wampis are not a static and subjugated people, but very active people with their own voice—people who, by the way, speak a very beautiful and very sophisticated language. 881 APPENDIX A ABBREVIATIONS In general, I have used abbreviations which follow the Leipzig Glossing Rules (2008). An equal symbol “=” separates an enclitic from a stem. I follow Overall (2007) in using a plus symbol “+” for phonologically identi"able markers in “fused” forms. An “\” symbol means that the morpheme received by the stem consists of a vowel switching process (for marking possession) or that the morpheme is autosegmental (e.g. high tone or nasalization). 1 ‘First person’ 2 ‘Second person’ 3 ‘Third person’ ACC ‘Accusative’ ADD ‘Additive’ ADJZ ‘Adjectivalizer’ AFF ‘A$ective’ APPR ‘Apprehensive’ APPL ‘Applicative’ ATT ‘Attenuative’ ATTRIB ‘Attributive’ AUG ‘Augmentative’ COM ‘Comitative’ CON ‘Conditional’ 882 CONCESS ‘Concessive’ COP ‘Copula’ DEF ‘De"nite’ DIST ‘Distal/Distant’ DISTR ‘Distributed action aktionsart’ DS ‘Di$erent subject’ DTR ‘Detransitivizer’ EXCLAM ‘Exclamative’ FOC ‘Focus’ FRUST ‘Frustrative’ GEN ‘Genitive’ HESIT ‘Hesitation’ HIAF ‘High a$ectedness aktionsart’ HORT ‘Hortative’ IDEO 'Ideophone' IMM ‘Immediate’ INFER ‘Inferential’ INT ‘Interrogative’ INTERJ ‘Interjection’ INTENS ‘Intensive aktionsart’ INTs ‘Intensi"er’ IPFV ‘Imperfective’ LOAF ‘Low a$ectedness aktionsart’ LOC ‘Locative’ MIR ‘Mirative’ NARR ‘Narrative’ 883 NMLZ ‘Nominalizer’ NON.VIS ‘Non-visible’ OBJ ‘Object’ PL ‘Plural’ PRES ‘Present’ PRO ‘Pronoun’ PROH ‘Prohibitive’ PT ‘Past’ PURP ‘Purpose’ Q ‘Question marker’ REC ‘Recent’ RECP ‘Reciprocal’ REDUP ‘Reduplicative’ REFL ‘Re/exive’ RESTR ‘Restrictive’ RHET ‘Rhetorical question’ SG ‘Singular’ SIM ‘Simultaneous’ SIMIL 'Similative' SPEC ‘Speculative’ SS ‘Same subject’ SUD.REALZ ‘Sudden realization’ TAG ‘Tag question’ TR ‘Transitivizer’ VBZ ‘Verbalizer’ 884 APPENDIX B SELECTED TEXTS Text 1: Pear !lm story This is a Pear Film story (Chafe 1980). The speaker was asked to watch the "lm and then relate it to another speaker. Speaker: Atilio Nuningo; Age: 30 Community of origin: Puerto Galilea Date: September 2014, Duration: 2 min. 02 sec. (1) húu nankaḿas ahúka hu nankama-sa a=hu=ka PROX happen-ATT PROX=FOC ‘This is what happened’ (2) tikit́͡ʃik ʃuaŕ ahari ̃ń puhús jurankɨń ara ́huwaẃɛi tikit ͡ʃiki ʃuara aha-ri=̃ni ̃ puhu-sa ̃ one person farm-1PL/2PL/3.POSS=LOC live/be-SUB\3SG.SS iurankɨ=na ara ̃ ́ huu-a-ua-i fruit=ACC plant+HIAF\3SG.SS gather-IPFV-3.SBJ-DECL ‘One person, being in his farm, having planted fruits, was gathering [them]’ (3) túman jamɛíkikʲa tikit́͡ʃik úun kuntińtri ̃taḱakũ tuma ̃ iamai=ki=ka tikit ͡ʃiki uunta happen.thus\3.SS now=RESTR=FOC one elder 885 kuntina-tu-ri ̃ taḱa-kũ animal-COM-1PL/2PL/3.POSS have+IPFV-SIM\3SG.SS ‘Thus, at the beginning one elder that had his animal’ (4) naŋkamaḱmatɛɛ ̃ija ́ijaḱua iɲańkaki nankama-ka-matai ̃ iia ii-a-kaua ̃ inanka-ki ̃ happen-INTENS-1SG/3.DS REDUP see-IPFV-REDUP\3SG.SS CAUS-pass-WHILE.MOVING\3SG.SS ‘when that happened, looking and looking, having passed’ (5) núna iiś naŋkańɨak nu=na ii-sa ̃ nankanɨ-a-kũ NON.VIS=ACC see-SUB\3SG.SS end.IPFV-IPFV-SIM\3SG.SS ‘having "nished seeing that’ (6) ataḱʃa jaki ́warúk atakʃa iaki ́ ua-ru-ka ̃ again above\LOC climb.up-APPL-INTENS\3SG.SS ‘again having climbed [on the tree]’ (7) jurańkɨn huúk t͡ʃankińtrin ɛiḿʲak puhóon tikit́͡ʃik út͡ʃi tari ́ iurankɨ=na huu-ka ̃ fruit=ACC gather-INTENS\3SG.SS t͡ʃankina-ri=ni ̃ aima-kũ puha-ũ=na basket-1PL/2PL/3.POSS=LOC "ll+IPFV-SIM\3SG.SS live/be+IPFV-NMLZ=ACC tikit ͡ʃiki ut͡ʃi ta-ru-i ̃ one child arrive-APPL-3.PFV+DECL ‘Having gathered fruit, to that one "lling his basket (i.e. the old man) one child arrived’ (8) tikit́͡ʃik t͡ʃankiń ʔɛiḿkʲamuanúna hurukni ̃ ́ tikit ͡ʃiki t͡ʃankina aima-ka-mau=a=nú=na one basket "ll.up-INTENS-NMLZ=COP=NON.VIS=ACC 886 hu-ru-kini ̃ take-APPL-WHILE.MOVING+3.PFV ‘He (i.e. the child) took one basket that was "lled’ (9) nútikʲa núu út͡ʃi huruki ́wɨsatahkama ̃ ́ nutika ̃ nu ut͡ʃi hu-ru-ki ̃ do.that+HIAF\3SG.SS NON.VIS child take-APPL-WHILE.MOVING\3SG.SS wɨ-sa-tahkama ̃́ go-ATT-FRUSTR\3SG.SS ‘Having done that, that child when he wanted to go having taken [the basket]’ (10 tikit́͡ʃ kumpari ́miɲóon tikit ͡ʃi kumpa-ri ̃ uina-u=na other friend-1PL/2PL/3.POSS come+IPFV-NMLZ=ACC ‘to his other friend that was coming’ (11) miɲóohɛɛ ̃tukúniawar núu huki ́wɨaḿuanuna ukaŕ uina-u-hai ̃ tukuni-au-ara nu COME+IPFV-NMLZ-COM collide-HIAF-3PL.SS NON.VIS hu-ki ̃ uɨ-́a-mau=a=nu=na uka-ra ̃ take-WHILE.MOVING\3SG.SS go-IPFV-NON.SBJ.NMLZ=cop=ana=acc spill-DISTR\3SG.SS ‘the one who was coming collided with the one who was going having taken [the bas- ket], and [the "rst child] spilled [the content]’300 (12) ukaŕmatɛɛ ̃kumpari ́ɛɛ́ɲa núu jɛi ̃ńkʲarɛɛ uka-ra-matai ̃ kumpa-ri ̃ a-ina nu spill-DISTR-1SG/3.DS FRIEND-1PL/2PL/3.POSS COP-PL.IPFV NON.VIS 300. The verb 'collide' marks its object with the accusative =na, but since the action is reciprocal, the other participant receives the comitative =hai.̃ 887 iai-̃ka-ara-i help-INTENS-PL-3.PFV+DECL ‘when he spilled it, his friends helped him’ (13) huúk t͡ʃumpiámun nútikʲa ̃kumparih́ɛɛ ̃inkunirar huu-ka ̃ t͡ʃumpi-a-mau=na nútika ̃ gather-INTENS\3SG.SS put.in-IPFV-NON.SBJ.NMLZ=ACC do.that+HIAF\3SG.SS kumpa-ri-̃hai ̃ inku-nai-ra-ara friend-1PL/2PL/3.POSS-COM meet-RECP-DISTR-3PL.SS ‘having gathered what he put in (i.e. the fruits in the basket), having done that, having met with their friends’301 (14) (núka út͡ʃikʲa ...)302 núu ukanaŕu anúna húkiar juaẃar kumpamnɛíja ʃiaḱarɛɛ (nu=ka ut͡ʃi=ka) nu uka-na-ra-u a=nu=na NON.VIS=FOC child=FOC NON.VIS spill-DTR-DISTR-NMLZ COP=NON.VIS=ACC hu-ki-ara iu-aú-ara kumpa-ma-naía-a ̃-́ara take-WHILE.MOVING-3PL.SS eat-HIAF-3PL.SS friend-VBZ-INCHO-RECP-HIAF-3PL.SS ʃia-ka-ara-i leave.PL-INTENS-PL-3.PFV+DECL ‘having taken what that was spilled, having eaten, having become friends, they left.’ (15) túma úun núu jurańkɨn huwaẃa nuka tuma uunta nu iurankɨ=na huu-a-u=a nu=ka thus elder NON.VIS fruit=ACC gather-IPFV-NMLZ=COP NON.VIS=FOC ‘Thus, that elder who was gathering the fruit’ 301. This passage is a little confusing, because the verb is in plural, suggesting that the action is referring to both children and not the (single) protagonist child. The same occurs in the next sentence. 302. The speaker hesitates at this point, and prefers to re-start with núu ukańaru... . 888 (16) aiḿkamapha túsa ijahkama ́tikit́͡ʃik t͡ʃankiń atsóon wɛiɲ́ak aima-ka-ma-api-ha tu-sa ̃ ii-a-hkama ̃́ "ll.up-INTENS-REC.PT-SUD.REAL-1SG.SBJ+EXCL say-SUB\3SG.SS see-IPFV-TERM\3.SG.SS tikit ͡ʃiki t͡ʃankina a-tsa-u=na uaina-ka one basket exist-NEG-NMLZ=ACC see-INTENS ‘‘upon thinking,303 saying: “I "lled it (i.e. the basket) up!”, having seen that there was one basket missing’ (17) ataḱʃa út͡ʃi wakɨt́ki miniɲ́an iís atakʃa ut͡ʃi wakɨ-tu-ki uini-ina-na ii-sa ̃ again child walk-APPL-WHILE.MOVING come-PL.IPFV-NMLZ see-SUB\3SG.SS ‘Again, seeing that the children were coming walking’ (19) nitʲat́͡suk kasaḿturkaruiti túsa túu ɨnɨńtɛimʲar naŋkaŋ́kowɛitʲi nita=t͡su=kɨ kasama-tu-ru-ka-ara-u=iti tu-sa ̃ 3PL=INFER=RESTR steal-APPL-1SG.OBJ-INTENS-PL-NMLR=COP.3+DECL say-SUB\3SG.SS tu ɨnɨntai-̃ma-ra ̃ nankana-ka-u=aiti thus heart-VBZ-DISTR\3SG.SS end-INTENS-NMLZ=COP.3+DECL ‘having thought thus, saying: “they must have robbed me!”, he was one to "nish.’ (20) húka naŋkaḿasuiti hu=ka nankama-sa-u=iti PROX=FOC happen-ATT-NMLZ=COP.3+DECL ‘This is what happened’ 303. The verb ii ‘see’ is translated here as ‘think’ by the speaker. It is frequent in Wampis to use the verb ‘see’ with a subordinated verb ‘say’ to express what a person thinks, so the construction literally translates as “seeing, saying “...”’, but it means “thinking “...” ”. 889 (21) tikit́͡ʃik ahańam úun ni ̃ŋ́ki puhaḿunam tikit ͡ʃiki aha=nama uunta ni=̃kɨ puha-mau=nama one farm=LOC elder 3SG=RESTR live-NMLZ=LOC ‘in a farm where an elder, he only, lived’ (22) patɛíñkiʃa wɛiɲ́akt ͡ʃaji patai=̃na=kɨ=ʃa uaina-ka-t ͡ʃa-ji relative\1PL/2PL/3.POSS=ACC=RESTR=ADD see-INTENS-NEG-3.PT+DECL ‘his relative(s) has(have) not been seen’ (23) aj́atɨk úun ni ̃ŋ́ki ahari ̃ń takaŕmas aíatɨkɨ uunta ni=̃kɨ aha-ri=̃ni ̃ taka-ru-ma-sa ̃ only elder 3.SG=RESTR farm-1PL/2PL/3.POSS=LOC work-APPL-REFL-SUB\3.SS ‘only one elder, he alone, working in his farm.’ (24) núna núnikuiti nu=na nuni-ka-u=iti NON.VIS=ACC do.that-INTENS-NMLZ=COP.3+DECL ‘That is what he did.’ 890 Text 2: A woman encounters a Tijai in the forest This is a narrative that tells the story of a woman that encounters a Tijai in the forest as she goes to hunt. Tijai are human-like people that inhabit the forest in the Wampis folklore. They are said to be tall and have uneven feet (usually they are described as having one foot shorter than the other). The Tijai are consider “people of the forest” by the Wampis. The protagonist is a woman, Inchis. The story is described as being part of “present history”, it did not happen in the mythological past. The speaker, Óscar Jimpikit, says that the protagonist Inchis was a real person who was known by his father. Speaker: Óscar Jimpikit; Age: 42 Community of origin: Candungos Date: October 2013; Duration: 3 min. 30 sec. (1) tikit́͡ʃin óuhmatsathɛɛ tihɛɛ́ pat͡ʃiśan tikit ͡ʃi=na aúhumatu-sa-ta-ha-i tihai pat͡ʃi-sa-nu one=ACC tell-ATT-IMM.FUT-1SG.SBJ-DECL Tijai mention-SUB-1SG.SS ‘I am going to tell one, mentioning the Tijai.’ (2) miɲ́a papaŕ uhaḱmia núna uhaḱtathamɨ haiḿito jatsurú mina papa-ru uha-ka-mia nu=na 1SG.GEN father-1SG inform-INTENS-DIST.PT NON.VIS=ACC 891 uha-ka-tata-hamɨ Jaimito iatsu-rú inform-INTENS-DEF.FUT-1SG>2SG+DECL Jaimito brother-1SG\VOC ‘My father informed this, I am going to tell you, Jaimito, my brother.’ (3) núka jamɛíjɛiti jamɛija hui ̃joúnt͡ʃukiant ͡ʃu iḿatikʲa ̃ nu=ka iamai=aiti iamai=a hui ̃ NON.VIS=FOC now=COP.3+DECL now=COP here iaunt͡ʃukɨ=a=nu-t͡ʃau iḿatika ̃ long.ago=COP=NON.VIS=NEG.NMLZ do.much\3SG.SS ‘That is [from] now, it is here now, it is not from so much long ago’ (4) jamɛíhũ mat͡samki ́wiɲah́ hui ̃ ́ iamai=hũ mat͡sama-ki uina-hi hui ̃ now=PROX live.together=WHILE.MOVING come+IPFV-1PL.SBJ here ‘right now here where we are going on living.’ (5) nuwa ́nuwɛítʲak uumi ́ampúntin tiḿaji nua nua=ita-kũ uumi ́ ampu-tu-inu timaji woman woman=COP.3+IPFV-SIM\3SG.SS blowgun\ACC blow-APPL-NMLZ NARR ‘The woman, being a woman, she used the blowgun.’304 (6) ińt͡ʃis naaŕtin int͡ʃisu naa-ri-̃tinu Inchis name-1PL/2PL/3.POSS-ATTRIB ‘her name [was] Inchis’ 304. In the Wampis culture, men go to hunt (hunting is done with the blowgun), so the speaker is explaining that Inchis, the protagonist, had the characteristic of mastering the blowgun. Note the use of the simultaneous construction as a concessive—a more free translation of the sentence could be: ‘the woman, [though] being a woman, knew how to use the blowgun’. 892 (7) naaŕiŋ̃ka núka núwaka ińt͡ʃis naa-ri=̃ka nu=ka nua=ka int͡ʃisu name=1PL/2PL/3.POSS=FOC NON.VIS=FOC woman=FOC Inchis ‘The name of that woman [was] Inchis.’ (8) miɲ́a papaŕuka nukúr ińt͡ʃis tiḿaji núnaka mina papa-ru=ka nuku-ru int͡ʃisu ti-ma-ji nu=na=ka 1SG.GEN father-1SG=FOC mother-1SG Inchis say+LOAF-REC.PT-3.PT+DECL NON.VIS=ACC=FOC ‘My father called her “my mother Inchis”.’ (9) nútikʲamu núu nukuchúr305 ińt͡ʃisuanu t͡sɨasań pɛiŋ́ nutika-mau nu nukut͡ʃi-ru int͡ʃisu=a=nu do.that-NMLZ NON.VIS grandmother-1SG Inchis=COP=NON.VIS t͡sɨasa=na painta ̃ curare=ACC apply.poison.PFV\3SG.SS ‘Thus, that my grandmother who is Inchis, having applied curare [to darts],’ (10) mitʲaj́u wɨka ́hakú tiḿaji jóunt͡ʃukʃa mitaiu uɨka hak-u timaji iaunt͡ʃukɨ=ʃa hunting walk HAB.PT-NMLZ NARR long.ago=ADD ‘she was one to go hunting long ago too.’ (11) nútikʲa tukɨ ́wɨkɛín tukú júu nutika ̃ tukɨ ́ wɨka-inu tuku-u iu-u do.that.PFV\3SG.SS always walk-NMLZ shoot-NMLZ eat-NMLZ ‘Having done that, she always walked, she was one to hunt and eat.’ (Lit: ‘always a walker, shooter, eater’) (12) pakińt͡ʃa tukú aʃi ́ma ́unuiḿaru ampúntin aśa paki=na=ʃa tuku-u aʃi ma peccary=ACC=ADD shoot-NMLZ all INTERJ 305. There is a metathesis in the surface form of this word. 893 unuima-ra-u ampu-tu-inu asa ̃ learn-DISTR-NMLZ blow-APPL-NMLZ COP-SUB\3SG.SS ‘[she] shot peccaries too, all [i.e. other animals that she found], she had learned, be- cause she knew how to blow’ (Lit.: ‘[she was] one who learned, being that she was a blower’.) (13) mitʲaj́u wɨt́asa ̃t͡sɨasa ́pɛiŋ́kʲu aśa ̃ mitaiu uɨ-tasa ̃ t͡sɨasa ́ painta-ka-u a-sa ̃ hunting go-PURP curare\ACC apply.poison-INTENS-NMLZ COP-SUB\3SG.SS ‘To go hunting, because she was one to prepare [the darts with] curare,’ (14) kaʃ́ik hiiŋki ́wɨú tiḿaji uumi ́huki ̃ ́ kaʃiki hiina-ki ̃ uɨ-u timaji early.morning go.out-WHILE.MOVING\3SG.SS go-NMLZ NARR uumi ́ hu-ki ̃ blowgun\ACC take-WHILE.MOVING\3SG.SS ‘early morning, having gone out, she went, having taken the blowgun.’ (15) nútikʲa wɨɨ́ ̃iḿʲɛɛ mitaju wɨkat́ɛɛr̃i ̃hɨa ́ nutika ̃ wɨ ̃ do.that.PFV\3SG.SS go.PFV\3SG.SS iḿai mitaiu wɨka-tai-̃ri ̃ hɨã ̃́ far.over.there hunting walk-NMLZ-1PL/2PL/3.POSS arrive+HIAF\3SG.SS ‘Thus, having gone, [she] having arrived where she walked to hunt, far over there’ (Lit.: having arrived to her walking to hunt’) (16) maʃ́u t͡ʃit͡ʃa ́mat͡sat́u wɛíŋkʲu tiḿaji maʃu t͡ʃit͡ʃa mat͡satu-u uaina-ka-u timaji curaSSow speak+IPFV live.together-NMLZ see-INTENS-NMLZ NARR ‘she found curaSSow that were singing.’ 894 (17) piʃ́ piʃ́ hɛiɲ́a tiḿaji maʃ́u nútikʲak piʃ́ piʃ́ ha-ina timaji maʃu nutika-kũ IDEO IDEO speak.animal-PL.IPFV NARR curassow do.that+IPFV-SIM\3SG.SS ‘“pish pish” the curaSSow were saying, while she was doing that (i.e. observing),’ (18) miɲ́aka wɛit͡ʃitʲatmapi tú puhaẃa túsa ija ́tiḿaji mi=na=ka uai-t ͡ʃi-tata-ma-api tu puha-ua 1SG=ACC=FOC see-1SG.OBJ-NEG-SUDD.REALZ like.that live/be+IPFV-3.SBJ tu-sa ̃ ii-a ̃ ́ timaji say-SUB\3SG.SS see-HIAF\3SG.SS NARR ‘she thought, saying: “it is not seeing me! It is there like that”.’ (19) nɛińɲum haś iḿʲɛɛ hapóutanum hɨãri ̃ ́iḿani ija ́tiḿaji nainta=numa has iḿai hapauta=numa hill=LOC become far.over.there slope=LOC hɨã-ru-i ̃ iḿani ii-a timaji arrive-APPL-LOAF\3SG.SS INTENS.DEM.ADV see-IPFV NARR ‘In the hill, far over there in the slope, having arrived, she was observing very carefully.’ (20) nui ̃ ́tihɛɛ́ʃa wahaś ií wahat́u tiḿaji núna maʃ́u tuiɲ́akɛɛ ̃ nui ̃ tihai=ʃa uaha-sa ̃ ii uaha-tu-u timaji nu=na there Tijai=ADD stand-SUB\3SG.SS see stand-APPL-NMLZ NARR NON.VIS=ACC maʃu tu-ina-ka-i ̃ curaSSow say-PL.IPFV-SIM-DS ‘A Tijai also standing there kept looking at those (i.e. the curaSSow), while the curaSSow were singing.’ (21) nútikamtɛɛ ̃ʃuaraṕitʲa túsa nutika-matai ̃ ʃuara=api=ita tu-sa ̃ do.that-1SG.3.DS person-=SUD.REALZ=COP.3+DECL say-SUB\3SG.SS “Having done that, saying: “it is a person!” 895 (22) núwaka intʲaʃ́iŋ̃kʲa hútikʲa ̃tɨntɨa ́ɨt͡sɨḿa nahańa wɨka ́aśa nua=ka intaʃi=̃na=ka hutika ̃ tɨntɨa ́ woman=FOC hair\1PL/2PL/3.PoSS=ACC=FOC do.this.PFV/3SG.SS make.circle-HIAF\3SG.SS ɨt͡sɨma nahana wɨka ́ a-sa ̃ tie.up.hair make wɨka+IPFV COP-SUB\3SG.SS ‘the woman, because she walked having made her hair in a circular form,’ (23) núnaka kuiki ̃ ́awaś nu=na=ka kui-ki ̃ aua-sa ̃ NON.VIS=ACC=FOC unfasten-WHILE.MOVING\3SG.SS untie.hair-SUB\3SG.SS ‘having unfastened it (i.e. her hair) untying it’306 (24) nuiɲ̃a ́tarat́͡ʃriñt͡ʃa huni ́itiṕruma wɨka ́aśa ̃ nui=̃ia tarat ͡ʃi-ri=̃ʃa there=ABL dress-1PL/2PL/3.POSS=ACC=ADD hu=ni itipiruma wɨka asa ̃ PROX=ALL wear.itipi+IPFV307 walk COP-SUB\3SG.SS ‘in addition, because she went around wearing her dress in this way’308 306. A cultural note for better comprehension of this part of the story: traditionally, men used to style a long hair that they would tie up. The woman of the story is portrayed as having several characteristics of men: she is able to hunt, fashions her hair like men and she uses her dress as an itipi [itiṕ]—a skirt-like garment only worn by men (see next line). 307. itipi = A skirt-like garment worn by men. 308. The speaker makes gestures signaling that the woman is wearing her dress as a skirt, hanging it from her waist. 896 (25) nunaʃ́a kuiki ̃ ́nuŋkurú tiḿaji nu=na=ʃa kui-ki ̃ ́ nunku-ra-u timaji NON.VIS=ACC=ADD unfasten-WHILE.MOVING\3SG.SS put.cloths.on-DISTR-NMLZ NARR ‘having unfastened that too (i.e. her dreSS that was tied up around her waist), she put it on.’ (26) núwa núnis wahaśu tiḿaji nua nuni-sa ̃ uaha-sa-u timaji woman do.that-SUB\3SG.SS stand-ATT-NMLZ NARR ‘The woman, doing that, she [remained] standing.’ (27) túra uúmriŋ̃kʲa atúsu tiḿaji tura uumi-ri=̃na=ka atu-sa-u timaji and blowgun-1PL/2PL/3.POSS=FOC lean.against-ATT-NMLZ NARR ‘and she left her blowgun leaning (i.e. against a trunk)’ (28) núna ɛíʃmaŋka tihɛɛ́ wɛiɲ́ak sapih́mak nu=na aiʃmanku tihai uaina-ka ̃ sapija-ma-ka ̃ NON.VIS=ACC man Tijai see-INTENS\3SG.SS fear-VBZ-INTENS\3SG.SS ‘The Tijai-man having seen her, having become nervous,’ (29) nútikʲa wahaśunak tihɛíka nɨkaṕrin aśa ̃nɨkapaŕ309 warukú tiḿaji nutika waha-sa-u=na=kɨ tihai=ka nɨkapɨ-ra-inu a-sa ̃ thus stand-ATT-NMLZ=ACC=RESTR Tijai=FOC feel-DISTR-NMLZ COP-SUB\3SG.SS nɨkapɨ-ra ̃ ua-ru-ka-u timaji feel-DISTR\3SG.SS climb-APPL-INTENS-NMLZ NARR ‘thus, to that (i.e. the woman) who was standing alone, the tijai, because he is one who feels, having sensed [her], he climbed toward her’ 309. Notice the metathesized pronunciation on the surface form. 897 (30) warúk iʃa ́iʃaḿak hɨa ̃ńtou tiḿaji ua-ru-ka ̃ iʃa ́ iʃama-kũ climb-APPL-INTENS\3SG.SS REDUP be.afraid+IPFV-SIM\3SG.SS hɨa-̃tu-a-u timaji arrive-APPL-IPFV-NMLZ NARR ‘Having climbed, while being afraid, he was one to approach her.’ (31) túra hɨa ̃ń jɛítʲam túsa ̃tóu tiḿaji núwa tura hia ̃ ia=ita-mɨ tu-sa ̃ta-u timaji nua and arrive.PFV\3SG.SS who=COP-2SG.SBJ say-SUB\3SG.SS NARR woman ‘And having arrived, the woman said “who are you?”’ (32) ińt͡ʃisua núu iɲiá tiḿaji int͡ʃisu=a nu inia timaji Inchis=COP NON.VIS ask+IPFV NARR ‘The one who was Inchis was asking.’ (33) wiḱʲa kampuiɲúnmaja ʃuaŕɛithʲɛɛ tóu tiḿaji ui=ka kampuinu=numa=ia ʃuara=aita-ha-i ta-u timaji 1SG=FOC forest=LOC=ABL person=COP-1SG.SBJ-DECL say+IPFV-NMLZ NARR ‘He said: “I am a man from the forest”.’ (34) nútikʲan kaḿɨ yɛítʲam túsa tóu tiḿaji nutika ̃ kamɨ ia=ita-mɨ tu-sa ̃ ta-u timaji do.that.3SG.SS INTERJ who=COP-2SG.SBJ say-SUB\3SG.SS say+IPFV-NMLZ NARR ‘Having done that, he said: “who are you?”.’ (35) ɨt͡sɨŕɨat͡ʃu wiḱʲa wij́ɛithɛɛ tóu tiḿaji ɨt͡sɨrɨ-a-t͡ʃau ui=ka ui=aita-ha-i ta-u timaji declare.IPFV-IPFV-NEG.NMLZ 1SG=FOC 1SG=COP-1SG.SBJ-DECL say+IPFV-NMLZ NARR ‘She was not one to say the truth, she said: “I am I”.’ 898 (36) túra iʃaḿrukɛipʲa túsa tóu tiḿaji tih́ɛɛ iʃaḿrukɛipʲa tura iʃama-ru-ka-ai-pa tu-sa ̃ ta-u timaji tihai and be.afraid-1SG.OBJ-INTENS-APPR-PROH say-SUB\3SG.SS say+IPFV-NMLZ NARR Tijai iʃama-ru-ka-ai-pa be.afraid-1SG.OBJ-INTENS-APPR-PROH ‘Then, the Tijai said “don’t be afraid of me!”, “don’t be afraid of me!”’ (37) wij́ɛɛ ̃mɨt́ɨkat͡ʃitʲam ui=hai ̃ mɨtɨka-t͡ʃi=ita-mɨ 1SG=COM equal-DIM=COP-2SG.SBJ+DECL “you are equal with me” (38) kuntúram itaḱia apat́ka iistah́mɨ kuntu-ramɨ ita-kia apatu-ka ii-sa-ta-́hamɨ arm-2SG ita-FAM.IMP put.together-INTENS see-ATT-IMM.FUT-1SG>2SG+DECL “bring your arm, I’ll see you comparing” (39) tusa ̃taḿa tu-sa ̃ ta-ma say-SUB\3SG.SS say+IPFV-NON.SBJ>SBJ ‘when he was saying that to her’ (40) nuwaʃ́a kuntúriñ hútikʲas iwɛińtus nua=ʃa kuntu-ri=̃na hutika-sa ̃ i-uana-tu-sa ̃ woman=ADD arm-1PL/2PL/3.poSS=ACC do.that-SUB\3SG.SS CAU-see-APPL-SUB\3SG.SS ‘the woman too, doing that, showed him her arm.’ (41) iiśu tiḿaji ii-sa-u timaji see-ATT-NMLZ NARR ‘He saw.’ 899 (42) nútikʲa iwɛińtus iiś nutika ̃ i-uaina-tu-sa ̃ ii-sa ̃ do.that.PFV\3SG.SS caus-see-APPL-SUB\3.SS see-ATT\3.SG.SS ‘Having done that, having seen,’ (43) wih́ɛi ̃mɨtɨḱɛitmɨ ui=hai ̃ mɨtɨka=ita-mɨ 1SG=COM equal=COP-2SG.SBJ+DECL “You are equal to me” (44) aḿɨka wií ʃuaŕɛitmɨ amɨ=ka ui ʃuara=ita-mɨ 2SG=FOC 2SG person=COP-2SG.SBJ+DECL “You are a person like me.” (45) tihɛɛ́ ʃuaŕɛitmɨ túsa tóu tiḿaji tijai ʃuara=ita-mɨ tu-sa ̃ ta-u timaji tijai person=COP-2SG.SBJ+DECL say-SUB\3SG.SS say+IPFV-NMLZ NARR ‘He said, saying: “you are a Tijai person”,’ (46) taḿa núwaʃa t͡ʃií túsa aj́atɨk kaḿɨ sapih́mak huaḱu tiḿaji ta-ma nua=ʃa t͡ʃii tu-sa ̃ say+IPFV-NON.SBJ>SBJ woman=ADD uh-huh say-SUB\3SG.SS aiatɨkɨ kamɨ sapihama-kũ hua-ka-u timaji only INTERJ be.afraid+IPFV-SIM\3SG.SS stay-INTENS-NMLZ NARR ‘when he was saying that to her, the woman, saying “uh-huh”, stayed afraid alone.’ (47) nútikak wiḱa hũú wɨáhɛɛ nutika-kũ ui=ka hũ uɨ-a-ha-i do.that+IPFV-SIM\3SG.SS 1SG=FOC PROX go-IPFV-1SG.SBJ-DECL ‘While doing that, “I am going this way”’ 900 (48) amɨʃ́a wɨkaśa wakɨtkitʲa ́tóo tiḿaji amɨ=ʃa uɨka-sa ̃ uakɨ-tu-ki-ta ́ ta-u timaji 2SG=ADD walk-ATT\3SG.SS return-APPL-WHILE.MOVING-IMP say+IPFV-NMLZ NARR ‘“You too having walked around, go back”, he said.’ (49) turaḱ núwaka sapih́maku aśa aińt͡san hɨɨ̃ ̃ń wakɨt́ki tura=kɨ nua=ka sapihama-ka-u a-sa ̃ and=RESTR woman=FOC be.afraid-INTENS-NMLZ COP-SUB\3SG.SS aint ͡sana ̃ hɨɨ̃ ̃=́ni ̃ wakɨ-tu-ki ̃ in.that.way\3SG.SS house\1PL/2PL/3.POSS=LOC return-APPL-WHILE.MOVING\3SG.SS ‘And the woman, because she had become afraid, in that way, she having returned to her house,’ (50) núna ɨt͡sɨrkou tiḿaji ʃuara ́kuaḿkaja shuara ́wɛiŋ́kahɛɛ túsa nu=na ɨt͡sɨra-ka-u timaji NON.VIS=ACC relate-INTENS-NMLZ NARR ʃuara ́ kuamaka=ia ʃuara ́ uaina-ka-ha-i tu-sa ̃ person\ACC forest=ABL person\ACC see-INTENS-1SG.SBJ-DECL say-SUB\3SG.SS ‘she told about that, saying “I’ve just seen the person from the forest”.’ (51) núkɨti nu=kɨ=iti NON.VIS=RESTR=COP.3+DECL ‘That is it.’ 901 Text 3: How to prepare banana drink This is a procedural text about how to make t͡samau, a refreshing drink made by mashing and fermenting ripe banana (the Amazonian Spanish name for this drink is ). T͡samau also means ‘ripe banana’. Speaker: Dina Ananco; Age: 29 Community of origin: Huabal Date: August, 2013; Duration: 1 min 38 sec (1) t͡samóo ikariaŕ umaŕtasarka t͡samau i-kari-a-́ri uma-ra-tasa-ri=ka banana CAUS-ferment-HIAF-1PL.SS drink.PFV-DISTR-PURP-1PL.SS=FOC ‘To drink banana that has been made fermented’ (2) paańtam úun iísar tikit́͡ʃik rasiḿo hukiŕ ahaḱar hukiŕ paantama uunta ii-sa-ri tikit ͡ʃiki rasimo hu-ki-ri banana big see-SUB-1PL.SS one bunch take-WHILE.MOVING-1PL.SS aha-ka-ri hu-ki-ri fell-INTENS-1PL.SS take-WHILE.MOVING-1PL.SS ‘having chosen one big banana bunch, having taken one bunch, having cutting it and taken it,’ (3) kaḿɨ ʃiír t͡samaḱmatɛɛ ̃ kamɨ ʃiira t͡sama-ka-matai ̃ INTERJ very ripen-INTENS-1SG/3.DS ‘when they [the bananas] are very ripe’ 902 (4) tiʃiŕar ója úunnum tiʃi-ra-ri oja uunta=numa peel-DISTR-1PL.SS pot big=LOC ‘having peeled [the bananas] in a big pot’ (5) kaḿɨ t͡samóo pakaŕar ója úunnum kamɨ t͡samau paka-ra-ri oja uunta=numa INTERJ banana skin-DISTR-1PL.SS pot big=LOC ‘having peeled the bananas in a big pot’ (6) t͡ʃimpiaŕ hií jumiíʃa ahũńtar t͡ʃimpi-a-́ri hii iumi=ʃa ahũnta-ri "ll.in.recipient-HIAF-1PL.SS "re water=ADD add.water.PFV-1PL.SS ‘having "lled them in the pot, in the "re, having added water too’ (7) hiiʃ́a kaḿɨ ikaṕarmoo asaḿtɛɛ ̃nui ̃ ́ɨkɨŋ́kar ukútɛiɲ̃ɛiti hii=ʃa kamɨ ikapa-ra-mau a-sa-matai ̃ "re=ADD INTERJ set.alight-ATTRIB-NMLZ COP-SUB-1SG/3.DS nui ̃ ɨkɨna-ka-ri uku-tai=̃aiti there put.on."re-INTENS-1PL.SS boil-NMLZ=COP.3+DECL ‘the "re too, because it is lit, having put [the pot with bananas] on the "re, we boil [the bananas].’ (8) hiíka iḿanik kaḿɨ hii=ka iḿani=kɨ kamɨ "re=FOC so.much=RESTR INTERJ ‘With a lot of "re,’ (9) nɨẃa nɨẃaka ɨmah́t͡ʃatɛiɲ̃ɛiti nɨua nɨua=ka ɨmaha-t͡ʃa-tai=̃aiti /ame /ame=FOC progress-NEG-NMLZ=COP.3+DECL ‘with a lot of /ame, we do not do it.’ 903 (10) aj́atɨk iḿatika awah́tʲukt͡suk aíatɨk imatika auahitʲa-t͡su=kɨ only do.much fan+IPFV-NEG=RESTR ‘[We prepare the drink] without fanning the /ames too much’ (11) nɨntɨaŕ aḿounam ɨkɨŋ́kar nɨntɨa-́ri a-mau=nama ɨkɨna-ka-ri calm.down.PFV-1PL.SS COP-NMLZ=LOC put.on."re-INTENS-1PL.SS ‘having put it [the pot with bananas] on the "re when there are embers’310 (12) kuakti ́túsa iɲaítɛiɲ̃ɛiti waʃukti ́tóo asaŕ kua-ka-ti tu-sa inai-tai=̃aiti boil-INTENS-JUSS say-SUB leave-NMLZ=COP.3+DECL uaʃu-ka-ti ta-u a-sa-ri become.brown-INTENS-JUSS say+IPFV-NMLZ COP-SUB-1PL.SS ‘saying “let it boil”, we leave it, being ones who say “let them become brown”’ (13) narúkmatɛiʃ̃a iɲɛísar ukuwá puhútɛiɲ̃ɛiti washukti ́túsar naru-ka-matai-̃ʃa inai-sa-ri uku-a puhu-tai=̃aiti cook-INTENS-1SG/3.DS-CONCESS leave-ATT-1PL.SS boil-IPFV live/be-NMLZ=COP.3+DECL uaʃu-ka-ti tu-sa-ri become.brown-INTENS-JUSS say-SUB-1PL.SS ‘Even though [the banana] is cooked, we leave it boiling so that it becomes brown (Lit.: ‘we leave it boiling, saying: “let them become brown”).’ (14) túra núu aʃi ́waʃúkmatɛi ̃paańtam aʃi ́waʃúkmatɛiŋ̃kʲa tura nu aʃi uaʃu-ka-matai ̃ and NON.VIS all become.brown-INTENS-1SG/3.DS 310. Nɨntɨaŕi amaunama was translated as ‘cuando hay brasas’ ‘when there are embers’; a more literal translation is ‘when [the /ames are] calmed down’. 904 paatanama aʃi uaʃu-ka-matai=̃ka plantain all become.brown-INTENS-1SG/3.DS=FOC ‘And when all that has become brown, when all the plantain has become brown,’ (15) ɨnɨḱar ukútɛiɲ̃ɛiti mikirti ́tusar ɨnɨ-ka-ri uku-tai=̃aiti miki-ra-ti tu-sa-ri take.out.of."re-INTENS-1PL.SS leave-NMLZ=COP.3+DECL cool-DISTR-JUSS say-SUB-1PL.SS ‘we take it out of the "re, saying: “let it cool”.’ (16) nihaḿat̃͡ʃ ihúaj ɛítkasar ihútɛiɲ̃ɛiti t͡samóoʃa nihamat̃͡ʃi ihu-a-ji aitika-sa-ri ihu-tai=̃aiti manioc.beer mash-IPFV-1PL.SBJ do.like.that-SUB-1PL.SS mash-NMLZ=COP.3+DECL t͡samau=ʃa banana=ADD ‘Doing like we mash the manioc beer, we mash the banana too.’ (17) nútikʲar ihúrika aʃi ́kaḿɨ ʃiír nutika-ri ihu-ri=ka aʃi kamɨ ʃiira do.that.PFV-1PL.SS mash.PFV-1PL.SS=FOC all INTERJ well ‘Having done that, having mashed all well,’ (18) ihú ihú ɨmaśar ihu ihu ɨma-sa-ri mash mash progress-SUB-1PL.SS ‘doing it well mashed’ (19) ukútɛiɲ̃ɛiti naań uku-tai=̃aiti naa=na leave-NMLZ=COP.3+DECL HESIT.PRO=ACC ‘we leave it, umm,’ 905 (20) kaḿɨ ʃiír kariaḿtɛɛ umaŕmi túsar kamɨ ʃiira kari-a-́matai ̃ uma-ra-mi tu-sa-ri INTERJ very ferment-HIAF-1SG/3.DS drink.PFV-DISTR-HORT say-SUB-1PL.SS ‘saying: “let’s drink when it is very fermented”’. (21) túra kaʃi ́t͡sawańtɛik̃ʃa umaḱrikʲa umúmɛiɲɛiti jumiń umaŕtah takúrka túra tura kaʃińi t͡sauanta=i=̃kɨ=ʃa uma-ku-ri-ka and tomorrow day=LOC=RESTR=ADD drink+IPFV-SIM-1PL.SS-COND umu-mai-na=iti iumintu uma-ra-tah ta-ku-ri-ka tura drink-POT-NMLZ=COP.3+DECL sweet drink.PFV-DISTR-DES say+IPFV-SIM-1PL.SS-COND and ‘Or the next day, if we want to drink, we can drink sweet, if we want, and’ (Lit: ‘Or the next day, if we want to drink, we can drink while saying I want to drink sweet, and’)311 (22) nukaṕ t͡sawańkɨʃa ukumɛíɲɛiti ʃiír karia ́umaŕtah takúrka nukapɨ t͡suanta=kɨ=ʃa uku-mai-na=iti several day=RESTR=ADD leave-POT-NMLZ=COP.3+DECL ʃiira kari-a ́ uma-ra-tah ta-ku-ri-ka very ferment-HIAF drink.PFV-DISTR-DES say+IPFV-1PL.SS-COND ‘we can leave [the banana drink] several days too if we want to drink it very fermented.’ (Lit: ‘we can leave [the banana drink] several days too while saying I want to drink very fermented.’) (23) nukɨti nu-kɨ-iti NON.VIS-RESTR-COP.3+DECL ‘That is it.’ 311. Tura usually means ‘and’ or ‘then’, but this instance was translated as ‘or’. 906 REFERENCES CITED Abad Gonzáles, Luisa. 2003. Etnocidio y resistencia en la amazonía peruana. Cuenca: Ediciones de la Universidad de Castilla-La Mancha. Adelaar, Willem F. H. (in collaboration with Pieter Muysken). 2004. Languages of the Andes. Cambridge: Cambridge University Press. Aikhenvald, Alexandra Y. 2012. Languages of the Amazon. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. 2003. Typological parameters for the study of clitics, with special reference to Tariana. In R. M. W. Dixon & Alexandra Y. Aikhenvald (Eds.), Word: A cross-linguistic typology (pp. 42-78). Aissen, J. 2003. Di$erential object marking: Iconicity vs. economy. Natural Language & Linguistic Theory, 21(3), 435-483. Ameka, Felix K., Alan Dench, & Nicholas Evans (Eds.). 2006. Catching language: The standing challenge of grammar writing. Berlin: Mouton de Gruyter. Ameka, Felix K., & Stephen C. Levinson. 2007. Introduction: The typology and semantics of locative predicates: postural, positionals, and other beasts. Linguistics, 45(5/6), 847-871. Anaya, James. 2009. Observaciones sobre la situación de los pueblos Indígenas de la Amazonía y los sucesos del 5 de junio y días posteriores en las provincias de Bagua y Utcubamba, Perú. Retrieved 04-30, 2014 from http://unsr.jamesanaya.org/ special-reports/observaciones-sobre-la-situacion-de-los-pueblos-indigenas-de-la- amazonia-y-los-socesos-del-5-de-junio-y-dias-posteriores-en-las-provincias-de-bagua-y- utcubamba-peru-2009 Anderson, Gregory D. S. 2006. Auxiliary verb constructions. Oxford: Oxford University Press. 907 Anderson, Stephen R. 2005. Aspects of the theory of clitics. Oxford: Oxford University Press. Beasley, David, & Kenneth L. Pike. 1957. Notes on Huambisa phonemics. Lingua Posnaniensis, 6, 1-8. Berlin, Brent. 1992. Ethnobiological classi)cation: Principles of categorization of plants and animals in traditional societies. Princeton, NJ: Princeton University Press. Berlin, Brent. 1994. Evidence for pervasive synesthetic sound symbolism in ethnozoological nomenclature. In Leanne Hinton, Johanna Nichols, & John J. Ohala (Eds.), Sound symbolism (pp. 76-93). Cambridge: Cambridge University Press. Berlin, Brent, & Elois Ann Berlin. 1983. Adaptation and ethnozoological classi"cation: Theoretical implications of animal resources and diet of the Aguaruna and Huambisa. In Raymond B. Hames & William T. Vickers (Eds.), Adaptive responses of native Amazonians (pp. 301-325). New York: Academic Press. Berlin, Brent, & John P. O’Neill. 1981. The pervasiveness of onomatopoeia in Aguaruna and Huambisa bird names. Journal of Ethnobiology, 1(2), 238-261. Berlin, Brent, & Elois Ann Berlin. 1975. Aguaruna color categories. American Ethnologist, 2(1), 61-87. Bloom"eld, Leonard. 1926. A set of postulates for the science of language. Language, 2, 153-164. Bloom"eld, Leonard. 1933. Language. New York: Henry Holt and Company. Bolinger, Dwight. 1967. Adjectives in English: Attribution and predication. Lingua, 18, 1-34. Bossong, Georg. 1983-1984. Animacy and markedness in Universal Grammar. Glossologia, 2–3, 7-20. Boster, James S. 2003. Blood feuds and table manners: A neo-Hobessian approach to Jivaroan warfare. Antropológica, 99–100, 153-164. 908 Bresnan, Joan, & Lioba Moshi. 1990. Object asymmetries in comparative Bantu syntax. Linsguistic Inquiry, 147–185. Bybee, Joan. 1985. Morphology: A study of the relation between meaning and form. Amsterdam: John Benjamins. Bybee, Joan, Revere Perkins, & William Pagliuca. 1994. The evolution of grammar: Tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press. Campbell, Lyle. 1997. American Indian languages: The historical linguistics of Native America. Oxford: Oxford University Press. Campbell, Lyle. 2012. Classi"cation of the indigenous languages of South America. In Lyle Campbell & Verónica Grondona (Eds.), The indigenous languages of South America: A comprehensive guide (pp. 59-166). Berlin: Mouton de Gruyter. CATAPA. 2009. 2 Belgian volunteers were witnesses of a bloody con/ict in North Peru [Press Release]. Retrieved 04-30, 2014 from http://www.catapa.be/en/ news/340 Cerrón-Palomino, Rodolfo. 1987. Lingüística Quechua. Cusco: Centro Bartolomé de las Casas. Cerrón-Palomino, Rodolfo. 1994. Quechua sureño: Diccionario uni)cado, Quechua- Castellano, Castellano-Quechua. Lima: Biblioteca Nacional del Peru. Chafe, Wallace (Ed.). 1980. The pear stories: Cognitive, cultural, and linguistic aspects of narrative production. Norwood: Ablex. Chafe, Wallace L. 1970. Meaning and the structure of language. Chicago: The University of Chicago Press. Chaumeil, Jean-Pierre. 1988. Le Huambisa défenseur. La "gure de l’Indien dans le chamanisme populaire (région d’Iquitos, Pérou). Recherches amérindiennes au Québec, 18, 115-126. 909 Chomsky, Noam. 1965. Aspects of the theory of syntax. Cambridge: MIT Press. Clark, Eve V. 1978. Locationals: Existential, locative and possessive constructions. In Joseph H. Greenberg, Charles A. Ferguson, & Edith A Moravcsik (Eds.), Universals of human language (Vol. IV: Syntax, pp. 85-126). Standford: Standford University Press. Comrie, Bernard. 1976. Aspect: An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press. Comrie, Bernard. 1978. Ergativity. In Winfred P. Lehmann (Ed.), Syntactic typology: Studies in the phonomenology of language (pp. 329-334). Austin: Universtin of Texas Press. Comrie, Bernard. 1985. Language typology and syntactic description. In Timothy Shopen (Ed.), The Dative and related phenomena (Vol. III: Grammatical categories and the lexicon, pp. 308-348). Cambridge: Cambridge University Press. Comrie, Bernard. 1989. Language universals and linguistic typology (2 ed.). Chicago: University of Chicago Press. Corbera Mori, Ángel. 1978. Fonología aguaruna (jíbaro). Bachelor’s Degree thesis. Universidad Mayor de San Marcos, Lima. Corbera Mori, Ángel. 1980. Vocabulario breve del idioma huambisa. Panorama Amazónico, 4, 67-71. Corbera Mori, Ángel. 1994. Fonologia e gramática do Aguaruna (Jívaro). Doctoral Dissertation. Universidade Estadual de Campinas, Campinas, Brazil. Costales, Piedad, & Alfredo Costales. 2006. Historia de la nación Shuar (1). Quito: Abya-Yala. Cotlow, Lewis. 1953. Amazon headhunters. New York: Holt. Cristofaro, Sonia. 2003. Subordination. Oxford/New York: Oxford University Press. 910 Croft, William. 1984. Semantic and syntactic correlates to syntactic categories. In D Testen, V Mishra, & J Drogo (Eds.), Papers from the parasession on lexical semantics, Twentieth Regional Meeting of the Chicago Linguistic Society (pp. 53-71). Chicago: Chicago Linguistic Society. Croft, William. 1994. Voice: Beyond control and a#ectednes. In Barbara Fox & P. J. Hopper (Eds.), Voice: Form and function (pp. 89-117). Amsterdam: John Benjamins. Croft, William. 2001. Radical construction grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press. Croft, William. 2003. Typology and universals (2 ed.). Cambridge, UK: Cambridge University Press. Cuesta, José M., S.J. 1984. Jaén de Bracamoros (2). Lima: Librería Studium. Czaykowska-Higgins, Ewa. 2009. Research models, community engagement, and linguistic !eldwork: Re"ections on working within Canadian indigenous communities. Language Documentation & Conservation, 3(1), 15-50. Retrieved from http:/ /scholarspace.manoa.hawaii.edu/bitstream/handle/10125/4423/ czaykowskahiggins.pdf?sequence=1 DeLancey, Scott. 1981. An interpretation of split ergativity and related patterns. Language, 53(3), 626-657. DeLancey, Scott. 1994. Grammaticalization and linguistic theory. In Jule Gomez de Garcia & David S. Rood (Eds.), Proceedings of the 1993 Mid-America Linguistics Conference and Conference on Siouan/Caddoan Languages (pp. 1-22). Boulder: Department of Linguistics, University of Colorado. DeLancey, Scott. 1997. Mirativity: The grammatical marking of unexpected information. Linguistic Typology, 1, 33-52. DeLancey, Scott. 2001a. LSA Summer Institute. UC Santa Barbara lectures on functional syntax. Retrieved 02-25, 2014 from http://www.uoregon.edu/~delancey/sb/ fs.html 911 DeLancey, Scott. 2001b. The mirative and evidentiality. Journal of Pragmatics, 33(3), 369-382. DeLancey, Scott. Forthcoming. Deictic and sociopragmatic e#ects in Tibeto- Burman SAP indexation. Descola, Philippe. 1986. La nature domestique: Symbolisme et praxis dans l’écologie des Achuar. Paris: Editions de la Maison des Sciences de l’homme. Dietrich, Wolf. 2006. La categoría del ‘frustrativo’ en las lenguas tupí-guaraníes. In Wolf Dietrich & S. Haralambos (Eds.), Guaraní y Mawetí-Tupí-Guaraní: Estudios históricos y descriptivos sobre una familia lingüística de América del Sur (pp. 71-85). Münster: LIT-Verlag. Dingemanse, Mark. 2014. Making new ideophones in Siwu: Creative depiction in conversation. Pragmatics and Society, 5(3), 384-405. Dixon, R. M. W. 1982. Where have all the adjectives gone? And other essays in Semantics and Syntax. Berlin: Mouton. Dixon, R. M. W. 1997. The rise and fall of languages. Cambridge: Cambridge University Press. Dixon, R. M. W. 2002. Australian languages: Their nature and development. Cambridge: Cambridge University Press. Dixon, R. M. W., & Alexandra Y. Aikhenvald. 1999. Introduction. In R. M. W. Dixon & Alexandra Y. Aikhenvald (Eds.), The Amazonian languages (pp. 1-22). Cambridge: Cambridge University Press. Doke, Clement M. 1935. Bantu linguistic terminology. London: Longmans, Green & Co. Dryer, Mathew S. 2006. Descriptive theories, explanatory theories and Basic Linguistic Theory. In Felix K. Ameka, Alan Dench, & Nicholas Evans (Eds.), Catching language: The standing challenge of grammar writing (pp. 207-234). Berlin: Mouton de Gruyter. 912 Espinoza Soriano, W. 1973. Los grupos étnicos en la cuenca del Chuquimayo, siglos XV y XVI. Bulletin de l’Institute Français de Études Andines, 2(3), 19-73. Espinoza Soriano, W. 2004. La etnia guayaqundo en la sierra piurana. Boletín de Arqueología de la Ponti!cia Universidad Católica del Perú, 8, 133-150. Fabre, Alain. 2005 [modi!ed 07-22-2007]. Diccionario etnolingüístico y guía bibliográ!ca de los pueblos indígenas sudamericanos. Retrieved 02-20, 2013 from http://www.ling.!/Entradas%20diccionario/Dic=Jivaro.pdf Fast, Gerhard. 1975a. Análisis tentativo del sistema fonológico del idioma jíbaro del río Corrientes. Yarinacocha: Instituto Lingüístico de Verano. Fast, Gerhard. 1975b. Sistema fonológico del idioma Achual. Lima: Instituto Lingüístico de Verano. Fast, Gerhard, Ruby Warkentin de Fast, & Daniel Fast Warkentin. 1996. Diccionario Achuar-Shiwiar~Castellano (2 ed.). Yarinacocha: Ministerio de Educación/ Instituto Lingüístico de Verano. Fleck, David W. 2008. Sugerencias metodológicas para realizar trabajo de campo lingüístico en la Amazonía. Lexis, 32(2), 251-180. Fleischman, Suzanne. 1990. Tense and narrativity: From medieval performance to modern !ction. Austin: University of Texas Press. FONCODES. 2006. Mapa de la pobreza 2006. Retrieved 02-12, 2014 from http:/ /www.foncodes.gob.pe/portal/documentos/mapapobreza/ Foncodes2006_mapapobreza_PRES.pdf François, A. 2003. La sémantique du prédicat en mwotlap. Leuven/Paris: Peeters. García-Rendueles, Manuel, S.J. 1996–1999. Yaunchuk…: Universo mítico de los huambisas (1–2). Lima: Centro Amazónico de Antropología y Aplicación Práctica. Gast, Volker, & Johan Van der Auwera. 2011. Scalar additive operators in the languages of Europe. Language, 87(1), 2-54. 913 Genetti, Carol, A. R. Coupe, Ellen Bartee, Kristine Hildebrandt, & Lin You-Jing. 2008. Syntactic aspects of nominalization in !ve Tibeto-Burman languages of the Himalayan area. Linguistics of the Tibeto-Burman Area, 31(2), 97-143. Gil, David. 2001. Escaping eurocentrism: Fieldwork as a process of unlearning. In Paul Newman & Martha Ratli# (Eds.), Linguistic !eldwork (pp. 102-132). Cambridge: Cambridge University Press. Gildea, Spike. 1998. On reconstructing grammar: Comparative Cariban morphosyntax. New York: Oxford University Press. Givón, T. 1983. Topic continuity in discourse: The functional domain of switch- reference. In John Haiman & Pamela Munro (Eds.), Switch reference and universal grammar (pp. 51-82). Amsterdam: John Benjamins. Givón, T. 1994. The pragmatics of de-transitive voice: Functional and typological aspects of inversion. In T Givón (Ed.), Voice and inversion (pp. 3-44). Amsterdam/Philadelphia: John Benjamins. Givón, T. 1971. Historical syntax and synchronic morphology: An archaeologist’s !eld trip. In Papers from the Seventh Regional Meeting of the Chicago Linguistic Society (pp. 394-415). Chicago: Chicago Linguistic Society. Givón, T. 1979. On understanding grammar. New York: Academic Press. Givón, T. 2002a. Syntax: An introduction. (I). Amsterdam: John Benjamins. Givón, T. 2002b. Syntax: An introduction. (II). Amsterdam: John Benjamins. Gleason, Henry A. 1961. An introduction to descriptive linguistics. New York: Holt, Rinehart and Winston. Gnerre, Maurizio. 1973. Sources of Spanish Jívaro. Romance Philology, 27(2), 203-204. 914 Gnerre, Maurizio. 1975. L’utilizzazione delle fonti documentarie dei secoli VXI e XVII per la storia linguistica Jibaro. In E Cerulli & G. Della Raggione (Eds.), Atti del XL Congresso Internazionale degli Americanisti (pp. 79-86). Genoa: Thilger. Gnerre, Maurizio. 2000. From headhunters to writers: A Shuar myth and an oration. In Kay Sammons & Joel Sherzer (Eds.), Translating native Latin American verbal art: Ethnopoetics and ethnography of speaking (pp. 197-210). Washington: Smithsonian Institution Press. Gnerre, Maurizio. 2010. Per!l descriptivo e histórico-comparativo de una lengua amazónica: el shuar (Jíbaro) (Roberta Gozzi, Trans.). Bilbao: Euskal Herriko Unibertsitatea (Universidad del País Vasco). Gnerre, Maurizio. n.d. Los salesianos y los shuar. Unpublished manuscript. Godoy, Lina, & Rocío Faya. 2009, June 6. Masacre en la Amazonía. Diario La Primera. Retrieved from http://www.diariolaprimeraperu.com/online/politica/masacre- en-la-amazonia_39866.html Goldenberg, Gideon. 1998. Tautological in!nitive. In Studies in Semitic linguistics: Selected writings (pp. 36-85). Jerusalem: Magnes Press, Hebrew University. Good, Je#, & Michael Cysouw. 2013. Languoid, doculect, and glossonym: Formalizing the notion ‘Language’. Language Documentation & Conservation, 7, 331-359. Green, Shane. 2009. Customizing indigeneity: Paths to a visionary politics in Peru. Palo Alto: Standford University Press. Greenberg, Joseph H. 1987. Language in the Americas. Stanford: Stanford University Press. Grinevald, Colette. 1998. Language endangerment in South America: A programmatic approach. In Lenore A. Grenoble & Lindsay J. Whaley (Eds.), Endangered languages: Language loss and community response (pp. 124-159). Cambridge: Cambridge University Press. 915 Grinevald, Colette. 2006. The expression of static location in a typological perspective. In Maya Hickmann & Stéphan Robert (Eds.), Space in languages: Linguistic systems and cognitive categories (pp. 29-58). Amsterdam: John Benjamins. Guallart, José M. 1990. Entre pongo y cordillera: Historia de la etnia aguaruna- huambisa. Lima: Centro Amazónico de Antropología y Aplicación Práctica. Gu#roy, Jean. 2006. El Horizonte corrugado: correlaciones estilísticas y culturales. Bulletin de l’Institute Français de Études Andines, 35(3), 347-359. Guillaume, Antoine. 2006. La catégorie du ‘movement associé’ en cavineña: apport a une typologie de l’encodage du movement et de la trajectoire. Bulletin de la Societé de Linguistique de Paris, 101(2), 415-436. Gussenhoven, Carlos. 2004. The phonology of tone and intonation. Cambridge: Cambridge University Press. Haiman, John. 1978. Conditionals are topics. Language, 54(3), 564-689. Haiman, John. 1983. Iconic and economic motivation. Language, 59, 781-819. Haiman, John & Pamela Munro (Eds.). 1983. Switch-reference and universal grammar. Amsterdam: John Benjamins. Hale, Ken. 2001. Ulwa (Southern Sumu): The begginings of a language research project. In Paul Newman & Martha Ratli# (Eds.), Linguistic !eldwork (pp. 76-101). Cambridge: Cambridge University Press. Hale, Ken, Michael Krauss, Lucille J. Watahomigie, Akira Y. Yamamoto, Colette Craig, LaVerne Masayesva Jeanne, & Nora C. England. 1992. Endangered languages: On endangered languages and the safeguarding of diversity. Language, 68(1), 1-42. Haspelmath, Martin. 2002. Understanding morphology. London: Arnold. Haspelmath, Martin (Ed.). 2004. Coordinating constructions. Amsterdam/ Philadelphia: John Benjamins. 916 Haspelmath, Martin. 2007. Pre-established categories don’t exist: consequences for language description and typology. Linguistic Typology, 11, 119-132. Haspelmath, Martin. 2012. How to compare major word-classes across the world’s languages. In Thomas Graf, Denis Paperno, Anna Szabolci, & Jos Tellings (Eds.), Theories of everything: In honor of Edward Keenan (pp. 109-130). Los Angeles: UCLA. Hayes, Bruce. 1995. Metrical stress theory: Principles and case studies. Chicago: University of Chicago Press. Heine, Bernd, Ulrike Claudi, & Friederike Hünnemeyer. 1991. Grammaticalization: A conceptual framework. Chicago: University of Chicago Press. Hengeveld, Kees. 1992. Non-verbal predication: Theory, typology, diachrony. Berlin: De Gruyter. Hockett, C. F. 1958. A course in modern linsguistics. New York: Macmillan. Hocquenghem, Anne Marie. 1989. Los guayacundos de Caxas y la sierra piuruana: siglos XV y XVI. Lima: Centro de Investigación y Promoción del Campesinado/Instituto Francés de Estudios Andinos. Hogg, Richard, & C. B. McCully. 1987. Metrical phonology: A coursebook. Cambridge: Cambridge University Press. Hopper, Paul. 1979. Aspect and foregrounding in discourse. Syntax and semantics, 12, 213-241. Hopper, Paul J., & Sandra Thompson. 1984. The discourse basis for lexical categories in Universal Grammar. Language, 60, 703-752. Houaiss, Antônio, & Mauro Villar. 2001. Dicionário Houaiss da Língua Portuguesa. Rio de Janeiro: Objetiva. Hyman, Larry M. 2006. Word-prosodic typology. Phonology, 23(2), 225-257. Hyman, Larry M. 2009. How (not) to do phonological typology: the case of pitch-accent. Language Sciences, 3(2–3), 213-238. 917 Hyman, Larry M. 2010. Amazonia and the typology of tone systems. Paper presented at Amazónicas III Conference: The Structure of Amazonian Languages, Bogotá, Colombia. Retrieved from http://linguistics.berkeley.edu/phonlab/annual_report/ documents/2010/Hyman%20Amazonia%20Bogota%20PLAR.pdf INEI. 2010. Mapa de pobreza provincial y distrital 2009. Lima: INEI. Jakway, Martha, Horacio Lorenzo C., & Arturo Antonio A. 1987. Vocabulario huambisa (2 ed.). Yarinacocha, Pucallpa: Ministerio de educación/Instituto Lingüístico de Verano. Jansen, Joana, Janne Underriner, & Roger Jacob. 2013. Revitalizing languages through place-based language curriculum: Identity through learning. In Elena Mihas, Bernard Perley, Gabriel Rei-Doval, & Kathleen Whealey (Eds.), Responses to Language Endangerment: In honor of Mickey Noonan. New directions in language documentation and language revitalization (pp. 221-242). Amsterdam/Philadelphia: John Benjamins. Jespersen, Otto. 1992 [1924]. The philosophy of grammar. Chicago: University of Chicago Press. Karsten, Rafael. 1935. The head-hunters of Western Amazon: The life and culture of the Jibaro indians of Eastern Ecuador and Peru. Societas Scientiarum Fennica, Commentationes Humanarum Litterarum, 7(1). Katan Jua, Tuntiak. 2011. Ii chichame unuimiamu / Investigando nuestra lengua / Investigating our language: Shuar Chicham. In Marleen Haboud & Nicholas Ostler (Eds.), Endangered languages: voices and images. Proceedings of the FEL XV Annual Conference (pp. 103-105). Bath, England: Foundation for Endangered Languages. Kaufman, Terrence. 1990. Language history in South America: What we know and how to know more. In Doris L. Payne (Ed.), Amazonian linguistics: Studies in lowland South American languages (pp. 13-73). Austin: University of Texas Press. Kaufman, Terrence (with the help of Brent Berlin). 2007. South America. In Christopher Moseley & R. E. Asher (Eds.), Atlas of the World’s Languages (2 ed.). London: Routledge. 918 Kemmer, Suzanne. 1993. The middle voice. Amsterdam/Philadelphia: John Benjamins. Kemmer, Suzanne. 1994. Middle voice, transitivity and the elaboration of events. In Barbara Fox & Paul J Hopper (Eds.), Voice: Form and function (pp. 179-230). Amsterdam: John Benjamins. Koch, Harold. 1984. The category of “associated motion” in Kaytej. Language in Central Australia, 1, 23-34. Kohlberger, Martin. 2014. Predicative uses of nominalisations in Shiwiar (Jivaroan). Paper presented at the 11th High Desert Linguistic Society Conference, University of New Mexico. König, Ekkehard. 1991. The meaning of focus particles: A comparative perspective. London: Routledge. Koptjevskaja-Tamm, Maria. 1993. Nominalizations. Oxford/New York: Routledge. Krifka, Manfred. 1998. Additive particles under stress. In Devon Strolovitch & Aaron Lawson (Eds.), Proceedings of Semantics and Linguistic Theory 8 (pp. 111-128). New York: CLC Publications. Ladefoged, Peter. 2005. Vowels and consonants: An introduction to the sounds of languages (2 ed.). Malden: Blackwell. Langacker, Ronald W. 1976. Semantic representation and the linguistic relativity hypothesis. Foundations of Language, 14, 307-357. Langacker, Ronald W. 1987. Foundations of cognitive grammar (Vol. I: Theoretical prerequisites). Standford: Standford University Press. Langacker, Ronald W. 1988. A view of linguistic semantics. In Brygida Rudzka- Ostyn (Ed.), Topics in cognitive linguistics (pp. 49-90). Amsterdam: John Benjamins. Larson, Mildred L. 1955–1957. Comparación de los vocabularios Aguaruna y Huambisa. Tradición, Revista Peruana de Cultura (suplement), 19-20. 919 Larson, Mildred L. 1963. Emic classes which manifest the obligatory tagmemes in major independent clause types of Aguaruna (Jivaro). Studies in Peruvian Indian languages 1 (pp. 1-36). Norman: Summer Institute of Linguistics of the University of Oklahoma. Larson, Mildred L. 1978. The functions of reported speech in discourse. Dallas: Summer Institute of Linguistics. Lathrap, Donald W. 1970. The Upper Amazon. London: Thames & Hudson. Lazard, Gilbert. 1992. Y at-il des catégories interlangagières? In Susanne Anschütz (Ed.), Texte, Sätze, Wörter und Moneme: Festschrift für Klaus Heger zum 65. Geburtstag (pp. 427-434). Heidelberg: Heidelberger Orientverlag. Lehmann, Christian. 1985. Grammaticalization: Synchronic variation and diachronic change. Lingua e Stile, 20, 303-318. Lehmann, Christian. 2008. Roots, stems and word classes. Studies in Language, 32(3), 546-567. Levinson, Stephen C., & D. P Wilkins. 2006. The background to the study of the language of space. In Stephen C Levincon & D. P. Wilkins (Eds.), Grammars of space: Explorations in cognitive diversity (pp. 1-23). Cambridge: Cambridge University Press. Lichtenberk, F. 1995. Apprehensional epistemics. In Joan Bybee & S. Fleischmann (Eds.), Modality in grammar and discourse (pp. 293-327). Amsterdam/ Philadelphia: John Benjamins. Lindblom, Björn, & Ian Maddieson. 1988. Phonetic universals in consonant systems. In Larry M. Hyman & Charles N. Li (Eds.), Language, speech and mind (pp. 62-78). London: Routledge. Longacre, Robert E. 1970. Sentence structure as a statement calculus. Language, 46, 783-815. Longacre, Robert E. 1976. An anatomy of speech notions. Lisse: Peter de Ridder Press. 920 Longacre, Robert E. 2007. Sentences as combinations of clauses. In Timothy Shopen (Ed.), Language typology and syntactic description (Vol. II: Complex constructions, pp. 372-420). Cambridge: Cambridge University Press. Loukotka, Čestmír. 1968. Classi!cation of South American indian languages. Los Angeles: University of California. Maddieson, Ian, & Kristin Precoda. 1992. Syllable structure and phonetics models. Phonology, 9(1), 45-60. Maroni, Pablo S.J. 1988 [1738]. Noticias aunténticas del famoso río Marañón. Iquitos: IIAP/CETA. Matissof, James A. 1975. Rhinoglottophilia: The mysterious connection between nasality and glottality. In Charles A. Ferguson, Larry M. Hyman, & John J. Ohala (Eds.), Nasálfest: Papers from a symposium on nasals and nasalization (pp. 265-268). Standford: Language Universals Project. Michael, Lev. 2011. The interaction of tone and stress in the prosodic system of Iquito (Zaparoan). Amerindia, 35, 53-74. Miller, George, & Philip Johnson-Laird. 1976. Language and perception. Cambridge: Harvard University Press. Mithun, Marianne. 1993. Switch-reference: Clause combining in Central Pomo. International Journal of American Linguistics, 59(2), 119-136. Morales Chocano, Daniel. 1998. Chambira: una cultura de sabana árida en la selva peruana. Investigaciones Sociales, 2, 61-75. Moseley, Christopher (Ed.). 2010. Atlas of the World’s languages in danger (3 ed.). Paris: UNESCO Publishing. Muysken, Pieter. 2010. The copula in Ecuadorian Quechua. In Eithne B. Carlin & Simon van de Kerke (Eds.), Linguistics and archaeology in the Americas: The historization of language and society (pp. 191-206). Leiden: Brill. 921 Muysken, Pieter. 2013. Change, contact, and ethnogenesis in Northern Quechua: Structural phylogenetic approaches to clause-embedding predicates. In Alf Hornborg & Jonathan D. Hill (Eds.), Ethnicity in ancient Amazonia: Reconstructing past identities from Archaeology, Linguistics, and Ethnohistory. Boulder: University Press of Colorado. Myers, Thomas. 2005. Variación regional y cultural en la cerámica Jíbaro. Unay Runa, 7, 137-151. Neves, Eduardo. 2001. Indigenous historical trajectories in the Upper Rio Negro basin. In Colin McEwan, Christina Barreto, & Eduardo Neves (Eds.), Unknown Amazon: Studies in visual and material culture (pp. 266-286). London: The British Museum Press. Noonan, Michael. 2007. Complementation. In Timothy Shopen (Ed.), Language typology and syntactic description (Vol. II: Complex constructions, pp. 52-150). Cambridge: Cambridge University Press. Nuckolls, Janice B. 1996. Sounds like life: sound-symbolic grammar, performance and cognition in Pastaza Quechua. New York: Oxford University Press. Overall, Simon. 2007. A grammar of Aguaruna. Ph.D. Dissertation. La Trobe University, Melbourne, Australia. Overall, Simon. 2008. On the non-phonemic status of the velar nasal /ŋ/ in Jivaroan. Liames, 8, 45-59. Overall, Simon. 2011. The semantics of clause linking in Aguaruna. In R.M.W. Dixon & Alexandra Y Aikhenvald (Eds.), The semantics of clause linking: A cross-linguistic typology (pp. 167-192). Oxford: Oxford University Press. Payne, David L. 1981. Bosquejo fonológico del Proto-Shuar-Candoshi: Evidencias para una !liación genética. Revista del Museo Nacional, 45, 323-377. Payne, David L. 1989. Extrametricalidad y acento tonal en el aguaruna. In Rodolfo Cerrón-Palomino & Gustavo Solís Fonseca (Eds.), Temas de lingüística amerindia (pp. 251-272). Lima: Consejo Nacional de Ciencia y Tecnología. 922 Payne, David L. 1990a. Accent in Aguaruna. In Doris L Payne (Ed.), Amazonian linguistics: Studies in lowland South American languages (pp. 161-184). Austin: University of Texas Press. Payne, David L. 1990b. Some widespread grammatical forms in South American languages. In Doris L Payne (Ed.), Amazonian linguistics: Studies in lowland South American languages (pp. 75-88). Austin: University of Texas Press. Payne, David L. 2008. Nasalidad en Aguaruna (3 ed.). Lima: Instituto Lingüístico de Verano. Payne, Doris L. 1985. In"ection versus derivation: Is there a di#erence? In Scott DeLancey & Russel S. Tomlin (Eds.), Proceedings of the First Annual Paci!c Linguistics Conference (pp. 247-260). Eugene: University of Oregon. Payne, Doris L. 1999a. Review of The Amazonian languages, ed. by R. M. W. Dixon and Alexandra Aikhenvald. Language, 77(3), 594-598. Payne, Doris L. 1999b. What counts as an explanation? A functional approach to word order. In Michael Darnell, Edith Moravcsik, Frederick J. Newmeyer, Michael Noonan, & Kathleen Wheatley (Eds.), Functionalism and Formalism in Linguistics (pp. 135-163). Amsterdam: John Benjamins. Payne, Doris L. 2009. Cuestiones tipológicas de la transitividad. Dimensión Antropológica, 47, 7-37. Payne, Doris L., & Immanuel Barshi. 1999. External possession: What, where, how and why. In Doris L. Payne & Immanuel Barshi (Eds.), External possession (pp. 3-33). Amsterdam: John Benjamins. Payne, Doris L., & Thomas E. Payne. 1990. Yagua. In Desmond Derbyshire & Geo#rey Pullum (Eds.), Handbook of Amazonian languages (Vol. 2, pp. 249-474). Berlin: Mouton de Gruyter. Payne, Doris L. & Shahar Shirtz (Eds.). 2015. Beyond aspect: The expression of discourse functions in African languages. Amsterdam: John Benjamins. 923 Payne, Thomas E. 1997. Describing morphosyntax: A guide for !eld Linguists. Cambridge: Cambridge University Press. Payne, Thomas E. & David Weber (Eds.). 2006. Perspective on grammar writing. Amsterdam/Philadelphia: John Benjamins. Pellizaro, Siro M. 1969. Shuar: Apuntes de gramática. Sucúa, Ecuador: Federación Provincial de Centros Shuar de Morona Santiago. Pellizaro, Siro M., & Fausto Oswaldo Náwech. 2005. Chicham: Diccionario shuar- castellano. Quito: Abya-Yala. Pellizzaro, Siro M. 1969. Shuar: Apuntes de gramática. Sucúa: Federación Provincial de Centro Shuar de Morona Santiago. Peña, Jaime. 2015. The expression of locational relations in Wampis. Paper presented at the Society for the Study of Indigenous Languages of the Americas (SSILA) Annual Meeting, Portland. Peña, Jaime. Forthcoming. Constructions with has(a) in Wampis. In Simon Overall, Rosa Vallejos, & Spike Gildea (Eds.), Non-verbal predication in Amazonian languages. John Benjamins. Peterson, David A. 2007. Applicative constructions. Oxford: Oxford University Press. Pike, Kenneth L., & Mildred L. Larson. 1964. Hyperphonemes and non- systematic features of Aguaruna phonemics. In A.H. Marckwardt (Ed.), Studies in languages and linguistics in honor of Charles C. Fries (pp. 55-67). Ann Arbor: English Language Institute. Pitman, N., E. I. Ruelas, D. Alvira, C. Vriesendorp, D. K. Moskovits, Á. del Campo, . . . R. C. Smith (Eds.). 2012. Peru: Cerros de Kampankis. Chicago: The Field Museum. 924 Queixalós, Francesc. 2009. La posture du corps dans la classi!cation et la localisation: l’exemple du sikuani. In L Wetzels (Ed.), The Linguistics of Endangered Languages (pp. 151-170). Utrech: LOT. Raimondi, Antonio. 1862. Apuntes sobre la provincia litoral de Loreto. Lima: Tipografía Nacional. Ramírez, Luis Hernán. 2003. El español amazónico hablado en el Perú. Lima: Juan Gutemberg Editores. Regan, Jaime. 1999. Mito y rito: una comparación entre algunas imágenes Mochicas y Jíbaras. Investigaciones Sociales, 3, 27-46. Renard-Casevitz, France-Marie, Thierry Saignes, & Anne-Christine Taylor. 1986. L’Inca, l’Espagnol et les Sauvages. Rapports entre les sociétés amazoniennes et andines du XVe au XVIIe siècle. Paris: Éditions Recherches sur les Civilizations. Rice, Keren. 2006. Let language tell its story? The role of linguistic theory in writing grammars. In Felix K. Ameka, Alan Dench, & Nicholas Evans (Eds.), Catching language: The standing challenge of grammar writing (pp. 235-268). Berlin: Mouton de Gruyter. Rivet, Paul. 1934. Population de la province de Jaén. Equateur. In Congrès international des sciences anthropologiques et ethnologiques: Compte-rendu de la première session (pp. 245-247). London: Royal Institute of Anthropology. Ross, Jannet B. 1988. A balance of deaths: revenge feuding among the Achuarä Jívaro of the northwest Peruvian Amazon. Washington: Columbia University. Rostain, Stéphen. 2010. Cronología del valle del Upano (Alta Amazonía Ecuatoriana). Bulletin de l’Institute Français de Études Andines, 39(3), 667-681. Saad, George Michel. 2014. A grammar sketch of Shuar: With focus on the verb phrase. M.A. Thesis. Radboud Universiteit Nijmegen. Santos-Granero, Fernando. 1992. Etnohistoria de la Alta Amazonía: siglos XV- XVIII. Quito: Abya-Yala. 925 Santos-Granero, Fernando, & Frederica Barclay. 2011. Bundles, stampers, and "ying gringos: Native perceptions of capitalist violence in Peruvian Amazonia. The Journal of Latin American and Caribbean Anthropology, 16(1), 143–167. Schachter, Paul, & Timothy Shopen. 2007. Parts-of-speech systems. In Timothy Shopen (Ed.), Language typology and syntactic description (Vol. I: Clause structure, pp. 1-60). Cambridge: Cambridge University Press. Schiering, René. 2006. Cliticization and the evolution of morphology: A cross- linguistic study on phonology in grammaticalization. Ph.D. Dissertation. Universität Konstanz, Germany. Shibatani, Masayoshi. 1985. Passive and related constructions: A prototype analysis. Language, 61(4), 821-848. Shibatani, Masayoshi. 2009. Elements of complex structures—where recursion isn’t: The case of relativization. In T. Givón & Masayoshi Shibatani (Eds.), Syntactic complexity: Diachrony, acquisition, neuro-cognition, evolution (pp. 163-198). Amsterdam: John Benjamins. Shibatani, Masayoshi, & Prashant Pardeshi. 2002. The causative continuum. In Masayoshi Shibatani (Ed.), The grammar of causation and interpersonal manipulation (pp. 85-126). Amsterdam/Philadelphia: John Benjamins. Siewierska, Anna. 2003. Person agreement and the determination of alignment. Transactions of the Philological Society, 101, 339-370. Silverstein, Michael. 1976. Hierarchy of features and ergativity. In R. M. W. Dixon (Ed.), Grammatical categories in Australian languages (pp. 112-171). Canberra: Australian Institute of Aboriginal Studies. Siverts, Henning. 1978. Jívaro headhunters in a headless time. In Regina E. Holloman & Serghei A. Arutiunov (Eds.), Perspective on ethnicity (pp. 127-138). The Hague: Mouton. 926 Stark, Louisa R. 1985. Indigenous languages of lowland Ecuador: History and current status. In South American indian languages: Retrospect and prospect (pp. 157-193). Austin: University of Texas Press. Stark, Louisa R., & Pieter Muysken. 1977. Diccionario Español-Quichua, Quichua- Español. Quito: Publicaciones de los Museos del Banco Central de Ecuador. Stassen, Leon. 1985. Comparison and universal grammar. Oxford: Basil Blackwell. Stassen, Leon. 2009. Predicative possession. Oxford: Oxford University Press. Steel, Daniel. 1999. Trade goods and Jívaro warfare: The Shuar 1850–1957, and the Achuar, 1940–1978. Ethnohistory, 46(4), 745-776. Stirling, Lesley. 1993. Switch-reference and discourse representation. Cambridge: Cambridge Univesity Press. Suárez, Jorge. 1974. South American indian languages. In Encyclopaedia Britannica (15 ed., pp. 105-112, vol 17). Chicago: Encyclopaedia Britannica. Swadesh, Morris. 1959. Mapas de clasi!cación lingüística de México y las Américas. México: UNAM. Talmy, Leonard. 2007. Lexical Typologies. In Timothy Shopen (Ed.), Language typology and syntactic description (Vol. III: Grammatical categories and the lexicon, pp. 66-168). Cambridge: Cambridge University Press. Taylor, Anne-Christine. 1991. Les Palta - Les Jivaro Andins précolombiens à la lumière de l’ethnographie contemporaine. Bulletin de l’Institute Français de Études Andines, 20(2), 439-459. Taylor, Anne-Christine. 1994. Una categoria irreductible en el conjunto de las naciones indígenas: los jívaro en las representaciones occidentales. In Blanca Muratorio (Ed.), Imágenes e imagineros: Representación de los indígenas ecuatorianos, siglos XIX y XX (pp. 75-109). Quito: FLACSO-Ecuador. 927 Taylor, Anne-Christine, & Philippe Descola. 1981. El conjunto jívaro en los comienzos de la conquista española del Alto Amazonas. Bulletin de l’Institute Français de Études Andines, 10(3–4), 7-54. Thiesen, Wesley. 1969. The Bora signal drum. Lore, 19, 101-103. Thompson, Sandra. 1988. A discourse approach to the cross-linguistic category “Adjective”. In John A Hawkins (Ed.), Explaining language universals (pp. 167-185). Oxford: Blackwell. Thompson, Sandra, Robert Longacre, & Shin Ja Hwang. 2007. Adverbial clauses. In Timothy Shopen (Ed.), Language typology and syntactic description (Vol. II: Complex constructions, pp. 237-300). Cambridge: Cambridge University Press. Timberlake, Alan. 2007. Aspect, tense, mood. In Timothy Shopen (Ed.), Language typology and syntactic description (Vol. III: Grammatical categories and the lexicon, pp. 280-333). Cambridge: Cambridge University Press. Torero, Alfredo. 1993. Lenguas del nororiente Peruano: la hoya de Jaén en el siglo XVI. Revista Andina, 2, 447-472. Torero, Alfredo. 2002. Idiomas de los Andes: Lingüística e historia. Lima: Instituto Francés de Estudios Andinos. Turner, Glen D. 1958a. Alternative phonemicizing in Jivaro. International Journal of American Linguistics, 24, 87-94. Turner, Glen D. 1958b. Jivaro phonology and morphology. Ph.D. Dissertation. Indiana University, Bloomington. Turner, Glen D. 1992. Una breve gramática del shuar. Quito: Instituto Lingüístico de Verano. Uwarai Yagkug, Abel, Isaac Paz Suikai, & Jaime Regan. 1998. Diccionario Aguaruna-Castellano, Awajún Chicham Apáchnaujai. Lima: Centro Amazónico de Antropología y Aplicación Práctica. 928 Valdez, Francisco. 2013. Primeras sociedades de la Alta Amazonía: La cultura Mayo Chinchipe–Marañón. Quito: Instuto Nacional de Patrimonio Nacional/Institute de Recherche pour le Développement. Valenzuela, Pilar. 2010. Applicative constructions in Shipibo-Konibo (Panoan). International Journal of American Linguistics, 76(1), 101-144. Valeš, Miroslav. 2013. Relaciones de lengua en el sector occidental de la familia lingüística Jívaro. Études Romanes de Brno, 34(2), 171-181. Vallejos, Rosa. 2014. Peruvian Amazonian Spanish: uncovering variation and deconstructing stereotypes. Spanish in Context, 11(3), 425-453. Van der Auwera, Johan, & Vladimir Plungian. 1998. Modality’s semantic map. Linguistic Typology, 2(1), 79-124. Vuillermet, Marine. 2012. A grammar of Ese Ejja (Takana), a language of the Bolivian Amazon. Doctoral Dissertation. Université Lumière Lyon 2, Lyon, France. Watkins, Laurel J. 1984. The discourse function of Kiowa switch-reference. International Journal of American Linguistics, 59(2), 137-164. Whitten, Norman E., Jr. 1976. Sacha runa: Ethnicity and adaptation of Ecuadorian jungle Quichua. Urbana: University of Illinois Press. Whitten, Norman E., Jr. 2013. Ethnogenesis and interculturality in the “Forest of Canelos”: The wild and the tame revisited. In Alf Hornborg & Jonathan D. Hill (Eds.), Ethnicity in ancient Amazonia: Reconstructing past identities from Archaeology, Linguistics, and Ethnohistory (pp. 321-334). Boulder: University of Colorado Press. Wipio D., Gerardo, Alejandro Paati Antunce, & Martha Jakway. 1996. Diccionario aguaruna-castellano, castellano-aguaruna. Lima: Ministerio de Educación/ Instituto Lingüístico de Verano. Wise, Mary Ruth. 2002. Applicative a'xes in Peruvian Amazonian languages. In Mily Crevels, Simon van de Kerke, Sergio Meira, & Hein van der Voort (Eds.), Current studies in South American languages (pp. 329-344). Leiden: CNWS. 929 Wise, Mary Ruth. 2014. Rasgos desconcertantes de contactos entre idiomas y culturas a lo largo de los contrafuertes orientales de los Andes del Perú. In Willem E.H. Adelaar, Pilar Valenzuela Bismarck, & Roberto Zariquiey Biondi (Eds.), Estudios sobre lenguas andinas y amazónicas: Homenaje a Rodolfo Cerrón-Palomino (pp. 305-326). Lima: Fondo Editorial de la Ponti!cia Universidad Católica del Perú. Wise, Mary Ruth. 1999. Small language families and isolates in Peru. In R.M.W. Dixon & Alexandra Y Aikhenvald (Eds.), The Amazonian languages (pp. 307-340). Cambridge: Cambridge Univerity Press. Zavala, Roberto. 1999. External possessor in Oluca Popoluca (Mixean): Applicatives and incorporation of relational terms. In Doris L. Payne & Immanuel Barshi (Eds.), External Possession (pp. 339-373). Amsterdam: John Benjamins. Zuñiga, Fernando. 2006. Deixis and alignment: Inverse systems in indigenous languages of the Americas. Amsterdam/Philadelphia: John Benjamins. Zwicky, Arnold M. 1977. On clitics. Bloomington: Indiana University Linguistic Club. Zwicky, Arnold M., & Geo#rey K. Pullum. 1983. Cliticization vs in"ection. Language, 59, 502-513. 930