AN INVESTIGATION OF VARIOUS LINGUISTIC CHANGES IN CHINESE AND NAXI by JUNG-YAO LU A DISSERTATION Presented to the Department of Linguistics and the Graduate School of the University of Oregon in partial fulfillment of the requirements for the degree of Doctor of Philosophy September 2012 ii DISSERTATION APPROVAL PAGE Student: Jung-yao Lu Title: An Investigation of Various Linguistic Changes in Chinese and Naxi This dissertation has been accepted and approved in partial fulfillment of the requirements for the Doctor of Philosophy degree in the Department of Linguistics by: Dr. Scott C. DeLancey Chairperson Dr. Eric W. Pederson Member Dr. Cynthia M. Vakareliyska Member Dr. Zhuo Jing-Schmidt Outside Member and Kimberly Andrews Espy Vice President for Research & Innovation/Dean of the Graduate School Original approval signatures are on file with the University of Oregon Graduate School. Degree awarded September 2012 iii © 2012 Jung-yao Lu iv DISSERTATION ABSTRACT Jung-yao Lu Doctor of Philosophy Department of Linguistics September 2012 Title: An Investigation of Various Linguistic Changes in Chinese and Naxi This dissertation investigates the diachronic development of Chinese and Naxi, focusing particularly upon six linguistic puzzles that are likely to be associated with the various linguistic changes in most areas of the grammar, including sound/phonological changes, semantic/meaning changes, syntactic/sentence-structure changes, and contact- induced changes. This dissertation’s primarily purpose is to provide new perspectives in order to solve these puzzles on the basis of typological and diachronic evidence. The dissertation will analyze cross-linguistic data from Chinese and Tibeto-Burman languages in order to reconstruct various diachronic developments in Chinese and Naxi. The main body of the dissertation from Chapter II to Chapter V will examine the six linguistic puzzles successively, as follows: (1) tonal splits in proto-checked syllables and subgrouping of Loloish, (2) semantic development of RETURN—还 in Chinese, (3) semantic development of TAKE—把 in Chinese, (4) development of agentive passive markers in Mandarin, (5) definiteness and nominalization, relativization, and genitivization in Chinese, and (6) development of nominalization, relativization, and genitivization in Naxi. v My approach is a rather elaborate attempt to pursue a new framework for comparative reconstruction of historical linguistics. In my study, comparative analysis of historical linguistics focuses on reconstructing ancient patterns based on diachronic records and/or typological data from several languages or dialects in a language group. The ultimate aim of the comparative reconstruction is to demonstrate the historical process of language change. A historical linguist, like a competent detective, must possess acute vision and strong reasoning skills to be able to reconstruct the whole story of language change, and admissible evidence is of upmost importance. In order to discover the solution to the aforementioned linguistic puzzles, the linguist must rely on three key types of clues: typological evidence, historical evidence, and linguistic theories. The basic assumption behind the comparative reconstruction is that the diverse synchronic, linguistic patterns in the same language group were diachronically derived from an identical origin. The common origin of these linguistic differences could be a sound, a meaning, a function word, a syntactic structure, etc., depending on the linguistic field in question. Between the origin and synchronic diversity is a series of diachronic processes. Therefore, the framework of the comparative reconstruction should consist of at least three basic elements: (1) synchronic diversity in a language group, (2) the original pattern or form of diversity, and (3) diachronic processes from the origin to the diversity. vi CURRICULUM VITAE NAME OF AUTHOR: Jung-yao Lu GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED: University of Oregon, Eugene University of Hawaii at Manoa, Honolulu National Cheng Kung University, Tainan, Taiwan DEGREES AWARDED: Doctor of Philosophy, Linguistics, 2012, University of Oregon Master of Arts, Linguistics, 2005, University of Hawaii at Manoa Bachelor of Arts, Chinese Language and Literature, 1997, National Cheng Kung University AREAS OF SPECIAL INTEREST: Historical linguistics PROFESSIONAL EXPERIENCE: Teaching Assistant, Department of Linguistics, University of Oregon, Eugene, 2006-2008 Teaching Assistant, Department of Linguistics, University of Hawaii at Manoa, Honolulu, 2003 vii ACKNOWLEDGMENTS This work could not have been completed without the support of many people. To those people who were willing to spend their time and energy to help me complete this dissertation, my appreciation is beyond words. I would especially like to express my deep appreciation to Scott DeLancey, chair of my committee, for his support of my work over a period of many years. Special thanks go to Eric Pederson, Cynthia Vakareliyska, and Zhuo Jing-Schmidt for serving on my committee and for their comments and probing questions on my research all along the way. Finally, I would like to extend a special note of thanks to my family for their patience, understanding, and willingness to encourage me to go on. viii For Joshua Yichen Lu (吕奕辰) & Yuqing Yang (杨喻清) ix TABLE OF CONTENTS Chapter Page I. INTRODUCTION .................................................................................................... 1 1.1. Methodology ................................................................................................... 2 1.2. Cross-linguistic Data ....................................................................................... 5 1.3. Abstract and Organization .............................................................................. 9 II. SOUND CHANGE: TONAL SPLIT ...................................................................... 14 2.1. Tonal Splits in Chinese *Checked Syllables .................................................. 18 2.2. Previous Studies about Subgroupings of Loloish ........................................... 33 2.2.1. James A. Matisoff and David Bradley ................................................... 34 2.2.2. Kang Chen and Yongsui Li.................................................................... 39 2.3. Tonal Splits in Lolo-Burmese *Checked Syllables ........................................ 45 2.3.1. Tonal Splits in Yi (Lolo) *Checked Syllables ....................................... 46 2.3.2. Tonal Splits in Possible Loloish and Burmish *Checked Syllables ...... 52 2.4. Discussion ....................................................................................................... 62 III. SEMANTIC CHANGE: SCHEMATIC EFFECT ................................................. 68 3.1. The Motion Verb RETURN—还.................................................................... 69 3.1.1. Yeh’s Argument on hai 还 ..................................................................... 70 3.1.2. Yang’s Argument on hai 还 ................................................................... 73 3.1.3. Unsolved Questions on hai 还 ............................................................... 75 3.1.4. Image Schema of RETURN................................................................... 77 3.1.4.1. An Inherent Rotative Concept in RETURN—huan 还 ................ 79 x Chapter Page 3.1.4.2. Metaphorical Extensions of RETURN ......................................... 83 3.1.4.3. Repetitious Sense—“Again” ......................................................... 84 3.1.4.4. Opposite Sense—“Unexpectedly”, “on the Contrary” ................. 84 3.1.4.5. Sequential Sense—“Also/in Addition”, “More”, “Or”, “Next”, “Moreover” .............................................................................. 85 3.1.4.6. Continuant Sense—“Still” ............................................................ 88 3.1.5. Discussion .............................................................................................. 88 3.2. The Disposal Marker BA—把 ........................................................................ 90 3.2.1. Type 1: Thing-transferred Event ............................................................ 104 3.2.2. Type 2: Thing-located Event ................................................................. 106 3.2.3. Type 3: Thing-transformed Event .......................................................... 110 3.2.4. Type 4: Instrumental Causative Event ................................................... 113 3.2.5. Discussion .............................................................................................. 120 IV. SYNTACTIC CHANGE: STRUCTURAL REANALYSIS ................................. 123 4.1. Previous Studies .............................................................................................. 127 4.1.1. Bennett’s (1981) Observations .............................................................. 128 4.1.2. Newman’s (1996) Considerations.......................................................... 129 4.1.3. Wu’s (1999) “Stages of Grammaticalization” ....................................... 130 4.1.4. Xing’s (2006) “Subjectification” ........................................................... 132 4.1.5. Shi & Wang’s (2009) Arguments .......................................................... 134 4.2. Previous Studies: Causative-to-passive Development .................................... 135 4.2.1. Yap & Iwasaki’s (2003) Grammaticalization Pathway ......................... 136 xi Chapter Page 4.3. Questions on Previous Studies ........................................................................ 140 4.4. Development of Agentive Passive Markers .................................................... 142 4.4.1. Three Types of Causativity .................................................................... 143 4.4.1.1. Permissive Causative .................................................................... 145 4.4.1.2. Instrumental Causative.................................................................. 146 4.4.1.3. Manipulative Causative ................................................................ 147 4.4.2. Topicalization ........................................................................................ 148 4.4.3. Omission of Primary Causer .................................................................. 150 4.4.4. Structural Reanalysis ............................................................................. 152 4.5. Discussion ....................................................................................................... 155 V. CONTACT-INDUCED CHANGE: BORROWING .............................................. 157 5.1. Definiteness and Nominalization, Relativization, and Genitivization in Chinese ....................................................................................................... 158 5.1.1. Development of di (底)/de (的) in Mandarin ........................................ 158 5.1.1.1. Development of kai in Chaozhou Southern Min .......................... 163 5.1.1.2. Development of ge3 in Cantonese ................................................ 164 5.1.2. Definiteness............................................................................................ 165 5.1.2.1. The First Definiteness Type: “That + Classifier” ......................... 167 5.1.2.2. The Second Definiteness Type: Phonological Fusion of [That + Classifier] .................................................................... 171 5.1.2.3. The Third Definiteness Type: A Bare “That” ............................... 173 5.1.2.4. The Fourth Definiteness Type: A Bare Classifier......................... 177 xii Chapter Page 5.1.3. Definiteness of Conditional Subordinator, Cleft, and Sentence-final Marker .............................................................................................. 180 5.1.3.1. Definiteness of Conditional Subordinator de hua/的话 ................ 182 5.1.3.2. Definiteness of Sentence-final Marker de..................................... 183 5.1.3.3. Definiteness of Cleft shi~de/是~的 .............................................. 183 5.1.4. Discussion .............................................................................................. 184 5.2. Development of Nominalization, Relativization, and Genitivization in Naxi ...................................................................................................... 185 5.2.1. Nominalization, Relativization, and Genitivization in Possible Loloish ............................................................................................. 185 5.2.2. Migration and Language Contact........................................................... 187 5.2.3. Nominalization, Genitivization, and Relativization in Naxi, Labo, and Mosuo ........................................................................................ 190 5.2.3.1 Nominalization-relativization Syncretism ..................................... 191 5.2.3.2 Contact-induced Change in Naxi ɡə33............................................ 198 5.2.4. Discussion .............................................................................................. 201 VI. CONCLUSION...................................................................................................... 203 APPENDICES ............................................................................................................. 214 A. LIST OF ABBREVIATIONS ........................................................................... 214 B. REFERENCES & RECONSTRUCTIONS FOR *VOICED INITIALS IN CHINESE ................................................................................................ 215 C. REFERENCES & RECONSTRUCTIONS FOR *VOICELESS INITIALS IN CHINESE............................................................................... 216 D. IPA DESCRIPTIONS FOR WU DIALECTS (*VOICED INITIALS)............ 217 E. IPA DESCRIPTIONS FOR WU DIALECTS (*VOICELESS INITIALS)...... 218 xiii F. IPA DESCRIPTIONS FOR HAKKA & MIN DIALECTS (*VOICED INITIALS) .................................................................................................... 219 G. IPA DESCRIPTIONS FOR HAKKA & MIN DIALECTS (*VOICELESS INITIALS) .................................................................................................... 220 H. IPA DESCRIPTIONS FOR MANDARIN DIALECTS (*VOICED INITIALS) .................................................................................................... 221 I. IPA DESCRIPTIONS FOR MANDARIN DIALECTS (*VOICELESS INITIALS) .................................................................................................... 222 J. IPA DESCRIPTIONS FOR CANTONESE DIALECTS (*VOICED INITIALS) .................................................................................................... 223 K. IPA DESCRIPTIONS FOR CANTONESE DIALECTS (*VOICELESS INITIALS) .................................................................................................... 224 L. IPA DESCRIPTIONS FOR MANDARIN DIALECTS, XIANG, AND GAN (*VOIVED INITIALS) ....................................................................... 225 M. IPA DESCRIPTIONS FOR MANDARIN DIALECTS, XIANG, AND GAN (*VOIVELESS INITIALS)................................................................. 226 N. REFERENCES & RECONSTRUCTIONS FOR *VOICED INITIALS IN YI AND LOLOISH ................................................................................. 227 O. REFERENCES & RECONSTRUCTIONS FOR *VOICELESS INITIALS IN YI AND LOLOISH ............................................................... 228 P. IPA DESCRIPTIONS FOR YI DIALECTS (*VOICED INITIALS) ............... 229 Q. IPA DESCRIPTIONS FOR YI DIALECTS (*VOICELESS INITIALS)........ 231 R. REFERENCES FOR *VOICED INITIALS IN NAXI AND MOSUO ............ 232 S. REFERENCES FOR *VOICELESS INITIALS IN NAXI AND MOSUO ...... 233 T. IPA DESCRIPTIONS FOR POSSIBLE LOLOISH & BURMISH (*VOICED INITIALS) ................................................................................. 234 U. IPA DESCRIPTIONS FOR POSSIBLE LOLOISH & BURMISH (*VOICELESS INITIALS) .......................................................................... 236 REFERENCES CITED ................................................................................................ 238 xiv LIST OF FIGURES Figure Page 2.1. From Proto-Lolo-Burmese to Proto-Lahoid ......................................................... 36 2.2. Bradley’s Genetic Relationship between Naxi and *L ......................................... 39 2.3. Tonal Splits in the *Checked Syllable from *LB ................................................. 60 2.4. Tonal Splits in the *Checked Syllable from *C.................................................... 61 2.5. Genetic Distribution Based on Patterns of Tonal Splits ....................................... 63 3.1. The Historical Changes of RETURN (还) ............................................................ 73 3.2. Sweetser’s Image Schema of AGAIN .................................................................. 78 3.3. Prototypical Image Schema of RETURN ............................................................. 79 3.4. Semantic Extensions from the Image Schema of 还 ............................................ 89 3.5. Wu’s Grammaticalization Process of BA Construction ....................................... 91 3.6. Image Schema of the Thing-transferred Event ..................................................... 104 3.7. Image Schema of the Thing-located Event ........................................................... 107 3.8. Image Schema of the Thing-transformed Event ................................................... 110 3.9. Image Schema of the Instrumental Causative Event ............................................ 114 3.10. Image Schema of the Disposal Event ................................................................. 117 3.11. Development of Constructions with TAKE ........................................................ 121 4.1. Causative-to-passive Process of gei, rang, na, and jiao ....................................... 156 5.1. Grammaticalization Pathways of Mandarin di (底)/de (的) ................................. 160 5.2. Grammaticalization Pathways of Chaozhou kai ................................................... 164 5.3. Grammaticalization Pathways of Cantonese ge3 .................................................. 164 5.4. Mono-syllabified Process of [That + Cl] .............................................................. 184 xv LIST OF TABLES Table Page 2.1. Middle Chinese Tone Categories .......................................................................... 19 2.2. Eight-tone System in Chinese ............................................................................... 20 2.3. The *Checked Syllable with Voiced Initial in Dialects of Wu ............................. 22 2.4. The *Checked Syllable with Voiceless Initial in Dialects of Wu ......................... 23 2.5. The *Checked Syllable with Voiced Initial in Dialects of Hakka and Min .......... 24 2.6. The *Checked Syllable with Voiceless Initial in Dialects of Hakka and Min...... 25 2.7. The *Checked Syllable with Voiced Initial in Dialects of Mandarin ................... 26 2.8. The *Checked Syllable with Voiceless Initial in Dialects of Mandarin ............... 27 2.9. The *Checked Syllable with Voiced Initial in Dialects of Mandarin ................... 28 2.10. The *Checked Syllable with Voiceless Initial in Dialects of Mandarin ............. 29 2.11. The *Checked Syllable with Voiced Initial in Dialects of Cantonese ................ 30 2.12. The *Checked Syllable with Voiceless Initial in Dialects of Cantonese ............ 30 2.13. The *Checked Syllable with Voiced Initial in Mandarin, Xiang, & Gan ........... 32 2.14. The *Checked Syllable with Voiceless Initial in Mandarin, Xiang, & Gan ....... 32 2.15. The Short/Long Contrast of *Vowel................................................................... 41 2.16. A Combination of Conditioning Factors............................................................. 42 2.17. Secondary Split in *Short Vowel ........................................................................ 43 2.18. The Voiced/Voiceless Contrast of the Root-initial Consonant ........................... 44 2.19. The *Checked Syllable with Voiced Initial in Yi Dialects ................................. 49 2.20. The *Checked Syllable with Voiceless Initial in Yi Dialects ............................. 50 2.21. The *Checked Syllable with Voiced Initial in Possible Loloish ........................ 54 xvi Table Page 2.22. The *Checked Syllable with Voiceless Initial in Possible Loloish .................... 55 2.23. The *Checked Syllable with Voiced Initial in Possible Burmish ....................... 58 2.24. The *Checked Syllable with Voiceless Initial in Possible Burmish ................... 59 2.25. Matrix of Tonal Split Patterns in Languages ...................................................... 64 3.1. Serial Verb Constructions with the Verb TAKE .................................................. 100 4.1. A Shared Form of Passive /Disposal Markers in Certain Chinese Dialects ......... 125 4.2. Position Class in Causativity of gei, jiao, rang, or na .......................................... 144 4.3. Thematic Roles in Causativity of gei, jiao, rang, or na ........................................ 145 4.4. Position Class after Topicalization of NP3 ........................................................... 149 4.5. Position Class after Omission of NP1 ................................................................... 151 4.6. Position Class in Passive Constructions ............................................................... 153 5.1. Nominalizer, Genitive Marker, and Relativizer in Chinese .................................. 166 5.2. Definiteness in Chinese Genitive Marker, Relativizer, and Nominalizer ............. 168 5.3. Four Types of Constituents for the Expression of Definiteness ........................... 170 5.4. Nominalizer, Genitive Marker, and Relativizer in Possible Loloish Languages ........................................................................................................ 186 5.5. Nominalization-relativization Syncretism ............................................................ 191 xvii LIST OF MAPS Map Page 1.1. Linguistic Groups in Mainland China and Taiwan ............................................... 6 1.2. Tibeto-Burman Languages in Yunnan Province of China .................................... 7 5.1. The Mosuo and the Naxi Regions ......................................................................... 189 1 CHAPTER I INTRODUCTION It is not the purpose of this dissertation to discuss the mechanisms or causes which have been said to explain linguistic changes. The aim is rather to consider the diachronic development of the Chinese language and the Naxi language, focusing particularly upon six debatable linguistic puzzles that are likely to be associated with the various linguistic changes in most areas of the grammar, including sound/phonological changes, semantic/meaning changes, syntactic/sentence-structure changes, and contact- induced changes. These are historical mysteries that need to be resolved. The dissertation‘s primary purpose is to provide new methods and perspectives through which to solve these puzzles on the basis of typological and diachronic evidence. The dissertation will analyze cross-linguistic data from Chinese and the Tibeto-Burman languages in order to reconstruct various diachronic developments in Chinese and Naxi. The main body of the dissertation from Chapter II to Chapter V will examine the six linguistic puzzles successively, as shown in the following (1) to (6): (1) Tonal splits in proto-checked syllables and subgrouping of Loloish, (2) Semantic development of RETURN—还 in Chinese, (3) Semantic development of TAKE—把 in Chinese, (4) Development of agentive passive markers in certain dialects of Chinese, (5) Definiteness and nominalization, relativization, and genitivization in Chinese, and (6) Development of nominalization, relativization, and genitivization in Naxi. 2 I will briefly introduce my methodology in Section 1.1, map the cross-linguistic data in Section 1.2, and provide an abstract for each puzzle and the organization of the dissertation in Section 1.3. 1.1. Methodology My approach is a rather elaborate attempt to pursue a new framework for comparative reconstruction of historical linguistics. In my study, comparative analysis of historical linguistics focuses on reconstructing ancient patterns based on diachronic records and/or typological data from several languages or dialects in a language group. My approach requires significant cross-linguistic observation and theoretical reasoning. The ultimate aim of the comparative reconstruction is to demonstrate the historical process of language change. In most cases, it is more challenging to reveal why a language changed than to simply describe the outcome. Without support from a strong linguistic theory or reliable evidence from historical records, my study would have been impossible. The six linguistic puzzles mentioned above are used as linguistic cases to evaluate the feasibility of my methodology. With my approach, I believe more and more historical linguistic mysteries will be unraveled. A historical linguist, like a competent detective, must possess acute vision and strong reasoning skills to be able to reconstruct the whole story of language change, and admissible evidence is of upmost importance. In order to discover the solution to the aforementioned linguistic puzzles, the linguist must rely on three key types of clues. The first one is typological evidence which is related to the synchronic typological comparison of languages. Fieldwork is necessary if the researched language is undocumented or less-documented. Without any support from historical evidence, the 3 original form must be reconstructed based on synchronic comparative reconstruction in a language group. The second one is concerned with historical evidence. The original form in most languages has been lost in the mists of antiquity. In order to receive more support from historical records, I choose Chinese as a research language. Unlike most languages of the world, Chinese is retained in a large number of historical records written in Chinese characters. The unique writing system of Chinese has preserved plentiful linguistic information through the ages, and it makes diachronic documents readable. Even though their pronunciations may differ across various Chinese dialects, the written forms used in these dialects are normally identical. 1 It is very important to understand that Chinese characters were created after the colloquial/spoken forms. The origins of the colloquial forms are less-known and most of them were conventionalized before the creation of written forms. The Chinese writing system was used to record events and activities in a formal style; therefore, the written forms used in Old Chinese are not necessary the same as the spoken forms. The third type of clue involves applying appropriate theories to the study of historical linguistics. The reference value of linguistic theories to a linguist is like that of criminal psychology or criminology to a detective. It should be noted that applying theories is the easiest way to narrow down possible hypotheses and conclusions, but it is never persuasive enough without evidence. Relying only on theory is like convicting a man of a crime based on criminology alone; fortunately, this never happens in court. A 1 Because they share a very similar writing system, different linguistic groups in China are defined as ―dialects‖ rather than ―languages‖ of Chinese. 4 historical linguist must piece together all the facts and available evidence to build up a theory that is indisputable. The basic assumption behind the comparative reconstruction is that the diverse synchronic, linguistic patterns in the same language group were diachronically derived from an identical origin. The common origin of these linguistic differences could be a sound, a meaning, a function word, a syntactic structure, etc, depending on the linguistic field in question. Between the origin and synchronic diversity is a series of diachronic processes. Therefore, the framework of the comparative reconstruction should consist of at least three basic elements: (1) synchronic diversity in a language group, (2) the original pattern or form of diversity, and (3) diachronic processes from the origin to the diversity. There has been historical evidence of language change. If it is available, a historical linguist must draw upon historical evidence to defend her theory. If it is unavailable, a historical linguist must piece together the whole story of language change from synchronic data of a language group. On the basis of available evidence, the diachronic process from the origin to diversity is the result of theoretical reasoning. A more plausible reconstruction of the origin of synchronic diversity will reveal more possible diachronic processes. With more plausible arguments about diachronic processes, it is easier to explain why synchronic diversity could have happened over time. Literature summary is also an important part of my methodology. Most of the linguistic puzzles discussed in my dissertation are long overdue for a public debate. Linguists have been continually vexed by these puzzles, and have put forward various theories. This dissertation summarizes the findings of existing studies for the purpose of demonstrating different methods and perspectives. Previous studies by others also serve 5 as important sources of linguistic data, including both synchronic and diachronic examples. 1.2. Cross-linguistic data Cross-linguistic data in Chinese dialects and several Tibeto-Burman languages will be presented in this study. All of the Chinese dialects will be investigated. The rough geographic areas in which the dialects are most frequently encountered are provided in Map 1.1. 2 Note that Map 1.1 shows not only Chinese linguistic groups but also non- Chinese linguistic groups. Chinese has been classified into seven mutually unintelligible dialect groups. (1) Mandarin (普通话), spoken by over seventy percent of all Chinese speakers, is used everywhere north of the Yangzi River, as well as in parts of the southwestern provinces of China, including Guangxi, Guizhou, Hunan, Sichuan, and Yunnan. (2) Min (闽语) is mainly spoken in Fujian, Taiwan, southern Zhejiang, and some coastal areas of Guangdong. (3) Hakka (also known as Kejia) (客家话) is spoken in eastern Guangdong and scattered areas of Fujian, Guangxi, Hunan, Jiangxi, Sichuan, and Taiwan. (4) Yue (also known as Cantonese) (粤语) is spoken in Guangdong, Hong Kong, as well as southeastern Guangxi. (5) Gan (赣语) is spoken in the province of Jiangxi. (6) Xiang (湘 语) is spoken in the province of Hunan. (7) Wu (吴语) is spoken in southern Jiangsu and Zhejiang (Norman 2003:72-81 and Thurgood 2003:6). 2 Map 1.1 is downloaded from a website called Wikipedia, a free online encyclopedia. It is obvious that this map is made based on a collection of Chinese-version maps called Zhongguoyuyandituji (中国语言地图集) published by Langwen (朗文) in Hong Kong in 1987 and 1990. 6 Map 1.1. Linguistic groups in mainland China and Taiwan 7 Several Tibeto-Burman languages, such as Tibetan, Burmese, Naxi, Na/Mosuo, Lahu, Hani, Lisu, and Yi, spoken in the province of Yunnan (云南), will be discussed. Yunnan is located in southwestern China (see Map 1.1). The geographic areas of these Tibeto-Burman languages in Yunnan Province are provided in Map 1.2. 3 Map 1.2. Tibeto-Burman languages in Yunnan Province of China 3 Map 1.2 is taken from A Descriptive Grammar of Yongning Na (Mosuo) written by Liberty A. Lidz (2010:5). 8 Among these languages, Naxi and Mosuo must be introduced beforehand because they are less-known and they raise some important issues in Chapter II and Chapter V. The Mosuo and the Naxi people, mostly living in the Lijiang (丽江) Naxi Autonomous County in northwestern Yunnan, have been described in the Chinese historical records under the name ―Moso‖ for centuries; however, neither the Mosuo nor the Naxi speak of themselves as ―Moso‖. Instead, both groups call themselves ―Na‖ (Mathieu 2003:2). The Mosuo and the Naxi have divided by the Jinsha River for centuries. About 210,000 Naxi people are mostly settled on the western banks of the river in Lijiang County, in settlements such as Fengke (奉科), Baoshan (宝山), Yulong Snow Mountain (玉龙雪山), Lunan (鲁南), and Lijiang Ancient Town (shown as pink areas in Map 1.2).4 About 30,000 Mosuo people live on the eastern banks of the river. Mosuo are the great majority inhabitants in the Yongning basin and the Lugu Lake (泸沽湖) regions including Muli (木里), Yongning, and Labo (拉伯) (shown as red areas in Map 1.2).5 In Naxiyu Jianzhi (A grammar of Naxi (纳西语简志)), He and Jiang (1985:104) classify ―Mosuo‖ as the eastern dialect of the Naxi language group, and ―Naxi‖ itself as the western dialect of the language group. Naxi and Mosuo are mutually unintelligible. The Naxi language encompasses three sub-groups, which are partially mutually intelligible: Dayanzhen, Lijiang, and Baoshan (1985:104). The Mosuo language also includes three sub-groups, which are mutually unintelligible: Yongning, Beiqu, and Guabie (1985:107). 4 This number is based on China‘s 2000 census. The nationwide population of Naxi is listed as 308,839, and approximately seventy percent of the Naxi people live in the Lijiang region. 5 This number is based on Mathieu‘s (2003:1) personal report. In China‘s 1990 census, the Yongning-Lugu Lake regions had about 15,000 Mosuo people. Today, their population may be closer to 40,000 nationwide. 9 The genetic classification of Mosuo has not yet been determined, but it is likely that Naxi and Mosuo can be classified in the same language branch. There exist various opinions about the genetic position of Naxi. It has been viewed as unsubgrouped within Tibeto-Burman (posed by Thurgood (2003:20)), excluded from Loloish (posed by Matisoff (1972:8)), included in Loloish (posed by Chen (1993:26)), excluded from Lolo- Burmese (posed by Bradley (1975:93)), and included in Lolo-Burmese (posed by Li (1999:25)). 6 Issues related to the genetic position of Naxi will be discussed in Chapter II. A contact-induced grammatical change that has occurred in Naxi will be explored in Chapter V. 1.3. Abstract and organization Each chapter from II to V focuses on one or two puzzles in one particular area of linguistic change. Chapter II deals with puzzle (1), and focuses on the diachronic phenomenon of tonal splits in checked syllables cross-linguistically. Chapter III deals with puzzles (2) and (3), emphasizing a cognitive effect of image schemas during semantic changes. Chapter IV deals with puzzle (4) and is relevant to a process of structural reanalysis. Chapter V deals with puzzles (5) and (6), discussing a borrowing of a structural feature from a contacted language. Chapter II aims to investigate the relationship between ―tonal splits of checked syllables in Proto-Loloish‖ and ―subgrouping of the Loloish language branch‖. Most of the linguists in the Tibeto-Burman field agree that there is a common phenomenon of the tonal split corresponding to the Proto-Loloish checked syllables conditioned by the 6 In Lidz‘s dissertation entitled A Descriptive Grammar of Yongning Na (Mosuo), the author mentions that the genetic position of Mosuo has been classified by Matisoff (1972), Bradley (1975), Thurgood (2003) and etc. This is misleading. Matisoff (1972), Bradley (1975) and Thurgood (2003) never analyze Mosuo; the language they tried to include in their research is Naxi, not Mosuo. 10 voicing and voicelessness of the root-initial consonants. A general phonological rule of this bipartite contrast is: voiced initial consonants in proto-checked syllables tend to develop lower-pitched tones, while voiceless initial consonants in proto-checked syllables tend to develop higher-pitched tones. Typological evidence shows that the Loloish tonal split corresponding to Proto- Loloish checked syllables did not simply fall into a two-class contrast, which Matisoff has characterized as ―low-checked‖ and ―high-checked‖. Most of the patterns of tonal splits, such as ―regularity‖, ―flip-flop‖, and ―irregularity‖, occurred in different dialects of Chinese and Yi (also known as ‗Lolo‘, a part of Loloish). This fact shows that a language with an irregular pattern of tonal splits in proto-checked syllables can be categorized with patterns of ―regularity‖ and ―flip-flop‖ in the same language group. In addition, it seems that the behavior of the tonal split in checked syllables of the proto- language has nothing to do with the subgrouping of the modern languages because it often fails to establish a more reliable genetic relationship among languages. This is why I conclude that Matisoff‘s methodology as used in The Loloish tonal split revisited (1972) is weak. Chapter III proposes a cognitive principle in terms of ‗schematization‘ on the semantic development of a motion verb RETURN—还 and an action verb TAKE—把 in Chinese. The chapter consists of two sections, Section 3.1 and 3.2. In Section 3.1, I will claim that all of the versatile uses of 还 pronounced as /hai/ in Mandarin, such as ―again (再)‖, ―still (仍然)‖, ―unexpectedly (竟然)‖, ―also (也)‖, ―or (或)‖ and ―more (更加)‖ were conceptually abstracted from the prototypical image schema of RETURN. The meaning of the continuant sense (namely ―still‖) of 还 was not 11 derived from the repetitious sense (namely ―again‖); it originated directly from the rotative concept embedded in the image schema of RETURN. The opposite sense (namely ―unexpectedly‖) of 还 was abstracted from the retraversal trajectory of the image schema of RETURN, in that its direction is opposite of the presupposed, previously traversed path. Section 3.2 is concerned with a diachronic problem related to the most studied construction—so-called BA sentences (把字句 [sentences with ba (把)]) or ‗disposal‘ constructions (处置式) in Chinese. I will argue that the disposal function of 把 (represented as BA) pronounced as /ba/ in Mandarin was derived from an ‗instrumental causative event‘ in which BA functions as a lexical verb meaning ‗to take‘ or ‗to hold‘. In order to understand the semantic development of the BA construction, it is necessary to realize how Chinese speakers have used the verb meaning ―to take‖ or ―to hold‖ (represented as TAKE) from ancient times. The written records of ancient Chinese show that the BA construction in Mandarin originated from four individual and independent serial verb events that involve TAKE; they are ‗thing-transferred event‘, ‗thing-located event‘, ‗thing-transformed event‘, and ‗instrumental causative event‘. In the instrumental causative event, the image schema of the verb TAKE inherently consists of two senses: ―manipulativeness‖ and ―affectedness‖. The most-common grammatical meaning of BA functioning as a marker of the causee is conceptually abstracted from the sense of affectedness. The disposal construction, including the BA construction, is derived from the instrumental causative event due to the omission of the instrument. Chapter IV focuses on a long-term puzzle concerning the agentive passive marker. It aims to provide a new perspective from which to explore the development of 12 the agentive passive marker in certain dialects of Chinese, such as Mandarin and Xiang. In these dialects, speakers use a single morpheme to function as not only the passive/agent marker but also the disposal/patient marker. In my hypothesis, it will be suggested that the development of agentive passive markers, such as gei (给) ‗to give‘, na (拿) ‗to take‘, jiao (教/叫) ‗to call‘, and rang (让) ‗to let‘, is a natural linguistic process. The agentive passive constructions arise from a causative construction, the configuration of which can be demonstrated as: [NP1 primary causer + gei/ na/ jiao/ rang + NP2 secondary causer + V + NP3 causee]. 7 This causative-to-passive process involves three steps: (1) NP3 causee is topicalized to the sentence-initial position, (2) NP1 primary causer is omitted by the speaker, and (3) the final configuration [NP3 causee + gei/ na/ jiao/ rang + NP2 secondary causer + V] is reanalyzed as a passive construction by the listener. Chapter V proposes a possible explanation for the development of nominalization, relativization, and genitivization in Chinese and Naxi. From a synchronic point of view, most of the Chinese dialects and Naxi apply a single morpheme as a genitive marker, a relativizer, and a nominalizer; for example, Beijing Mandarin uses de (的) and Naxi uses ɡə. In order to account for the origin of Naxi ɡə, it is important to first consider and evaluate the pan-Chinese function of nominalization, relativization, and genitivization. This chapter consists of two sections, Sections 5.1 and 5.2. In Section 5.1, typological and historical evidence will show that the functions of nominalization, relativization, and genitivization are united by a common value of definiteness in Chinese; this shared value is the primary explanation for why Chinese 7 NP refers to ‗noun phrase‘ and V refers to ‗verb‘. For an explanation of all abbreviations used in the text, please see Appendix A. 13 speakers use an identical morpheme as a nominalizer, a relativizer, and a genitive marker. Diachronically speaking, Cantonese ge3, Chaozhou Southern Min kai, Beijing Mandarin de, Kunming Mandarin nə, and Taiwanese he were derived from a common construction [that + Cl] in which ―Cl‖ is usually the most general-purpose classifier. Section 5.2 provides a possible explanation for the development of Naxi ɡə. The origin of Naxi ɡə is puzzling because it is clear that Labo Naxi, Mosuo, and Naxi were derived from the same ancestor—Proto-Na; however, they show divergent evolutions in their nominalization, genitivization, and relativization. Naxi ɡə is used as a genitive marker, a relativizer, and a nominalizer, but Labo Naxi and Mosuo use a cognate marker i for nominalization and relativization, and use other different markers na and bɯ33 for genitivization, respectively. According to the historical records and typological dissimilarities, the development of Naxi ɡə is due to language contact with authoritative Chinese language in the forms of Late Medieval, Pre-Modern, and Modern Chinese (from 960 to 1900 A.D.); ɡə was a loanword borrowed from the most general-purpose classifier ɡə (个) in Mandarin. The last chapter, Chapter VI, will not only contain brief concluding remarks for each chapter but also provide research sketches for several less-discussed topics. The present introductory chapter has provided an abstract for each puzzle. The organization of the rest of the dissertation will be provided as follows: Chapter II: ‗sound change: tonal split‘; Chapter III: ‗semantic change: schematic effect‘; Chapter IV: ‗syntactic change: structural reanalysis‘; Chapter V: ‗contact-induced change: borrowing‘; and Chapter VI: ‗conclusion‘. 14 CHAPTER II SOUND CHANGE: TONAL SPLIT Since 2003 I have been working on the languages of the Lolo-Burmese (=Burmese-Lolo; hereafter LB) branch of the Tibeto-Burman (TB) family, especially Naxi (纳西) and Mosuo (摩梭). The genetic relationship between Loloish languages and Naxi or Mosuo is confusing, and has resulted in a long-term debate between Western and Chinese linguists. 8 The Loloish languages of the LB branch, such as Lahu (拉祜), Hani (哈尼), Lisu (傈僳), etc., seem to have two well-split sets of tonal values corresponding to the checked (or stop-finalled) syllables of Proto-Loloish (hereafter *L). 9 The voiced initial consonant of *L checked syllables tend to develop lower-pitched tones in modern Loloish, while the voiceless initial consonant of *L checked syllables tend to develop higher-pitched tones (please see Matisoff 1970, 1971, 1972, and 1973). 10 Matisoff (1972) was the first TB linguist who grouped Loloish based on the above correspondences of Loloish tonal split. David Bradley (1975), who is the author of the book entitled Proto-Loloish (1978), has approved of Matisoff‘s methodology for subgroupings of the Loloish. According to their studies, Naxi has often been excluded 8 In this study, it is assumed that Naxi (纳西) and Mosuo (摩梭) have the same ancestor called ―Proto-Na‖. The ethnic group of Proto-Na could have been a nomadic tribe living in the mountainous plateau of the northwest part of Sichuan Province, China. Naxi and Mosuo are mutually unintelligible in that they have considerable lexical and grammatical differences; however, it could be reasonably assumed that Naxi and Mosuo were derived from an ancient language called ―Proto-Na‖ because Naxi and Mosuo are more similar to each other than either one is to any of the other Tibeto-Burman languages (Jiang 1993:43-48). 9 The checked syllable is a syllable which ends in an unreleased voiceless stop, such as /-p/, /-t/, /-k/, or (in some variants represents a neutralized form) glottal stop /-ʔ/. 10 An asterisk mark [*] in a sentence indicates a proto-form. For example, a ‗*stop-final syllable‘ means a ‗proto-stop-final syllable‘. 15 from the Loloish languages because of its large number of irregular patterns of tonal split corresponding to the checked syllables of *L. 11 However, post 1980s Chinese linguists have contributed to a growing body of literature on tonal splits, reconstruction of proto-tones, and subgroupings of Loloish. Chen (陈康) (1993:24-26) and Li (李永燧) (1996:16 and 2008:53) have proposed on identical grounds that Naxi has a two-way tonal contrast conditioned by a distinction between voiceless versus voiced syllable-initial consonants corresponding to the checked syllables in *L. The higher-pitched tone [55] derived from *L checked syllables with voiceless initial consonants, while the lower-pitched tone [31] derived from *L checked syllables with voiced initial consonants. 12 This leads us to believe that Naxi is certainly a Loloish language since it follows Matisoff‘s pattern of Loloish tonal splits. Given that we now have a great abundance of cross-linguistic data gathered through the dialect surveys available in the Chinese and Loloish languages, how could we still agree with Matisoff‘s monograph of Loloish tonal split without any suspicion about his methodology? In this study, I will do what Matisoff did in the book entitled The Loloish tonal split revisited (1972) using newer data to see if I can achieve the same result. 11 Due to the limited data, Matisoff (1972:5) makes a wrong statement in that Moso (sometimes called Nakhi) tends to involve an initial pre-nasalized stop in all checked syllables. In fact, the initial pre- nasalized stops in Naxi occur not only in the checked syllables but in the non-checked syllables, and the words with an initial pre-nasalized stop are only a small portion of all checked syllables. In addition, it is very important to understand that the initial pre-nasalized stops (such as /mb-/, /nd-/, or /ŋɡ -/) are only used in Naxi (纳西). The Mosuo (摩梭) language does not have any initial pre-nasalized stop in its contemporary sound system. Therefore, the language Matisoff discussed in 1972 is Naxi, not Mosuo, even though he called it ―Moso‖. 12 The pitch of tonal values follows a conventional five-point digital scale first introduced by Y-R Chao (1930:25). The numeral 5 represents the highest pitch and 1 represents the lowest pitch. A short tone is indicated by a single-digit number while a complex tone may be indicated by two or three digital numbers. For instance, [55] and [31] stand for a high-level and a mid-falling tone, respectively. 16 For the purpose of evaluating Matisoff‘s methodology, there are four typological terms that need to be defined. The rule of Matisoff‘s two-way tonal splits is a precondition in which voiced initial consonants of *checked syllables tend to develop lower-pitched tones, while voiceless initial consonants of *checked syllables tend to develop higher-pitched tones. When a language applies this rule in its tone system, it is defined as displaying a pattern of ‗regularity‘. When a language applies a reverse condition of the rule, it is defined as displaying a pattern of ‗flip-flop‘. When a language does not apply this rule or applies another rule (such as a three-way split), it is defined as having a pattern of ‗irregularity‘. When a language maintains an un-split tone from *checked syllables, it is defined as having a pattern of ‗un-split‘. According to Matisoff‘s hypothesis in 1972, different tonal split patterns can reflect the genetic distribution of Lolo-Burmese. The group of ―un-split‖ is distinct from the groups of ―two-way contrast‖ and ―irregularity‖ because it did not undergo tonal splits in *checked syllables. The groups of ―two-way contrast‖ and ―irregularity‖, which have a closer genetic relationship due to a common feature in terms of the diachronic tonal splits, are siblings, equidistant to their common ancestor. The group of ―two-way contrast‖ can be further subdivided into two smaller daughter-level groups: ―regularity‖ and ―flip-flop‖. The typological evidence shows that the Loloish tonal split corresponding to *L checked syllables did NOT simply fall into a two-class contrast, which Matisoff has characterized as ―low-checked‖ and ―high-checked‖. Most of the patterns of tonal splits, such as ―regularity‖, ―flip-flop‖, and ―irregularity‖, occur in different dialects of Chinese 17 and Yi (also known as ‗Lolo‘ which belongs to a part of Loloish).13 This fact shows that a language with an irregular pattern of tonal splits in proto-checked syllables can be categorized with patterns of ―regularity‖ and ―flip-flop‖ in the same language group. In addition, it seems that the behavior of the tonal split in checked syllables of the proto- language has nothing to do with the subgrouping of the modern languages because it often fails to predicate a more reliable genetic relationship among languages. This is why I will conclude that Matisoff‘s methodology as used in The Loloish tonal split revisited is weak. The present chapter is organized as follows. First comes the typological discussion of tone splits in Chinese in Section 2.1, because the development of Chinese tonal splits in *checked syllables, which is supported by historical records, is uncontroversial. Then, some introductory remarks on the methodology of previous studies about the genetic position of Naxi and subgroupings of Loloish or Lolo-Burmese from Matisoff (1971 and 1972), Bradley (1975), Chen (1993 and 1997), and Li (1996, 1999 and 2008) will be provided in Section 2.2. The typological evidence from Yi, and possible evidence from Loloish and Burmish will be provided to evaluate the utility of tone-split patterns for the establishment of genetic relationship among languages in Section 2.3. In Section 2.4, several questions will be discussed based on the typological evidence shown in Sections 2.1 and 2.3. 13 The term ―flip-flop‖, adopted from William S.-Y. Wang (1967:102), refers to the expression reversal of pitch value in which a higher pitch later developed into a lower pitch and vice versa in certain linguistic environments. 18 2.1. Tonal splits in Chinese *checked syllables Before Matisoff hypothesized the mechanism of tonogenesis in 1973, the synchronic phenomenon of the two-way tone contrast in Chinese had been observed for a long time. In Haudricourt‘s (1961:58) article, the author mentioned that as early as 1931, Roman Jakobson has reported the following terms concerning the two-way splitting of the tonal system in Chinese dialects: ―In certain Chinese dialects voiced and voiceless consonants have merged. The phonemic feature of voicing which distinguished one series of consonants from another series is replaced by the phonemic distinction of pitch level in the following vowels: low tone of the vowel is substituted for the voicing of the preceding consonant, high tone on the other hand corresponds to voicelessness of the consonant in question. The difference of pitch level, at first a [non-phonetic] combinatory variation, has become a phonetic feature which distinguishes two series of vowels.‖ Because the tonal development is a diachronic phenomenon, in this section, I will aim to present various diachronic patterns of tonal splits in *checked syllables rather than provide a plain synchronic description of tonal splits for Chinese dialects. These patterns will thus come from reconstructed forms of Archaic and Ancient Chinese. All of the Chinese dialects, including Mandarin, Min, Hakka, Yue, Gan, Xiang, and Wu, will be presented to validate my conclusion. Looking at Chinese historically, there is evidence from the Qieyun rhyme dictionary (切韵) reflecting a stage of Middle Chinese (approximately from 200 to 900 A.D.) during which there were four tonal categories: Ping (平), Shang (上), Qu (去), and Ru (入).14 The checked syllables (ending in -p, -t, and -k) belong to a category traditionally known as ‗entering tone (入)‘ (Tone IV); other non-checked syllables belong 14 Qieyun rhyme dictionary (切韵), published in 601 A.D. during the Sui Dynasty, was edited by Fayan Lu (陸法言). 19 to any of the three other tone categories, called ‗level tone (平)‘ (Tone I), ‗rising tone (上)‘ (Tone II), or ‗departing tone (去)‘ (Tone III), as shown in Table 2.1 (Chen 2000:5). Note that the Roman numerals (I, II, III, and IV) represent the four tones of Middle Chinese. Table 2.1. Middle Chinese tone categories (Chen 2000:5) Tone Traditional Name Gloss I Ping (平) ―level tone‖ II Shang (上) ―rising tone‖ III Qu (去) ―departing tone‖ IV Ru (入) ―entering tone‖ There is also evidence from another rhyme book called Zhongyuan Yinyun (中原 音韵) that reflects three phonological changes from Middle Chinese to Mandarin: (1) the disappearance of final stops, (2) the tonal split of Ping (平) tone (namely ‗level tone‘), and (3) the regrouping of the tones. 15 The entering tone (入声字) with a checked ending - p, -t, or -k in Middle Chinese disappeared when developing into most of the Mandarin dialects. The Ping tone has divided into two groups: Yin-Ping (阴平) and Yang-Ping (阳 平); this phenomenon is known in Chinese as ‗平分阴阳‘. The Ru tone (namely ‗entering tone) has been reassigned to the other three tones in Middle Chinese: Ping, Shang, and Qu; this phenomenon is known in Chinese as ‗入派三声‘. The two-way split of the Ping tone in Middle Chinese was conditioned by the distinction between voiceless and voiced initials. If the initial consonant had been ―voiceless‖ (termed Yin (阴) in Chinese), the syllable now carries a higher-pitched tone. 15 Zhongyuan Yinyun (中原音韵) edited by Qingde Zhou (周清德) was published in 1324 A.D. in the Yuan Dynasty. 20 If the initial consonant had been ―voiced‖ (termed as Yang (阳) in Chinese), the syllable now carries a lower-pitched tone. For instance, in Mandarin the high level tone [55] is correspondent to *Yin-Ping ‗level tone with voiceless initial‘, while the high-rising tone [35] is correspondent to *Yang-Ping ‗level tone with voiced initial‘. Based on this process, Wang (1967:95) proposes that each of the medieval tone categories has split into two contrasting tones; the tone in the Yin subgroup was conditioned by voiceless initials (清声) and the tone in the Yang subgroup was conditioned by voiced initials (浊声). Therefore, the ideally conventional eight-tone system in Chinese can be shown in Table 2.2. Note that the real situation of the development of each tone may involve a merger or a three-way diachronic tonal split. Table 2.2. Eight-tone system in Chinese Tone Traditional Name Gloss I Yin-Ping (阴平) ―level tone with voiceless initial‖ II Yang-Ping (阳平) ―level tone with voiced initial‖ III Yin-Shang (阴上) ―rising tone with voiceless initial‖ IV Yang-Shang (阳上) ―rising tone with voiced initial‖ V Yin-Qu (阴去) ―departing tone with voiceless initial‖ VI Yang-Qu (阳去) ―departing tone with voiced initial‖ VII Yin-Ru (阴入) ―entering tone with voiceless initial‖ VIII Yang-Ru (阳入) ―entering tone with voiced initial‖ The following demonstration will only focus on entering syllables, Yin-Ru (阴入) and Yang-Ru (阳入), in Chinese.16 In order to distinguish more actual situations concerning patterns of tonal split in *checked syllables, three Chinese references are 16 Again, the purpose of this chapter is to evaluate Matisoff‘s methodology concerning subgrouping of Loloish. Since Matisoff only focuses on *checked syllables, it is not necessary to discuss the other *non- checked syllables. 21 scanned for data. (1) Han zi gu jin yin hui (汉字古今音汇—A Pronouncing Dictionary of Chinese Characters in Archaic & Ancient Chinese, Mandarin & Cantonese (Zhou 1973)), edited by Fagao Zhou (周法高), provides the phonetic interpretation of Archaic and Ancient Chinese reconstructed by Bernhard Karlgren (高本汉). (2) Hanyu fangyan cihui (汉語方言詞汇—Lexicon of Chinese Dialects (Yuan 1964)), edited by the Department of Chinese Language and Literature at Peking University (北京大学), provides the common vocabularies for seven Chinese dialects. (3) Jiangsusheng he Shanghaishi fangyan gaikuang (江苏省和上海市方言槪况—A Dialectal Survey in Province of Jiangsu and Shanghai City (1960)) provides lexicon for different dialects, particularly Wu, spoken in the Province of Jiangsu and Shanghai City. 17 There are twenty words used to exemplify each pattern of tonal split in *checked tones. Ten of these had been a checked syllable co-occurring with a voiced initial; they are ‗thin (薄), pull out (拔), butterfly (蝶), wax (腊), deer (鹿), stone (石), honey (蜜), wheat (麦), hot (热), and sun; day (日). The other ten had been a checked syllable co- occurring with a voiceless initial; they are pen (笔), peel (剥), pick (摘), bamboo (竹), color (色), snow (雪), rule (尺), holiday (节), pigeon (鸽), and horn (角). When selecting these words, different places and manners of articulation have been taken into consideration. The reconstruction of the ancient forms for these twenty words done by Karlgren and the source of references are provided in Appendices B and C. The pattern of tonal ‗regularity‘ has been purely preserved in the dialects of Wu. In the Wu dialect of Suzhou (苏州), the *voiced root-initial syllables are under the low- 17 This book was edited by a group called ‗fang yan diao cha zhi dao zu‘ (方言调查指导组). 22 checked tone [23]; the *voiceless root-initial syllables are under the high-checked tone [4]. In the Wu dialect of both Wuxi (无锡) and Shanghai (上海), the *voiced root-initial syllables are under the low-checked tone [2]; the *voiceless root-initial syllables are under the high-checked tone [5]. In the Wu dialect of Wenzhou (温州), the *voiced root- initial syllables are low-rising [12], and the *voiceless root-initial syllables are mid-rising [23]. Please see Tables 2.3 and 2.4 for details. Note that there is a basic division among the Wu dialects: the final stop (-p, -t, and -k) in Proto-Chinese (hereafter *C) have merged to a single manifestation, interpreted as a glottal stop in certain Wu dialects, such as Suzhou, Wuxi, and Shanghai; the checked syllables in other Wu dialects, such as in Wenzhou dialect, have completely lost the final stop and have become an open syllable CV. The detailed phonetic descriptions for those cognate sets provided in Tables 2.3 and 2.4 below are available in Appendices D and E. Table 2.3. The *checked syllable with voiced initial in dialects of Wu Suzhou Wuxi Shanghai Wenzhou 1 thin (薄) 23 2 2 12 2 pull out (拔) 23 2 2 12 3 butterfly (蝶) 23 2 2 12 4 wax (腊) 23 2 2 12 5 deer (鹿) 23 2 2 12 6 stone (石) 23 2 2 12 7 honey (蜜) 23 2 2 12 8 wheat (麦) 23 2 2 12 9 hot (热) 23 2 2 12 10 sun; day (日) 23 2 2 12 23 Table 2.4. The *checked syllable with voiceless initial in dialects of Wu Suzhou Wuxi Shanghai Wenzhou 1 pen (笔) 4 5 5 23 2 peel (剥) 4 5 5 23 3 pick (摘) 4 5 5 23 4 bamboo (竹) 4 5 5 23 5 color (色) 4 5 5 23 6 snow (雪) 4 5 5 23 7 rule (尺) 4 5 5 23 8 holiday (节) 5 5 9 pigeon (鸽) 4 5 5 23 10 horn (角) 4 5 5 23 The pattern of tonal ‗flip-flop‘ can be defined as ―the expression reversal of pitch value for cases in which a higher pitch later developed into a lower pitch and vice versa (Yue-Hashimoto 1986:162)‖. It is of special interest that the tonal flip-flop in Chinese dialects most often occurs with checked syllables (1986:169). The pattern of tonal ‗flip-flop‘ occurs in the dialects of Hakka and Min. In the Hakka dialect of Meixian (梅县), the *voiced root-initial syllables are under the high- checked tone [4]; the *voiceless root-initial syllables are under the low-falling checked tone [21]. 18 In the Min dialect of Xiamen (厦门), the *voiced root-initial syllables are under the high-checked tone [5]; the *voiceless root-initial syllables are under the mid- falling checked tone [32]. In the Min dialect of Chaozhou (潮州), the *voiced root-initial syllables are high-checked tone [4], and the *voiceless root-initial syllables are low- 18 The word /ɲit21/ ‗sun/day‘ in the Hakka dialect of Meixian represents an exception to tonal flip-flop, but this should not affect the statement. 24 falling checked tone [21]. In the Min dialect of Fuzhou (福州), the *voiced root-initial syllables are high-checked tone [4], and the *voiceless root-initial syllables are mid-rising checked tone [23]. Please see Tables 2.5 and 2.6.for details. The development and mechanism of tonal flip-flop in Chinese dialects are puzzling. Although there have been a few articles concerning the linguistic nature of flip- flop, I will not provide any discussion about this issue since it is not relevant to the purpose of the current study. Some of the stop endings (-p, -t, and -k) in Hakka and Min dialects are intact, and some of them have decayed into a glottal stop /-ʔ/. The detailed phonetic descriptions for those cognate sets provided in Tables 2.5 and 2.6 below are available in Appendices F and G. Table 2.5. The *checked syllable with voiced initial in dialects of Hakka and Min Meixian Xiamen Chaozhou Fuzhou 1 thin (薄) 4 5 4 4 2 pull out (拔) 44 5 4 4 3 butterfly (蝶) 5 4 4 4 wax (腊) 4 5 4 4 5 deer (鹿) 4 5 4 4 6 stone (石) 4 5 4 4 7 honey (蜜) 4 5 4 4 8 wheat (麦) 4 5 4 4 9 hot (热) 4 5 4 10 sun; day (日) 21 5 4 4 25 Table 2.6. The *checked syllable with voiceless initial in dialects of Hakka and Min Meixian Xiamen Chaozhou Fuzhou 1 pen (笔) 21 32 21 23 2 peel (剥) 21 32 21 23 3 pick (摘) 21 32 21 23 4 bamboo (竹) 21 32 21 23 5 color (色) 21 32 21 23 6 snow (雪) 21 32 21 23 7 rule (尺) 21 32 21 23 8 holiday (节) 21 32 21 9 pigeon (鸽) 21 32 23 10 horn (角) 21 32 21 23 The voicing-triggered split of checked syllables into two tonal categories is not always symmetrical, especially in the dialects of Mandarin. The pattern of tonal ‗irregularity‘ has been found in many dialects of Mandarin, which exhibit different subtypes in *checked tones (without a two-way distinctive categories). The correspondence between Middle Chinese entering tone and its modern phonetic value in these dialects is not always predictable, particularly in the *voiceless-initial subgroup. However, the entering tone with a *voiced initial has commonly involved a phonological parameter conditioned by the contrast between sonorant (including liquid, nasal, and empty-initial) and obstruent initials (Chen 2000:8). For example, in the Mandarin dialect of Beijing (北京), syllables with a *voiced obstruent initial are under the high-rising tone [35]; syllables with a *sonorant initial are under the sharp falling tone [51]; syllables with a *voiceless initial are scattered unpredictably among all four tonal categories [35], [51], [214], and [55]. In the Mandarin 26 dialect of Shenyang (沈阳), syllables with a *voiced obstruent initial are under the high- rising tone [35]; syllables with a *sonorant initial are under the falling tone [41]; syllables with a *voiceless initial are distributed into the tonal categories [35], [33], and [213] without any clear condition for the split. 19 These varieties of tones in the *voiceless group are evidently not to be accounted for in terms of regular tonal splits. Please see Tables 2.7 and 2.8 for details. Note that all of the *checked syllables have lost their ending stop entirely in Mandarin dialects. The detailed phonetic descriptions for the twenty cognate sets provided in Tables 2.7 and 2.8 below are available in Appendices H and I. Table 2.7. The *checked syllable with voiced initial in dialects of Mandarin Karlgren (高本汉) Beijing Shenyang 1 thin (薄) *buɑk 35 35 2 pull out (拔) *buɑt 35 35 3 butterfly (蝶) *diɛp 35 213 4 wax (腊) *lɑp 51 41 5 deer (鹿) *luk 51 41 6 stone (石) *dziæk 35 35 7 honey (蜜) *miɪt 51 41 8 wheat (麦) *muək 51 41 9 hot (热) *niæt 51 41 10 sun; day (日) *niɪt 51 41 19 The word ‗butterfly‘ with tone [213] in the Mandarin dialect of Shenyang could be an exception. The tone [41] is absent in the Shenyang dialect, but it will probably be found if more data can be gathered. 27 Table 2.8. The *checked syllable with voiceless initial in dialects of Mandarin Karlgren (高本汉) Beijing Shenyang 1 pen (笔) *piuet 214 213 2 peel (剥) *pok 55 33 3 pick (摘) *thiek 55 35 4 bamboo (竹) *tiuk 35 35 5 color (色) *siɪk 51 213 6 snow (雪) *siuæt 214 213 7 rule (尺) *tshiæk 214 213 8 holiday (节) *tsiɛt 35 213 9 pigeon (鸽) *kəp 55 33 10 horn (角) *kok 214 33 Another pattern of tonal ‗irregularity‘ has been found in other dialects of Mandarin. In this pattern, the conditioning factor, the voiced obstruent initial, only functions in the group of *voiced initials; syllables with a *voiceless initial seem maintain unchanged, as shown in Tables 2.9 and 2.10. Therefore, there are two different tonal values in the *voiced-initial group, and only one tonal value in the *voiceless-initial group. For example, in the Mandarin dialect of Xi‘an (西安), syllables with a *voiced obstruent initial go to the mid-rising tone [24]; syllables with a *sonorant initial and with a *voiceless initial merge with the low-falling tone [21]. In the Mandarin dialect of Xuzhou (徐州) and Jinan (济南), syllables with a *voiced obstruent initial go to the high- level tone [55]; syllables with a *sonorant initial and with a *voiceless initial merge with the falling-rising tone [313]. 28 In the Mandarin dialect of Pixian (邳县), the tonal split seems to be conditioned by a three-way contrast between voiceless, sonorant, and voiced obstruent initials without any exception. Syllables with a *voiced obstruent initial are found in the falling tone [42]; syllables with a *sonorant initial are found in the low-falling tone [21], and syllables with a *voiceless initial are found in the falling-rising tone [213]. Note that all *voiced root- initial obstruents have all become voiceless initials in Mandarin dialects. The detailed phonetic descriptions for the twenty cognate sets provided in Tables 2.9 and 2.10 below are available in Appendices H and I. Table 2.9. The *checked syllable with voiced initial in dialects of Mandarin Xi‘an Xuzhou Jinan Pixian Karlgren (高本汉) 1 thin (薄) 24 55 55 42 *buɑk 2 pull out (拔) 24 55 55 42 *buɑt 3 butterfly (蝶) 55 55 42 *diɛp 4 wax (腊) 21 313 313 21 *lɑp 5 deer (鹿) 21 313 313 21 *luk 6 stone (石) 24 55 55 42 *dziæk 7 honey (蜜) 313 313 21 *miɪt 8 wheat (麦) 21 313 313 21 *muək 9 hot (热) 21 313 313 21 *niæt 10 sun; day (日) 21 313 313 21 *niɪt 29 Table 2.10. The *checked syllable with voiceless initial in dialects of Mandarin Xi‘an Xuzhou Jinan Pixian Karlgren (高本汉) 1 pen (笔) 21 313 313 213 *piuet 2 peel (剥) 21 313 313 213 *pok 3 pick (摘) 21 313 313 213 *thiek 4 bamboo (竹) 21 313 313 213 *tiuk 5 color (色) 21 313 313 213 *siɪk 6 snow (雪) 21 313 313 213 *siuæt 7 rule (尺) 21 313 313 213 *tshiæk 8 holiday (节) 21 313 313 213 *tsiɛt 9 pigeon (鸽) 21 313 313 *kəp 10 horn (角) 313 313 213 *kok The last example related to the pattern of tonal ‗irregularity‘ has been found in the dialects of Cantonese, as shown in Tables 2.11 and 2.12. It has been argued that in the Cantonese dialect of Guangzhou (广州), the tonal values [2/22], [33] and [5] in the checked syllables are in complementary distribution. Syllables with a *voiced initial are under the low-level tone [2/22]; syllables with a *voiceless initial have separated and are conditioned by the contrast of vowel length (Yue-Hashimoto 1972:176). Therefore, syllables with a short vowel in the *voiceless-initial group are under the high tone [5], while syllables with a long vowel in the same group are under the mid-level tone [33]. In the Cantonese dialect of Yangjiang (阳江), the situation is more obscure and unpredictable. Syllables with a *voiced initial are under the rising-falling tone [454], but syllables with a *voiceless initial are divided into the rising tone [24] and the low-falling tone [21] without any predictable conditioning factor. 30 Note that Cantonese dialects preserved the stop endings and change the voiced *initial obstruents to their voiceless counterparts. The detailed phonetic descriptions for the twenty cognate sets provided in Tables 2.11 and 2.12 below are available in Appendices J and K. Table 2.11. The *checked syllable with voiced initial in dialects of Cantonese Karlgren (高本汉) Yangjiang Guangzhou 1 thin (薄) *buɑk 454 33 2 Pull out (拔) *buɑt 2 3 butterfly (蝶) *diɛp 454 22 4 wax (腊) *lɑp 454 22 5 deer (鹿) *luk 454 2 6 stone (石) *dziæk 454 22 7 honey (蜜) *miɪt 454 2 8 wheat (麦) *muək 454 2 9 hot (热) *niæt 454 22 10 sun; day (日) *niɪt 454 22 Table 2.12. The *checked syllable with voiceless initial in dialects of Cantonese Karlgren (高本汉) Yangjiang Guangzhou 1 pen (笔) *piuet 24 5 2 peel (剥) *pok 24 33 3 pick (摘) *thiek 33 4 bamboo (竹) *tiuk 24 5 5 color (色) *siɪk 24 5 6 snow (雪) *siuæt 33 7 rule (尺) *tshiæk 24 33 8 holiday (节) *tsiɛt 21 33 31 9 pigeon (鸽) *kəp 21 33 10 horn (角) *kok 21 33 The pattern of ‗un-split‘ in the *checked tone has been found in several dialects; this is the case with dialects of Mandarin, such as those of Yangzhou (扬州), Nanjing (南 京), Chengdu (成都), and Kunming (昆明), Xiang and Gan, as shown in Tables 2.13 and 2.14. Their original checked tones did not experience any split and have merged with the other tone category, usually the Yang-Ping ‗level tone with voiced initial‘. For example, in the Mandarin dialect of Yangzhou, Nanjing, Chengdu, and Kunming, the groups of *voiced-initial and *voiceless-initial share the common tonal value [4], [5], [31], and [31], respectively. In the Xiang dialect of Changsha (长沙), both groups maintain the same pitch value [24]. In the Gan dialect of Nanchang (南昌), both groups preserve the same pitch value [5]. Note that the Mandarin dialects of Yangzhou and Nanjing have no surviving original final stop, and the only syllable-final consonant is a glottal stop -ʔ which was neutralized from *-p, *-t, and *-k. The original final stops in the Mandarin dialects of Chengdu and Kunming and the Xiang dialect of Changsha have disappeared altogether. The Gan dialect of Nanchang has preserved -t and -k from the origin final stops, but *-p has been replaced by -t. The detailed phonetic descriptions for the twenty cognate sets provided in Tables 2.13 and 2.14 below are available in Appendices L and M. 32 Table 2.13. The *checked syllable with voiced initial in Mandarin, Xiang, & Gan Yangzhou Nanjing Chengdu Kunming Changsha Nanchang 1 thin (薄) 4 5 31 31 24 5 2 Pull out (拔) 4 5 31 31 24 5 3 butterfly(蝶) 4 5 31 31 24 5 4 wax (腊) 4 5 31 31 24 5 5 deer (鹿) 4 5 31 31 24 5 6 stone (石) 4 5 31 31 24 5 7 honey (蜜) 4 5 31 24 5 8 wheat (麦) 4 5 31 31 24 5 9 hot (热) 4 5 31 31 24 5 10 sun;day(日) 4 5 31 31 24 5 Table 2.14. The *checked syllable with voiceless initial in Mandarin, Xiang, & Gan Yangzhou Nanjing Chengdu Kunming Changsha Nanchang 1 pen (笔) 4 5 31 31 24 5 2 peel (剥) 4 5 31 31 24 5 3 pick (摘) 4 5 31 31 24 5 4 bamboo(竹) 4 5 31 31 24 5 5 color (色) 4 5 31 31 24 5 6 snow (雪) 4 5 31 31 24 5 7 rule (尺) 4 5 31 31 24 5 8 holiday (节) 4 5 31 31 24 9 pigeon (鸽) 4 5 31 31 24 5 10 horn (角) 4 5 31 31 24 5 33 In summary, based on the discussion above, it is clear that all of the patterns of tonal splits including ―regularity‖, ―flip-flop‖, ―irregularity‖, and ―un-split‖ occur in Chinese. The following sections provide other typological introductions to these tonal split patterns in the Lolo-Burmese language group. 2.2. Previous studies about subgroupings of Loloish The two-way or three-way splitting of the tonal development conditioned by the phonetic nature of the initial consonant in all syllable types seems to have occurred widely across Southeast Asia and the Far East some centuries ago (see Haudricourt (1961)). 20 In Chinese (which was discussed in the previous section), Loloish, Tai, Vietnamese, and Yi, for instance, a syllable came to be pronounced as a lower pitch with a voiced initial than with a voiceless one. Let us start with introductory remarks on the methodology of previous studies about the genetic position of Naxi and subgroupings of Loloish or Lolo-Burmese. I would like to briefly outline the fundamental differences between Western and Chinese linguists‘ hypotheses at the beginning. Western linguists, like Matisoff and Bradley, argue: (1) Naxi should not be included within *L, (2) tonal splits in the *L checked syllable were conditioned by a voiced/voiceless distinction for its syllable-initial consonants, and all of the Loloish languages should share this two-way tonal contrast corresponding to the *L checked syllables, and 20 Syllable types can be classified based on their ending elements, such as open vowel, nasal ending, and obstruent ending. Different languages have different types of syllables and different developments of syllable endings. 34 (3) Naxi does not show this two-way tonal split. 21 Chinese linguists, like Chen and Li, argue: (1) Naxi is certainly a Loloish language (Chen 1993:26), or at least it must be included within Lolo-Burmese (Li 1999:25), (2) a ‗tense tone‘ was developed when the Loloish languages dropped their root- final stopped consonant. Tonal splits from the *tense tone were motivated by qualities of vocalic nucleus, such as ‗long‘ vs. ‗short‘, and/or the distinction between voiceless versus voiced syllable-initial consonants, and (3) Naxi does show a regular pattern of the two-way tonal split corresponding to the *L checked syllables. It appears that the Chinese linguists‘ opinions are opposite those of the Western linguists (at UC-Berkeley). The following section will provide brief introductions to their specific hypotheses. 2.2.1. James A. Matisoff and David Bradley The mechanism of ―tonogenesis‖ in Southeast Asia was developed by Matisoff between 1970 and 1973. 22 Based on his investigation of Lahu and its close relatives in the Loloish group, Matisoff (1972:3-4) assumes that tonogenesis was a compensatory mechanism used to introduce tonal contrasts into the sound systems, during which the 21 Matisoff (1972) calls Naxi ―Moso‖ and Bradley (1975) calls Naxi ―Nahsi‖. No matter what they call Naxi, their studies have nothing to do with Mosuo. 22 I will only focus on Matisoff‘s argument related to subgrouping of Loloish. For more details about tonogenesis, please read the following references: ―Glottal dissimilation and the Lahu high-rising tone (1970)‖; ―The tonal Split in Loloish checked syllables (1971)‖; ―The Loloish tonal split revisited (1972)‖; and ―Tonogenesis in Southeast Asia (1973)‖. 35 rich variety of prefixes preceding the root along with most consonantal contrasts at the end of syllables in the proto-language gradually disappeared. 23 The two-way tonal split in *L checked syllables, which did not occur in other TB languages, was ultimately conditioned by the voiced vs. voiceless contrast of the root- initial consonants. In other words, tones in Loloish arise through the influence of syllable-initial consonants (including ‗stops and affricates‘, ‗spirants‘, ‗nasal‘, and ‗resonants‘ (1972:12-26)); the conditioning factor for the tonogenetic mechanism is that voiced initial syllables tend to join lower-pitched tones and voiceless initial syllables tend to join higher-pitched tones. Based on the documentary support from Written Burmese, which shows no evidence of tonal distinctions in checked syllables at all, Matisoff assumes that there is no tonal contrast in checked syllables of *LB, and that the Loloish branch had separated from *LB due to certain phonetic conditioning factors operating to trigger a two-way tonal split in *L (Matisoff 1972:3). Those languages that keep the two checked tone- classes distinction in checked syllables descending from *L constitute the contemporary Loloish languages (Matisoff 1972:5), as shown in Figure 2.1. Figure 2.1 below shows that *LB, which involved no check tone-contrast, is the parent language of two subgroups: Proto-Burmish (*B) and Proto-Loloid. 24 Neither of these subgroups developed tone-contrast in checked syllables. The Proto-Loloid can be subdivided into three language branches: Proto-Mosoid, Proto-Nasoid, and Proto-Lahoid. 23 Note that the proto-languages of Southeast Asia had three significant types with regard to their syllables with final segments: (1) those ending in an open vowel or nasal, (2) those ending in voiceless spirants *-s, and (3) those ending in some sort of stop (Matisoff 1973:74). Again, I will only focus on the third type in this study. 24 There will be only brief discussion about Burmish in Section 2.3.2, since it does not have a tonal contrast in its *checked syllables. 36 The languages in the Mosoid group, such as Nakhi (spelled ‗Naxi‘ (纳西) in China) and Chiang (spelled ‗Qiang‘ (羌语) in China), show ―no tonal split‖ in checked syllables. It is very important to understand that the category ―no tonal split‖ (in Matisoff‘s terminology) does not reflect the real situation in Naxi or Mosuo. Based on my observation, the more appropriate term would be ―no regular tonal split‖. I will use ―irregularity‖ to replace ―no tonal split‖ in the rest of the chapter. Figure 2.1. From Proto-Lolo-Burmese to Proto-Lahoid (Matisoff 1972:8) Note that Chiang is excluded from the current study. I am uncertain if Chiang shows ―no regular tonal split‖ in its *checked syllables. I guess that Matisoff‘s decision to put Nakhi and Chiang in the same language branch is due to the fact that some Chinese linguists have argued that the Ancient Chiang (古羌人) were the ancestors of the Nakhi (see Fang and He (1979:33-41)). But it is important to know that the Ancient Chiang was a collective name, in that it referred to hundreds of different ethnic groups in the past. 37 The contemporary Chiang is just one of the ethnic groups of the Ancient Chiang. Naxi/Mosuo and Qiang may not actually have a close genetic relationship. 25 In Matisoff‘s analysis, the two-way tonal contrast in *checked syllables can be placed at least as far back as the *L period. Proto-Lahoid and Proto-Nasoid, which inherited the two checked-tone classes from *L, show two kinds of tonal split patterns. The languages (including Lahu, Lisu, Hani, etc.) in the former group apply the regular rule of Matisoff‘s Loloish tonal splits in that the *voiced initial syllables are lower- pitched, while the *voiceless initial syllables are higher-pitched. The languages (including Nasu and the Yi dialect of Luchuan (spelled ‗Luquan‘ in China)) in the latter group apply the tonal flip-flop, so that the *voiceless initial syllables are lower-pitched than the *voiced initial ones. Furthermore, Proto-Mosoid, the sibling language of Proto- Loloish, did not follow the rule; therefore, the two-way tonal contrast in *checked syllables does not occur in Nakhi (Naxi) and Chiang. In addition, Proto-Burmish, which is the sibling language of Proto-Loloid, maintained an un-split tone from *checked syllables. According to Figure 2.1, Proto-Lahoid belongs to the pattern of regularity; Proto- Nasoid belongs to the pattern of flip-flop; Proto-Mosoid belongs to the pattern of irregularity; and Proto-Burmish belongs to the pattern of un-split. It is very obvious that Nakhi (Naxi) in the Mosoid group is coordinate with the Loloish group under the more inclusive Loloid family, but Naxi is definitely excluded from the Loloish branch. In 1975, the genetic relationship between Nahsi (spelled ‗Naxi‘ (纳西) in China) and *LB was discussed again by Bradley when he had almost completed his doctoral 25 Ge and Jiang (1990:71) point out that the percentage of words that are cognates between Mosuo and Qiang is about 26.9 percent when excluding loanwords. 38 thesis entitled ―Lahu dialects and Proto-Loloish‖. Having expertise in reconstructions of forms in *L and *LB, Bradley posed two viewpoints to support the idea that Nahsi should not be included within *L. Looking at the question from the Nahsi point of view: (i) there is a lack of adequate systematic correspondence when Nahsi forms are compared directly with the *L and *LB reconstructions based on their initials and prefixes. The limited examples of shared systematic regularity of correspondence, such as pre-nasalized stops /mb-/, /nd-/, or /ŋɡ-/, may be due to language contact (Bradley 1975:99), and (ii) there is no clear conditioning factor for its tonal splits in *stop-final syllables. It seems clear that the regular rule of Matisoff‘s Loloish tonal splits is not reflected in Nahsi. The distribution of the high-level tone [55] is less predicable or even unpredictable, whether the syllable is stop-final or not (Bradley 1975:100). Thus, there appears to be relatively little genetic relationship between Nahsi and *L. Nahsi may be very close to *LB within *TB, but it is certainly not a Loloish language, and probably not a Burmish language either (Bradley 1975:99). To demonstrate Bradley‘s genetic relationship between Nahsi and *L, it is necessary to define a language called ―Proto-Na‖ (my terminology) which is the parent language of Naxi and Mosuo. As shown in Figure 2.2, the common ancestor of both *LB and *Na can be subdivided into three language branches: Proto-Loloish, Proto-Burmish, and Proto-Na. *LB and *Na are siblings, equidistant from their common ancestor. Therefore, the Naxi language, which was derived from Proto-Na, is excluded from not only Loloish but also Lolo-Burmese. 39 Figure 2.2. Bradley‟s genetic relationship between Naxi and *L 2.2.2. Kang Chen and Yongsui Li The majority of TB linguists who graduated from the Department of Minority Language and Literatures at Minzu University of China (中央民族大学, formerly known as the Central University for Nationalities), including Kang Chen and Yongsui Li, have formed a consensus regarding the mechanism of tonogenesis in Loloish. Most of them agree that the primarily conditioning factor for the tonal split in *L is the tense/lax contrast of vowels. 26 That is, the syllables have developed a ‗tense tone (紧调)‘ when dropping their root-final stopped consonant, while a ‗lax tone (松调)‘ was developed in syllables that formerly had a root-final non-stopped consonant or an open vowel. The *lax tone descending from syllables with a non-stopped final consonant is distinctively lower in pitch than the *tense one, which is derived from syllables with a stopped final consonant. After the first binary tonal split, other conditioning factors, such as the voiceless/voiced contrast in syllable-initial consonants and the distinction between short 26 The phonological contrasts between tense vowels and lax vowels can be principally described based on their different phonation types, namely the tense or lax states of the larynx. The tense vowel usually involves a [+stiff] or [-slack] vocal cords, while the lax vowel usually involves a [+slack] or [-stiff] vocal cords (Shi & Zhou 2005:60-63). Concerning phonetic and acoustic descriptions of tense vowel and lax vowel in Hani, Yi, Jingpho, and Wa, please read Maddieson and Ladefoged‘s (1985:447) article. 40 vs. long vowels, began to trigger the secondary split from the *lax and *tense tone by doubling the number of tones from two to four or more (Chen 1991:27 and Li 1996:3). Much detailed work postulating this two-step tonal split in Loloish has been done by Chen (1986, 1988, 1991, 1993, and 1997) and Li (1992, 1996, 1999, 2000, and 2008). Considering the tonal development of the *L checked syllables (namely the *tense tone in their terminology), there should be a fundamental difference between Chen‘s opinion and Li‘s. Chen argues that the two-way split of the *tense tone was conditioned mainly by the voiced/voiceless contrast of the root-initial consonants (1993:24). However, Li argues that the *tense tone was split through a combination of conditioning factors, either the voiced/voiceless contrast of the root-initial consonants and/or the short/long contrast of the vowels (1996:16 and 2008:53). It is very important to understand that all of Chen and Li‘s research is not focused on the subgrouping of Loloish or Lolo-Burmese. It appears that those languages, which were defined as the Lolo-Burmese family by Li, or as the Loloish branch by Chen, rely completely on their lexical affinity. However, as far as I know, the researchers did not explain their methodologies regarding lexical affinity among those languages and usually did not provide overall cognate sets to support their hypotheses. But it is not difficult to ascertain their attitudes concerning the genetic relationship between Naxi and *L. Chen (1993:26) points out that Naxi is certainly a Loloish language (cf. Matisoff 1972). Li (1999:25) argues that Naxi must be included within Lolo-Burmese (cf. Bradley 1975). Both of Chen and Li agree that Naxi underwent a regular two-way tonal split corresponding to the *checked syllables. 41 In the following, I will introduce a few examples selected from a possible Loloish group based on Li‘s conclusions (1996:16 and 2008:53). Table 2.15 below shows that the two-way tonal split corresponding to the *checked syllables in Hani, Jinuo (基诺), Northern Yi (北彝), Eastern Yi (东彝), Western Yi (西彝), and Southern Yi (南彝) was conditioned by the short/long contrast of the vowel. For example, the long vowel in *checked syllables has developed into a mid- level tone [33], while the short vowel in *checked syllables has developed into a falling tone [31] in modern Hani. It is obvious that, according to Li‘s analysis, the rule of Matisoff‘s Loloish tonal split did not hold true in these languages. However, in the case of Hani, it appears that a reverse pattern exists because a syllable with a short vowel is more likely to involve a higher pitch than one with a long vowel. In addition, based on my observation, Hani (no matter whether spoken in Luchun or Mojiang) could have applied a ‗regular‘ tonal pattern corresponding to its *checked syllables in which its tonal split seems to be conditioned by the rule of Matisoff‘s Loloish tonal split perfectly. The details will be demonstrated when discussing the tonal split of Hani in Section 2.3.2. Table 2.15. The short/long contrast of *vowel Conditioning Factors for the Tonal Splits *Long Vowel *Short Vowel Hani 33 31 Jinno 42 55 Northern Yi 33 55 Eastern Yi 33 13 Western Yi 33 21 Southern Yi 33 21 42 Table 2.16 below shows that the two-step split corresponding to the *checked syllables in Lisu was affected by a combination of conditioning factors. The primary conditioning factor is the short/long contrast of the vowel; the secondary one is the voiced/voiceless contrast of the root-initial consonant. Therefore, the long vowel with a voiceless initial in *checked syllables has developed into a high-rising tone [35]; the long vowel with a voiced initial in *checked syllables has developed into a level tone [44]; the short vowel with a voiceless initial in *checked syllables has developed into a high-level tone [55]; and the short vowel with a voiced initial in *checked syllables has developed into a falling tone [42] in modern Lisu. Note that, based on my observation, Lisu could have applied a ‗regular‘ tonal pattern corresponding to its *checked syllables. The details will be demonstrated when discussing the tonal split of Lisu in Section 2.3.2. Table 2.16. A combination of conditioning factors Conditioning Factors for the Tonal Splits *Long Vowel *Short Vowel *Voiceless initial *Voiced initial *Voiceless initial *Voiced initial Lisu 35 44 55 42 Table 2.17 below shows that the tonal split corresponding to the *checked syllables in Lahu, Central Yi (中彝), and Sani Yi (撒尼彝) involved a combination of conditioning factors. The primary conditioning factor is the short/long contrast of the vowel; the secondary one is the voiced/voiceless contrast of the root-initial consonant, which was only applied to the *short vowel. In Lahu, for example, the long vowel in *checked syllables has developed into a falling tone [54]; the short vowel with a voiceless initial in *checked syllables has developed into a high-rising tone [35]; and the 43 short vowel with a voiced initial in *checked syllables has developed into a low-falling tone [21]. Matisoff (1970:13-44) has posed a possible diachronic explanation for the development of Lahu high-rising tone (the same as tone [35] in Li‘s transcription) in which it descends by ―glottal dissimilation‖ from a ―glottal-incident‖ syllable with a voiced glottalized or voiceless sibilant initial and a final stop in *LB. I will explain Matisoff‘s idea of glottal dissimilation when discussing the tonal split of Lahu in Section 2.3.2. Unlike Li, Matisoff never accepts that the trigger of the tonogenetic phenomena can be related to any feature of vocalic nucleus. It is necessary to point out that Central Yi could have applied an ‗irregular‘ tonal pattern corresponding to its *checked syllables, in that its tonal split seems to be conditioned by a three-way contrast between voiceless, C-prefixed voiced, and non-C- prefixed voiced initials (when applying Bradley‘s reconstruction of *L). The details will be demonstrated when discussing the tonal split of the Central Yi dialect of Nanhua in Section 2.3.1. Table 2.17. Secondary split in *short vowel Conditioning Factors for the Tonal Splits *Long Vowel *Short Vowel *Voiceless initial *Voiced initial Lahu 54 35 21 Central Yi 33 55 21 Sani Yi 44 55 2 Table 2.18 below shows that the two-way split corresponding to the *checked syllables in Naxi was regularly conditioned by the voiced/voiceless contrast of the root- 44 initial consonant. Therefore, the voiceless initial in *checked syllables has developed into a high-level tone [55]; and the voiced initial in *checked syllables has developed into a mid-falling tone [31] in modern Naxi. However, based on my analysis, Naxi could have applied a ‗flip-flop‘ pattern corresponding to its *checked syllables. The details will be outlined when discussing the tonal split of Naxi in Section 2.3.2. Table 2.18. The voiced/voiceless contrast of the root-initial consonant Conditioning Factors for the Tonal Splits *Voiceless initial *Voiced initial Naxi 55 31 So far, I have provided summaries for several previous studies about the theory of tonal splits in Loloish and the genetic identification of the Naxi language. Clearly, on this issue, agreement between Western and Chinese linguists is rare. The Naxi language has been variously influenced in its phonology and syntactic structure by prolonged contact with Chinese; therefore, attempting to subgroup Naxi is not easy to achieve. Although Thurgood (2003:19-20) states that the genetic position of Naxi is still unsubgrouped within Tibeto-Burman, it is reasonable to believe that there are several possibilities for the genetic position of Naxi, which are as follows: (i) Naxi is excluded from Loloish (posed by Matisoff), (ii) Naxi is included in Loloish (posed by Chen), (iii) Naxi is excluded from Lolo-Burmese (posed by Bradley), and (iv) Naxi is included in Lolo-Burmese (posed by Li). 45 Any of the above possibilities could be correct, but I will not make any conclusions regarding the genetic position of Naxi in this study. 27 The following section will focus on my disagreement with Matisoff‘s methodology, which often fails to predicate a more reliable genetic relationship among languages. I will argue that the tonal split in checked syllables of the proto-language may have nothing to do with the subgrouping of the modern languages. Moreover, newer data could prove that Bradley‘s conclusion is less likely to be true. Before doing that, I want to explain why I refrain from discussing Chen and Li‘s research assumptions. There are two reasons: (i) uncertainty regarding their methodologies in subgrouping languages; for instance, I have no idea how they include those languages in the subgroup of Loloish or Lolo-Burmese; 28 and (ii) uncertainty regarding their theoretical understanding of tonal splits; for instance, I am wondering how the distinction between long and short vowel can interact with tone in *checked syllables and whether this process the same as the tonal split in Cantonese. Even though their conclusions about the genetic position of Naxi could be right, it is very difficult for me to agree or disagree with them because of those uncertainties. 2.3. Tonal splits in Lolo-Burmese *checked syllables In this section, the data will show that all of the patterns of tonal splits including ―regularity‖, ―flip-flop‖, ―irregularity‖, and ―un-split‖ occur in Lolo-Burmese. The patterns, such as ―regularity‖, ―flip-flop‖, and ―irregularity‖ are found in Yi and Loloish. 27 The genetic position of Naxi is a perfect topic for the further research. I could probably present my own solution for this when the evidence I need is sufficient. 28 Chinese linguists particularly like to form a ‗consensus,‘ an accumulative achievement of several generations of research, even though those achievements might not be made on the same theoretical ground. 46 Note that the Yi language is a part of Loloish. The Loloish language does not involve the pattern of ―un-split‖ because languages maintaining an un-split tone in their *checked syllables are often categorized as Burmish. The data will also show that Lolo-Burmese, like Chinese, applies not only a two-way but also a three-way tonal split in *checked syllables. In the following, I will explore the tonal split in Yi *checked syllables in Section 2.3.1, and I will observe the tonal split in possible Loloish and Burmish *checked syllables in Section 2.3.2. 2.3.1. Tonal splits in Yi (Lolo) *checked syllables Before investigating the pattern of tonal splits in *L checked syllables, it would be more desirable to observe the behavior of tonal splits in the largest group of the Loloish—Yi (also known as ‗Lolo‘ (彝语)). Historically, ‗Lolo‘ is a derogatory name used by the Chinese to refer to various mountain groups in Sichuan and Yunnan. Several Yi languages that have been reported and classified as Loloish by Bradley (1979), Chen (1993 and 1997), and Matisoff (1972) will be selected and discussed in this section. I hope that eliciting the pattern of tonal splits in Yi *checked syllable can help us understand the pattern of tonal splits in *L checked syllables, as the Yi language is absolutely a part of Loloish. Within the Yi language spoken in China, dialects can be approximately divided into six mutually unintelligible subgroups as follows (Huang 1992:664): (1) ‗Northern Yi‘, spoken by about 1.8 million people in the mountainous region of southern Sichuan and adjoining parts of northern Yunnan and Western Guizhou; (2) ‗Western Yi‘, spoken by about 900,000 people in western Yunnan; (3) ‗Central Yi‘, spoken by about 500,000 people in southern and central Yunnan; (4) ‗Eastern Yi‘ (traditionally called ‗Black 47 Yi‘/‗Black Lolo‘) spoken by about 1.5 million people scattered throughout northeastern Yunnan, southern Sichuan, western Guizhou, and western Guangxi; (5) ‗Southeastern Yi‘, spoken by about 400,000 people in southeastern Yunnan; and (6) ‗Southern Yi‘, spoken by about 800,000 people in southern Yunnan. Note that each dialect of Yi spoken in China has an alternative ―native name‖ which may appear in other literature. 29 For example, the native name used to refer to Northern Yi is ‗Nuosu‘; Western Yi is ‗Lalo‘; Central Yi is ‗Lolopo‘; Southern Yi is ‗Nisu‘; Southeastern Yi is ‗Wusa‘, and Eastern Yi is ‗Nasu‘. My argument would be better supported if data from the six dialects of the Yi languages could be found, but no data from Southern Yi is available to me. Furthermore, other groups called ‗White Lolo‘ or ‗Tonkin Lolo‘, who live in Vietnam (as reported by Bradley (1979:52)), are not included due to the shortage of data at hand. Therefore, I will provide comparative data for five dialects of Yi: the Northern Yi dialect of Xide (喜德); the Western Yi dialect of Weishan (巍山); the Central Yi dialect of Nanhua (南华); the Eastern Yi dialect of Wuding (武定), Luquan (禄劝); and the Southeastern Yi dialect of Sani (撒尼), Mile (弥勒) and Lunan (路南). There are thirty-four words used to exemplify each pattern of tonal splits in *checked syllables for each Yi dialect, twenty-one of which had been a checked syllable with a *voiced initial, as shown in Table 2.19. The additional thirteen words had been a checked syllable with a *voiceless initial, as shown in Table 2.20. The reconstructed proto-forms of these words rely on two sources. The forms of *LB were reconstructed by Matisoff (1972), and the reconstructed forms of *L were done by Bradley (1978), as 29 The native name usually indicates what the different Yi groups call themselves in their mother tongues. 48 shown in Appendices M and N. Because the purpose of this study is to evaluate Matisoff‘s theory, I will apply Matisoff‘s reconstruction in all cases. Bradley‘s reconstruction will be used only when Matisoff‘s reconstructed form is inapplicable or when Bradley‘s reconstruction can bring about a different result. The sources of these words are also provided in Appendices N and O. The IPA transcriptions for the words in Xide, Weishan, Nanhua, Wuding, and Sani are taken from The Tibeto-Burman Lexicon (Huang 1992 (藏缅语族语言词汇)), published by Minzu University of China (中央民族大学). It collects the largest amount of lexical material for Tibeto-Burman languages spoken in China. The IPA transcriptions for the words in Luquan, Mile, and Lunan are taken from Chen‘s 1986, 1988, 1991, 1993, and 1997 articles. I believe their transcriptions, including tonal values, can be trusted because all of the providers had been trained as professional linguistic fieldworkers in the Department of Minority Language and Literatures at Minzu University of China. The detailed phonetic descriptions of the thirty-four cognate sets of the five Yi dialects are available in Appendices P and Q. Note that the names of dialects of Yi mentioned above are abbreviated in Tables 2.19 and 2.20 as follows: WS refers to Weishan; SN refers to Sani; ML refers to Mile; LN refers to Lunan; XD refers to Xide; WD refers to Wuding; LQ refers to Luquan, and NH refers to Nanhua. 49 Table 2.19. The *checked syllable with voiced initial in Yi dialects WS SN ML LN XD WD LQ NH 1 hand 21 2 2 2 55 55 55 21 2 pig 21 2 2 2 55 55 55 21 3 year 21 3 2 2 55 55 55 55 4 enough 21 2 2 2 55 55 55 21 5 six 21 2 2 2 55 55 55 55 6 waist 21 2 2 2 55 55 55 21 7 lick 21 2 2 2 55 55 55 21 8 vomit 21 2 2 2 55 55 55 55 9 needle 21 2 2 2 55 55 55 21 10 sleep 21 2 55 55 21 11 goat 21 2 55 55 55 55 12 bite 21 2 55 55 55 55 13 hungry 21 2 55 55 55 21 14 brain 21 44 2 2 55 55 55 21 15 stand 21 2 2 2 55 55 55 21 16 thirsty 21 2 55 55 55 55 17 eight 21 2 55 55 55 55 18 new 2 55 55 55 19 boil 21 55 55 55 55 20 kill 21 55 55 21 sew 21 2 55 55 21 50 Table 2.20. The *checked syllable with voiceless initial in Yi dialects WS SN ML LN XD WD LQ NH 1 chicken 33 44 33 4 33 2 2 33 2 black 33 44 33 4 33 2 2 33 3 bean 33 44 33 4 33 2 2 33 4 come out 33 44 33 4 33 2 2 33 5 be afraid 33 44 33 4 33 2 2 55 6 shoot 33 44 33 4 33 2 2 33 7 eye 33 44 33 2 33 8 blow 33 44 33 2 33 9 ascend/ go up 33 44 33 4 33 2 2 33 10 cold 33 44 33 2 33 11 sharp 55 44 33 4 33 2 2 33 12 tree 33 44 33 4 33 2 2 33 13 bird 33 33 33 2 33 Based on tonal values shown in Tables 2.19 and 2.20 above, three patterns can be found from these Yi dialects. The pattern of tonal ‗regularity‘ has been preserved in Western Yi and Southeastern Yi. In the Western Yi dialect of Weishan, the *voiced root-initial checked syllables are under the low-falling tone [21]; the *voiceless root-initial checked syllables are under the mid-level tone [33]. In the Southeastern Yi dialect of Sani, Mile and Lunan, the *voiced root-initial checked syllables are under the low tone [2]; the *voiceless root- initial checked syllables are under the middle or high tone [44], [33], and [4], respectively. 30 30 There could be four exceptions here: ‗brain‘ and ‗boil‘ from Sani; ‗sharp‘ from Weishan and ‗bird‘ from Sani. They are small in number and should not affecttonal pattern results. 51 The pattern of tonal ‗flip-flop‟ has been found in Northern Yi and Eastern Yi. In the Northern Yi dialect of Xide, the *voiced root-initial checked syllables are under the high-level tone [55]; the *voiceless root-initial checked syllables are under the mid-level tone [33]. In the Eastern Yi dialect of Wuding and Luquan, the *voiced root-initial checked syllables are under the high-level tone [55]; the *voiceless root-initial checked syllables are under the low tone [2]. The pattern of tonal ‗irregularity‘ occurs in Central Yi. In the Central Yi dialect of Nanhua, checked syllables with a *voiceless initial are found in the mid-level tone [33], but checked syllables with a *voiced initial are divided into the high-level tone [55] and the low-falling tone [21] without any predictable conditioning factor. Note that this result is based on Matisoff‘s reconstruction of *LB. However, when using Bradley‘s (1978) reconstruction of *L, the tonal split in Nanhua seems to be conditioned by the three-way contrast between voiceless, voiced C- prefixed, and voiced non-C-prefixed initials. Therefore, checked syllables with a *voiced C-prefixed initial are found in the high-level tone [55]; checked syllables with a *voiced non-C-prefixed initial are found in the low-falling tone [21], and checked syllables with a *voiceless initial are found in the mid-level tone [33]. 31 Note that *C-prefix is a cover- symbol used by Matisoff (1972:14) which indicates voiced *b, *d, *g, *r, and *l prefixes in *TB. It is amazing that Bradley‘s reconstruction can show the distinction between *voiced C-prefixed initial and *voiced non-C-prefixed initial, which reflects the two-way tonal split of the *voiced group in the Yi dialect of Nanhua. The pattern of tonal ‗un-split‘ has not been found in the Yi language. It is uncertain if Southern Yi, White Lolo or other dialects of Yi could be of this type. 31 The word ‗be afraid‘ in the Central Yi dialect of Nanhua could be an exception. 52 2.3.2. Tonal splits in possible Loloish and Burmish *checked syllables The subgrouping of Loloish languages has not been completely accepted due to several disagreements between Chinese and Western linguists; however, there are four languages which are most likely to be Loloish because they show numerous lexical similarities. They are: (1) Yi (Lolo), Chen‘s main research language, was investigated in the previous section. The Yi language presents three kinds of tonal split patterns in the *checked syllables, including ‗regularity‘, ‗flip-flop‘, and ‗irregularity‘. (2) Lahu, Bradley and Matisoff‘s main research language, is one of the best- documented and most widespread languages in Loloish. (3) Hani, Li‘s main research language, is one of the Akha dialects spoken in Yunnan, China. (4) Lisu, one of the best-recorded languages, has been defined as a Loloish language by Bradley, Chen, Li, Matisoff, and Xu, Mu, and Ge (1986:1). In addition to these four languages, I would like to evaluate Naxi, Mosuo, and Jinuo (基诺) because they have been classified as Loloish languages by several authors. Before Chen, Dai (1981:65) and He & Jiang (1985:3) argued that Naxi is a Loloish language. In He and Jiang‘s Naxiyu Jianzhi (A grammar of Naxi 纳西语简志), it is claimed that the Mosuo language belongs to the Eastern dialect of the Naxi language; in other words, Naxi and Mosuo have a sibling relationship and are derived from the same ancestor. Ge (1982:227 and 1986:2) has pointed out that Jinuo lies within the Loloish branch. In addition, Thurgood (2003:8) has posited that Jinuo belongs to the Central Loloish branch of Loloish. 53 In addition to the seven languages mentioned above, it is necessary to evaluate languages that did not involve tonal splits in their *checked syllables. Matisoff (1972:3) mentions that ―the most striking points of divergence between the Burmish and Loloish branches of the Lolo-Burmese is the fact that Burmese checked (or stop-finalled) syllables are all under the same tone‖. Thus, I will also examine Written Burmese, Modern Burmese, and Achang in this study. The Achang language has been classified as a Burmish language by Dai & Cui (1985:1) and Thurgood (2003:8). In order to maintain consistency, the words that were used to evaluate the Yi dialects in the previous section will be used again for the possible Loloish and Burmish languages in this section. Twenty-one had been a checked syllable with a *voiced initial, as shown in Tables 2.21 and 2.23. The other thirteen words had been a checked syllable with a *voiceless initial, as shown in Tables 2.22 and 2.24. The reconstructed forms for theses words are mainly taken from Matisoff (1972). The phonetic transcriptions for the words from Lahu, Hani (Luchun), Hani (Mojiang), Lisu, Jinuo, Written Burmese, Burmese (Yangon), and Achang are taken from the same source—The Tibeto-Burman Lexicon (Huang 1992). The sources of these words are provided in Appendices N and O. The phonetic transcriptions for the words from Naxi and Mosuo are mainly taken from Naxiyu Jianzhi (A grammar of Naxi (He and Jiang 1986)), The Tibeto-Burman Lexicon (Huang 1992), and A Descriptive Grammar of Yongning Na (Mosuo) (Lidz 2010). The sources of the words from Naxi and Mosuo are provided in Appendices R and S. The detailed phonetic descriptions of the thirty-four cognate sets of these languages are available in Appendices T and U. 54 Table 2.21. The *checked syllable with voiced initial in possible Loloish Lahu Hani (Luchun) Hani (Mojiang) Lisu Jinuo Naxi Mosuo 1 hand 21 31 31 31 55 31 31 2 pig 21 31 31 31 55 31 13 3 year 21 31 31 31 55 13 4 enough 21 31 31 31 55 31 31 5 six 21 31 31 31 44 55 13 6 waist 21 31 31 31 55 55 13 7 lick 21 31 31 31 55 8 vomit 21 55 55 13 9 needle 21 31 31 55 31 13 10 sleep 21 31 31 55 55 13 11 goat 21 31 31 31 55 55 13 12 bite 21 31 31 31 55 55 13 13 hungry 21 31 31 31 55 14 brain 21 31 31 55 55 13 15 stand 35 31 31 31 55 55 13 16 thirsty 35 31 31 55 17 eight 35 31 31 31 44 55 13 18 new 35 31 31 31 55 55 13 19 boil 35 31 31 55 55 55 13 20 kill 31 31 31 55 55 13 21 sew 31 31 31 55 55 Table 2.22. The *checked syllable with voiceless initial in possible Loloish Lahu Hani (Luchun) Hani (Mojiang) Lisu Jinuo Naxi Mosuo 1 chicken 54 33 33 33 42 31 13 2 black 54 33 33 33 42 31 13 3 bean 54 33 33 33 31 31 4 come out 54 33 33 33 42 33 5 be afraid 54 33 33 33 33 33 13 6 shoot 54 33 33 33 42 7 eye 54 33 33 33 42 31 31 8 blow 54 33 33 31 13 9 ascend/go up 54 33 33 33 42 33 33 10 cold 54 33 33 11 sharp 54 33 33 35 42 55 13 12 tree 54 35 44 33 13 bird 54 35 42 Based on tonal values shown in Tables 2.21 and 2.22 above, three patterns can be found in the possible language group of Loloish. The pattern of tonal ‗regularity‘ has been found in Hani, Lisu, and Lahu. In the Hani dialect of both Luchun (绿春) and Mojiang (墨江), the *voiced root-initial checked syllables are under the mid-falling tone [31]; the *voiceless root-initial checked syllables are under the mid-level tone [33]. In Lisu, the *voiced root-initial checked syllables are under the mid-falling tone [31]; the *voiceless root-initial checked syllables are under the mid-level tone [33]. 32 There are a few exceptions with a high-rising tone [35] in the group of *voiceless-initials, 32 The Lisu words ‗brain‘ and ‗boil‘ could be exceptions. These should not affect the result of the type of tonal patterns. 56 as shown in Table 2.22 (no.11-13). Matisoff (1970:15) mentions that the checked syllable with a high tone in Lisu has suffered a further split; it became either [33] or [35], according to whether the *initial consonant was ‗non-aspirated‘ or ‗aspirated and sibilant‘, respectively. Unlike Matisoff, I can not find any possible conditioning factor for the two- way split of [33] and [35] based on the words provided in Table 2.22. The words ‗sharp‘, ‗tree‘, and ‗bird‘ with [35] are still treated as exceptions. In Lahu, the *voiced root-initial checked syllables are under the low-falling tone [21]; the *voiceless root-initial checked syllables are under the high-falling tone [54]. However, the high-rising tone [35] does not follow the regular rule because it occurs with the *voiced root-initial syllable, as shown in Table 2.21 (no.15-19). To explain this irregularity, Matisoff (1970) argues that the Lahu high-rising tone in the *checked syllable had been a causative form, which could have involved a root-initial glottalized consonant. 33 The initial glottal stop /*ʔ-/, which functions as a causative prefix, tends to influence the pitch value in the checked syllable (in which *-p, *-t, or *-k has neutralized into a root-final *-ʔ). In other words, the Lahu high-rising tone arose in a phonetic environment that both began and ended with a glottal stop /ʔ/. Note that this environment ―ʔ_ʔ‖ is called two ―glottal incidents‖ in Matisoff‘s terminology. Therefore, the simplex form with tonal value [21] in the checked syllable became [35] when it was used in the causative context. 34 33 Matisoff (1970:18) points out that causativization was optionally marked by a sibilant prefix reconstructed as *s- or *sə- in a verb-root in *TB; however, the causative prefix in *LB had fused with the initial consonant and had become a glottalized consonant. 34 If Matisoff‘s solution for the Lahu high-rising tone in the *checked syllable is correct, the Lahu word ‗eight‘ with [35] should be an exception. It is hard to believe that the numeral ‗eight‘ can have its causative form at the *L stage. 57 The pattern of tonal ‗flip-flop‟ has been found in Jinuo and Naxi. In Jinuo, the *voiced root-initial checked syllables are under the high-level tone [55] or [44]; the *voiceless root-initial checked syllables are under the mid-falling tone [42]. 35 In Naxi, the *voiced root-initial syllables are under the high-level tone [55]; the *voiceless root-initial checked syllables are under the falling tone [31]. 36 Note that the tonal value of the words ‗come out‘, ‗be afraid‘, and ‗ascend‘ with mid-level tone [33] in the *voiceless-initial group could be a heritage succeeding from their corresponding *causative forms, because all of their reconstructed forms involve a *voiceless root-initial glottalized consonant. In other words, the simplex form with tonal value [31] became [33] when it involved a root- initial glottal stop -ʔ in order to constitute a causative form. Since causativization (in terms of *-ʔ) is an optional causative prefix in the proto-language, the occurrence of tone [33] is a secondary process. The pattern of tonal ‗irregularity‘ has been found in Mosuo. In Mosuo, checked syllables with a *voiced initial are under the mid-rising tone [13]; checked syllables with a *voiceless initial are scattered unpredictably among three tonal categories [13], [31], and [33]. 37 In the final part of this section, it is necessary to demonstrate the ‗un-split‘ pattern in *checked syllables, which has been defined as a fundamental tonal feature of Burmish, as shown in Tables 2.23 and 2.24. 35 In Jinuo, the very small amount of tone [44] in the *voiced-initial group could be a derivative from the original tone value [55]. The words ‗be afraid‘ and ‗tree‘ in the *voiceless-initial group could be exceptions. 36 In Naxi, the words ‗hand‘, ‗pig‘, ‗enough‘ and ‗needle‘ with tone [31] in the *voiced-initial group could be exceptions. The word ‗sharp‘ with tone [55] in the *voiceless-initial group could be an exception, too. These should not affect the result of the type of tonal patterns. 37 In Mosuo, the words ‗hand‘ and ‗enough‘ with tone [31] in the *voiced-initial group could be exceptions. These should not affect the result of the type of tonal patterns. 58 Table 2.23. The *checked syllable with voiced initial in possible Burmish Written Burmese Burmese (Yangon) Achang 1 hand 4 4 55 2 pig 4 4 55 3 year 4 4 55 4 enough 4 4 5 six 4 4 55 6 waist 7 lick 4 4 55 8 vomit 9 needle 4 4 55 10 sleep 4 4 11 goat 4 4 55 12 bite 4 4 13 hungry 55 14 brain 4 4 31 15 stand 4 4 55 16 thirsty 4 4 55 17 eight 4 4 55 18 new 4 4 55 19 boil 4 4 20 kill 4 4 55 21 sew 4 4 55 59 Table 2.24. The *checked syllable with voiceless initial in possible Burmish Written Burmese Burmese (Yangon) Achang 1 chicken 4 4 55 2 black 4 4 55 3 bean 31 4 come out 4 4 55 5 be afraid 4 4 55 6 shoot 4 4 55 7 eye 4 4 55 8 blow 4 4 55 9 ascend/go up 4 4 35 10 cold 55 11 sharp 4 4 55 12 tree 4 4 13 bird 4 4 55 Table 2.23 and Table 2.24 show that Written Burmese, Burmese dialect of Yangon, and Achang do not undergo tonal splits in their *checked syllables. For example, the groups of *voiced-initial and *voiceless-initial share the common tonal value [4] in Written Burmese and Burmese (Yangon). In Achang, both groups maintain the same pitch value [55]. 38 38 In Achang, the word ‗brain‘ with tone [31] in the *voiced-initial group could be an exception. Moreover, the words ‗bean‘ with tone [31] and ‗ascend‘ with tone [35] could be exceptions, too. These should not affect the result of the type of tonal patterns. 60 Based on the discussion in Sections 2.3.1 and 2.3.2, the development of tonal splits in the * checked syllable from *LB to modern languages can be illustrated in Figure 2.3. Comparing Figure 2.1 with Figure 2.3, the crucial difference is that I put Naxi in the subgroup of flip-flop, while Matisoff put it in the subgroup of irregularity. In addition, Mosuo, which was not included in Matisoff‘s study in 1972, is put in the subgroup of irregularity. Figure 2.3. Tonal splits in the *checked syllable from *LB Furthermore, with a better understanding about the way Matisoff subgrouped the Loloish languages in 1972, the development of tonal splits in the * checked syllable of Chinese can be illustrated in Figure 2.4. Figure 2.4 is established based on the discussion provided in Section 2.1. 61 Figure 2.4. Tonal splits in the *checked syllable from *C Figure 2.4 shows that *C, which has not involved any contrast in the *checked tone, is the parent language of two groups. One group includes languages that have never experienced any tone change in their *checked syllables and belong to the ―un-split‖ pattern. The other one includes languages that have experienced tonal splits in their *checked syllables and can be divided into three subgroups in terms of ―regularity‖, ―flip- flop‖, and ―irregularity‖. The groups of ―regularity‖ and ―flip-flop‖ show two-way tonal contrast in *checked syllables, but the group of ―irregularity‖ does not. 62 2.4. Discussion One of Matisoff‘s (1972) great contributions is the way in which his analysis can show four patterns of tonal splits in *checked syllables from several proto-languages within *LB: ―regularity‖, ―flip-flop‖, ―irregularity‖, and ―un-split‖. The pattern of ―un- split‖ maintains one tonal value from its *checked syllables; in other words, it did not involve any tone contrast in *checked syllables. The patterns of ―regularity‖, ―flip-flop‖, and ―irregularity‖ present at least two or more tonal values in their *checked syllables; among these patterns, the patterns of ―regularity‖ and ―flip-flop‖ are restricted to a ―two- way tonal contrast‖ in *checked syllables. Matisoff‘s other great contribution is the genetic distribution of these patterns as shown in a tree diagram. If the methodology Matisoff applied is valid, the only reasonable way to assign the genetic position for languages that apply these four patterns is shown in Figure 2.5. The group of ―un-split‖ is distinct from the groups of ―two-way contrast‖ and ―irregularity‖ because it did not undergo tonal splits in *checked syllables. The groups of ―two-way contrast‖ and ―irregularity‖, which have a closer genetic relationship due to a common feature in terms of the diachronic tonal splits, are siblings equidistant from their common ancestor. This common ancestor, in Matisoff‘s treatment, did not display tonal splits. The group of ―two-way contrast‖ can be further subdivided into two smaller daughter-level groups: ―regularity‖ and ―flip-flop‖, which are placed in the lowest level of the tree diagram. 63 Figure 2.5. Genetic distribution based on patterns of tonal splits Matisoff‘s findings and analyses seem invincible if the patterns of tonal splits in the *checked syllables can really be used as a judgment by which to subgroup languages. However, there are several proofs showing that his methodology is misleading. First, Matisoff‘s method cannot be applied across all languages. For instance, the Chinese languages cannot be subgrouped based on the patterns of tonal splits in *checked syllables. It is very obvious that the dialects of Mandarin are divided into two groups. One group belongs to the pattern of ―irregularity‖ and includes the Mandarin dialects of Beijing, Shenyang, Xi‘an, Xuzhou, Pixian, and Jinan; the other belongs to the pattern of ―un-split‖ and includes the Mandarin dialects of Yangzhou, Nanjing, Chengdu, and Kunming, as shown in Table 2.25 below. Table 2.25 is built based on data discussed in the previous sections. This division of Mandarin is unacceptable because it groups some Mandarin dialects with Yue and others with Xiang and Gan, and we know on other grounds that all Mandarin dialects have a common ancestor that was more recent than their common ancestor with any of the other Sinitic languages. The dialects of Mandarin cannot be placed in two groups that exhibit a less-relevant genetic relationship. In his 64 study, Matisoff did not explain why the universal phonetic mechanism of tonogenesis is unique to Loloish. Table 2.25. Matrix of tonal split patterns in languages Regularity Flip-flop Irregularity Un-split Chinese Wu: Suzhou (苏州); Wuxi (无锡); Shanghai (上海); Wenzhou (温州) Hakka: (Meixian (梅 县)); Min: Xiamen (厦门); Chaozhou (潮 州); Fuzhou (福 州) Yue: (Yangjiang (阳 江); Guangzhou (广州) Mandarin: Beijing (北京); Shenyang (沈阳); Xi‘an (西安); Xuzhou (徐州); Pixian (邳县); Jinan (济南) Xiang: Changsha (长沙); Gan: Nanchang (南昌); Mandarin: Yangzhou (扬州); Nanjing (南京); Chengdu (成都); Kunming (昆明) Loloish Hani Lahu Lisu Yi: Weishan (巍山); Mile (弥勒); Lunan (路南); Sani (撒尼) Jinuo Naxi Yi: Xide (喜德); Wuding (武定); Luquan (禄劝) Mosuo Yi: Nanhua (南华) Burmish Burmese Achang Second, Matisoff‘s method fails to establish a more reliable genetic relationship among languages. For instance, the most problematic situation is found in the Yi dialect of Nanhua, which belongs to the pattern of ―irregularity‖, as shown in Table 2.25 above. Table 2.25 shows that the Central Yi dialect of Nanhua and Mosuo are categorized in the same subgroup. It is very difficult to believe that the genetic relationship between Nanhua and Mosuo is closer than that between Nanhua and Xide, even though the tonal splits in 65 the *checked syllables of the Yi dialects of Nanhua and Xide were two parallel, independent developments. It is also very difficult to believe that the genetic relationship between Nanhua and Mosuo is closer than that between Naxi and Mosuo. Ge and Jiang (1990:67, 73) point out that the percentage of words which are cognates between Naxi and Mosuo is over sixty percent when excluding loanwords. Clearly, Naxi and Mosuo should be put in the same language branch within *LB. Now let us return to the arguments about Naxi posed by Bradley (1975:99). Bradley claims that the source of pre-nasalized stops, such as /mb-/, /nd-/, or /ŋɡ-/, in Naxi may be due to language contact. It is necessary to make clear that the majority of the Naxi dialects do not involve pre-nasalized stops in their consonant systems. Only about ten percent of the Naxi speakers, particularly those living in the Lijiang Ancient Town, use pre-nasalized stops. In other words, the Naxi language is losing its pre- nasalized stops diachronically. In addition, Mosuo, which is a sibling language of Naxi, does not have pre-nasalized stops. Furthermore, it is very difficult to locate the donor that has given those pre- nasalized stops to Naxi. According to Dai (1992:42), the pre-nasalized stops in Naxi correspond to the [ɦ] and [m] initials in Written Tibetan; thus, those pre-nasalized stops in Naxi could be an inherited characteristic rather than a borrowed one. It should be noticed that pre-nasalized stops are also found in the Northern Yi dialect of Xide, and the Eastern Yi dialects of Wuding, Luquan, and Dafang (大方) (Dai 1992:42).39 These dialects of Yi, like Naxi, are subcategorized in the tonal-split group of flip-flop. 39 The Yi dialect of Dafang is spoken in Guizhou Province. 66 This could be a shared diachronic tendency in Tibeto-Burman. The same thing happens in Tibetan languages; most of them are losing or have lost prenasalized stops. This is not surprising. Prenasalized stops are typologically unusual, and languages that have them often lose them. 40 Furthermore, there is no evidence to prove that pre- nasalized stops in Naxi were borrowed because the Mosuo language spoken in the area between Naxi, Eastern Yi, and Northern Yi does not have pre-nasalized stops. Ge and Jiang (1990:73) point out that the percentage of words that are cognates between Mosuo and the Northern Yi dialect of Xide is about forty percent when excluding loanwords; therefore, the genetic relationship between Mosuo, Naxi and Northern Yi could not be too distant. Bradley (1975:100) also points out that the distribution of the high-level tone [55] in Naxi is less predictable or even unpredictable in stop-final syllables. However, according to Tables 2.20 and 2.21 above, the tone [55] in Naxi is predicable because most of the words with tone [55] occur in the *voiced-initial checked syllable. These facts give us little reason to believe Bradley‘s supposition in that Naxi is excluded from the *LB. Let us focus on what the comparative reconstruction in this chapter shows. First of all, the synchronic tonal values corresponding to the *checked tone can be classified into four patterns: ‗regularity‘, ‗flip-flop‘, ‗irregularity‘, and ‗un-split‘. This classification is based on the universal phonetic mechanism of tonogenesis posed by Matisoff. Secondly, the origin form of the *checked syllable was a single tonal value. The supporting evidence can be found in Chinese historical documents. Finally, the diachronic development from the *checked tone to the four synchronic patterns is 40 Information about Tibetan and the universal idea were provided by Scott DeLancey via an in-text comment. 67 uncertain. The languages that exhibit the regular pattern are a very small group among all of the Sino-Tibetan languages. The issues about the patterns of ‗flip-flop‘, ‗irregularity‘ and ‗un-split‘ that are used by the most speakers of the Sino-Tibetan languages are less- discussed. 68 CHAPTER III SEMANTIC CHANGE: SCHEMATIC EFFECT An image schema is a cognitive representation that can reflect why a speaker uses a word or a sentence and how a listener understands it. When an expression is used by a speaker to convey an abstract concept, the image schema‘s conceptual representation reflects the speaker‘s cognition. The image schema also helps the speaker switch meaning from a concrete domain to an abstract domain. If a miscommunication does not occur after the first use of the abstract expression, this means that the listener can process the image schema based on his cognition. Essentially, a successful semantic change requires a shared image schema between the speaker and the listener. This cognitive interaction is the focus of this chapter. Specifically, I will discuss a schematic effect during the semantic change. The image schema of RETURN in Chinese is a typical example that can be used to explain semantic extensions. The motion verb ―to return‖ is not a simple concept because it requires both a retraversal path and an unmarked path-traversed portion. The most interesting question is how a Chinese speaker chooses a sound to express the meaning of ―to return‖. Speakers use the sound /huan/ to indicate ―to return‖ because it can be conceptualized as an image schema which is similar to the refresh symbol . Based on the shared understanding of the schema between the speaker and the listener, the sound /huan/ (written as (还) in Modern Chinese and as (环) meaning ‗a circle‘ in Ancient Chinese) was adopted as the verb ‗to return‘ to profile a returning movement. I will provide a related discussion about the semantic extension from the image schema of RETURN in Section 3.1. 69 The image schema of TAKE in Chinese is unique because it barely occurs on its own. The action verb ―to take‖ normally co-occurs with another verb to constitute a serial verb event. There exist image schemata that can conceptualize several serial verb events with the verb ―to take‖, such as ‗thing-transferred event‘, ‗thing-located event‘, ‗thing- transformed event‘, and ‗instrumental causative event‘. The image schemata of these events will be discussed in Section 3.2. I will claim that the so-called ―disposal construction‖ was derived from the ‗instrumental causative event‘ because they share common senses in terms of ―manipulativeness‖ and ―affectedness‖ in their image schemata. 3.1. The motion verb RETURN—还 The Chinese character 还 in Mandarin has polysemous functions. When used as a lexical verb, it is pronounced /huan/ with a raising tone [35], meaning ―to return‖; in its other non-verbal senses, such as ―again‖ or ―still‖, it is pronounced /hai/ with a raising tone [35]. There are two different bodies of literature concerning the diachronic development of 还 from /huan/ to /hai/. Yeh (1998:236-280) argues that the meaning of ―again‖ for hai was grammaticalized from the meaning of ―to return‖ for huan around the 5-6th century; the meaning of ―still‖ for hai was further derived from the meaning of ―again‖ in the 8th century. However, Yang (2004:207-216) argues that both of the meanings of ―still‖ and ―again‖ for hai were derived from the meaning of ―to return‖ for huan around the 5-6 century. I will briefly introduce the two studies concerning 还 in the succeeding sections, from Sections 3.1.1 to 3.1.2. Several questions about 还 will be 70 posed in Section 3.1.3. My personal hypothesis, which accounts for the image schema of RETURN, will be discussed in Section 3.1.4. 3.1.1. Yeh‟s argument on hai 还 Yeh (1998:242) points out that huan (还) ‗to return‘ was originally a verb, meaning ―to go/come back‖ and its meaning is extended to ―to return an object‖. The Chinese character 还 had begun to involve a motion event indicating ―to return/come back home‖ or ―to return an object‖ by the 3rd century B.C., as shown in (1). (1) 《庄子》 Zhuang Zi 369-286 B.C. (Yeh 1998:242) 相与还而观之 xiang yu huan er guan zhi together with come-back and look it ‗Come back/return together and take a look at it.‘ In addition, Yeh argues that the historical changes of 还 from huan to hai consist of several stages. In the first stage, the sense of repetition—―to go back to a previous situation‖ or ―to go to a previous state again‖—had been fully developed from the original verbal meaning ―to return home‖ by the 6th century (1998:244), as shown in (2). (2) 徐陵《长相思》Chang Xiang Si 507-583 A.D. (Yeh 1998:244) 柳絮飞还聚，游丝断复结 liu xu fei huan ju, you si duan fu jie willow catkins fly again gather, wondering spider-web break again connected ‗The willow catkins float away and come together again; the spider web breaks and is connected again.‘ In the second stage, two grammatical meanings had been separately derived from the repetitious sense (namely ―again‖) by the 8th century (1998:246). One is a textual connective/cohesive marker, meaning ―also‖ or ―moreover‖; the other implies temporal continuation (―still‖) and describes a durative situation internally. Examples are shown in (3) and (4), respectively. 71 (3) 魏征《述怀》 Shu Huai 580-643 A.D. (Yeh 1998:245) 既伤千里目，还惊九折魂 ji shang qian li mu, hai jing jiu zhe hun as-well-as hurt thousand mile eye, also frighten nine turn soul ‗Not just the view of the long trail makes you sad, the winding turns also frighten you.‘ (4) 杜甫《秋兴》 Qiu Xing 712-770 A.D. (Yeh 1998:247) 信宿渔人还泛泛 xin-su yu-ren hai fan-fan stay-two-night fishermen still float ‗The fishermen stayed over two nights; their boats are still anchored on the river.‘ In the final stage, the continuant meaning (namely ―still‖), which functions as a scalar operator, was extended to account for the newer uses in Middle Mandarin around the10th-16th century, such as haishi (还是), the indication of expectation contradiction (namely ―unexpectedly‖), the comparative use, and the moderate sense (1998:249-252).41 Based on Yeh‘s treatment, all of the extended uses derived from the continuant sense can be translated as ―still‖, as shown from (5) to (8). Yeh (1998:250) also mentions that hai in examples (6) and (7) not only expresses a continuing situation, but also indicates an ―expectation contradiction‖.42 Note that the development of hai (还) and haishi (还是) could be independent; therefore, I will not discuss the latter in the study. 41 The morpheme shi (是) is a copula in Chinese. The compound haishi (还是) can express the meaning, such as ―or‖ or ―still‖, depending on the context. Yang (2004:215) mentions that the earliest example of hai (还) meaning ―or‖ is found in Zu Tang Ji《祖堂集》 (952 A.D.), as shown in the following. The example is adopted from Mei (1978, 2000). 只划得这个，还划得那个摩？ Zhi hua de zhe ge, hai hua de nei ge mo? only mow de this Cl., or mow de that Cl. Ques. ‗Will you only mow this one, or mow that one?‘ 42 It is also very odd for Yeh to argue that hai in example (6) can express both a continuing situation and an expectation contradiction, because there are no ancient people alive to judge this sentence. 72 (5) 张先 《一从花》 Yi Cong Hua 990-1078 A.D. (Yeh 1998:249) …黄昏后，又还是，新月帘拢，沉恨细思 huang-hun hou, you hai shi, xin yue lian long, chen hen xi si sun-setting after, again still shi, new moon curtain window, deep regret sensitive thought ‗After sunset, it is still the scene that the new moon is outside the curtain window, I am thinking with deep regret.‘ (6) 《水浒传》 Shui Hu Chuan (Yeh 1998:249) 还敢应口 hai gan ying-kou still dare talk-back ‗Dare you still talk back!‘ (7) 《西游记》 Xi You Ji (Yeh 1998:250) 这场祸比天还大 zhe chang huo bi tian hai da this Cl disaster compare sky hai big ‗This disaster is still bigger than the sky.‘43 (8) 《西游记》 Xi You Ji (Yeh 1998:250) 你还像个人样，那三个丑的断然是怪 ni hai xing ge ren yang, na san ge chou-de duan-ran shi guai you still resemble Cl human look, that three Cl ugly definitely copula monster ‗You still look like a human being, but those three are definitely monsters.‘44 The development of hai (还) can be demonstrated in Figure 3.1 (Yeh 1998:251): 43 It is very odd to gloss hai as ‗still‘ in (7). The more appropriate gloss for this would be ‗more‘. 44 Yeh incorrectly translates guai (怪) as an adjective meaning ‗odd‘. I have made a correction in that guai (怪) should be a noun meaning ‗monster‘. 73 Figure 3.1. The historical changes of RETURN (还) 3.1.2. Yang‟s argument on hai 还 In order to show the diachronic development of grammatical meanings derived from the motion verb huan (还), Yang (2004) uses the following examples to support her arguments. 45 First, the original meaning of huan (还) is ―to return‖, as shown in (9). (9) 《诗 小雅 何人斯》 Shi Jing 1100-771 B.C. (Yang 2004:209) 尔还而入。 er huan er ru you return then enter ‗if you return and enter my house‘ Second, the meanings of ―still‖ and ―again‖ of hai (还) occurred in Qi Ming Yao Shu 《齐民要术》around the 5-6 century, as shown in (10) and (11), respectively. Based 45 The main purpose of Yang‘s paper is to explore the semantic developments of ku (故) in Min dialects. She argues that the adverbial usages of ―still‖ and ―once more‖ of ku (故) in Min dialects came from Old Chinese ku (故) which originally meant ―because‖. In the paper, she also investigates the words fu (复) and huan (还) in Chinese; I only cite examples from this section (2004:207-216). 74 on these findings in Qi Ming Yao Shu, Yang (2004:209) argues that both of the meanings of ―still‖ and ―again‖ for hai (还) were directly derived from the meaning of ―to return‖ for huan (还); there is no logical prerequisite needed when talking about the diachronic development from ―again‖ to ―still‖ in Chinese. (10) 《齐民要术 卷七》Qi Ming Yao Shu Vol.7 534-550 A.D. (Yang 2004:210) 唯以渐加米，还得满瓮。 wei yi jian jia mi, hai de man weng only use gradually add rice, again get full urn ‗Only by adding rice gradually will you get a full urn of wine again.‘ (11) 《齐民要术 卷四》Qi Ming Yao Shu Vol.4 534-550 AD (Yang 2004:210) 阳中者还种阳地，阴中者还种阴地。 yang-zhong-zhe hai zhong yang di, yin-zhong-zhe hai zhong jin di sun-loving-plant still plant sunny place, shade-grown-plant still plant shade place ‗The sun-loving-plant is still planted in a sunny place; the shade-grown-plant is still planted in a shaded place.‘ In addition to the meanings of ―again‖ and ―still‖ of hai, Yang further exemplifies several ‗more-grammaticalized‘ meanings of hai, such as ―more‖, ―selective marker (还 是)‖, and ―unexpected situation (竟然)‖, as shown from (12) to (14), respectively. Note that the function of hai in (12) is the same as that in (7); however, it is clear that Yang and Yeh have different interpretations; Yeh translates it as ―still‖ but Yang uses ―more‖ for the gloss of 还 in (12). (12) ―more‖ (更加) in Mandarin (Yang 2004:214) 张三比李四还啰唆。 Zhang-san bi Li-si hai luo-suo NAME compare NAME more talkative ‗Zhang-san is more talkative than Li-si.‘ It is worth noting that haishi (还是) in example (5) carries the meaning of ―still‖; it differs from haishi (还是) in example (13), which involves a selection in the question sentence. Yeh does not provide any discussion about the latter haishi. 75 (13) selective marker ―or‖ haishi (还是) in Mandarin (Yang 2004:214) 你是去还是不去? ni shi qu hai-shi bu qu you copula go or not go ‗Will you go or not?‘ It should be noticed that Yang‘s ―unexpected sense‖ of hai in (14) is the same as Yeh‘s terminology—‗expectation contradiction‘, as shown in (6) and (7) above. (14) ―unexpected situation‖ (竟然) in Mandarin (Yang 2004:215) 这样的菜还卖得出去！ zhe yang de cai hai mai de chu qu this kind de vegetable hai sell de come out „It is unexpected that this kind of vegetable can be sold out!‘ 3.1.3. Unsolved questions on hai (还) According to Yeh and Yang‘s investigations, it can be concluded that there are at least six grammaticalized meanings of hai (还) derived from the verb huan (还); they are ‗again‘, ‗still‘, ‗also‘, ‗unexpectedly‘, ‗more‘, and ‗or‘, as shown in the following: 1. again (再) (Yeh 1998:244) (Yang 2004:210) 2. still (仍然) (Yeh 1998:247) (Yang 2004:210) 3. also (也) (Yeh 1998:245) 4. unexpectedly (竟然) (Yeh 1998:249) (Ynag 2004:215) 5. more (更加) (Yang 2004:214) 6. or (或) (Yang 2004:214) Yeh‘s discussion does not include the meaning of ―more‖ and ―or‖ for hai, and Yang‘s examples do not involve the meaning of ―also‖ for hai. In addition, there are several points that need to be considered. First, Yeh (1998:250) mentions that hai in (6) and (7) not only expresses a continuing situation, but also indicates an ―expectation contradiction‖; however, it is obvious that the use of the continuant sense of hai—―still‖ in (6) and (7) may be inappropriate. In addition, it is unnecessary for the continuant sense and the unexpected 76 sense of hai to coexist in a sentence. Obviously, we do not have access to ancient people who could judge these sentences. However, in my opinion, example (6) may only carry the sense of ―expectation contradiction‖, while example (7) may only carry the sense of ―more‖. The sense of ―expectation contradiction‖ may not originate from the continuant sense of hai (cf. Yeh 1998:251). Second, although Yang has pointed out that the meaning of ―still‖ for hai was not derived from the meaning of ―again‖ for hai because both of the meanings occurred in Qi Ming Yao Shu 《齐民要术》 around the 5-6 century. It would be more sound to apply theoretical evidence to prove that both of the meanings of ―still‖ and ―again‖ for hai (还) were derived from the meaning of ―to return‖ for huan (还). Third, both Yang and Yeh provide a text-based explanation for the historical development of the grammaticalized meanings of hai (还), such as ―still‖, ―again‖, ―also‖, ―unexpectedly‖, ―more‖, and ―or‖. However, they stand for different arguments based on the different classical documents from Ancient Chinese. The sequence of the occurrence of hai in classical Chinese might not correspond to the process of grammaticalization of hai. The way in which these meaning shifts developed through the ages is still unclear. From the following sections, I will try to apply the theoretical concept of image schema as a possible solution to the questions. I will argue that all of the versatile meanings of hai (还) mentioned by Yeh (1998) and Yang (2004), such as ―to return‖, ―again‖, ―still‖, ―also‖, ―unexpectedly‖, ―more‖, and ―or‖, originate from the prototypical image schema of RETURN (see Section 3.1.4). The etymological meaning of huan (还) ―to return‖ involves an inherent circulative or rotative concept (see Section 3.1.4.1). All of the meaning shifts from huan 还 can be abstracted via metaphorical extensions from 77 the source of the repetitious sense, opposite sense, sequential sense, or continuant sense embedded in the prototypical image schema of RETURN (see Sections 3.1.4.2-3.1.4.6). 3.1.4. Image schema of RETURN An image schema, which can reveal how the world is perceived, is a simple and abstract cognitive structure derived from our daily experiences (Lakoff 1987:275). Our understanding of motion events is based on an abstract image schema which originally includes a starting point, or ―source‖, a given trajectory, or ―path‖, an endpoint, or ―goal‖, and a certain ―direction‖ as its structural elements. The image schema provides a conceptual basis for the metaphorical mapping process; through the conceptual metaphor, a schematic mapping from the source concept onto another target concept is generated and a new metaphorical meaning will thus be derived. Before depicting an image schema for RETURN, I would like to review a part of Sweetser‘s (1988:398-399) article concerning the image-schematic structure of AGAIN in modern English, which is relevant to that of RETURN. She proposes that there is a diachronic development of the current sense of AGAIN that can account for its metaphorical extensions. First, she takes Traugott‘s (1982:250-251) examples and provides a linear semantic development for AGAIN, as shown in the following: ―against, facing, opposite to‖ ―in response to, reply‖ ―return an object‖ ―repetition, again‖. Second, she suggests that ―an image-schematic treatment can readily bring out regularities in this development (1988:398)‖; in other words, all the senses of AGAIN, including opposing, replying, returning an object, and repetition, share an identical image-schematic structure. 78 Finally, she provides an image-schematic structure of AGAIN, as shown in Figure 3.2, and use the verb ―to reply‖ to map into this image schema. She mentions that ―the relevant image-schematic structure of a reply is traversal of a path between Speaker and Hearer, with the presupposition of previous traversal of a path from Hearer to Speaker‖. In this sequence, the action of reply ―carries with it the understanding (shown in Figure 3.2 as a dotted line) that a previous utterance has gone the opposite direction in the speech world‖; in addition, the motion event of reply (the solid line) is ―aligned relative to that previous utterance‖. Figure 3.2. Sweetser‟s image schema of AGAIN Following Sweetser‘s assumption, it could be suggested that the prototypical image schema of the motion verb RETURN should be similar to her image-schematic structure of AGAIN, as shown in Figure 3.3. In Figure 3.3, the circle A and B indicate source and goal, respectively. A previously traversed path (from A to B) is represented by a dotted line, and the retraversal path (from B to A) is depicted as a solid line. The profiled potion of RETURN is the returning movement including source and the retraversal path in the motion event. The path-traversed portion and the goal, which carry the presupposition of the previous motion event, are usually unmarked in the speech act. 79 Figure 3.3. Prototypical image schema of RETURN ―RETURN‖ A: SOURCE B: GOAL The etymological meaning of huan (还) in Chinese is a combined concept which employs the prototypical image schema of RETURN. Mei (1992:328) mentions that huan (还) expresses a concept that is identical to the Chinese idioms ―zhou er fu shi (周 而复始)‖ and ―wu ji bi fan (物极必反)‖. The phrase ―zhou er fu shi‖ means ―to make a circle and start again from the original starting point, then repeat this process again and again‖. The phrase ―wu ji bi fan‖ means ―a thing turns into its opposite when it reaches the extreme‖. The former idiom implies the sense of continuance and repetition; the latter implies the sense of opposition and sequence. Based on Mei‘s explanation, it can be understood that all of the senses (including the repetitious sense, opposite sense, sequential sense, and continuant sense) abstracted from the prototypical image schema of RETURN are conceptually embedded in the original meaning of huan (还) in Chinese. 3.1.4.1. An inherent rotative concept in RETURN—huan (还) The inherently conceptual understanding of RETURN—huan (还) in Chinese crucially involves the concept of circulation or rotation. Evidence can be found from Chinese characters and Taiwanese, as follows: First, concerning the traditional Chinese character huan written as 還, the left- side of 還 is ―辶‖ indicating a movement; the right-side of 還 is ―睘‖ meaning a ‗circle‘. 80 Therefore, huan is used to indicate a turn-round action. Considering that most of the Chinese characters involving ―睘‖, such as 環 ‗jade ring‘, 繯 ‗noose‘, 鐶 ‗metal ring‘, and 圜 ‗circle‘, have a circulative sense; these words could be paronymous (Mei 1992:328). Second, another word indicating the meaning of RETURN is hui (回). The Chinese character hui (回) ―to return‖, 回 was one of the hieroglyphs that depicts the rotative shape of vortices. It is possible that two circles rather than two squares were intended to be drawn, but circles were more difficult to carve on the oracle bones and bamboo strips. An example of hui ―to return (home)‖ in Mandarin is shown in (15a). Based on its etymology, it is possible that the Chinese character hui involves a sense of rotation. (15a) RETURN in Mandarin Ta hui jia qu le. 他 回 家 去 了 3S return home go Asp ‗He returned home.‘ It should be noticed that the other counterpart used to express the meaning of ‗to return‘ in Chinese is fan (返) ‗to return‘ which is synonymous with huan (还) ‗to return‘, as shown in (15b). The etymology of fan ‗to return‘ is a noun, fan (反), meaning a ‗reverse side‘. Its image schema is very similar to the recessive symbol ← which only profiles an opposite sense. (15b) RETURN in Mandarin Ta fan xiang qu le. 他 返 乡 去 了 3S return hometown go Asp ‗He returned to his hometown.‘ 81 Third, the word used for ―to return (home)‖ in Taiwanese is dəŋ (转). Example (16a) shows that the original meaning of dəŋ is ‗to rotate‘. The word dəŋ can be used in (16b) to mean ‗to return‘. This shows that the meaning of RETURN in Taiwanese involves the concept of rotation. (16a) dəŋ ‗to rotate‘ in Taiwanese giu de dəŋ ball Prog rotate ‗the ball is rotating‘ (16b) dəŋ ‗to return‘ in Taiwanese i dəŋ khi tshu ɑ. 伊 转 去 厝 啊 3S rotate go home Prt ‗He returned home.‘ The way we understand the recycling symbol on the recycle bin is very similar to what Chinese people would understand from the image schema of RETURN in Figure 3.3 above. There are two major temporal senses that can be schematized from the image schema of RETURN—huan (还): (1) sequential sense: event A follows event B to constitute a complete movement, and (2) continuant sense: the same movement (or state) occurs again and again. I will exemplify these temporal senses in Sections 3.1.4.5 and 3.1.4.6, respectively. Supporting evidence can be found in a Tibeto-Burman language. Mei (1992:328) mentions that there are two cognate Tibetan words—'khor/skor and gor/sgor—which originated from the same etymon meaning ―circle‖, ―to return‖, and ―to turn round‖, as shown in (17). Note that the prefix s- in skor and sgor denotes the syntactic functions, such as causativization, denominative, or directive (Mei 1992:328). There could be a 82 parallel development of grammaticalization between the morphemes kor/gor in Tibetan and huan in Chinese, because both involve the concept of rotation. (17) Tibetan (Mei 1992:328) 'khor circle, circumference 'khor-ba to turn round, to go around in a circle skor circle, repetition skor-ba to surround, encircle, to return skyor-ba to repeat, enclosure, fence gor-gor round, circle gor-ma round, circle sgor to turn on a lathe sgor-ma round, a circle, a globe In addition to Tibetan, the rotative sense of RETURN can also be found in other Tibeto-Burman languages, such as Labo Naxi and Labo Mosuo. 46 In Labo Mosuo and Labo Naxi, le 33 , le 13 , or le 33 u 13 is a lexical verb meaning ―to return‖ or ―to turn around‖.47 Examples from Labo Mosuo are provided in (18) and (19), respectively. (18) tʂhɯ33lɑ31 o33bu33ne33 le33u13 xə31. 3P oneself return go (Perf) ‗They returned (home) by themselves.‘ (19) ɲɑ33 u31ly33 le33u13 li33, tʂhɯ33 thi33 mə33 dʝu31 ze31. 1S head turn around look 3S Pref Neg exist Prt ‗When turning around my head and looking, (I found) he had disappeared.‘ From an etymological perspective, the lexical verb *le 33 in Proto-Na has its original concept meaning ―to turn around‖ or ―to turn to a different direction‖; it was usually used in a specific context where a person was turning around a horse when riding on its back. In Labo Mosuo and Labo Naxi, when speakers ―le33‖a horse, they turn the 46 Labo (拉伯) is an isolated mountainous area where the Naxi people and the Mosuo people are in contact. The language spoken by the Naxi people in Labo is defined as Labo Naxi; when spoken by the Mosuo people, it is defined as Labo Mosuo. 47 The lexical compound le33u33 ―to return‖ in Labo Naxi or le33u13 in Labo Mosuo was a suffixation derived from the root le33 ―to turn around‖. It is not clear when the compound le33u33 was created; the suffix u33 meaning ―self‖ was attached to le33 to form a more precise meaning of RETURN in that u33 ―self‖ indicates the previous position mapping onto the ―source‖ of the image schema of RETURN. 83 horse in the opposite direction; meanwhile, they feel a physical rotation when riding on the horse‘s back.48 In what follows, I will show how metaphorical extension occurs via the image-schematic structure of RETURN in Mandarin, Labo Mosuo, and Labo Naxi. 3.1.4.2. Metaphorical extensions of RETURN The image-schematic structure of RETURN provides the conceptual foundation for all of its meaning shifts. Starting with a cognitive basis, humans tend to understand the image schema of RETURN in terms of ―Figure‖, ―Ground‖, and the figure‘s ―trajectory‖ in the spatial movement. The trajectory of the motion verb RETURN involves a retraversal path; and the Figure, a moving object, moves toward the Ground, referring to the previous location of starting point. Given the assumption that metaphorical mapping of the image schema of RETURN can structure semantic changes, it can be concluded that the meaning of ―again‖ is derived from its repetitious sense (see Section 3.1.4.3), the meaning of ―on the contrary‖ or ―unexpectedly‖ is derived from its opposite sense (see Section 3.1.4.4), the meaning of ―next‖, ―also‖, ―or‖, ―moreover‖, or ―more‖ is derived from its sequential sense (see Section 3.1.4.5), and the meaning of ―still‖ is derived from its continuant/rotative sense (see Section 3.1.4.6). I will use both Mandarin hai (还) and Labo Mosuo and Labo Naxi le 33 to exemplify these points in the following sections. 48 I received the information about the etymology of *le33 in Proto-Na via a personal conversation with Jiren He (和即仁) in 2003. He is one of the authors of Naxiyu Jianzhi (A grammar of Naxi 纳西语简志 (1985)). 84 3.1.4.3. Repetitious sense—“again” The repetitious sense (namely ―again‖) in the image schema of RETURN is abstracted from the re-arrival at a previous location of starting point. The newly derived meaning has replaced the sense of a physical motion with a sense of repetition, and starts to carry the sense of ―a return to the previous location/status‖. An example from Mandarin is shown in (20). (20) hai ―again‖ in Mandarin 他明年还要来。 ta ming-nian hai yao lai he next year again want come ‗He wants to come again next year.‘ In Labo Mosuo, the new meaning can also carry the sense of ―redoing the ever- done action‖, as shown in (21). (21) le 13 ―again‖ in Labo Mosuo tʂhɯ33 le13 le33 mu33sɯ33 ze33. 3S again Pref sleep Prt ‗He falls asleep again.‘ 3.1.4.4. Opposite sense—“unexpectedly”, “on the contrary” The opposite sense (namely ―on the contrary‖, or ―unexpectedly‖) is abstracted from the retraversal trajectory of RETURN in that its direction is opposite of the presupposed previously traversed path in the image schema of RETURN. When mapping the opposite sense onto the target domain, the new meaning, which is similar to ―on the contrary‖, usually indicates an unexpected, contradictory, or contrary condition. An example of hai from Mandarin is shown in (22). 85 (22) hai ―unexpectedly‖ in Mandarin 他信用差还能借到钱! ta xin-yong cha hai neng jie-dao qian he credit bad unexpectedly can borrow money ‗It is unexpected that he can borrow money even though his credit is bad.‘ In Labo Naxi, when le 33 occurs in sentences, such as (23) and (24), it indicates a contradictory condition in which the reality is in opposition to the speaker‘s presupposition. In (23), le 33 ―on the contrary‖ implies that the teacher only teaches a few things, but gives many written assignments. In (24), le 33 ―on the contrary‖ implies that the eldest son is supposed to come rather than the second-born son. (23) le 33 ―on the contrary‖ in Labo Naxi so 33 i 33 mbi 31 mə33 tʂɯ33 , pər33 i33 le33 mə33 nɯ33. learn Nml many Neg count write Nml on the contrary Neg few ‗There are only a few teachings; however, there are many writing assignments.‘ (24) le 33 ―on the contrary‖ in Labo Naxi tʂhɯ33 nɑ31 zo33 ndɯ33 mə33 tshɯ31; zo33 ly31 le33 3S Gen son big Neg come (Perf) son secondary on the contrary pɑ55 tshɯ31 iɑ55. arrive come (Perf) Prt ‗His eldest son did not come; on the contrary, the second-born son came.‘ 3.1.4.5. Sequential sense—“also/in addition”, “more”, “or”, “next”, “moreover” The sequential sense (namely ―also‖, ―next‖, ―or‖, ―moreover‖, or ―more‖) in the image schema of RETURN is abstracted from the two sequent events in which they constitute a complete movement of RETURN. The presupposed previous motion event is followed by the retraversal motion event. Examples that involve sequential senses usually have two independent clauses (or arguments) indicating two required and different events. The new derived meaning from the sequential sense can be translated 86 into ―in addition‖ or ―also‖, as shown in (25) and (26) from Mandarin and Labo Mosuo, respectively. (25) hai ―also‖ in Mandarin 他会弹琴还会修电脑 ta hai tan qin hai hui xiu dian-nao he can play piano also can repair computer ‗He can play piano; in addition, he can repair a computer.‘ (26) le 13 ―also‖ in Labo Mosuo tʂhɯ33 zɯ33 tʂɯ33 ɡə13 le13 ba13 le13 tsho33. 3S grass ground up also run also jump ‗He is running and (also) jumping on the grassland.‘ The new derived meaning can carry the meaning of ―more‖ in a two-argument comparative construction. An example of hai from Mandarin is shown in (27). Note that the function of hai (还) in the comparative construction is very similar to geng (更) meaning ―more‖, but the former cannot be used in a sentence that involves more than two arguments (Shen 2001:483). 49 (27) hai ―more‖ in Mandarin 煮面比煮饭还方便 zhu mian bi zhu fan hai fang-bian cook noodle compare cook rice more convenient ‗Cooking noodles is more convenient than cooking rice.‘ 49 When Shen introduces Lu‘s (1980:191) arguments about the difference between hai ‗more‘ (还) and geng ‗more‘ (更); he mentions that geng can be used in a three-argument comparative construction, while hai cannot. Examples: a. 长江比黄河长，比淮河就更长了。 Chang-jiang bi huang-he chang, bi hai-he jiu geng chang le Chang-jiang compare Huang-he long, compare Hai-he JIU more long Asp. ‗Chang-jiang is longer than Huang-he, and it is even longer than Hai-he.‘ b. 长江比黄河长，比淮河就还长了。 Chang-jiang bi huang-he chang, bi hai-he jiu hai chang le Chang-jiang compare Huang-he long, compare Hai-he JIU more long Asp. ‗Chang-jiang is longer than Huang-he, and it is even longer than Hai-he.‘ 87 In Labo Mosuo, the new derived meaning of le 33 u 13 from the sequential sense can denote a meaning which is similar to the temporal concept of ―next‖, as shown in (28) and (29). The occurrence of the second event is preceded by the first event. (28) le 33 u 13 ―next‖ in Labo Mosuo tʂhɯ33 ʂɯ31 tʂhɯ33 tʂha33; le33u13 thɯ33 ʂɯ31 ɲɑ33 tʂha33 bi33. this time 3S wash next that time 1S wash go (Impfv) ‗He washes it this time; I will wash it next time.‘ (29) le 33 u 13 ―next‖ in Labo Mosuo ɲɑ33 tʂɯ33 pu13 mə33 ly13 , le33u13 thɯ33 ʂɯ31 1S money carry Neg enough next that time xuɑ33 tsa55 ɲi33. buy could Cop ‗I didn‘t bring enough money; we could buy it next time. The new derived meaning from the sequential sense can function as a temporal connective meaning ―after (that)‖ or ―moreover‖, as shown in (30) and (31) from Labo Mosuo le 33 u 13 and Mandarin hai, respectively. The occurrence of the first event is followed by the second event. (30) e 33 u 13 ―moreover‖ in Labo Mosuo da 33 dʝɯ33 dɯ33 se31 le33u13 dɯ13 ly31ly33. flood occur finish moreover earth shake ‗After flooding subsided, the earth shook.‘ (31) hai ―moreover‖ in Mandarin 他杀了人还焚尸。 ta sha le ren hai fen shi he kill Asp. people moreover burn corpse ‗He killed the man; moreover, he burned the corpse.‘ 88 3.1.4.6. Continuant sense—“still” The continuant sense (namely ―still‖) in the image schema of RETURN is abstracted from the circulative or rotative path. The grammatical meaning derived from the continuant sense in Mandarin hai can be translated as ―still‖, as shown in (32). (32) hai ―still‖ in Mandarin 都三点了你们还不睡觉 dou san dian le ni men hai bu shui-jiao already three o‘clock Perf. you Pl still not sleep ‗It is 3 o‘clock and you still do not want to go to bed.‘ In Labo Mosuo, le 13 ―still‖ is used to describe a maintained status or a repeating action, as shown in (33) and (34). (33) le 13 ―still‖ in Labo Mosuo ɲɑ33 le33u13 le33 su33dɯ33 lɑ31 le13 su33dɯ33 mə33 thu33. 1S again Pref think even still think Neg arrive ‗Even though I have thought about it over and over again, I still cannot think of it.‘ (34) le 13 ―still‖ in Labo Mosuo ɲɑ33 le33u13 le33 çi33 lɑ31 le13 çi33 mə33 do31. 1S again Pref search even still search Neg see ‗Even though I have searched for it over and over again, I still cannot find it.‘ 3.1.5. Discussion Based on the discussion above, the image schema of 还 is very similar to that of the refresh symbol . It was originally applied to the motion verb ―to return‖ in Ancient Chinese; therefore I use ―the image schema of RETURN‖ to represent the image schema of 还. The image schema of RETURN consists of four senses: the repetitious sense, opposite sense, sequential sense, and continuant sense, as shown in Figure 3.4. 89 Figure 3.4. Semantic extensions from the image schema of 还 All of the synchronic uses of 还 in Mandarin, such as ―to return (回)‖, ―again (再)‖, ―still (仍然)‖, ―unexpectedly (竟然)‖, ―also (也)‖, ―or (或)‖ and ―more (更加) could have been conceptually abstracted from the prototypical image schema of RETURN. The meaning of ―again‖ is derived from the repetitious sense; the meaning of ―unexpectedly‖ is derived from the opposite sense; the meaning of ―also‖, ―or‖ or ―more‖ is derived from the sequential sense; and the meaning of ―still‖ is derived from the continuant sense. Diachronically speaking, the order of appearance of these four senses is still unknown, but the occurrence of the verbal meaning ―to return‖ was likely earlier than the four grammatical senses. Note that there is a synonym for huan (还) in Chinese—fan (返) ‗to return‘. The etymology of fan ‗to return‘ is a noun fan (反), meaning a ‗reverse side‘.50 The verbal function of fan ‗to return‘ derived from the noun meaning a ‗reverse side‘ (written as 反) due to a metaphorical extension. Unlike huan ‗to return‘, fan ‗to return‘ was never used 50 The Chinese character fan (返) ‗to return‘ is a new form. It should be written as 反, which is identical to fan ‗reverse side‘ in Old Chinese. 90 to convey the senses of repetition, sequence, and continuance, because its image schema, which is very similar to the recessive symbol ←, only profiles an opposite sense. Although they share a synonymous meaning, ‗to return‘, they exhibit different outcomes after semantic extensions, because fan ‗to return‘ and huan ‗to return‘ originated from two different etyma. 3.2. The disposal marker BA—把 Section 3.2 is concerned with problems related to the most studied construction— so-called BA sentences (把字句) or ‗disposal‘ constructions (处置式) in the Chinese language. The concept of ‗disposal‘ (chuzhi (处置) in Chinese) is introduced in a context involving certain disposal verbs in relation to the use of BA (把) before certain preverbal objects in Chinese. The structure of ‗disposal‘ in Contemporary Chinese can be represented as ―NP1-Agent+BA+NP2-Patient+VP‖. Wang (1947:161) first described the notion of the disposal form in the following way: ―The disposal form states how a person is handled, manipulated, or dealt with; how something is disposed of, or how an affair is conducted. Since it is specifically designed for disposing, the disposal form cannot be used unless the action possesses the quality of disposal‖. 51 This section attempts to analyze the origin and evolution of this construction from a cognitive perspective. It starts out by introducing previous treatments contributing to the evolution of the BA construction. Then, I will point out inadequacies of previous studies and provide possible solutions. Among hundreds of articles related to the BA construction, Wu‘s (2003) investigation is most interesting. He combined advantageous 51 English translation is taken from Y-C Li (1974:200). The original texts are: 处置式是把人怎样安排,怎 样支使,怎样对付,或把物怎样处理,或把事情怎样进行。它既然专为处置而设，如果行为不带处置 性质，就不能用处置式。 91 findings from several previous studies into a refined grammaticalization process, as shown in Figure 3.5 below. Figure 3.5. Wu‟s grammaticalization process of BA construction Figure 3.5 shows that Wu‘s grammaticalization process can be divided into three consequent steps: (1) from a serial verb construction to an instrumental construction, (2) from an instrumental construction to ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ constructions, and (3) from ‗giving disposal‘ and ‗placing disposal‘ constructions to the ‗simple disposal‘ construction. The partial idea in the first step may be adopted from Zhu (1957:17), Wang (1958:410-412), Bennett (1981:65), or especially Peyraube (1988:619-627; 1996). Peyraube (1996:168-170) hypothesizes a categorical change that would grammaticalize BA into an instrumental or an object marker from serial verb constructions. He argues that BA (把) meaning ‗to take, to hold‘ was a verb in Archaic Chinese; in the Early Medieval period, it was used as the first verb (represented as V1) in two different serial verb constructions: ‗V1+O1+V2+ (O2)‘ and ‗V1+O+V2‘. Then, the V1 in terms of BA was grammaticalized and became a preposition (Prep. BA) in an instrumental construction and a disposal construction between the 7th and the 9th centuries. Thus, two different diachronic changes can be demonstrated. Due to the process of 92 grammaticalization, one of the processes turns BA from a verb into an instrumental preposition, as shown in (i), and the other turns BA from a verb into an object marker, as shown in (ii). (i) V1-BA+O1+V2 (+O2) > Prep. BA+O1 +V (+O2) [instrumental construction] (ii) V1-BA +O+V2 > Prep. BA+O+V [disposal construction] It is reasonable to believe that the first process in Peyraube‘s treatment was adapted by Wu in his approach. Generally speaking, the linguist who agrees that BA was grammaticalized from a verb meaning ‗to take, to hold‘ accepts that the BA construction was derived from a serial verb construction. Note that it is very hard to agree with Peyraube‘s grammaticalization process because he mixes up different types of BA constructions with each other when providing examples for the two grammatical processes. I will provide suitable examples for each type later. However, I do believe that the grammaticalized functions of BA were derived from not just one or two but four types of serial verb constructions individually and independently (please see Table 3.1 below for details). The partial idea in the second step may be adopted from Ye (1988:56) or particularly Mei (1990 and 2000b). Significantly, Mei (2000b:192-199) first divides those BA constructions used in the Late Medieval period into four categories: the ‗simple disposal‘ construction, the ‗giving disposal‘ construction, the ‗placing disposal‘ construction, and the ‗regarding disposal‘ construction.52 The four BA constructions are apparent in Mandarin. 52 In Wu‘s (2003:4-5) terminology, the ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ constructions are defined as ―broad-sense‖ disposal construction (广义处置式). The ‗simple disposal‘ construction is defined as ―narrow-sense‖ disposal construction (狭义处置式). 93 The ‗simple disposal‘ involving an object with a final-verb occurs in the ‗(S)+BA+O+V‘ construction. For example: (37a) Simple disposal construction in Mandarin 老板把我炒了。 Laoban ba wo chao le. boss BA I fire Perf. O V ‗My boss has fired me.‘ The ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ under the label of ‗double-object disposals‘ must occur in the ‗(S)+BA+O1+V+O2‘ construction which involves two objects. For example: (37b) Giving disposal construction in Mandarin 我把钱还老板了。 Wo ba qian huan laoban le. I BA money return boss Perf. O1 V O2 ‗I have returned money to my boss.‘ (37c) Placing disposal construction in Mandarin 我把钱放在老板桌上了。 Wo ba qian fang zai laoban zhuoshang le. I BA money put on boss desk Perf. O1 V O2 ‗I have put the money on my boss‘ desk.‘ (37d) Regarding disposal construction in Mandarin 老板把我看作外人。 Laoban ba wo kanzuo wairen. Boss BA I look-as outsider O1 V O2 ‗My boss viewed me as an outsider.‘ Note that Mei has not concentrated his investigation only on BA (把), but also looks at other disposal forms, such as JIANG (将) and YI (以), which have ultimately become extinct. The fundamental difference between the theories of Wu and Mei is that Mei does not agree that grammaticalization is the right reason for the evolution of BA; 94 instead, he argues that certain BA used in the ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ construction during the Tang Dynasty (619-907 A.D.) were all inheritors of YI, which had already essentially become JIANG during the Sui Dynasty (581-619 A.D.). 53 Therefore, his basic assumption for the emergence of the BA construction in Middle Chinese is that the oldest disposal form YI was replaced by JIANG; then the newer form BA replaced JIANG via a process of lexical replacement. The mechanism of lexical replacement in the BA construction has many advocates, such as Chen (1983:201-205) and Her (1990:277). The linguist who believes that the emergence of BA was derived from the process of lexical replacement is more unlikely to accept grammaticalization. 54 I believe that the process of lexical replacement is possible and the oldest disposal form YI may play an important role in the establishment of BA forms. However, Mei‘s analyses are misleading. I will make corrections later. Considering the second step of Wu‘s grammaticalization process, the development from the instrumental construction to the ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ constructions seems to be an innovation. However, the following discussions will also show that such a process is problematic. The partial idea in the third step may be adopted from Ding (2001:102-128). Exploring the relation between the ‗giving disposal‘, ‗placing disposal‘, and ‗simple 53 Based on his observation of ancient documents during the 5th and 6th centuries, Mei (2000b:205) argues that certain disposal forms were added to patient-subject sentences shoushi zhuyu ju (受事主語句) to obtain disposal expression. In other words, a correspondent patient-subject sentence is the prerequisite for the origin of BA disposal forms. 54 There are a few exceptions. Bennett (1981:63-65) hypothesizes that the YI-instrumental and YI-disposal constructions may have served as the source for the grammaticalization of JIANG. Huang (1986:43) proposes that BA came from JIANG by a process of lexical replacement, however, prepositional JIANG derived from the verb via grammaticalization in serial verb constructions. 95 disposal‘, Ding points out that the ‗giving disposal‘ and ‗placing disposal‘ correspond well with the semantic extension and semantic shift of ‗verb-final BA sentence‘ (‗simple disposal‘ in Mei‘s terminology) in that ―the act of giving something to somebody or placing something in some place typically involves the holding of a concrete object‖ and ―the whole situation itself represents some sort of manipulation, with the transfer of ownership or the change of locus for existence (2001:114).‖ In other words, the development from ‗giving disposal‘ and ‗placing disposal‘ to ‗simple disposal‘ is simply a process of metaphorical extension. The sense of manipulating a concrete object transfers to the sense of an act of disposal. However, Ding did not explain why the three- argument sentence, such as the ‗giving disposal‘ and ‗placing disposal‘ constructions, can become a two-argument sentence in terms of the ‗simple disposal‘ construction via a metaphorical process. Furthermore, there is a significant difference between the ‗giving disposal‘ and the ‗placing disposal‘ and ‗simple disposal‘ in that the ‗simple disposal‘ construction often involves a malefactive causee. Where does the sense of malefactive come from? I will provide solutions to these problems in Section 3.2.4. Let us consider the third step of Wu‘s grammaticalization process again. There is another crucial factor that makes Wu put the ‗simple disposal‘ construction at the end of the grammaticalization model. 55 Based on the diachronic documents from ancient Chinese, many linguists, such as Zhu (1957), Wang (1958), and Peyraube (1989), claim that the ‗simple disposal‘ in BA started to emerge about time of the Tang Dynasty (618- 907 A.D.). However, the ‗giving disposal‘ and the ‗regarding disposal‘ in YI can be dated to the period of Archaic Chinese (770-249 B.C.) and the emergence of the ‗placing 55 Note that Wu (2003) also points out a divergent development in which the ―causative disposal‖ construction is derived from ‗simple disposal‘, ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ constructions via a syntactic extension. 96 disposal‘ in YI can be dated to the Han Dynasty (206 B.C.-220 A.D.). Therefore, the expression of the ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ should develop before the expression of the ‗simple disposal‘. It appears that the third step in Wu‘s grammaticalization process is theoretically and historically possible. However, although it is acceptable that the structure of the ‗simple disposal‘ in BA did not appear until the Tang Dynasty, I will argue against the third step of Wu‘s grammaticalization process in that it involves both semantic illogicality and syntactic impossibilities. Wu‘s renewed grammaticalization pattern generates several problems that are worth further discussion. First of all, the whole process itself is semantically illogical. The instrumental construction (in the first step) normally involves a disposal and/or causative sense. In addition, the ‗giving disposal (see (37b))‘, ‗placing disposal (see (37c))‘, and ‗regarding disposal (see (37d))‘ constructions (in the second step) involve a very weak causative sense or even have nothing to do with the disposal sense. Furthermore, most of the ‗simple disposal (see (37a))‘ construction (in the third step) undoubtedly carries a disposal and/or causative sense. Therefore, I am wondering why the development of the disposal expression in Chinese necessarily underwent the process from ―disposal sense‖ to ―non-disposal sense‖, then from ―non-disposal sense‖ to ―disposal sense‖ in Wu‘s treatment. It is clear that the semantic change of the BA construction cannot follow the process ―disposal non-disposal disposal‖ because this is redundant. Secondly, syntactic observations show that the ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ constructions (in the second step) structurally involve 97 three arguments (including a subject and two objects), while the ‗simple disposal‘ construction (in the third step) only involves two arguments (including a subject and an object). Wu‘s treatment is not tenable because he fails to explain why the three-argument sentence can become a two-argument sentence via grammaticalization. It is unlikely that the ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ constructions can develop and gain an additional disposal sense (especially the malefactive sense) while losing an argument. Finally, several questions concerning the most troublesome construction, the ‗regarding disposal‘ construction, remain unanswered. The ‗regarding disposal‘ construction, as exemplified in (37d), is essentially a ‗non-disposal‘ three-argument constitution, even though its terminology involves ‗disposal‘. In my treatment, the ‗regarding disposal‘ cannot originate from the instrumental construction. Although Ding (2001:114) ambiguously points out that ―the metaphoric use of BA in Middle Chinese appears in the ‗regarding disposal‘ form, in which BA may take a person as its object‖, I am still uncertain of exact metaphorical process from the ‗regarding disposal‘ to the ‗simple disposal‘ in his argument. After a fifty-five-year debate, there is still no convincing explanation for the emergence of the ‗regarding disposal‘ construction. In fact, most of the linguists who have written articles about the BA construction unavoidably excluded the examples of the ‗regarding disposal‘ construction from their research. Furthermore, I do not see any theory that can be said to come close to accounting for all of the examples of the BA construction. Many different theories have been proposed, and a comprehensive review is beyond the scope of this section. I believe that a linguist who wants to use one theory to 98 explain all of the phenomena of the BA construction will never reach his goal because the synchronic BA construction is a diachronic polymer derived from multiplex sources. As discussed above, many previous research projects bring up important questions and contribute to our understanding of the nature of the BA construction. However, none of them seems to be able to provide an adequate solution to all the remaining problems. The purpose of the following sections is to give an alternative approach that can both overcome these unanswered problems and give a better explanation for the evolution of the BA construction. In order to understand the semantics of the BA construction, it is necessary to realize how Chinese speakers use the verb meaning ―to take‖ or ―to hold‖ (represented as TAKE, hereafter) diachronically. It should be noticed that I will not only concentrate on BA (把) but will also look at other words, such as YI (以), JIANG (将), ZHOU (捉), CHI (持), QU (取), and NA (拿), because all of these words have been used as a verb meaning TAKE throughout the history of Chinese. 56 I argue that the BA construction in Mandarin originates from four individual and independent serial verb constructions involving TAKE. It is well-known that in Sino- Tibetan languages, verb serialization is one of the preferred syntactic constructions or even preconditions for the process of grammaticalization. Considering the stages of grammaticalization, DeLancey (2004:1595) mentions that the pathway of grammaticalization ―from clause-chaining through verb serialization to auxiliarization‖ is widely attested. I will demonstrate how the grammaticalized functions 56 Note that in certain dialects of Chinese, such as Mandarin and Xiang, speakers use gei (给) ‗to give‘, jiao (教/叫) ‗to call‘, and rang (让) ‗to let‘ as the disposal marker. The development of these words may differ from that of TAKE discussed here. I will discuss related issues in Chapters IV and VI. 99 of TAKE emerge via metaphorical extensions in serial verb constructions in the following sections. The neutral representation of the serial verb construction with the verb TAKE (including YI (以), BA (把), JIANG (将), ZHOU (捉), CHI (持), QU (取), and NA (拿)) in Chinese can be generally characterized as ―NP1+V1_(TAKE)+NP2+V2+NP3‖, as shown in Table 3.1. NP1 is the subject of the serial verb construction. The first event consists of two constituents: V1_(TAKE) and NP2. The second event involves two components: V2 and NP3. There are four types of serial verb events that can use TAKE as their first verb (V1). The second verb (V2, which occurs in the second event, requires a specific class of verbs, such as transferring verbs, placing verbs, treating verbs, and disposal verbs. Different types of serial verb events assign different thematic roles to their nominals: NP1, NP2, and NP3. The first type (type 1) which is defined as a ‗thing- transferred event‘ involves a giver, a thing, and a recipient. The second type (type 2) which is defined as a ‗thing-located event‘ involves an agent, a thing, and a location. The third type (type 3) which is defined as a ‗thing-transformed event‘ involves an agent, a thing, and an instrument. The fourth type (type 4) which is defined as an ‗instrumental causative event‘ involves a causer, an instrument, and a causee. 100 Table 3.1. Serial verb constructions with the verb TAKE Serial verb construction First event Second event NP1 V1_(TAKE) NP2 V2 NP3 Type 1 giver yi (以) ba (把) jiang (将) zhou (捉) chi (持) qu (取) na (拿) thing transferring verb recipient Type 2 agent placing verb location Type 3 treating verb instrument Type 4 causer instrument disposal verb causee In comparing Wu‘s grammaticalization model (provided in Figure 3.4) with Table 3.1, it is very important to note that examples that qualify as a ‗thing-transferred event‘ are equivalent to those in Wu‘s ‗giving disposal‘ construction. Examples used in the ‗thing-located event‘ are equivalent to those in his ‗placing disposal‘ construction. Examples used in the ‗thing-transformed event‘ are equivalent to those in his ‗regarding disposal‘ construction. Examples used in the ‗instrumental causative event‘ are equivalent to those in his ‗instrumental‘ construction. Note that the ‗instrumental‘, ‗giving disposal‘, ‗placing disposal‘, and ‗regarding disposal‘ constructions in Wu‘s treatment are undergoing the grammaticalization process; in other words, the morpheme TAKE in these constructions is no longer a lexical verb. However, in my study, TAKE in the four types of serial verb constructions in Table 3.1 is still a lexical verb in the first event. Contrary to Wu‘s supposition that the BA construction originated from one serial verb construction, the present study argues that the nature of the evolution of TAKE (including BA) is derived from four different types 101 of serial verb events, as shown in Table 3.1 above, no matter which form of TAKE is used diachronically. Because the disposal marker BA (把) in Modern Chinese cannot occur as the predicate of a simple sentence, I use NA (拿) meaning ‗to take‘ to exemplify the four types of serial verb events in Mandarin, as shown in examples (38a-d). It should be feasible to use NA as an example, because the Wu dialect (吳語) of Chinese uses NA instead of BA as its disposal construction. (38) Four types of serial verb events in Mandarin a. Thing-transferred event 他拿弓给张三。 Ta na gong gei Zhangsan. giver thing recipient he TAKE bow give NAME ‗He took a bow and gave it to Zhangsan.‘ b. Thing-located event 他拿弓挂在背上。 Ta na gong gua zai bei shang. agent thing location he TAKE bow hang at back top ‗He took a bow and carried it on his back.‘ c. Thing-transformed event 他拿弓当作钓竿。 Ta na gong dang zuo diaogan agent thing instrument he TAKE bow as use fishing pole ‗He took a bow and used it as a fishing pole.‘ d. Instrumental causative event 他拿弓打张三。 Ta na gong da Zhangsan causer instrument causee he TAKE bow hit NAME ‗He took a bow to hit Zhangsan.‘ 102 The four types of serial verb events can be found in both Modern Chinese and Archaic Chinese, as illustrated in examples (39a-d), which involve an old form, YI (以). This shows that the variety of uses in the verb TAKE were fully developed by more than two thousand years ago. Indeed, I cannot find any syntactic and semantic change from these serial verb events when comparing example (38) with (39). Note that sentences in example (39) are selected from ancient written documents; therefore, the subject in these sentences is often omitted. Also note that most of the examples selected from ancient Chinese are from Mei (1990; 2000b) and Wang (1947). However, I have provided most of the translations and interpretations for these sentences. (39) Four types of serial verb events in Archaic Chinese a. Thing-transferred event 以物与人。 (in Zhuangzi 庄子 (369-286 B.C.)) Yi wu yu ren. thing recipient TAKE thing give people ‗Took extra belongings and gave them to other people.‘ b. Thing-located event 以弟子一人投河中。 (in Shiji 史記 (104 B.C.)) Yi dizi yi ren tou he zhong. thing location TAKE disciple one Cl throw river middle ‗Took a disciple and threw him into the river‘ c. Thing-transformed event 以其頭為飲器。 (in Shiji 史記 (104 B.C.)) Yi qi tou wei yinqi. thing instrument TAKE his head as drinking container ‗Took his head and used it as a drinking container.‘ 103 d. Instrumental causative event 以戈逐子犯。 (in Zuozhuan 左傳 (300-400 B.C.)) Yi ge zhu Zifan. instrument causee TAKE halberd evict NAME ‗(Someone) took a halberd to evict Zifan.‘ According to the examples selected from both Modern and Archaic Chinese, it is clear that the serial verb construction typically contains two or more verb phrases without any overt connective element. The serial verb expression involving two or more sequential actions is common in various language families scattered around the world. The semantic change of disposal meaning from the verb TAKE is attributed to the capability of humans to take an object in their hands (in the first event) and to control it at will (in the second event). To interact with the environment, humans use their hand to hold weapons or instruments, to give something to somebody, to place something in some place, or to use something to function as a tool. It is conceivable that the expression of the four serial verb events essentially requires the act of holding some objects. These kinds of activities are basic and common in daily life no matter whether they are in the past or the present. Taking the trend of semantic shift from concrete domains to abstract domains into consideration, it is proposed that the meaning of TAKE has been grammaticalized from ‗to take/hold a concrete thing in hand‘ in a serial verb event to ‗to indicate any manipulatable thing‘ in a single event. In the following, I will begin by describing the mechanism of the metaphorical extension in each type of serial verb events (Sections 3.2.1-3.2.4). Particularly, in Section 3.2.4 I will claim that the ‗instrumental causative event‘ is independently responsible for developing the two-argument ‗simple disposal‘ construction which occurred plentifully after the Tang Dynasty. 104 3.2.1. Type 1: thing-transferred event In the thing-transferred event, the second event is more salient than the first event; therefore, there are three profiled entities, GIVER, THING, and RECIPIENT, in its image schema, as shown in Figure 3.6. 57 The profiled portion of the thing-transferred event includes the circles representing the persons giving and receiving and the object being passed. Figure 3.6. Image schema of the thing-transferred event Giver Thing Recipient In Figure 3.6, the two bigger circles indicate the GIVER and the RECIPIENT, and the smaller circle in the middle indicates the THING. A traversed path indicating a motion from GIVER to RECIPIENT is represented by a solid line. The event begins with the GIVER, who is understood as the initiator of the event and controls the thing transferred. The RECIPIENT receiving the thing completes the act of transference. The THING is positioned on the intermediate point of the solid line, indicating that it is involved in a motion event. The arrow indicates the direction of the movement. Note that the giving verb used in the second event of this type may give rise to an idea with respect to the transference of possession; therefore, the RECIPIENT involved in some cases can be referred to as a BENEFACTIVE. Since the act of ‗taking‘ in the thing-transferred event is not profiled, the image schema of the first type is equivalent to the typical act of giving. More examples of this 57 Peyraube (1996:169) points out that the second verb (V2) is probably more important than the first verb (V1) in the serial verb construction; therefore, the first verb meaning TAKE is often bleached. It loses its full meanings and becomes a marker when grammatialization begins. 105 type from historical records of Mandarin are provided from examples (40) to (43). In these examples, the various forms of TAKE including BA (把), JIANG (将), CHI (持), and NA (拿) are still lexical verbs. (40) BA_TAKE in thing-transferred event 把米与鸡...。 (in Luoyangqielanji 洛阳伽蓝记 (534-550 A.D.)) Ba mi yu ji.... TAKE rice give chicken ‗(Someone) took rice and gave it to the chickens.‘ (41) JIANG_TAKE in thing-transferred event 将一大牛...卖与此城中人。 (in Foshuoshuiniujing 佛说水牛经 (265-317 A.D.)) Jiang yi da niu...mai yu ci cheng zhong ren TAKE one big buffalo...sell to this city inside people ‗(Someone) took a big buffalo and sold it to the people in the city.‘ (42) CHI_TAKE in thing-transferred event 复自破骨,持髓与之。 (in Daoxingbanruojing 道行般若经 (150-250 A.D.)) Fu zi po gu, chi sui yu zhi. then self break bone, TAKE marrow give him ‗(He) broke the bone by himself, then took the marrow and gave it to (somebody).‘ (43) NA_TAKE in thing-transferred event in Mandarin 我拿球棒递给张三。 Wo na qiubang di gei Zhangsan. I TAKE bat hand-over to NAME ‗I took a bat and handed it over to Zhangsan.‘ In order to reflect more human experiences, humans start to conceptualize the linguistic use from concrete domains in abstract domains. For instance, when the transferring THING is not a concrete entity in the thing-transferred event, the original concrete meaning of the verb TAKE has to be grammaticalized into a thing-marker, as shown in examples (44)-(47), selected from historical records or Mandarin. These examples involve various grammatical forms of TAKE, including YI (以), BA (把), and JIANG (将) to signify the effect of the abstract THING. 106 (44)YI in metaphorical thing-transferred event a. 楚子以灭舒鸠赏子木。 (in Zuozhuan 左传 (300-400 B.C.)) Chuzi yi mie Shujiu shang Zimu. NAME YI ruin NAME award NAME ‗Chuzi promised an award to Zimu for conquering the kingdom of Shujiu.‘ b. 尧以天下与舜。 (in Mengzi 孟子 (385-304 B.C.)) Yao yi tianxia yu shun NAME YI world give NAME ‗Yao gave the world to Shun.‘ (45) BA in metaphorical thing-transferred event 应把清风遗子孙。 (in Fangganshi 方干诗 (809-888 A.D.)) Ying ba qing feng yi zi sun should BA fresh breeze bequeath son grandson ‗One should bequeath (such a) fresh breeze to one‘s posterity.‘ (46) JIANG in metaphorical thing-transferred event 将佛言语传闻维摩。 (in Bianwen 变文 (400-900 A.D.)) Jiang fo yanyu chuanwen Weimo. JIANG Buddha words deliver NAME ‗(Someone) delivered Buddha's words to Weimo.‘ (47) BA in metaphorical thing-transferred event in Mandarin 张三把好运带给了大家。 Zhangsan ba hao yun dai gei le dajia. NAME BA good luck bring to Asp everybody ‗Zhangsan brought good luck to everybody.‘ 3.2.2. Type 2: thing-located event The image schema of the second type, the thing-located event, is somewhat compatible with that of the thing-transferred event if we treat AGENT as the GIVER, and LOCATION as an inanimate RECEIVER. The second event of this type is still more salient than the first event; therefore, it is not necessary to profile the first taking event. There are three profiled entities: AGENT, THING, and LOCATION, as shown in Figure 3.7. 107 Figure 3.7 is a representation of an act of thing-locating. The two bigger circles indicate the AGENT and the LOCATION, and the smaller circle in the middle indicates the THING. A traversed path indicating a movement from AGENT to LOCATION is represented by a solid line. The serial event begins with the AGENT, who is understood as the carrier of the event and controls the thing-locating. The LOCATION is the end point of the act of thing-locating. The THING is at the intermediate point of the solid line indicating that it is involved in a movement. The arrow indicates the direction of the movement. Figure 3.7. Image schema of the thing-located event Agent Thing Location The act of taking something and locating it somewhere is common. Examples can be shown in (48)-(51), which are selected from historical records or Mandarin. The THING object in these sentences can be held and/or operated in the hands; therefore, it is accurate to interpret YI (以), BA (把), JIANG (将), and NA (拿) as the lexical verb TAKE. The placing verbs used in the thing-located event are not necessarily verbs meaning ‗to put‘, ‗to place‘, or ‗to locate‘. They can be any action verb that involves a sense of thing-locating, such as suan ‗to hang‘ in (49), cha ‗to insert‘ in (50), or tie ‗to paste‘ in (51). (48) YI_TAKE in thing-located event 家人常以琴置灵床上。 (in Shishuoxinyu 世说新语 (420-479 A.D.)) Jia ren chang yi qin zhi ling chuang shang. house person often TAKE lute place holy bed on ‗The family members often took the lute and placed it on the bier.‘ 108 (49) BA_TAKE in thing-located event 把舜子头发悬在中庭树地。 (in Bianwen 变文 (400-900 A.D.)) Ba Shunzi toufa suan zai zhong ting shu di. TAKE NAME hair hang on central courtyard tree place ‗(He) took Shunzi‘s hair and hung it on the tree in the central countyard.‘ (50) JIANG_TAKE in thing-located event 将尼拘陀树一枝插于地上。 (in Fobenxingjijing 佛本行集经 (587-591 A.D.)) Jiang nijutuo shu yi zhi cha yu di shang. TAKE NAME tree one Cl insert at ground top ‗(He) took a stick of Nijutuo Tree and inserted it into the ground.‘ (51) NA_TAKE in thing-located event in Mandarin 他拿先发名单贴在门上。 Ta na xianfa mingdan tie zai men shang. he TAKE lineup list paste on door top ‗He took the lineup list and pasted it on the door.‘ When the interaction between the AGENT and the THING is irrelevant to the act of taking, the verb TAKE must be grammaticalized into a marker to indicate the THING, as shown in examples (52)-(55) selected from historical records or Mandarin. These examples involve various grammatical forms of TAKE, including JIANG (将), CHI (持), QU (取), and BA (把), respectively. Since the act of taking has been bleached out, the THING object in these sentences does not have to be a concrete entity, as shown in (55). (52) JIANG in metaphorical thing-located event 汝将我子置彼林内。 (in Fobenxingjijing 佛本行集经 (587-591 A.D.)) Ru jiang wo zi zhi pi lin nei. you JIANG my son locate that forest inside ‗You put my son in the forest.‘ (53) CHI in metaphorical thing-located event 持手着阿难肩上。 (in Daoxingbanruojing 道行般若经 (179 A.D.)) Chi shou zhuo Aji jian shang. CHI hand on NAME shoulder top ‗(He) put a hand on Aji's shoulder.‘ 109 (54) QU in metaphorical thing-located event 取父王闭在牢狱。 (in Zengyiahanjing 增壹阿含经 (500-599 A.D.)) Qu fuwang bi zai laoyu. Qu father imprison in jail ‗(He) imprisoned his father in jail.‘ (55) BA in metaphorical thing-located event in Mandarin 所有人把目光放在投手身上。 Suoyou ren ba muguang fang zai toushou shen shang every people BA gaze put on pitcher body top ‗Everyone gazes at the pitcher.‘ It is obvious a very similar schema has been used in the first type and second type. Both the thing-transferred event and the thing-located event denote a general notion of ‗change of spatial orientation of the THING‘. The THING object immediately after TAKE is obligatory and cannot be omitted from the image schema. In other words, the schematic structure must involve three arguments even though the co-referential AGENT or GIVER is not necessarily overt in all contexts. In addition, it is important to notice that the receiver (no matter whether an animate RECIPIENT or inanimate LOCATION) perceived as the end point of the action cannot be moved in front of the THING whenever before or after the process of metaphorical extension. Fronting the RECIPIENT or LOCATION will give rise to ungrammatical expressions. In other words, the word order in the first type and second type cannot be changed. When the metaphorical extension is completed in the first type and second type, the target domain is allowed to accept an intangible THING, as shown in examples (47) and (55). The animacy is no longer a restriction in LOCATION, as shown in example (55). The verb TAKE has moved towards the grammatical function indicating the THING object. 110 3.2.3. Type 3: thing-transformed event: The image schema of the thing-transformed event involves three profiled entities: AGENT, THING, and INSTRUMENT, as shown in Figure 3.8. The large middle circle representing the person who is doing the act of taking represents the AGENT. The smaller solid circle representing the object being taken and used refers to the THING. The smaller dotted circle representing a conceptual entity transformed from the THING refers to the INSTRUMENT. Since the AGENT controls the thing-transformed process from one thing to another, it can also be referred to as a TRANSFORMER. The arrows indicate the direction of the process. Figure 3.8. Image schema of the thing-transformed event Taking Transforming Thing-1 Agent/Transformer Instrument=>Thing-2 Unlike the first type and second type, the thing-transformed event profiles the first event (referring to the act of taking) as well as the second event (referring to the act of transforming). The act of taking, represented by the first solid line, involves the hands of the AGENT; the movement is initiated by the AGENT; and the THING ends up in the hands of the AGENT. The act of transforming represented by the second solid line begins with holding the THING (namely ‗Thing-1‘) in the hands; then using the THING as another conceptualized instrumental THING (namely ‗Thing-2‘). There is a close relationship between the manipulable THING and the instrumental THING in that both are associated with a shared function to some extent. Examples are shown in (56-57), 111 which are selected from historical records or Mandarin. These examples involve various forms of TAKE, including JIANG (将), and NA (拿), respectively. (56) JIANG_TAKE in thing-transformed event 将此茶芽为信。 (in Lidaifabaoji 历代法宝记 (700-800 A.D.)) Jiang ci cha ya wei xin. TAKE this tea bud to-make token ‗(He) took this tea bud and used it as a token.‘ (57) NA_TAKE in thing-transformed event 他拿球帽当作手套。 Ta na qiumao dang zuo shoutao. he TAKE ball cap treat as glove ‗He took a baseball cap and used it as a baseball glove.‘ When ‗Thing-1‘ is not a concrete entity, the meaning of TAKE starts to lose its verbal content and becomes a marker referring to the THING via a metaphorical extension. The treating verb used in this type is usually wei (为) ‗to use as‘, zuo (作) ‗to use as‘, or ~ cheng (成) ‗~ as‘. Examples are shown in (58-61), which are selected from historical records or Mandarin. These examples involve various grammatical forms of TAKE, including YI (以), BA (把), and CHI (持). (58) YI in metaphorical thing-transformed event a. 文王以民力为台。 (in Mengzi 孟子 (385-304 B.C.)) Wen wang yi min li wei tai. NAME Lord YI people strengh as tower ‗Lord Wen used people‘s strength as (his) tower.‘ b. 吾必以仲子为巨擘焉。 (in Mengzi 孟子 (385-304 B.C.)) Wu bi yi Zhongzi wei jubo yan. I must YI NAME as authority SFP ‗I will definitely treat Zhongzi as the authority.‘ (59) BA in metaphorical thing-transformed event a. 有人把椿树唤作白旃檀。 (in Hanshanshi 寒山诗 ((691-793 A.D.)) You ren ba Chunshu huan zuo Baizhantan. have people BA NAME call as NAME ‗Some people called Chunshu Baizhantan.‘ 112 b. 莫把边地比京都。 (in Wangjinshi 王缙诗 (700-781 A.D.)) Mo ba biandi bi jingdu. Don‘t BA frontier treat-as capital ‗Don‘t treat the frontier as the capital.‘ (60) CHI in metaphorical thing-transformed event 持无常作有常。 (in Foshuoyirimonibaojing 佛说遗日摩尼宝经 (220-265 A.D.)) Chi wuchang zuo youchang. CHI uncertainty treat-as certainty ‗Treat uncertainty as certainty.‘ (61) BA in metaphorical thing-transformed event in Mandarin 主审把全垒打看成界外球。 Zhushen ba quanleida kan cheng jiewaiqiu. umpire BA home-run look as foul-ball ‗The home plate umpire viewed a home run as a foul ball.‘ The image schema of the thing-transformed event is quite different from those of the former two types. The first type and second type only profile the second event, while the Type 3 profiles both the first event and the second event. The transformed THING is not a perceptible entity in all cases. But there are similarities; for instance, the metaphorical extension applied in all types is allowed to involve an abstract THING. All of the elements in the schema are obligatory. Based on the discussion in Sections 3.2.1, 3.2.2, and 3.2.3, it is obvious that the semantic nature of the THING being transferred, located, or transformed triggers the process of metaphorical extension. The meaning of TAKE can expand from holding a tangible entity to an intangible object. When the THING is not a concrete object held in the hands, the first verb TAKE in the serial verb event develops an abstract meaning, functioning as a marker of the THING object. The serial verb event has become a single event after metaphorical extension. 113 I have pointed out three different serial verb events involving a lexical verb TAKE (such as the following examples (i-iii)), and three different signal events which involve a metaphorical TAKE (such as the following examples (iv-vi)). (i) Lexical verb TAKE in thing-transferred event (i.e., (38a), (39a), (40-43)) (ii) Lexical verb TAKE in thing-located event (i.e., (38b), (39b), (48-51)) (iii) Lexical verb TAKE in thing-transformed event (i.e., (38c), (39c), (56-57)) (iv) Metaphorical TAKE in thing-transferred event (i.e., (44-47)) (v) Metaphorical TAKE in thing-located event (i.e., (52-55)) (vi) Metaphorical TAKE in thing-transformed event (i.e., (58-61)) The instrumental causative event will be discussed in the following sections. None of these examples above can originate from an instrumental construction because the instrumental causative event always involves the sense of ―manipulativeness‖ and ―affectedness‖. The affected entity in terms of CAUSEE in the instrumental causative event is more likely to be a malefactive rather than a benefactive. In addition, there is no evidence to show that the two-argument disposal event is derived from the first type, the second type, and the third type. All of the three arguments in the image schema of the first type, the second type, and the third type are restricted and obligatory. Removing the THING from their image schemata will cause semantic insufficiency and syntactic ungrammaticality. 3.2.4. Type 4: instrumental causative event The instrumental causative event is a causative force-transferring event that involves an instrument held in the hands. When completing an action with an instrument in hand, the causative force is transferred from the causer to the causee via the instrument. 114 In this sequence, the second event is more salient than the first event; therefore, there are three profiled entities, CAUSER, INSTRUMENT, and CAUSEE, in the image schema, as shown in Figure 3.9. The two bigger circles indicate the CAUSER and the CAUSEE, and the smaller circle in the middle indicates the INSTRUMENT. The two separated paths indicating a forceful movement from CAUSER to CAUSEE via INSTRUMENT are represented by two solid lines. The arrows indicate the direction of the movement. Figure 3.9. Image schema of the instrumental causative event Manipulating Affecting Causer Instrument Causee The instrumental causative event involves the sense of ―manipulativeness‖ and ―affectedness‖. The act of Manipulating represented by the first solid line involves a CAUSER who manipulates the INSTRUMENT. The act of affecting represented by the second solid line involves a CAUSEE who is affected by the CAUSER via the INSTRUMENT. Examples are shown in (62-66), which are selected from historical records or Mandarin. These examples involve various forms of TAKE including BA (把), JIANG (将), ZHOU (捉), CHI (持), QU (取), and NA (拿), respectively. (62) BA_TAKE in instrumental causative event a. 汤自把钺以伐昆吴。 (in Shiji 史记 (104 B.C.)) Tang zi ba yue yi fa Kunwu. NAME self TAKE axe use-to fight NAME ‗Tang himself took an axe to fight Kunwu.‘ b. 便把被衫揩拭面。 (in Bianwen 变文 (400-900 A.D.)) Bian ba pishan kaishi mian. then TAKE mantle wipe face ‗(Someone) then took a mantle to wipe the face.‘ 115 (63) JIANG_TAKE in instrumental causative event 将暖水浴菩萨身体。 (in Foben 佛本行集经 (587-591 A.D.)) Jiang nuan shui yu pusa shenti. TAKE warm water bathe Buddha body ‗(Someone) took warm water to bathe Buddha‘s body.‘ (64) JIANG_TAKE in instrumental causative event 伊便能捉杖打人。 (in Shishuoxinyu 世说新语 (420-479 A.D.)) Yi bian neng zhuo zhang da ren. he thus can TAKE staff hit people ‗He can thus take a staff to hit people.‘ (64) CHI_TAKE in instrumental causative event 右手持匕首揕之。 (in Zhanguoce 战国策 (452-216 B.C.)) You shou chi bishou zhen zhi. right hand TAKE dagger stab him ‗(Someone) took a dagger in the right hand to stab him.‘ (65) QU_TAKE in instrumental causative event 广行取儿弓射杀追骑。 (in Hanshu 汉书 (20-90 A.D.)) Guang xing qu ergong she sha zhui ji. NAME xing TAKE bow shoot kill chase rider ‗Guang took a bow to shoot and kill the chasers.‘ (66) NA_TAKE in instrumental causative event in Mandarin 拿刀杀牛。 Na dao sha niu. TAKE knife kill cow ‗(Someone) took a knife to kill a cow.‘ It is found that the instrumental causative constructions used two thousand years ago, such as in (39d), (62a), (64), and (65), tend to involve a MALEFACTIVE. This tendency is probably the origin of the sense of ―affectedness‖ because MALEFACTIVE is a kind of CAUSEE. It is also found that the lexical verbs (namely ‗disposal verbs‘) used in the instrumental causative constructions normally involve the sense of both ―manipulativeness‖ and ―affectedness‖. Even though the INSTRUMENT has been omitted from the sentences, they still carry the sense of ―manipulativeness‖ and ―affectedness‖, as shown in (67). Examples (67a-g) are sentences that have removed the 116 INSTRUMENT from examples (62-66). Without an overt INSTRUMENT, the derived two-argument sentences in (67) maintain grammaticality. 58 However, the word order has been changed in that all of these sentences involve a fronting of CAUSEE located at the previous position of INSTRUMENT after it is omitted. (67) Disposal construction a. 汤把昆吴伐 Tang ba Kunwu fa NAME TAKE NAME fight ‗Tang fought against Kunwu‘ b. 便把面揩拭 Bian ba mian kaishi then TAKE face wipe ‗(someone) then wiped the face‘ c. 将菩萨身体浴 Jiang pusa shenti yu. TAKE Buddha body bath ‗(someone) gave a bath to Buddha‘s body‘ d. 伊便能捉人打 Yi bian neng zhuo ren da he thus can TAKE people hit ‗He can thus hit people‘ e. 右手持之揕 You shou chi zhi zhen right hand TAKE him stab ‗(someone) stabbed him with right hand‘ f. 广取追骑射杀 Guang qu zhui ji she sha NAME TAKE chase rider shoot kill ‗Guang shot and killed the pursuants‘ g. 拿牛杀 Na niu sha. TAKE cow kill ‗kill a cow‘ 58 Examples in (67) are Old Chinese which has not been found in historical records yet. But as far as I can judge, these sentences are grammatical. 117 Therefore, the syntactic representation of the so called ―disposal construction‖ shares at least this common structure: ―NP1+Grammatical_(TAKE)+NP2+V‖. NP1 is the subject and NP2 is the direct object. The grammatical TAKE, including BA (把), JIANG (将), ZHOU (捉), CHI (持), QU (取), or NA (拿), functions as a marker of the direct object. The lexical verb, which can be applicable to the disposal construction, is called the ―disposal verb‖; its direct object (NP2) is depicted as having been affected by the CAUSER to some extent. The image schema of the disposal event represents a transfer of causative force from an ‗energy initiator‘—causer to an ‗energy receiver‘—causee, as shown in Figure 3.10. In Figure 3.10, the two circles indicate the CAUSER and the CAUSEE. The causative force from CAUSER to CAUSEE is represented by a solid line. The arrow indicates the direction of the energy flow. Figure 3.10. Image schema of the disposal event Causative Force Causer Causee The disposal construction was rarely found until the Tang Dynasty (618-907 A.D.), since the change of word order from the instrumental causative construction to the disposal construction was a gradual process. More examples can be seen in (68). Once the use of the disposal construction became full-developed, then the door was wide open for Mandarin sentence like (69). Example (69) normally involves a complement after the disposal verb to specify how the CAUSEE has been affected. 118 (68) BA_TAKE in disposal event a. 仰山便把茶树摇。 (in Zutangji 祖堂集 (952 A.D.)) Yangshan bian ba cha shu yao. NAME then TAKE tea tree shake ‗Yangshan then shook the tea tree.‘ b. 把契丹墓坟、宫室、庙像一齐烧了。 (in Sanchaobeimonghuibian 三朝北盟会编 Ba Qidan mufen gongshi miaoxiang yiqi shao le. (1194 A.D.)) TAKE Qidan grave palace temple together burn Perf. ‗(Someone) burnt the Qidan grave, palace, and temple.‘ (69) BA_TAKE in disposal event in Mandarin a. 那粒界外球把一位观众打晕了。 Na li jiewaiqiu ba yi wei guanzhong da yun le. that Cl foul-ball TAKE one Cl audience hit faint Perf. ‗That foul ball hit an audience member and made him faint.‘ b. 这场球赛把我累坏了。 Zhe chang qiusai ba wo lei huai le. this Cl game TAKE I tired badly Pref. ‗This game really tired me out.‘ c. 张三把球打得很远。 Zhangsan ba qiu da de hen yuan. NAME TAKE ball hit DE very far ‗Zhangsan hit the ball very far.‘ Without an INSTRUMENT, it makes little sense for the newly-derived disposal event to keep a full sense of ―manipulativeness‖. In other words, the form TAKE cannot not be interpreted as ‗to hold‘ or ‗to take‘ after losing the INSTRUMENT. In this consequence, the CAUSER in the disposal event is not necessary an animate entity, as shown in (69a); it is allowed to be an immaterial object, as shown in (69b). However, the sense of ―affectedness‖ is needed to be maintained in the disposal event. The CAUSEE is not necessarily animate either, but it must be a concrete object which can receive the causative force, as shown in (68) and (69c). The grammatical form of TAKE (namely ba (把)) in (68) and (69) is simply the marker of CAUSEE. 119 The disposal event considered so far involves two arguments—a CAUSER and a CAUSEE—used in the structure of ―NP1+Grammatical_(TAKE)+NP2+V/VP‖ with the obligate sense of ―affectedness‖. However, the major problem in Mandarin is that there are abundant examples that use the same structure but involve an intangible NP2, as shown in (70). The sense of ―affectedness‖ becomes futile when the NP2 is not a concrete entity. (70) BA_TAKE in ―agent-patient‖ event in Mandarin a. 洋基队把气势打出来了。 Yangjidui ba qishi da chulai le. Yankees TAKE vigor hit come Asp ‗The Yankees played the game with vigor.‘ b. 洋基队把比分追平了。 Yangjidui ba bifeng zhui ping le. Yankees TAKE score chase even Asp ‗The Yankees has evened the score.‘ Because the intangible NP2 cannot be affected physically, the image schema of a disposal event cannot be applied properly to example (70). Without transferring of a causative force, the grammatical form of TAKE in (70) has nothing to do with the marker of CAUSEE. Example (70) demonstrates a derived event, which involves an AGENT and a PATIENT (represented as an ―agent-patient‖ event). The grammatical form of TAKE in (70) is a marker of PATIENT. Because the ―agent-patient‖ event in Mandarin is very common and it is normally used in the SVO structure, the word order of (70) can be changed into SVO without any significant change to the cognitive meaning, as shown in (71). Note that examples (71a-b) demonstrate a transitive construction in Mandarin, which need not involve any marker for NPs. 120 (71) Agent-patient event in Mandarin a. 洋基队打出气势来了。 Yangjidui da chu qishi lai le. Yankees hit out vigor come Asp ‗The Yankees played the game with vigor.‘ b. 洋基队追平比分了。 Yangjidui zhui ping bifeng le. Yankees chase even score Asp ‗The Yankees has evened the score.‘ The fundamental difference between the disposal event (such as (69)) and the ―agent-patient‖ event (such as (70)) is that the former cannot change its word order into SVO. When the word order of (69) is switched into SVO, the sentences become ungrammatical, as shown in (72). This shows that the marker of CAUSEE in the disposal event is obligatory. (72) Ungrammatical word order in disposal event in Mandarin *a. 那粒界外球打晕一位观众了。 Na li jiewaiqiu da yun yi wei guanzhong le. that Cl foul-ball hit faint one Cl audience Asp ‗That foul ball hit an audience member and made him faint.‘ *b. 这场球赛累坏我了。 Zhe chang qiusai lei huai wo lei. this Cl game tired badly I Asp ‗This game really tired me out.‘ *c. 张三打球得很远。 Zhangsan da qiu de hen yuan. NAME hit ball DE very far ‗Zhangsan hit the ball very far.‘ 3.2.5. Discussion Based on the discussion of image schemata above, the construction with TAKE in Ancient Chinese was composed of four independent serial events in terms of the ―thing- transferred event‖, the ―thing-located event‖, the ―thing-transformed event‖, and the 121 ―instrumental causative event‖. However, the construction with TAKE in Modern Chinese consists of at least six basic events: the thing-transferred event, the thing-located event, the thing-transformed event, the instrumental causative event, the disposal event, and the agent-patient event, as shown in Figure 3.11. The latter two are only used in a single event. Figure 3.11. Development of constructions with TAKE The disposal event is a new form that was derived from the instrumental causative event by the Tang Dynasty due to the omission of the INSTRUMENT. The semantic relationship between the disposal event and the instrument causative event is very close because both events contain the senses of ―manipulativeness‖ and ―affectedness‖ in their image schemata, event though the serial event has become a single event. The other new form is the ―agent-patient event‖ which was derived from the disposal event when the nominal argument in the event is not a tangible object. The agent-patient event and the disposal event share the same syntactic structure, but they differ completely in semantics because TAKE is used to mark the PATIENT in the former but marks the CAUSEE in the latter. 122 Each event can involve a metaphorical function when the object in the event becomes abstract. The verb ―to take‖ becomes a marker of the THING object after metaphorical extension. Therefore, the construction with TAKE in Modern Chinese has numerous forms after the diachronic development. The origin of the puzzle of the disposal construction in terms of the BA construction is due to its complicated diachronic evolution in semantics as well as syntax. Therefore, before creating a theory to explain the disposal construction, it is necessary to understand its diachronic development. It is impossible to adopt only one theory in order to analyze all types of constructions with TAKE. 123 CHAPTER IV SYNTACTIC CHANGE: STRUCTURAL REANALYSIS As a native speaker of Taiwanese Southern Min, the use of Mandarin gei ‗to give‘ has confused me since I was young. When my classmate complained to our teacher by using the Mandarin sentence ‗ta gei wo da‘, which means ‗he beat me‘, I always laughed. For me, this sentence can only be interpreted as ‗he let me beat him‘ or ‗he was beaten by me‘, because Taiwanese ho ‗to give‘ used in the same sentence can only indicate a causative or passive expression. 59 As a young child, it was impossible for me to accept that Mandarin gei ‗to give‘ can express a disposal meaning.60 Even now, I still cannot easily accept it. My classmate‘s utterance reflects how he understood the use of gei ‗to give‘ in Mandarin. The use of Mandarin gei ‗to give‘ functions as a disposal/patient marker and is nothing but a natural metaphorical extension. On the contrary, the use of ‗to give‘ functioning as an agentive passive marker in Mandarin and Taiwanese conflicts the metaphorical process, because the motion of ‗to give‘ that involves a dative can be associated with a patient, not an agent. Why is it that I can accept an abnormal ‗passive‘ meaning but cannot accept a normal ‗disposal‘ meaning? 59 Although I gloss ho of Taiwanese as ‗to give‘ here, the etymology of ho in Taiwanese must be reconstructed carefully. The sound ho written as 互 in Taiwanese may not have meant ‗to give‘ during the ancient stage. The study of GIVE in Taiwanese requires collecting GIVE-sentences from all of the dialects of Southern Min. 60 The idea of the disposal construction (处置式) (known as the BA construction (把字句)) has been discussed in Chapter III. It was first proposed by Wang Li (1945; 1984:119-121). The construction ‗X ba Y Z‘ denotes ‗X disposes of Y in the way described by Z‘. 124 In Chinese, the agentive passive markers gei (给) and jiao (教/叫) are puzzling in that they are used to mark both a patient in the disposal construction and an agent in the passive construction. Why do two markers that seem to have opposite functions employ the identical form? This distinctive phenomenon has attracted the attention of several linguists interested in the grammar of Chinese, e.g., Bennett (1981), Jiang (2002, 2009), Newman (1996), Shi & Wang (2009), Wu (1999), Xing (2006), and Yap & Iwasaki (2003). It is necessary to provide a brief introduction to some of these works (see Sections 4.1.1-4.1.5 and Section 4.2.1) before proposing my personal hypothesis. According to Shi & Wang (2009) and Wu‘s (1999) cross-dialectal presentation of disposal and passive markers in dialects of Mandarin and Xiang, there are at least four types of lexical verbs that can serve as a single source of the passive and the disposal marker. 61 They are ‗GIVE type‘, ‗TAKE type‘, ‗CALL type‘, and ‗ALLOW type‘, as shown in Table 4.1. Table 4.1 provides some locations where speakers use a single morpheme for the passive and the disposal marker. It also includes the pronunciation and written characters of the morphemes in each dialect of Mandarin or Xiang (if applicable). 61 Xiang is one of the Chinese dialects spoken in Hunan Province. 125 Table 4.1. A shared form of the passive/disposal marker in certain Chinese dialects Location Pronunciation Character GIVE type Beijing (北京) /gei/ 给 Ningyuan, Hunan (湖南宁远) /ε/ 挨 Linwu, Hunan (湖南临武) /a/ 阿 Changning, Hunan(湖南长宁) /te/ 得 Changsha, Hunan (湖南长沙) /pa/ 把 Jiaocheng, Shanxi (山西交城) /kuɣʔ/ 给 Xiuning, Anhui (安徽休宁) /te/ 提 Susong, Anhui (安徽宿松) /ma/ TAKE type Lengshuijiang, Hunan (湖南冷水江) /la/ 拿 Shaodong, Hunan (湖南邵東) /pa/ 把62 Rucheng, Hunan (湖南汝城) /no/ 拿 CALL type Yexian, Henan (河南叶县) /tçiau/ 叫 Xinjiang, Shanxi (山西新绛) /tʂao/ 招63 Tancheng, Shandong (山东郯城) 叫 ALLOW type Xuchang, Henan (河南许昌) 让 Take Beijing Mandarin as an example. Its verb gei, meaning ―to give‖, can be used as a disposal marker as well as a passive marker, as shown in (1) and (2), respectively. Therefore, Beijing Mandarin belongs to the GIVE type. (1) gei ―to give‖ in the disposal construction (Shi & Wang 2009:43) 我给电视机修好了。 Wo gei dianshiji xiu hao le. I give television repair good Asp ‗I have repaired the television.‘ 62 Note that the sound /pa/ spoken in Hunan Shaodong has two meanings: one is ‗to give‘; the other is ‗to take/to hold‘. 63 Shi and Wang (2009) do not classify the word /tʂao/ written as 招. Based on its pronunciation, I put it in the CALL type. 126 (2) gei ―to give‖ in the passive construction 花瓶给张三摔破了。 Huaping gei Zhangsan shuaipo le. vase give NAME fall break-down Asp ‗The vase was broken by Zhangsan.‘ In addition to basic information provided in Table 4.1, there are certain points that should be noted. First, the majority of Chinese dialects and their distributive locations use one morpheme as a disposal marker and the other as a passive marker. Based on previous studies (Shi & Wang 2009, and Wu 1999), I conclude that there are only approximately sixty locations where speakers employ the same morpheme for both the disposal marker and the passive marker. In other words, this puzzling phenomenon developed in the sixty locations does not represent an overwhelming tendency; it occurs sporadically in a few dispersive areas with small groups of speakers. Among these locations, over eighty-five percent apply the GIVE type. Second, in Table 4.1 several characters do not match their lexical meaning, especially in the GIVE type. It should be noted that many of the Chinese characters in dialects are created via a strategy of sound borrowing because most Chinese dialects are colloquial and did not have their own written forms until modern times. When a dialect speaker wanted to record colloquial sounds, he borrowed the existent characters from Standard Chinese or available written forms into his dialect based on identical or similar pronunciations. For example, the sound /te/ in Changning is written as 得 meaning ―to receive‖, while in Xiuning it is written as 提 meaning ―to carry‖. It is more likely that the sound /te/ could have nothing to do with the meaning ―to receive‖ or ―to carry‖ because 127 these written forms were created via a sound borrowing. The etymological meaning of /te/ should be ―to give‖. The other example is /pa/ written as 把 in Changsha, as shown in the GIVE-type. The Chinese character 把 which means ―to take‖ in Classical Chinese has an opposite meaning from the lexical verb ―to give‖. The sound /pa/ spoken in Changsha could have nothing to do with ―to take‖; its etymological meaning could be ―to give‖. It is not surprising that Chinese dialects have taken a p-initial or t-initial sound like /pa/ or /te/ meaning ―to give‖ from the Proto-Sino-Tibetan language. According to Matisoff‘s (2003:652) reconstruction of the verb ―to give‖ in Proto-Tibeto-Burman, both *bəy and *ter are possible sounds. Therefore, the lexical meaning of de and pa used in Changning and Xiuning, and Changsha Xiang, respectively, is ―to give‖. They belong to the GIVE type. 4.1. Previous studies To provide a starting point for the current study, I am going to introduce certain previous studies, which can be divided into two categories. The argument which does not support a ―causative-to-passive‖ development (posed by Bennett (1981), Newman (1996), Shi & Wang (2009), Wu (1999), or Xing (2006)) will be discussed in Sections 4.1.1- 4.1.5, while the argument which supports the ―causative-to-passive‖ development (posed by Jiang (1997) and Yap & Iwasaki (2003)) will be provided in Section 4.2. Several questions concerning these previous arguments will be discussed in Section 4.3. 128 4.1.1. Bennett‟s (1981) observations In 1981, Bennett noticed a surprising phenomenon in Chinese: gei (给) ‗to give‘ in Mandarin can be used as a lexical verb, a dative marker, as well as a passive marker; in some Chinese dialects it can even be used as a disposal marker (1981:80-82). He mentions that the derivation of the dative preposition gei from the lexical verb meaning ‗to give‘ is a natural process via syntactic reanalysis, which occurs commonly and cross- linguistically. However, it is certainly striking that Mandarin also uses gei as a passive marker indicating an agent because the development of gei ‗to give‘ into an agent marker in the passive construction is unexpected and represents a dramatic reversal. For example, ―John gave me a beating‖ should mean ―John beat me‖ rather than ―John was beaten by me‖. It is hard to see exactly how the syntactic reanalysis could have occurred such that Mandarin gei ‗to give‘ developed from a verb to a passive marker; therefore, Bennett suggests using different kinds of explanations for this unusual phenomenon called ―verb- to-preposition shift‖ (his terminology). Moreover, he suggests that the process of ―verb- to-preposition shift‖ requires reference to specific Chinese facts in that there is a very close relationship between passive and disposal constructions in some Chinese dialects and this apparently applies to some speakers when using gei in Mandarin. Note that it is very difficult to understand Bennett‘s statement because it is hard to find examples to prove that there is really a very close relationship between the passive and the disposal construction. In fact, the relationship between the passive and the causative construction in Chinese is very close because a GIVE-type passive always has a corresponding causative expression, which can cause an ambiguity of meaning. However, 129 the relationship between the passive and the disposal construction is distant because these two expressions are opposite in all dialects of Chinese. 4.1.2. Newman‟s (1996) considerations In 1996, Newman mentions the same puzzle, where gei (给) ‗to give‘ (represented as GEI) is used as a marker of the agent in a passive construction in Mandarin, and he suggests a number of factors that motivate this intriguing use of GEI (1996:196-198). Newman stresses that, in order to understand the existence of this construction, more than one factor needs to be considered in accounting for the GEI-passive in Mandarin. First, gei ‗to give‘ can be used in contexts involving an enablement sense. The enablement sense attached to GEI helps to motivate the agentive marking function in that the noun phrase to the right of GEI, which functions as a trajector with respect to the following predicate. Secondly, there has been a general trend in the history of Mandarin syntax involving a word order shift from SVO to SOV, and this historical fact is surely relevant in accounting for the agentive marker GEI in the passive construction. 64 Thirdly, in a topic-comment language such as Mandarin, there is a certain slipperiness to the semantic role played by a noun phrase at the beginning of a sentence, and it is possible for initial noun phrases to be interpreted as either the trajector or the landmark of the following verbal predicate. This variability in the semantic function of the initial noun phrase would obviously help to smooth the way for the development of 64 The word order shift from SVO to SOV in Chinese is usually associated with the occurrence of disposal constructions. I have provided my own explanation in Chapter III. In my opinion, it is the omission of an instrument from the instrumental causative construction that brings about the word order change. 130 the GEI-passive construction in which the initial noun phrase would function like a patient rather than an agent. Finally, in Mandarin, there is a benefactive prepositional use of GEI ―for the benefit of‖ whereby the benefactive phrase is typically also the motivation for an act. The NP in the benefactive phrase (gei + NP) refers to a person who not only benefits from some act but is also the cause of the act. In other words, the benefactive sense attached to GEI makes the benefactive NP comparable to the agentive sense, since both of the agent of an act and the person for whose benefit an act is carried out play a part in setting the act in motion. Note that none of the factors mentioned above can be reasonably illustrated, except the first one. Newman‘s ―enablement sense‖ is equivalent to the sense of ―permissive causative‖ (in my terminology). I will define this term in Section 4.4.1.1. 4.1.3. Wu‟s (1999) “stages of grammaticalization” In 1999, Wu tried to solve the puzzle concerning the GIVE-passive construction based on her investigation of certain dialects of Xiang spoken in Hunan Province. She points out that the disposal and the passive markers are derived from a pair of synonyms meaning ―to give‖ or ―to take‖ in some dialects of Xiang. She mentions that a single form used in both the disposal and passive markers is unnatural because, in Xiang, the disposal marker is used to introduce the patient and the passive marker is used to introduce the agent. It is obvious that the two markers have opposite functions. Wu proposes that the word pa meaning ―to give‖ used in both the disposal and the passive construction in Xiang dialects has experienced several stages of grammaticalization (1999:110). 131 In the first stage, pa ‗to give‘ is used in a serial verb construction, which involves two pa ‗to give‘, as shown in (a). (a) S (agent) + verb „to give‟ + indirect O (patient) + V (to give) +direct O In the second stage, the first lexical verb pa ―to give‖ in (a) has been syntactically reanalyzed into a preposition introducing the relationship between subject (agent) and indirect object (patient), as shown in (b). (b) S (agent) + prep „to give‟ + indirect O (patient) + V (to give) +direct O In the third stage, the preposition pa is used to express an adverse situation in that it turns to introduce the relationship between subject (patient) and object (agent), and pa becomes an agentive marker in the passive construction, as shown in (c). (c) S (patient) + prep „to give‟ + O (agent) + V (an adverse situation only) In the fourth stage, the preposition pa is used in the disposal construction to introduce the relationship between the subject (agent) and the object (patient), as shown in (d). (d) S (agent) + prep „to give‟ + O (patient) + V Wu finally concludes that the lexical meaning of pa ‗to give‘ has become weaker and weaker through the above four stages: ―lexical verb  dative marker  passive marker  disposal marker‖.65 Note that Wu‘s four stages of grammaticalization in pa ‗to give‘ are inexplicable because her analysis does not include the causative meaning. The term ―adverse situation‖ needs to be explained. She mentions that the passive construction in the 65 Wu (1999:109) mentions that when both disposal and passive constructions share the same marker, a sentence like [Zhang San + disposal/passive + Li Si + to beat + Comp] is definitely regarded as a passive sentence in the Xiang dialects. It cannot be used as a disposal sentence. In this sequence, she suggests that the disposal construction developed later than the passive construction. 132 dialects of Xiang can only be used for an adverse situation; in this sequence, she claims that the passive construction occurs when people want to express an adverse situation. In my opinion, when people use a passive construction (for instance, with a passive marker jian (见) or bei (被)) in Ancient Chinese), it normally involves a malefactive. Therefore, the passive form was conventionalized so as to be associated with an adverse situation. This process of conventionalization could have occurred in several dialects of Chinese, including Taiwanese Southern Min. 4.1.4. Xing‟s (2006) “subjectification” In 2006, Xing uses a new approach to solve the same puzzle concerning the gei- passive and the jiao-passive construction in Chinese. In her study, ―subjectification‖ relies primarily on ―the listener/reader‘s subjective understanding or interpretation of the intended meaning and is deducible from discourse, pragmatic, social and cultural context (2006:471)‖. She points out that the causative verb jiao ‗to call‘ (教/叫) and the dative verb gei ‗to give‘ (给) can be used to express a passive meaning (2006:472-473). Between the 7th and 9th centuries A.D., jiao was not only commonly used as a causative verb meaning ―to ask somebody to do something‖, but also to introduce an agent expressing a passive meaning, as shown in (3). In sentence (3a), jiao ‗to call‘ carries an ambiguous meaning in that it may be interpreted as having either the causative meaning ―to let somebody do something‖ or the passive meaning ―something done by somebody‖. In sentence (3b), it only conveys the passive meaning with its agent in the subsequent position. 133 (3) jiao in 9th century (A complete collection of Tang poems) (Xing 2006:472) a. 莫教人笑汝。 mo jiao ren xiao ru not is-let people laugh-at you ‗Don‘t let people laugh at you.‘ Or ‗(You) should not be laughed at by people.‘ b. 春思翻教阿母疑。 chunsi fan jiao ‗amu yi spring-thinking turn-around pass mother suspect ‗Thoughts of love were nevertheless suspected by mother.‘ During the 17th century, gei ‗to give‘ developed its passive meaning, as shown in (4). In sentence (4a), gei is a dative verb meaning ―to give somebody something‖, while in (4b), it introduces an agent, yielding a passive meaning. (4) gei in 17th Century (Honglou meng) (Xing 2006:472) a. 我给他带了许多好东西来了。 wo gei ta dai le xuduo hao dongxi lai le. 1sg give 3sg bring asp lot good stuff come part ‗I brought him a lot of good stuff.‘ b. 就是天也是给气运使唤着。 jiushi tian ye shi gei qiyun shihuan zhe even sky also is pass air use stat ‗Even the sky is used by air.‘ Her argument is that the unmarked feature of Chinese nouns and verbs opens a window for speakers or listeners to rearrange the word order or interpret jiao and gei based on their subjective understanding of the surrounding entities. For example, it is not always clear to the listener whether the grammatical marker jiao in (3) is causative or passive; therefore, subjectification is at work, depending on the listener‘s subjective understanding of the discourse when determining or interpreting jiao as either causative or passive. Note that although Xing provides historical records to support the idea of ‗subjectification‘, her solution has nothing to do with the diachronic development of the 134 agentive passive markers. The fundamental problem is that the interpretation of (3a) relies completely on Xing‘s personal judgment. Without any historical evidence, it is impossible to say that ancient people can interpret jiao as either causative or passive in (3a).The ambiguity between the causative expression and the passive expression is a synchronic fact of existence. Xing‘s ‗subjectification‘ only touches on a synchronic issue. 4.1.5. Shi & Wang‟s (2009) arguments In 2009, Shi and Wang attempted to explore why some Chinese dialects employ the same marker for both their disposal and passive constructions. They are concerned with the same puzzle in which the marked constituents in the disposal and passive construction in Chinese are exact opposites, where the former is a patient but the latter is an agent. To solve this puzzle, they argue that both jiao (叫) and gei (给) are potential candidates to be used as both the disposal and passive marker due to their multi-vocal meanings in syntactic environments. Using gei ‗to give‘ as an example, they argue that gei ‗to give‘ has a bidirectional meaning. In both (5) and (6), the NP Laowang ‗NAME‘ can be interpreted as a giver or a recipient depending on the context. If the omitted NP is a giver, then Laowang (NAME) is a recipient; if the omitted NP is a recipient, then Laowang ‗NAME‘ is a giver. Note that the patient liwu ‗gift‘ in (5) is topicalized, in the initial position of the sentence. (5) gei in topicalized and NP-omitted sentences (Shi & Wang 2009:50) (礼物,) 老王已经给了。 Liwu, Laowang yijing gei le. gift, NAME already give Asp ‗(Speaking of the gift,) Laowang has given it (to someone).‘ ‗(Speaking of the gift,) (someone) has given it to Laowang.‘ 135 (6) gei in NP-omitted sentences (Shi & Wang 2009:50) 老王已经给了礼物啦。 Laowang yijing gei le liwu la. NAME already give Asp gift Prt ‗Laowang has given the gift (to someone).‘ ‗(Someone) has given the gift to Laowang.‘ Therefore, gei ‗to give‘ can be understood as ―to give‖ or ―to receive‖ in a sentence. 4.2. Previous studies: causative-to-passive development The ―causative-to-passive‖ development of gei ‗to give‘ (给) and jiao ‗to call‘ (叫 /教) has been explored by a few linguists. Shaoyu Jiang (1997) mentions that jiao ‗to call‘ was originally used in the causative construction, and the passive function of jiao is derived from the causative jiao. The agentless construction of jiao brings about ambiguous meanings, which can trigger the semantic extension of the passive function (1997:302). In 2002, Jiang further argues that the passive gei is also derived from the causative gei, and the causative gei is grammaticalized from the lexical gei meaning ―to give‖ (2002:159). In 2009, Jiang points out that the causative functions of jiao ‗to teach‘ (教) and jiao ‗to call‘ (叫) developed individually from their lexical meanings. The ―causative-to-passive‖ development of jiao (教) occured several hundreds of years earlier than that of jiao (叫) (2009:87-89). Basically speaking, Jiang‘s claims for the causative-to-passive development of gei and jiao are based on his observations from historical documents in Classical Chinese. He does not provide a detailed procedure for these diachronic developments based on a theoretical ground. 136 The main purpose of this study is to find a possible linguistic process for explaining the development of agentive passive markers in Chinese. Before explaining my hypothesis, I would like to introduce the interesting idea of a grammaticalization pathway involved in the causative-to-passive development posed by Yap and Iwasaki (2003). 4.2.1. Yap & Iwasaki‟s (2003) grammaticalization pathway In 2003, Yap and Iwasaki propose a diachronic view for a possible development of give-passive constructions, which they support through cross-linguistic comparisons. First, they claim that the emergence of give-passive from give-causative is a natural and fairly robust phenomenon, and this development is generally mediated in the permissive and reflexive environment, as shown in the following grammaticalization pathway: Lexical „give‟ permissive causative „give‟ reflexive „give‟ passive „give‟ In Mandarin, for example, gei meaning ‗to give‘ can be used to indicate the permissive causative (X lets Y do Z), which emerges in the non-coercive context, as shown in example (7). (7) gei in permissive causative (Yap and Iwasaki 2003:421) 我给你猜个谜语。 Wo gei ni cai ge miyu. I GIVE you guess Cl riddle ‗I (will) let you guess a riddle.‘ The reflexive-causative use of gei (X gives self to be perceived by Y) can be seen in example (8). 137 (8) gei in reflexive-causative (Yap and Iwasaki 2003:422) 李四给张三看见了。 Lisi gei Zhangsan kanjian-le. Lisi GIVE Zhangsan see-Asp lit. ‗Lisi gave Zhangsan see (him).‘ = ‗Lisi was seen by Zhangsan.‘ The ordinary passive gei can specify an overt agent, as shown in (9a); it can also be used in an agentless construction, as shown in (9b). (9) gei in passive (Yap and Iwasaki 2003:422) a. 房子给土匪烧了。 Fangzi gei tufei shao le. House GIVE hooligan burn Asp ‗The house was burned down by the hooligans.‘ b. 房子给烧了。 Fangzi gei shao le. House GIVE burn Asp ‗The house was burned down.‘ Secondly, their comparative perspective shows differential degrees of grammaticalization for the ―causative-to-passive‖ development, in that give-passives are most common in languages such as Mandarin and Cantonese, where the subject-agency can be weakened or lost. However, in languages such as Thai, Khmer, and Vietnamese, which lack of this ―causative-to-passive‖ phenomenon, the give-involved constructions strongly favor a highly agentive and volitional subject and tend to block the emergence of give-passive interpretations (2003:431). Third, their study shows that the ‗let‘-type morpheme rang ‗to let‘ (让), which is used in the passive construction in Mandarin, has evolved through an identical ―causative-to-passive‖ development. Mandarin rang means ‗allow, permit, let‘; it can be used in a permissive causative construction, as shown in (10). 138 (10) rang in permissive causative (Yap and Iwasaki 2003:434) 你让我再想想。 Ni rang wo zai xiang xiang. You ALLOW me again think think ‗Let me think it over.‘ The reflexive-causative use of rang can be illustrated in (10). (11) rang in reflexive causative (Yap and Iwasaki 2003:434) 我让你吓坏了。 Wo rang ni xia huai le. I ALLOW you have-a-fright ruin Asp ‗I let you give me a terrible fright.‘66 = ‗I really got a fright from you.‘ The ordinary passive rang can be exemplified in (12). (12) rang in passive (Yap and Iwasaki 2003:434) 房子让水冲走了。 Fangzi rang shui chong zhou le. House ALLOW water wash go Asp ‗The house was washed away by the water.‘ Fourth, considering the other causative morpheme jiao derived from a verb meaning ―to call, holler, yell‖, Yap and Iwasaki (2003:438) argue that jiao cannot readily extend its reflexive-causative constructions into reflexive-passive contexts where subject agentivity and volitionality are low. Therefore, the ―causative-to-passive‖ development of jiao ‗to call‘ has probably evolved without the direct mediation or facilitation of reflexive contexts. The following schema shows that the ―causative-to-passive‖ development of gei ‗to give‘ and rang ‗to let‘ follows the same pathway: lexical  permissive causative  reflexive causative  reflexive passive  passive, but jiao ‗to call‘ does not. (a). gei ‗to give‘ and rang ‗to let‘ in Mandarin: lexical permissive causative reflexive causative reflexive passive passive 66 The first translation provided here is very odd. The causative meaning seems inapplicable in (11). 139 (b). jiao ‗to call‘ in Mandarin: lexical permissive causative reflexive causative (not available)  passive Since jiao ‗to call‘ does not follow the grammaticalization pathway above, Yap and Iwasaki claim that jiao ‗to call‘ may have acquired passive functions via analogy with other causative morphemes like gei ‗to give‘ and rang ‗to let‘. Considering Yap and Iwasaki‘s assumption, there are two important terms, ―reflexive causative‖ and ―reflexive passive‖, which need to be discussed here. At the beginning of this chapter, I mentioned the sentence ta gei wo da, as shown in (13), which can be understood as ‗he beat me‘, ‗he let me beat him‘, and ‗he was beaten by me‘ by the native speaker of Mandarin; however, the expression of ‗he beat me‘ cannot be accepted by the native speaker of Taiwanese. This is because, in Taiwanese, the equivalent sentence i ho ua pha cannot express the disposal meaning, as shown in (14). The disposal meaning in Taiwanese requires a ―comitative marker‖ ka/kã (Chappell 2000:277, and 2006:12-14), as shown in (15). 67 (13) gei ‗to give‘ in Mandarin 他给我打 ta gei wo da he GIVE me beat ‗he beat me‘ (disposal expression) ‗he let me beat him‘ (causative expressive) ‗he was beaten by me‘ (passive expressive) (14) ho ‗to give‘ in Taiwanese 伊互我打 i ho ua pha he GIVE me beat *‗he beat me‘ (disposal expression) ‗he let me beat him‘ (causative expressive) ‗he was beaten by me‘ (passive expressive) 67 Chappell (2000:277) defines ka/kã, written as 共, as a comitative marker because its lexical meaning is very similar to English ―with‖. 140 (15) Comitative marker ka/kã in disposal 伊共我打 i ka ua pha he KA me beat ‗he beat me‘ (disposal expression) *‗he let me beat him‘ (causative expressive) *‗he was beaten by me‘ (passive expressive) The causative expression ‗he let me beat him‘ in (13) and (14) does involve a reflexive sense because the agent is also the patient; it is reasonable for Yap and Iwasaki to call this ―reflexive causative‖. However, I cannot find any suitable example for the expression of ―reflexive passive‖. Note that Chinese tends to involve a malefactive (or an ‗adverse situation‘ in Wu‘s (1999) terminology) in the passive construction. It is very odd to see a reflexive expression co-occur with a malefactive in Chinese. 4.3. Questions on previous studies All of the previous studies more or less try to explain why two markers (namely ―passive marker‖ vs. ―disposal marker‖ or ―agentive marker‖ vs. ―dative marker‖) seem to have opposite functions but employ an identical form, such as gei ‗to give‘, jiao ‗to call‘, rang ‗to let‘, or na ‗to take‘. So far, I have discussed Wu‘s (1999) argument on adverse situation in passive constructions, Shi and Wang‘s (2009) idea of bidirectional meaning, and Yap and Iwasaki's (2003) opinion about the mediation of reflexive contexts. However, there are several inadequacies that need to be reconsidered. First, most of the aforementioned studies focus on the GIVE type (namely gei) and/or the CALL type (namely jiao). Only Yap and Iwasaki (2003) provide discussion of the ALLOW type (namely rang). None of the studies proposed any solution to the TAKE type (namely na or pa/ba), even though Wu (1999:107) has mentioned that ―the lexical source of the disposal and the passive markers is either ‗to give‘ or ‗to take‘‖. 141 Secondly, those linguists who do not support the ―causative-to-passive‖ development should explain why their analyses exclude the causative function of ―give‖. In fact, the use of the morpheme ―to give‖ to express causative functions is a pan-Chinese phenomenon. Although Newman (1996:196) mentions that gei ‗to give‘ can be used in contexts involving an enablement sense that can help gei motivate the agentive marking function, he does not believe that the passive gei can emerge straight from the causative gei. Third, regarding linguists who use a grammaticalization pathway as a solution (e.g., Wu‘s four stages: lexical verb dative marker passive marker disposal marker, and Yap and Iwasaki‘s causative-to-passive pathway: lexical ‗give‘ permissive causative ‗give‘ reflective ‗give‘ passive ‗give‘), it appears that the examples these researchers provide for each stage are individual and independent. In fact, it is unlikely that the disposal marker evolved from the passive marker (cf. Wu (1999)). In addition, the reflexive-passive construction seems semantically impossible in Chinese; I cannot even find a grammatical example of it. How can the ―reflexive-passive‖ construction be treated as the mediation during the ―causative-to-passive‖ development (cf. Yap and Iwasaki (2003))? Fourth, gei ‗to give‘, jiao ‗to call‘, rang ‗to let‘, and na ‗to take‘ could not carry a bidirectional meaning. It is semantically illogical that gei ‗to give‘ can be understood as ―to give‖ and ―to receive‖ in a sentence (cf. Shi and Wang (2009)). In fact, from examples (5) and (6) above, it is clear that the bidirectional sense happens when the sentence structure changes. It is the omission of the NP in sentences that causes the ambiguity of the construction. 142 Fifth, there is no evidence that can prove that the causative-to-passive development of gei ‗to give‘, jiao ‗to call‘, rang ‗to let‘, or na ‗to take‘ underwent different processes. It may be inappropriate to use ―analogy‖ as a solution until it is clear that no internal factor from the language can be found. In fact, the morphemes may acquire the passive function via the same process (cf. Yap and Iwasaki (2003)). In the next section, I will begin to explain my personal hypothesis. I claim that the agentive passive marker arises from three types of causative constructions; their configurations can be demonstrated as [NP1 primary causer + gei/ na/ jiao/ rang + NP2 secondary causer + V + NP3 causee] (see Section 4.4.1). This ―causative-to-passive‖ process involves three steps: (1) NP3 causee is topicalized to the sentence-initial position (see Section 4.4.2), (2) NP1 primary causer is omitted by the speaker (see Section 4.4.3), and (3) the final configuration [NP3 causee + gei/ na/ jiao/ rang + NP2 secondary causer + V] is reanalyzed as a passive construction by the listener (see Section 4.4.4). 4.4. Development of agentive passive markers In order to solve the puzzle of the agentive passive marker used in Chinese, it is necessary to understand what kind of sentences may have served as input for developing agentive passive constructions. The input is the causative construction. Let us get started by looking at John Newman‘s (1996:196) example, as shown in (16). In (16), mao ‗cat‘ is understood as the entity which did the eating. Although the passive sentence ―the goldfish was eaten by the cat‖ is grammatical, its semantics are insufficient. It appears that the ‗cat‘, being a creature which fears touching water, could not successfully eat the goldfish by itself. There should be a primary causer making the cat do the eating. Without the primary causer, the event of eating will never happen. 143 In a common speech act, the primary causer can be elided when it is already known to both the speaker and the listener. In natural discourse, a logical meaning of (16) could be ―the goldfish, I let the cat eat it‖ in which the ‗goldfish‘ is a topicalized causee, and ‗I‘ is the omitted primary causer. In comparison with the primary causer, the ‗cat‘ is defined as a secondary causer because it is not the chief entity which controls and causes the entire event. (16) Agentive marker gei in passive (Newman 1996:196) 金鱼给猫吃了。 Jinyu gei mao chi-le. Goldfish give cat was eaten-Asp ‗The goldfish was eaten by the cat.‘ Based on the observation above, I suggest that the process in the development of agentive passive markers involves three thematic roles (primary causer, secondary causer, and causee) in three types of causative constructions: permissive causative, instrumental causative, and manipulative causative construction. This classification is established completely based on typological observation of Chinese, since I have no historical evidence for it so far. Details are provided in the following sections. Note that in most cases, I will use Mandarin to furnish examples. 4.4.1. Three types of causativity All of the agentive passive markers used in Mandarin, such as gei, jiao, rang, and na, can also be used in causative constructions. They can be categorized into three causative types: permissive causative, instrumental causative, and manipulative causative. Gei ‗to give‘ and rang ‗to let‘ occur in the permissive causative construction (see Section 4.4.1.1); na ‗to take‘ occurs in the instrumental causative construction (see Section 144 4.4.1.2); jiao ‗to call‘ occurs in the manipulative causative construction (see Section 4.4.1.3). It is very important to realize that all of the three causative types share an identical position class, as shown in Table 4.2. The word order can be represented as [NP1 + gei/ na/ jiao/ rang + NP2 + V + NP3]. The sentence that involves this word order has served as input for developing agentive passive constructions. Table 4.2. Position class in causativity of gei, jiao, rang, or na It is also important to understand the thematic role assigned to each constituent in Table 4.2. Table 4.3 shows that NP1 is assigned as the primary causer, NP2 is the secondary causer (not restricted to humans or animates), and NP3 is associated with the causee (not restricted to humans or animates). The order of thematic roles in the construction can be represented as: [primary causer + gei/ na/ jiao/ rang + secondary causer + V + causee]. Note that in the clause [secondary causer + V + causee], as shown in (16) above, the secondary causer is also an agent, and the causee is also a patient. In order to stress the sense of causativity and to prevent confusion, I will use causer and causee rather than agent and patient. NP1 gei (给) rang (让) na (拿) jiao (叫) NP2 V NP3 145 Table 4.3. Thematic roles in causativity of gei, jiao, rang, or na According to the position class and thematic roles provided above, a concrete representation for the three types of causative constructions can be determined: [NP1 primary causer + gei/ na/ jiao/ rang + NP2 secondary causer + V + NP3 causee]. Examples of each type are given in the following sections. 4.4.1.1. Permissive causative Causativity is a complex notion that can at least be simplistically understood as a) event X causes event Y, b) person A causes event Y, c) person A causes some entity B (animate or inanimate) to change, or d) person A causes person B to do Y (Newman 1996:172). Conceptualizing the ‗giver‘ as a ‗causer‘ is a common metaphorical extension based on similarities between the giver and the causer. In particular, the sense of causativity is often ascribed to the meaning of lexical ‗to give‘ because the ‗giver‘ is said to be the entity which causes a situation in which the recipient comes to have the theme object (1996:172). Therefore, gei ‗to give‘ (给) can be used in the causative construction cross-linguistically. Mandarin sentences that involve causative gei are considered permissive causative, denoting ―person A enables person/animate B to do something‖. Examples are shown in (17). Primary Causer gei (给) rang (让) na (拿) jiao (叫) Secondary Causer V Causee 146 (17) gei in a permissive causative construction a. 我给猫咪吃掉了那尾金鱼。 Wo gei maomi chidiao le na wei jinyu. 1S give kitty eat-off Asp that Cl goldfish ‗I let the kitty eat that goldfish.‘ b. 我给张三预支了下个月的薪水 。 Wo gei Zhangsan yuzhi le xia ge yue de xinshui. 1S give Name draw-in-advance Asp next Cl month de salary ‗I let Zhangsan draw next month‘s salary in advance.‘ The Mandarin causative marker rang (让) meaning ‗to let‘ originates from its etymological meaning ‗to yield‘ via a metaphorical extension. Its function in the permissive causative is no different from that of the causative gei in (17). Indeed, gei and rang can be interchanged, as shown in (18). In (18), all of the components are the same as those in (17) except for the causative markers, gei and rang. (18) rang in a permissive causative construction a. 我让猫咪吃掉了那尾金鱼。 Wo rang maomi chi diao le na wei jinyu. 1S let kitty eat-off Asp that Cl goldfish ‗I let the kitty eat that goldfish.‘ b. 我让张三预支了下个月的薪水 。 Wo rang Zhangsan yuzhi le xia ge yue de xinshui. 1S let Name draw-in-advance Asp next Cl month de salary ‗I let Zhangsan draw next month‘s salary in advance.‘ In (17a) and (18a), for example, the first NP wo ‗I‘ is associated with the primary causer, the second NP maomi ‗kitty‘ is associated with the secondary causer, and the third NP jinyu ‗goldfish‘ is the causee. 4.4.1.2. Instrumental causative The instrumental causative involving na ‗to take‘ (拿) can serve to express the notion of ‗person A causes event X with an instrument (where ‗event‘ is usually referred 147 to action, process, or change of state)‖. The instrumental causative is a causative construction that involves an instrument. When completing an action with an instrument in hand, the causative force is transferred from the primary causer to the causee via the instrument. In this case, the instrument in terms of secondary causer controlled by the primary causer is also a contributing factor imposing the force on the causee in the later event X. Examples of the instrumental causative are shown in (19). (19) na in an instrumental causative construction a. 我拿石头堵死了那个老鼠洞。 Wo na shitou dusi le na ge laoshu dong. 1S take stone block Asp that Cl rat's hole ‗I took a stone to block the rat's hole.‘ b. 我拿手枪吓跑了那群土匪 。 Wo na shouqiang xiapao le na qun tufei. 1S take pistol scare-away Asp that Cl bandit ‗I took a pistol to scare those bandits away.‘ In (19b), for example, the first NP wo ‗I‘ is associated with the primary causer, the second instrumental NP shouqiang ‗pistol‘ is associated with the secondary causer, and the third NP tufei ‗brigand‘ is the causee. 4.4.1.3. Manipulative causative The manipulative causative involves the manipulation of another person in order to have that person to do something. This sense, represented as ―have someone to do something‖, can be carried by jiao ‗to call‘ (叫) in Mandarin.68 Manipulative causative can serve to express the interpersonal causative action of ―person A causes/asks person B to do something‖. Examples are shown in (20). 68 Note that the causative jiao could originate from its etymological meaning ‗to yell‘ via metaphorical extension. 148 (20) jiao in a manipulative causative construction a. 我叫工人拆掉了墙上的挂钟。 Wo jiao gongren chaidiao le qiangshang de guazhong. 1S call worker remove Asp wall-on Gen wall-clock ‗I called the worker to remove the wall clock.‘ b. 我叫张三修好了那台电视机 。 Wo jiao Zhangsan xiuhao le na tai dianshiji. 1S call Zhangsan fix Asp that Cl television ‗I called Zhangsan to fix the television.‘ In (20a), for example, the first NP wo ‗I‘ is associated with the primary causer, the second NP gongren ‗worker‘ is associated with the secondary causer, and the third NP guazhong ‗wall-clock‘ is the causee. In the following sections (Sections 4.4.2-4.4.4), the ―causative-to-passive‖ process will be shown to involve three steps, topicalization, omission of primary causer, and structural reanalysis, in all types of causativity. 4.4.2. Topicalization The development of agentive passive markers in Mandarin involves two stages of structural change and one stage of structural reanalysis. The first step is to topicalize the causee from sentences (17-20). The sentence-initial NP in a topic-comment language like Chinese is, as expected, a topic. The property of the topic in the three causative constructions is semantically a causee; it is always definite since it represents part of the presupposed information, and it can be separated from the rest of the sentence in which it overtly occurs preceding a pause particle. The later NPs are primary and secondary causers. In the following I will illustrate the topicalized causee, as shown from (21) to (24). The constituent order after 149 topicalization in all of these examples can be represented as [NP3 causee, NP1 primary causer + gei/ na/ jiao/ rang + NP2 secondary causer + V], as shown in Table 4.4. Table 4.4. Position class after topicalization of NP3 Example (21) shows the topicalized causee in the permissive causative construction with gei. The causee in sentence (17) above is fronted, in the sentence-initial position. (21) Topicalized causee from gei causative a. 那尾金鱼，我给猫咪吃掉了。 Na wei jinyu, wo gei maomi chidiao le. that Cl goldfish, 1S give kitty eat-off Asp ‗Speaking of that goldfish, I let the kitty eat (it).‘ b. 下个月的薪水，我给张三预支了。 Xia ge yue de xinshui, wo gei Zhangsan yuzhi le. next Cl month de salary, 1S give Zhangsan draw-in-advance Asp ‗Speaking of next month‘s salary, I let Zhangsan draw (it) in advance.‘ Example (22) shows the topicalized causee in the permissive causative construction with rang. The causee in sentence (18) above is fronted, in the sentence- initial position. (22) Topicalized causee from rang causative a. 那尾金鱼，我让猫咪吃掉了。 Na wei jinyu, wo rang maomi chi diao le. that Cl goldfish, 1S let kitty eat-off Asp ‗Speaking of that goldfish, I let the kitty eat (it).‘ b. 下个月的薪水，我让张三预支了。 Xia ge yue de xinshui, wo rang Zhangsan yuzhi le. next Cl month de salary, 1S let Zhangsan draw-in-advance Asp ‗Speaking of next month‘s salary, I let Zhangsan draw (it) in advance.‘ causee, (NP3,) primary causer (NP1) gei (给) rang (让) na (拿) jiao (叫) secondary causer (NP2) V 150 Example (23) shows the topicalized causee in the instrumental causative construction with na. The causee in the sentence (19) is fronted, in the sentence-initial position. (23) Topicalized causee from na causative a. 那个老鼠洞，我拿石头堵死了。 Na ge laoshu dong, wo na shitou dusi le. that Cl rat's hole, 1S take stone block Asp ‗Speaking of that rat's hole, I took a stone to block (it).‘ b. 那群土匪，我拿手枪吓跑了。 Na qun tufei, wo na shouqiang xia pao le. that Cl bandit, 1S take pistol scare-away Asp ‗Speaking of those bandits, I took a pistol to scare (them) away.‘ Example (24) shows the topicalized causee in the manipulative causative construction with jiao. The causee in the sentence (20) is fronted, in the sentence-initial position. (24) Topicalized causee from jiao causative a. 墙上的挂钟，我叫工人拆掉了。 Qiangshang de guazhong, wo jiao gongren chaidiao le. wall-on Gen wall-clock, 1S call worker remove Asp ‗Speaking of the wall clock, I called the worker to remove (it).‘ b. 那台电视机，我叫张三修好了。 Na tai dianshiji, wo jiao Zhangsan xiuhao le. that Cl television, 1S call Zhangsan fix Asp ‗Speaking of the television, I called Zhangsan to fix (it).‘ 4.4.3. Omission of primary causer The second step of the development of agentive passive markers is the omission of the co-referential NP that is semantically a primary causer, as shown in (25-28). Without breaking the topic continuity, the speaker is allowed to omit any co-referential noun phrase in natural discourse. The constituent order after the omission of the NP1 in 151 all the following examples can be represented as [NP3 causee, + gei/ na/ jiao/ rang + NP2 secondary causer + V], as shown in Table 4.5. Table 4.5. Position class after omission of NP1 Example (25) shows the sentence after the omission of the primary causer from the permissive causative construction with gei. The primary causer is omitted from the topicalized example (21) above. (25) Omission of primary causer in gei causative a. 那尾金鱼，给猫咪吃掉了。 Na wei jinyu, gei maomi chidiao le. that Cl goldfish, give kitty eat-off Asp ‗That goldfish, (I) let the kitty eat (it).‘ b. 下个月的薪水，给张三预支了。 Xia ge yue de xinshui, gei Zhangsan yuzhi le. next Cl month de salary, give Zhangsan draw-in-advance Asp ‗Next month‘s salary, (I) let Zhangsan draw (it) in advance.‘ Example (26) shows the sentence after the omission of the primary causer from the permissive causative construction with rang. The primary causer is omitted from the topicalized example (22) above. (26) Omission of primary causer in rang causative a. 那尾金鱼，让猫咪吃掉了。 Na wei jinyu, rang maomi chi diao le. that Cl goldfish, let kitty eat-off Asp ‗That goldfish, (I) let the kitty eat (it).‘ b. 下个月的薪水，让张三预支了。 Xia ge yue de xinshui, rang Zhangsan yuzhi le. next Cl month de salary, let Zhangsan draw-in-advance Asp ‗Next month‘s salary, (I) let Zhangsan draw (it) in advance.‘ Causee, (NP3,) ø (empty) gei (给) rang (让) na (拿) jiao (叫) secondary causer (NP2) V 152 Example (27) shows the sentence after the omission of the first primary causer from the instrumental causative construction with na. The primary causer is omitted from the topicalized example (23) above. (27) Omission of primary causer in na causative a. 那个老鼠洞，拿石头堵死了。 Na ge laoshu dong, na shitou dusi le. that Cl rat's hole, take stone block Asp ‗That rat‘s hole, (I) took a stone to block (it).‘ b. 那群土匪，拿手枪吓跑了。 Na qun tufei, na shouqiang xia pao le. that Cl bandit, take pistol scare-away Asp ‗Those bandits, (I) took a pistol to scare (them) away.‘ Example (28) shows the sentence after the omission of the primary causer in the manipulative causative construction with jiao. The primary causer is omitted from the topicalized example (24) above. (28) Omission of primary causer in jiao causative a. 墙上的挂钟，叫工人拆掉了。 Qiangshang de guazhong, jiao gongren chaidiao le. wall-on Gen wall-clock, call worker remove Asp ‗The wall clock, (I) called the worker to remove (it).‘ b. 那台电视机，叫张三修好了。 Na tai dianshiji, jiao Zhangsan xiuhao le. that Cl television, call Zhangsan fix Asp ‗That television, (I) called Zhangsan to fix (it).‘ 4.4.4. Structural reanalysis The final step of the development of agentive passive markers is a structural reanalysis due to neglect of the pause particle by listeners. The neglect of the pause particle makes the final configuration [NP3 causee + gei/ na/ jiao/ rang + NP2 secondary causer + V] carry a passive reading, as shown in Table 4.6. 153 The secondary causer can be understood as a ―causer‖ after the construction has been reanalyzed from causative to passive by the listener. Note that without any historical evidence, it is uncertain if ambiguity between causative and passive readings as interpreted by the listener arose during this stage. All of the agentive passive markers (including gei, jiao, rang, and na) can be replaced by bei (被), the most common passive marker in Mandarin, without any significant change of cognitive meaning. The passive marker bei always occurs in the configuration [causee + bei + causer + V], in which it completely matches the semantic pattern provided in Table 4.6. Once the passive sense of gei, jiao, rang, and na becomes possible with sentences like (29-32), then the door is wide open for any productive passive sentence that employs gei, jiao, rang, or na as a passive marker. Table 4.6. Position class in passive constructions Example (29) is a gei passive construction. It is reanalyzed from example (25) via the neglect of the pause particle by listeners. (29) Passive gei a. 那尾金鱼给猫咪吃掉了。 Na wei jinyu gei maomi chidiao le. that Cl goldfish give kitty eat-off Asp ‗That goldfish was eaten by the kitty.‘ b. 下个月的薪水给张三预支了。 Xia ge yue de xinshui gei Zhangsan yuzhi le. next Cl month de salary give Zhangsan draw-in-advance Asp ‗Next month‘s salary was drawn in advance by Zhangsan.‘ causee (NP3) gei (给) rang (让) na (拿) jiao (叫) (secondary) causer (NP2) V 154 Example (30) is a rang passive construction. It is reanalyzed from example (26) via the neglect of the pause particle by listeners. (30) Passive rang a. 那尾金鱼让猫咪吃掉了。 Na wei jinyu rang maomi chi diao le. that Cl goldfish let kitty eat-off Asp ‗That goldfish was eaten by the kitty.‘ b. 下个月的薪水让张三预支了。 Xia ge yue de xinshui rang Zhangsan yuzhi le. next Cl month de salary let Zhangsan draw-in-advance Asp ‗Next month‘s salary was drawn in advance by Zhangsan.‘ Example (31) is a na passive construction. It is reanalyzed from example (27) via the neglect of the pause particle by listeners. Note that examples (31a-b) with passive reading are not acceptable sentences in standard Mandarin (Beijing Mandarin), but could be grammatical in other dialects of Mandarin. According to Wu (1999), the na passive construction occurs in a few dialects of Xiang, such as those of Lengshuijiang and Rucheng, spoken in Hunan Province, as mentioned in Table 4.1 above. (31) Passive na a. 那个老鼠洞拿石头堵死了。 Na ge laoshu dong na shitou dusi le. that Cl rat hole take stone block Asp ‗That rat‘s hole was blocked by the stone.‘ b. 那群土匪拿手枪吓跑了。 Na qun tufei na shouqiang xia pao le. that Cl bandit, take pistol scare-away Asp ‗Those bandits were scared away due to the pistol.‘ Example (32) is a jiao passive construction. It is reanalyzed from example (28) via the neglect of the pause particle by listeners. 155 (32) Passive jiao a. 墙上的挂钟叫工人拆掉了。 Qiangshang de guazhong jiao gongren chaidiao le. wall-on GEN wall-clock call worker remove ASP ‗The wall clock was removed by the worker.‘ b. 那台电视机叫张三修好了。 Na tai dianshiji jiao Zhangsan xiuhao le. that CL television call Zhangsan fix ASP ‗That television was fixed by Zhangsan.‘ In summary, although the speaker wants to express the causative meaning, as in sentences (25-28), the listener will interpret those causative sentences as passive due to the neglect of the pause particle. The syntactic and semantic features in sentences (29-32) satisfy the structure of the typical passive pattern of Chinese. This fact may help the use of the agentive passive markers develop. 4.5. Discussion This chapter has pointed out several inadequacies in the previous studies on the development of agentive passive markers in certain Chinese dialects. I have tried to give an alternative that can overcome these inadequacies and give a better explanation for why the two markers, passive versus disposal, seem to have opposite functions but employ an identical form. The development of agentive passive markers, such as gei (给), na (拿), jiao (教/ 叫), and rang (让), is a natural causative-to-passive process, as shown in Figure 4.1 below. Diachronically speaking, these markers were used in causative constructions. Gei ‗to give‘ and rang ‗to let‘ occurred in the permissive causative construction; na ‗to take‘ occurred in the instrumental causative construction; and jiao ‗to call‘ occurred in the manipulative causative construction. 156 The passive meanings were derived from the three causative constructions. The ―causative-to-passive‖ process involves three steps, ―topicalization‖, ―omission of primary causer‖, and ―structural reanalysis‖; when the process is completed, the passive meaning starts to be conventionalized as a fixed passive construction in a few dialects of Mandarin and Xiang. Figure 4.1. Causative-to-passive process of gei, rang, na, and jiao The passive and disposal markers employ the same morpheme due to a mismatch of interpretation between the speaker and the listener. The speaker wants to express a causative meaning, but the sentence is understood as a passive construction by the listener. However, it should be noticed that this mismatch does not bring about any misunderstanding during the conversation. Both the speaker and the listener focus on the affected causee. The notion of ‗causee is affected by something/someone‘ as understood by the listener is very similar to the notion of ‗a person causes the causee being affected by something/someone‘ as expressed by the speaker when the primary causer ‗a person‘ is omitted. 157 CHAPTER V CONTACT-INDUCED CHANGE: BORROWING This chapter proposes a possible explanation for the development of nominalization, relativization, and genitivization in Chinese and Naxi. Most of the Chinese dialects and Naxi apply a single morpheme as a genitive marker, a relativizer, and a nominalizer; for example, Beijing Mandarin uses de (的) and Naxi uses ɡə. In order to account for the origin of Naxi ɡə, we should first consider and evaluate the pan- Chinese function of nominalization, relativization, and genitivization. I will show that the functions of nominalization, relativization, and genitivization are united by a common value of definiteness in Chinese; this shared value is the primary explanation for why Chinese speakers use an identical morpheme as a nominalizer, a relativizer, and a genitive marker. Diachronically speaking, Cantonese ge3, Chaozhou Southern Min kai, Beijing Mandarin de, Kunming Mandarin nə, and Taiwanese he were derived from a common construction [that + Cl] in which ―Cl‖ is usually the most general-purpose classifier. Details will be provided in Section 5.1. I will then provide a possible explanation for the development of Naxi ɡə. The origin of Naxi ɡə is puzzling because it is clear that Labo Naxi, Mosuo, and Naxi were derived from the same ancestor—Proto-Na; however, they show divergent evolution in their nominalization, genitivization, and relativization. Naxi ɡə is used as a genitive marker, a relativizer, and a nominalizer, but Labo Naxi and Mosuo use a cognate marker i for nominalization and relativization and use different markers na and bɯ33 for genitivization, respectively. Evidently, the development of Naxi ɡə is due to contact with 158 the authoritative Chinese language in its Late Medieval, Pre-Modern, and Modern forms (from 960 to 1900 AD); ɡə is a loanword borrowed from the most general-purpose classifier ɡə (个) in Mandarin. Details will be discussed in Section 5.2. 5.1. Definiteness and nominalization, relativization, and genitivization in Chinese Contemporary Chinese languages use an identical morpheme as a nominalizer, a relativizer, and a genitive marker; Beijing Mandarin uses de (的), Chaozhou (潮州) Southern Min uses kai, and Cantonese uses ge3. There has been extensive discussion in the literature concerning the development of Beijing Mandarin de (Feng 1990, 1992; Liu 2006; Lu 1943, 1984; Mei 1988; Shi & Li 1998, 2002; Wang 1958, 1980; Yap, Choi & Cheung 2007; Yap & Matthews 2008; Yap, Matthews & Horie 2004; Zhang 1915, and Zhu 1982), Chaozhou Min kai (Xu & Matthews (in press), and Yap & Matthews 2008), and Cantonese ge3 (Yap & Matthews 2008, and Sio 2009). I will briefly introduce these previous studies in the succeeding section. 5.1.1. Development of di (底)/de (的) in Mandarin The emergence of the present-day nominalizer de (的) in Mandarin is commonly linked to the lexical noun di (底) with the sense of ‗origin‘, ‗foundation‘, or ‗bottom‘ (Yap, Choi & Cheung 2007:1). 69 There are at least nine hypotheses about the origin and development of di (底)/de (的). Note that (1-6) below are adopted from Shi and Li (1998:165-167) and Jiang (1999:83-85). 69 Di written as 底 originated in the Tang Dynasty (618-907 AD); after Yuan Dynasty (1206-1370 AD), di was written as 的 in the literature (Shi & Li 1998:165; Liu 2006:59). 159 (1) Binglin Zhang 章炳麟 (1915) argues that the non-clause final di (底) (functioning as a relativizer and a genitive marker) was derived from zhi (之) via phonological changes, while the clause final di (底) (functioning as a nominalizer) was derived from zhe (者) via phonological changes.70 (2) Shuxiang Lu 吕叔湘 (1943, 1984) points out that phonological change is not a convincing argument for the development of di (底). Based on numerous structural correspondences between zhe (者) and di (底), di was derived from zhe. (3) Li Wang 王力 (1958, 1980) argues that di (底) located in all positions was derived from zhi (之) via phonological changes. (4) Minche Zhu 祝敏彻 (1982) points out that di (底) was derived from both zhi (之) and zhe (者) because of their similarities in syntactic function. (5) Tsulin Mei 梅祖麟 (1988) argues that zhi (之) became di (底) via phonological changes in colloquialism, then the clause-final di originated from the clause-final zhe (者) via a structural analogy. (6) Chuntian Feng 冯春田 (1990, 1992) argues that di (底) adopted its syntactic function from zhe (者) via a structural analogy; then, the function of zhe was replaced by di. The etymology of di could be a demonstrative pronoun. (7) Yuzhi Shi 石毓智 and Charles N. Li 李讷 (1998:179) mention that di (底), zhi (之) and zhe (者) do not have any etymological or cognate relationship. Di (底) developed 70 Zhi (之) is a demonstrative used in the texts of Old and Middle Chinese and it had evolved various adnominal functions such as genitive and relativizer (Yap, Choi and Cheung 2007:1). Zhe (者), which is usually located in the clause-final position, is a nominalizer used in the texts of Old and Middle Chinese. 160 independently and it was originally a demonstrative/interrogative pronoun. Shi and Li (2002:14) further argue that the grammaticalization of de (的) was triggered by the structural analogy as an ―optimization process‖. (8) Lansheng Jiang 江蓝生 (1999:91) argues that the etymology of di (底) was the locative di (底/地). The locative di succeeded the genitive function from locative nouns, such as suo (所) and xu (许), via a structural analogy. (9) Yap and Matthews (2008:327) point out a diachronic sketch of the grammaticalization pathway of Mandarin di (底)/de (的), as shown in Figure 5.1. Di was derived from the lexical noun di (底) via grammaticalization and it was influenced by zhi (之) and zhe (者). The pronominal function of di gave rise to the development of the genitive marker di/de and the nominalizer di/de. The nominalizer di/de further grammaticalized into a conditional subordinator de hua (的话), relativizer, cleft, and stance marker. Figure 5.1. Grammaticalization pathways of Mandarin di (底)/de (的) (taken from Yap and Matthews 2008:327) 161 To sum up the nine hypotheses above, there are five possible origins of di (底): (1) di was derived from zhi (之) (e.g. Mei 1988; Wang 1958, 1980; Zhang 1915, and Zhu 1982); (2) di was derived from zhe (者) (e.g. Lu 1943,1984; Mei 1988; Zhang 1915, and Zhu 1982); (3) di originated from the locative di (底) (e.g. Jiang 1999); (4) di was originally a demonstrative pronoun (e.g. Feng 1990, 1992, and Shi and Li 1998, 2002); and (5) di originated from the lexical noun di (底) (e.g. Yap & Matthews 2008). Considering the linguistic mechanism used in the nine hypotheses above, there are three theories used for explaining the diachronic development of di (底): (1) Sound change: Before 1950s, Chinese linguists generally agree that the occurrence of di (底) is due to the phonological change from zhi (之) and/or zhe (者) (e.g. Wang 1958, and Zhang 1915). (2) Structural/functional analogy: After the 1950s, more and more Chinese linguists found that sound change cannot explain many of the examples of di (底) in Classical Chinese; their arguments thus focused on a structural/functional analogy between di and other morphemes, such as zhi (之), zhe (者), suo (所), xu (许), di (地), ge (个), etc. The shared functions in the same structure and its analogy with the other morphemes paved the way for the occurrence of di (e.g. Feng 1990, 1992; Jiang 1999; Lu 1943; and Zhu 1982). After lexicon competition and replacement, di replaced the others. (3) Grammaticalization: More recent studies have been influenced by the theory of grammaticalization in western linguistics. Shi and Li (2002:8) argue that di (底) was grammaticalized from a demonstrative/interrogative pronoun; after di was grammaticalized around the 13th century, its graphic form was replaced by 的 162 pronounced /də/ with a neutral tone. Yap, Choi, and Cheung (2007:15) conclude that the grammaticalization of di (底)/de (的) was from a locative noun written as 底 meaning ‗bottom‘. It grammaticalized into a nominalizer and further grammaticalized from a nominalizer into a speaker mood marker, a stance marker or a sentence final particle. It is worth noting that not all of the Chinese linguists adopt only one theory for explaining the development of di (底). Mei (1988), for example, combines contributions of sound change (from Wang 1958) and structural/functional analogy (from Lu 1943). Shi and Li (2002:2) argue that analogy is responsible for motivating the grammaticalization of di (底)/de (的) and that there are optimal structures serving as models, triggering grammaticalization via analogy. As have been shown above, many studies have been done on Mandarin di (底)/de (的); however, only a few Chinese linguists have conducted research on Chaozhou Southern Min kai and Cantonese ge3. In more recent work, Yap and Matthews (2008:322-326) suggest that the classifier grammaticalized into the genitive/pronominal marker and further into the nominalizer and relativizer in these languages. In her 2009 work, Joanna Ut-Seong Sio (manuscript) points out that there are similarities of discourse property in Cantonese versatile ge3. All of the versatile functions of Cantonese ge3 have the features of [+matching] and [+contrastive]. I will discuss these studies in the following two sections, Sections 5.1.1.1 and 5.1.1.2. 163 5.1.1.1. Development of kai in Chaozhou Southern Min Xu and Matthews (in press) mention that kai in Chaozhou Southern Min can appear in a wide range of syntactic positions and can function as a classifier, adnominal (i.e. genitive, associative, and relativizer), pronominal, nominalizer, as well as sentence final particle. Concerning the etymology of kai, Xu and Matthews pointed out that ―kai is first and foremost a general-purpose numeral classifier in the Chaozhou dialect, and is a cognate of the modern Mandarin general classifier 个 (/ɡ ə/), which has its origin from a Classical Chinese specific classifier for ‗bamboo‘‖. Evidence for this assumption can be found in the Northwestern Mandarin dialect of Shangzhou (商州). In the urban area of Shangzhou, the general-purpose classifier 个 is pronounced as /kə55/; in the countryside, it is pronounced as /kuo 55 /, and its vernacular pronunciation is /kai/ with a neutral tone (Zhang 2007:159). Therefore, 个 (/ɡ ə/) and kai are classifiers that originate from the same etymon. Concerning the grammaticalization of kai, Xu and Matthews hypothesize the following relationships between the versatile functions of kai as follows: classifier kai adnominal pronominal nominalization stance marking In an earlier study in 2008 (323-326), Yap and Matthews provided a sketch of the grammaticalization pathway of Chaozhou kai, as shown in Figure 5.2. The genitive and pronominal kai that originated from the classifier give rise to the development of the topic marker and the nominalizer. The topic marker kai further grammaticalized into the non-verbal copula. The nominalizer kai further grammaticalized into a conditional subordinator kai ue, relativizer, cleft, and sentence final stance marker. 164 Figure 5.2. Grammaticalization pathways of Chaozhou kai (taken from Yap and Matthews 2008:326) 5.1.1.2. Development of ge3 in Cantonese Like Mandarin di/de and Chaozhou kai, Cantonese ge3 is used both adnominally and in sentence-final position. Following the same approach used in Figure 5.1 and Figure 5.2, Yap and Matthews (2008:327) also provide a sketch of the grammaticalization pathway of Cantonese ge3, as shown in Figure 5.3. The genitive and pronominal ge3 were originally derived from the classifier go3. Then they were used as the nominalizer ge3. The nominalizer ge3 further grammaticalized into a topic marker, conditional subordinator ge3 waa2, relativizer, cleft, and stance marker. Figure 5.3. Grammaticalization pathways of Cantonese ge3 (taken from Yap and Matthews 2008:327) 165 In her 2009 study, Joanna Ut-Seong Sio (manuscript) provides a new perspective, which differs from Yap and Matthews‘ grammaticalization pathway of Cantonese ge3. I find that Sio‘s conclusion is very similar to my idea about definiteness of nominalization, relativization, and genitivization in Chinese in that she argues that all of the versatile functions of Cantonese ge3 have discourse features of [+matching] and [+contrastive]. In other words, all of the uses of the versatile ge3 carry the sense of definiteness to some degree. In the following section, Section 5.1.2, I will illustrate the sense of definiteness of nominalization, relativization, and genitivization in Chinese. In Section 5.1.3, I will discuss the sense of definiteness applied to the conditional subordinator, cleft, and sentence-final marker in Mandarin. 5.1.2. Definiteness The majority of the Chinese languages use a single morpheme to function as a nominalizer, a relativizer, and a genitive marker, as shown in Table 5.1. 71 Table 1 presents three of the so called ―dialects of Sinitic language (汉语方言)‖: Mandarin (官 话), Min (闽), and Yue (粤). Mandarin includes Northern Mandarin of Beijing (北京) and Southwestern Mandarin of Kunming (昆明); Min includes Southern Min of Chaozhou (潮州) and Taiwanese; and Yue is Cantonese. All of the morphemes provided in Table 5.1 are not cognates; Beijing Mandarin uses de; Kunming Mandarin uses nə; Taiwanese uses he; Chaozhou Min uses kai; and Cantonese uses ge3. 71 The references from Chaozhou Southern Min and Cantonese in Table 5.1 are based on Yap and Matthews (2008:326-327). 166 Table 5.1. Nominalizer, genitive marker, and relativizer in Chinese Sinitic languages Mandarin (官话) Min (闽) Yue (粤) Beijing Kunming Taiwanese Chaozhou Cantonese Nominalizer di (底)/ de (的) nə he kai ge3 Gen. marker di (底)/ de (的) nə he kai ge3 Relativizer di (底)/ de (的) nə he kai ge3 Etymology that [that + Classifier] Classifier According to the hypothesis provided by Yap and Matthews (2008), it can be concluded that there is a common developmental pathway constituting ―classifier genitive & pronominal nominalizer relativizer‖ in Chaozhou Southern Min and Cantonese, as shown in Figures 5.2 and 5.3 above. This pathway is a process of grammaticalization. However, in the authors‘ paper, there is no explanation for why a bound classifier morpheme (such as kai in Chaozhou Southern Min or ɡo3 in Cantonese) can exist alone without co-occurrence with a demonstrative or a numeral after grammaticalization. It is clear that a bare classifier, whether in old Chinese or modern Mandarin, is uncommon and must occur in one of the following patterns of classifier constructions: (1) Numeral + Classifier; (2) Demonstrative + Classifier; and (3) Demonstrative + Numeral + Classifier. Based on Zhang‘s (2007:159) study on the Northwestern Mandarin dialect of Shangzhou (mentioned above), evidence supports that the etymology of Chaozhou kai is the same as Mandarin ɡ e (个) pronounced as /ɡ ə/, and also Cantonese ge3. These are all general-purpose numeral classifiers. However, in Table 5.1, Taiwanese he was formed by a phonological fusion in which the demonstrative hit ‗that‘ has coalesced with the 167 classifier e; furthermore, Beijing Mandarin di (底)/de (的) pronounced as /də/ and Kunming Mandarin nə may be simply accounted for as the demonstrative—that. Based on the data available in Chinese, I find that there is a simple theory to explain these etymological divergences; that is ―definiteness‖. 5.1.2.1. The first definiteness type: “that + classifier” In most of the languages I investigate, the use of nominalization, genitivization, and relativization applies only one robust function in terms of ―definiteness‖ among all of the dialects of Chinese. The sense of definiteness in nominalization, genitivization, and relativization may even be pan-Chinese. The definiteness sense is carried by a constituent ―that + classifier‖ in three constructions, as shown in Table 5.2. Table 5.2 provides position classes for Chinese genitive/associate phrases (in Li and Thompson‘s (1981:113) terminology), relative clause, and nominalized clause.72 All of the constructions involve ―that + classifier‖ in the second slot to express the sense of definiteness. In order to demonstrate the use of definiteness, I will take Beijing Mandarin nei 51 ge ‗that (one)‘ as an example shown in (1); the demonstrative nei51 (那) means ‗that‘ and ge (个) with the neutral tone is the most general-purpose classifier. 72 Li and Thompson (1981:113) define the associate phrase as a phrase that involves two NPs in which ―two noun phrases are ―associated‖ or ―connected‖ in some way; the precise meaning of the association or connection is determined entirely by the meaning of the two noun phrases involved‖. 168 Table 5.2. Definiteness in Chinese genitive marker, relativizer, and nominalizer Slot 1 Slot 2 Slot 3 Noun that + classifier Noun  genitive / associate phrase Verb/Adjective that + classifier Noun  relative clause Verb/Adjective that + classifier (empty)  nominalized clause Definiteness In Beijing Mandarin, the construction ―Noun + nei51 ge + Noun‖ is equivalent to the genitive phrase and associate phrase, as shown in (1a) and (1b), respectively. (1) a. genitive phrase in Beijing Mandarin wo 214 nei 51 ge di 51 di zai 51 du 35 da 51 xue 35 . 我 那 个 弟弟 在 读 大学 1S that Cl little brother Prog study college ‗My little brother is studying at the college.‘ b. associate phrase in Beijing Mandarin zuo 35 tian 55 nei 51 ge di 51 zhen 51 hao 214 kong 214 bu 51 . 昨天 那 个 地震 好 恐怖 yesterday that Cl earthquake very terrible ‗Yesterday‘s earthquake was very terrible.‘ The construction ―Verb + nei51 ge + Noun‖ is equivalent to the agentive relative clause, as shown in (1c). c. relative clause in Beijing Mandarin tiao 51 wu 214 nei 51 ge xiao 214 gu 55 niang 35 shi 51 wo 214 mei 51 mei. 跳舞 那 个 小 姑娘 是 我 妹妹 dance that Cl little girl Cop 1S little sister ‗The little girl who is dancing is my little sister.‘ The construction ―Adjective + nei51 ge + Noun‖ is equivalent to the relative clause, as shown in (1d). d. relative clause in Beijing Mandarin pang 51 du 55 du 55 nei 51 ge xiao 214 hai 35 mei 35 qian 35 mai 214 shu 55 . 胖嘟嘟 那 个 小孩 没 钱 买 书 fat that Cl child Neg money buy book ‗The child who is very fat does not have money to buy books.‘ 169 The construction ―Verb + nei51 ge‖ is equivalent to the headless nominalized clause, as shown in (1e). e. nominalized clause in Beijing Mandarin tiao 51 wu 214 nei 51 ge shi 51 wo 214 mei 51 mei. 跳舞 那 个 是 我 妹妹 dance that Cl Cop 1S little sister ‗That dancing person is my little sister.‘ The construction ―Adjective + nei51 ge‖ is also equivalent to the headless nominalized clause, as shown in (1f). f. nominalized clause in Beijing Mandarin pang 51 du 55 du 55 nei 51 ge mei 35 qian 35 mai 214 shu 55 . 胖嘟嘟 那 个 没 钱 买 书 fat that Cl Neg money buy book ‗That fat person does not have money to buy books.‘ All of the examples in (1a-f) are common in Beijing Mandarin. I find that when nei 51 ge ‗that (one)‘ is used in nominalization, genitivization, and relativization, it does not have versatile functions. Its function is simply an expression of definiteness. Chinese speakers use an identical phrase nei 51 ge ‗that (one)‘ to modify three different constructions for a very straightforward reason; that is, the single function of nei 51 ge ‗that (one)‘ is semantically compatible in the three constructions. Note that de (的) with the neutral tone has long been known as the marker in nominalization, genitivization, and relativization in Beijing Mandarin. In example (1), nei 51 ge ‗that (one)‘ can be replaced by de (的) without changing the meaning. The etymology of de (的)/di (底) will be discussed later. Shi and Li (1998:171; 2002:9-11) have mentioned that the emergence of the classifier system in the new syntactic numeral phrase ―Numeral + Classifier + Noun‖ (during 1200-1400 A.D.) played a significant role in the rise of de/di. They argue that 170 there are two major aspects supporting that the development of de is parallel to that of classifiers: ―First, de began to appear as a morphosyntactic particle at the point when the new form ‗number + classifier + head‘ became predominant. Second, in the 15th century, both of the diachronic processes, ‗number + classifier +noun‘ replacing ‗number + noun‘ and ‗VP + de + noun‘, reached completion‖ (2002:10-11). In other words, the highly frequent use of classifiers in Mid-Modern Chinese matured the ―Numeral + Classifier + Noun‖ pattern, which served as an optimal structure to trigger grammaticalization via analogy (2002:1). The morpheme de (的)/di (底) is the winner after analogization and lexicon replacement. 73 My hypothesis contradicts Shi and Li‘s diachronic assumption. Cross-linguistic investigation of the synchronic dialects of Chinese have shown that there are four types of constituents used for the expression of definiteness in nominalization, genitivization, and relativization, as shown in Table 5.3. The first definiteness type ―that + classifier‖ has been demonstrated in (1) in Section 5.1.2.1. Speakers of all of the Chinese dialects can apply the first type in colloquial conversation. Types 2-4 will be discussed in the following sections, from Sections 5.1.2.2 to 5.1.2.4. Table 5.3. Four types of constituents for the expression of definiteness Type 1 that + classifier Pan-Chinese Type 2 fusion of [that + classifier] Taiwanese (Southern Min) he Type 3 that Mandarin (Beijing di/de, Kunming nə) Literary Cantonese deʔ Type 4 classifier Chaozhou (Southern Min) kai; Cantonese ge3 73 More detailed discussion about Shi and Li‘s optimal structure will be provided in Section 5.1.2.3. 171 5.1.2.2. The second definiteness type: phonological fusion of [that + classifier] The second type involving ―phonological fusion of [that + classifier]‖ located in the Slot 2 of Table 5.2 occurs in Taiwanese. The demonstrative hit ‗that‘ has phonologically coalesced with the most general-purpose classifier e to form a new form he, as shown in (2). The constituent [that + classifier] hit e in (2a) can be replaced by he ‗that (one)‘ without changing the meaning, as shown in (2b). Note that phonological fusion is very common in Taiwanese. For example: sio ‗mutual‘ + kaŋ ‗same‘ siaŋ ‘the same’; si ‗four‘ + tsah ‗ten‘ siah ‗forty‘; hit ‗that‘ + e ‗classifier‘ he ‗that (one)‘; di ‗exist‘ + e ‗particle‘ de ‗progressive marker‘. (2) a. that + classifier: hit e hit e mai tsiaʔ. that Cl Neg eat ‗Don‘t eat that (one).‘ b. fusion of [that + classifier]: he he mai tsiaʔ. that (one) Neg eat ‗Don‘t eat that (one).‘ The following examples (3a-f) show that the first type [hit e] ―that + classifier‖ is, like Beijing Mandarin, very common in Taiwanese. Note that examples (1a-f) used in Beijing Mandarin are equivalent to Taiwanese examples (3a-f). In addition, [hit e] ‗that + classifier‘ can be replaced by the coalesced form he ‗that (one)‘ without changing the meaning. In Taiwanese, the construction ―Noun + [hit e] + Noun‖ or ―Noun + [he] + Noun‖ is equivalent to the genitive phrase and associate phrase, as shown in (3a) and (3b), respectively. 172 (3) a. genitive phrase in Taiwanese ua [hit e] di-di de thak dai-hak. 1S that Cl little brother Prog study college ‗My little brother is studying at the college.‘ b. associate phrase in Taiwanese tsang [hit e] de-dang tsiok khiong-bo. yesterday that Cl earthquake very terrible ‗Yesterday‘s earthquake was very terrible.‘ The construction ―Verb + [hit e] + Noun‖ or ―Verb + [he] + Noun‖ is equivalent to the agentive relative clause, as shown in (3c). c. relative clause in Taiwanese thiao-bu [hit e] siə ko-nĩũ-ŋã si uan me-me. dance that Cl little girl Cop my little sister ‗The little girl who is dancing is my little sister.‘ The construction ―Adjective + [hit e] + Noun‖ or ―Adjective + [he] + Noun‖ is equivalent to the relative clause, as shown in (3d). d. relative clause in Taiwanese bui-tsut-tsut [hit e] ŋĩ-na bə tsĩ me tsheʔ. fat that Cl child Neg money buy book ‗The child who is very fat does not have money to buy books.‘ The construction ―Verb + [hit e]‖ or ―Verb + [he]‖ is equivalent to the headless nominalized clause, as shown in (3e). e. nominalized clause in Taiwanese thiao-bu [hit e] si uan me-me. dance that Cl Cop my little sister ‗That dancing person is my little sister.‘ The construction ―Adjective + [hit e]‖ or ―Adjective + [he]‖ is also equivalent to the headless nominalized clause, as shown in (3f). f. nominalized clause in Taiwanese bui-tsut-tsut [hit e] bə tsĩ me tsheʔ. fat that Cl Neg money buy book ‗That fat person does not have money to buy books.‘ 173 5.1.2.3. The third definiteness type: a bare “that” The third type of expression of definiteness in nominalization, genitivization, and relativization involves a bare demonstrative ―that‖ in the Slot 2 of Table 5.2. The typical examples can be found in Kunming Mandarin nə ‗that‘, as shown in (4). Note that examples (4a-f) of Kunming Mandarin are equivalent to Beijing Mandarin examples (1a- f) and Taiwanese example (3). Examples (4a-f) show that the first definiteness type ―that + classifier‖ (nə + gə) in Kunming Mandarin can be used in nominalization, genitivization, and relativization. In addition, nə gə ‗that (one)‘ can be replaced by the bare demonstrative nə ‗that‘ without changing the meaning. It is possible that nə ‗that‘ occurring in Slot 2 is derived via an omission of gə from nə gə ‗that (one)‘. Otherwise, the original tone of ‗that‘ in Kunming Mandarin involves a falling tone, which is pronounced nə51; when functioning as a single marker of definiteness without accompanying the classifier in (4a-f), nə51 ‗that‘ is pronounced as nə33 with mid-level tone. Therefore, it is also possible that nə33 is a phonological fusion of nə51 ‗that‘ and gə 33 ‗classifier‘. Note that phonological fusion can be found in Kunming Mandarin; for instance, bu ‗no‘ + xuyao ‗need‘bu xao ‗no need‘. In Kunming Mandarin, the construction ―Noun + nə gə + Noun‖ or ―Noun + nə + Noun‖ is equivalent to the genitive phrase and associate phrase, as shown in (4a) and (4b), respectively. (4) a. genitive phrase in Kunming Mandarin wo nə (gə) di-di zai du da-xio. 1S that Cl little brother Prog study college ‗My little brother is studying at the college.‘ 174 b. associate phrase in Kunming Mandarin zuo-tian nə (gə) di-zhen tai kong-bu. yesterday that Cl earthquake very terrible ‗Yesterday‘s earthquake was very terrible.‘ The construction ―Verb + nə gə + Noun‖ or ―Verb + nə + Noun‖ is equivalent to the agentive relative clause, as shown in (4c). c. relative clause in Kunming Mandarin tiao-wu nə (gə) xiao gu-niang shi wo mei-mei. dance that Cl little girl Cop 1S little sister ‗The little girl who is dancing is my little sister.‘ The construction ―Adjective + nə gə + Noun‖ or ―Adjective + nə + Noun‖ is equivalent to the relative clause, as shown in (4d). d. relative clause in Kunming Mandarin pang-du-du nə (gə) ua-ua mə də qian mai shu. fat that Cl child Neg have money buy book ‗The child who is very fat does not have money to buy books.‘ The construction ―Verb + nə gə‖ or ―Verb + nə‖ is equivalent to the headless nominalized clause, as shown in (4e). e. nominalized clause in Kunming Mandarin tiao-wu nə (gə) shi wo mei-mei. dance that Cl Cop 1S little sister ‗That dancing person is my little sister.‘ The construction ―Adjective + nə gə‖ or ―Adjective + nə‖ is also equivalent to the headless nominalized clause, as shown in (4f). f. nominalized clause in Kunming Mandarin pang-du-du nə (gə) mə də qian mai shu. fat that Cl Neg have money buy book ‗That fat person does not have money to buy books.‘ The way the speakers of Kunming Mandarin understand nə ‗that‘ in nominalization, genitivization, and relativization seems completely identical to the use of Beijing Mandarin de (的) because the etymological form of de (的) may be di (底), 175 which functioned as a demonstrative ―that‖ in Middle Modern Chinese, as shown in (5). Shi and Li (2002:8) point out that ―di214, which is pronounced with the third tone 底 in Modern Chinese, is the origin of the particle de with a neutral tone. After having been grammaticalized, the graphic form was replaced by 的 around the 13th century.‖ I find that this etymological assumption can be proven through analysis of the Hefei (合肥) dialect of Mandarin. In Hefei Mandarin, the demonstrative ―that‖ is pronounced as /ti53/ which is cognate with di (底) (Meng 1997:297). Therefore, di (底) could be the demonstrative ‗that‘. (5) di ―that‖ in Tang Dynasty (Shi & Li 2002:8) Di shi cang chun chu that be store Spring place ―That/this is a warm and pleasant place.‖ (Mo shan xi ci, c. 900A.D.) The Chinese character written as 底 indicates a lexical noun meaning ―bottom‖; however, the demonstrative sense of ―that‖ has nothing to do with the ―bottom‖. The people speaking Middle Modern Chinese may have borrowed the existent written character 底 for the meaning of ―that‖ under the following conditions: (1) the sound of 底 was the same as or similar to ―that‖ in that period; (2) the colloquial ―that‖ which had existed in a Chinese dialect did not have a written form; (3) a writer who spoke a dialect wanted to record the colloquial ―that‖; and (4) the writer borrowed the existent character 底 to express ―that‖ because of their identical or similar pronunciations. Therefore, it could be understood that the reason why the colloquial demonstrative ―that‖ was written as 底 is a strategy of sound borrowing. This kind of borrowing was very common in the documents of Classical Chinese, especially in the Dunhuang Pianwen (敦煌变文). 176 The sound of 底 in Middle Modern Chinese could have been pronounced as /de/ with a falling tone. Its pronunciation could be very similar to 底 in literary Taiwanese because literary Taiwanese has preserved the sound of Middle Modern Chinese. It is worth noting that deʔ in literary Cantonese is used in nominalization, genitivization, and relativization. Its function is identical to ge3, which has long been known as the marker used for the genitive/associate phrase, the relative clause, and the nominalized clause in colloquial Cantonese. Cantonese speakers have lost the etymology of deʔ, which could be related to the demonstrative ‗that‘ because deʔ was a loanword from 底 in Middle Modern Chinese. I am uncertain why literary Cantonese deʔ involves a checked tone. The sound of ‗that‘ in the Proto-Sino-Tibetan should not involve a checked syllable.74 In order to explain the diachronic development from di (底) to de (的), Shi & Li (2002:13) pose an ―optimal structure‖, which can serve as a model to trigger grammaticalization via analogy, as shown in the following: Step I: Lexical collocation: [relative clause + [didemonstrative + N]] Step II: Reanalysis: [[relative clause-de] + N] The authors explain that ―step II involves the phonological reduction of di into de without a tone and the boundary loss between the relative clause and de. Accordingly, the constituent hierarchy also changed: de forms a constituent with N in step I, but it forms a constituent with the preceding relative clause in step II (2002:13).‖ There are several points that need to be considered: First, the combination of the demonstrative and the classifier expressed as ‗that+Cl‘ is the most common candidate to express the sense of definiteness in Chinese 74 The form ‗that‘ in Proto-Tibeto-Burman is reconstructed as *day by Matisoff (2003:671). 177 and may even be pan-Chinese. Examples (1), (3), and (4) above have shown that there is only one function in terms of ―definiteness‖ applied in Chinese nominalization, genitivization, and relativization. Therefore, Beijing Mandarin de (的), like Kunming Mandarin nə ‗that‘, must maintain the function of definiteness. Second, it is also possible that de (的) with a neutral tone is derived from a phonological fusion of [de 51 + gə33] in which de51 (底) could be the demonstrative ―that‖ in Middle Modern Chinese, 75 and the gə33 (个) with a mid-level tone was the most general-purpose classifier in the same period. 76 It is clear that de 51 with a sharp falling tone is more unlikely to become /də/ with a neutral tone directly, while the new form /də33/ with a mid-level tone is most likely to survive. In addition, the pattern [de51 + gə33] in terms of [that + classifier] is the most common colloquial form used in nominalization, genitivization, and relativization in that period. The genesis of the sound change usually occurs from the most high-frequency patterns. In short, the new form /də/ (的) may be a phonological fusion of de51 (底) and gə33 (个). 5.1.2.4. The fourth definiteness type: a bare classifier The fourth type of expression of definiteness in nominalization, genitivization, and relativization involves a bare classifier, as shown in Slot 2 of Table 5.2. The most 75 The sound of底 in Middle Modern Chinese is reconstructed based on Guangyun (广韵): 都礼切. Note that when doing ―Fanqie‖ (反切) to reconstruct old sounds, scholars should use a dialect, such as Min (Taiwanese) or Yue (Cantonese), which have a stopped/checked tone (入声). ―Fanqie‖ is a method to determine the pronunciation of a character by using two other characters where the initial consonant is taken from the first character and the vowel is taken from the second character. 76 The word gə33 (箇), the most general-purpose classifier, is written as 個/个 in Modern Chinese. The sound of gə33 (箇) in the Middle Modern Chinese is reconstructed based on Tangyun (唐韵): 古贺切. The original lexical meaning of 箇 is ―bamboo stalk‖. 178 general-purpose classifier is the most likely candidate to be used in this type; it can modify several distinct nouns, including people, inanimate objects, and so on. Typical examples can be found in Chaozhou Southern Min, as shown in (6). In example (6), kai is used as a classifier, as shown in (6a), a genitive marker, as shown in (6b), a relative clause marker, as shown in (6c), and a nominalizer in a headless relative clause, as shown in (6d). Note that the first type of expression indicating ―that + classifier‖ (tsi kai) should be compatible in Chaozhou Southern Min nominalization, genitivization, and relativization because Taiwanese Southern Min applies the first type, as shown in (3) above. (6) a. classifier kai in Chaozhou (from Yap & Matthews 2008:323) ĩɯ tã i ai tsi kai me? 2S say 3S like this Cl Ques ‗Do you think she will like this (one)?‘ b. genitive phrase in Chaozhou (from Yap & Matthews 2008:323) i kai haʔ-sẽn 3S Gen student ‗his/her student‘ c. relative clause in Chaozhou (from Yap & Matthews 2008:323) lai tsio phaʔ-kaŋ kai naŋ loŋ si gua-sẽn kiã. come here work Rel people all Cop outside province worker ‗Those people who came here to work are all from other provinces.‘ d. nominalized clause in Chaozhou (from Yap & Matthews 2008:323) lai tsio phaʔ-kaŋ kai loŋ si gua-sẽn kiã. come here work Nml all Cop outside province worker ‗Those people who came here to work are all from other provinces.‘ Note that both Taiwanese and Chaozhou belong to the Southern Min dialect of Chinese, and they are partially mutually intelligible. As mentioned above, Taiwanese applies the second type of expression in which the demonstrative hit ‗that‘ has coalesced with the classifier e. I am uncertain about the diachronic developmental pathways of 179 Chaozhou kai. However, the mono-syllabified process seems pervasive in Chinese dialects as with in Taiwanese he (via a phonological fusion) and Kunming Mandarin nə ‗that‘ (via a deletion or a phonological fusion). It is possible that Chaozhou kai has replaced the definiteness function from ‗that+Cl‘ via the mono-syllabified process. The use of ―classifier-as-demonstrative‖ is attested. In other words, the concept of definiteness could be carried by a classifier. Taking Cantonese as an example, Wu and Bodomo (2009:495) point out that a [Cl-N] phrase in Cantonese ―can appear in subject position and receive a definite interpretation‖, as shown in (7). 77 (7) a. classifier-as-definiteness in Cantonese (from Wu & Bodomo 2009:495) gaa ce zo-zyu go ceothau. Cl car block-Cont Cl exit ‗The car is blocking the exit.‘ b. go leotsi jiu hou lek sin dak. Cl lawyer need very smart only-okay ‗The lawyer had better be very smart.‘ In 2009, Wu and Bodomo (495) pose several contextual-effect conditions for ―classifier-as-determiner‖; they mention that example (7) ―might lead one to conclude that classifiers in Cantonese are inherently definite, like the definite article the in English, and hence can indicate definiteness regardless of context. However, this is not true.‖ Wu and Bodomo further point out that there are two contextual conditions for the definite interpretation from a classifier phrase which ―arises (a) when the referent has already been mentioned in the discourse context (note that this is the typical case of anaphoric reference); (b) if not, the referent must be close by, so that the hearer can easily identify the referent‖ (2009:496). 77 Note that example (7) used by Wu and Bodomo is taken from Matthews and Yip (1994:89). 180 In my opinion, it is difficult to prove that the use of classifier-as-definiteness in Cantonese expresses definiteness in a way dissimilar from the use of the definite determiner ‗the‘ or the demonstrative ‗that‘ in English. The definiteness of the classifier in Cantonese is inherited from the most common definite pattern ―demonstrative + classifier‖, as shown in (8). Concerning examples (8a) and (8b), Wu and Bodomo (2009:495-498) mention that ―[Cl-N] phrases with a definite interpretation have been found to have the same distribution as [Dem-Cl-N] phrases and have been considered to be instances of omitting the demonstrative.‖ Therefore, example (7a) is derived from example (8a); example (7b) is derived from example (8b). (8) a. that + classifier-as-definiteness in Cantonese (from Wu & Bodomo 2009:495) (Go) gaa ce zo-zyu go ceothau. that Cl car block-Cont Cl exit ‗The car is blocking the exit.‘ b. (Ni) go leotsi jiu hou lek sin dak. this Cl lawyer need very smart only-okay ‗The lawyer had better be very smart.‘ In short, the [Cl-N] phrases carrying the sense of definiteness in Chaozhou Min and Cantonese is due to the omission of demonstrative, ‗this‘ or ‗that‘, from [Dem-Cl-N] phrases. In other words, the bare classifier could be the mono-syllabified form from ―that + classifier‖. 5.1.3. Definiteness of conditional subordinator, cleft, and sentence-final marker In Section 5.1.2, I have examined how the colloquial pattern in the structure of ―demonstrative + classifier‖ expresses ―definiteness‖ of nominalization, genitivization, and relativization in several dialects of Chinese. In this section, I will demonstrate that Mandarin conditional subordinator (de hua), cleft (shi~de), and the sentence-final marker 181 (de) in Figure 5.1 posed by Yap and Matthews (2008:327) are semantically related to the sense of definiteness. Unlike nominalization, genitivization, and relativization, de (的) functioning as a conditional subordinator, cleft, and sentence-final marker cannot be replaced by the pattern ―demonstrative + classifier‖ (nei ge (那个)). This fact shows that de has carried a more abstract meaning of definiteness via grammaticalization. The work of Jiang (2004:388-393) and Yap, Matthews, and Horie (2004:163) shows that the emergence of the conditional subordinator (de hua), cleft (shi~de), and sentence-final marker (de) was very late in Chinese history, occurring during the Qing Dynasty (1644-1911 A.D.), and all of these elements were only used in colloquial conversation. Since colloquialism was not intensively documented until the Yuan Dynasty (1206-1370 A.D.), it is difficult to understand the diachronic development of Mandarin de. Based on discussion provided in Section 5.1.2 above, the use of nominalization, genitivization, and relativization was derived from the colloquial pattern ―demonstrative + classifier‖ via the process of mono-syllabification. The demonstrative used in ―demonstrative + classifier‖ is definite and/or contrastive. In my treatment, the concept of definiteness of demonstratives involves broad pragmatic features including those that are ―contrastive‖, ―assertive‖, ―focus‖, and ―topical‖. The current study does not attempt to give a pragmatic account explaining the function of Mandarin de; therefore, I will still use the terminology ―definiteness‖ to indicate the abstract function of de. All of the contiguous and diachronic development of de is based on the semantic nature of definiteness. All of the functions shown in Figures 5.1, 5.2 and 5.3 carry the sense of definiteness. Joanna Ut-Seong Sio (2009, manuscript) has argued that all of the versatile 182 functions of Cantonese ge3 have the features of [+matching] and [+contrastive]. In the following sections, I will further examine Mandarin de. 5.1.3.1. Definiteness of conditional subordinator de hua/的话 Concerning the Mandarin conditional subordination (de hua/的话), Jiang (2004:395) claims that Xu and Liu‘s (1998:243) argument is credible in that de hua is related to ―topic‖. In other words, de hua involves definiteness in a condition. Mandarin de hua is used in a subordinate clause, as shown in (9b). Example (9b) is the response sentence to the question (9a). The compound de hua combining de (which expresses definiteness) with hua (which shows that this is a conditional sentence) functions as the conditional subordinator. Note that originally hua (话) is a lexical noun meaning ‗word‘ or ‗speech‘. Since hua (话) is a noun, the phrase ‗suppose that it is raining‘ constituting [VP + de + hua] (as shown in (9b)) is the same as the construction in the relative clause [VP + de + Noun]. We have proved that de (的) in the relative clause involves definiteness; therefore, de in the conditional subordinator de hua also carries the sense of definiteness. It can be interpreted as the demonstrative ‗that‘. (9) a. Question Ni qu bu qu ne? 你 去 不 去 呢 2S go Neg go Ques ‗Are you going or not?‘ b. Definiteness of conditional subordinator Xia yu de hua wo jiu bu qu le. 下 雨 的 话 我 就 不 去 了 fall rain Def saying 1S then Neg go Prt ‗If it is raining, I won‘t go.‘ 183 5.1.3.2. Definiteness of sentence-final marker de Most Chinese linguists give the Mandarin sentence final de a variety of definitions based on the different approaches they use. For example, the sentence-final de can function as a stance/pragmatic marker (e.g. Yap & Matthews 2008:327, and Yap, Matthews & Horie 2004:150), an evidential marker (e.g. Li, Thompson & Zhang 1998:96), or a focus marker (Yuan 2003:5). I find that sentences containing the sentence final de involve a strong topical, definite reading, as shown in (10b). Example (10b) is the response sentence to the question (10a). The sentence final de (的) in (10b) refers to the action kan dianying ‗to see a movie‘. It can be interpreted as the demonstrative ‗that‘. (10) a. Question Wo qing ni kan dianying keyi ma? 我 请 你 看 电影 可以 吗 1S invite 2S see movie can Ques ‗I‘ll invite you to see a movie, is that ok?‘ b. Definiteness on sentence final marking de Keyi de. 可以 的 can Def ‗That‘s ok!‘ 5.1.3.3. Definiteness of cleft shi~de/是~的 The sentence-final de can co-occur with a non-final copula shi (是) in forming a cleft construction, as shown in (11b). Example (11b) is the response sentence to the question (11a). The non-final copula shi (是) commonly serves as a topic or focus marker; when it combines with the sentence-final de, it gives rise to a sentence with a strong definite and assertive mood. The cleft construction shi~de can be interpreted as ‗that‘s 184 it/that‘. The morpheme de in the cleft construction carries the sense of definiteness in that it refers to the event ganhuo ‗to work‘. (11) a. Question Ni mei qu tian li ganhuo ma? 你 没 去 田 里 干活 吗 2S Neg go rice field inside work Ques ‗Didn‘t you go to the rice field to work?‘ b. Definiteness on cleft de Wo shi gan wan huo cai hui lai de. 我 是 干 完 活 才 回 来 的 1S Cop do finish job until return come Def ‗I did not come back until I finished the work, (that‘s it!).‘ 5.1.4. Discussion Based on the discussion above, it can be concluded that three possible mono- syllabified processes form the expression of definiteness in the frame [that + Cl], in which [Cl] is usually the most general-purpose classifier in Chinese. Via the mono- syllabified process, the expression of definiteness in the frame [that + Cl] has been replaced by a monosyllabic morpheme, which functions diachronically as a marker of nominalization, relativization, and genitivization, as shown in Figure 5.4. Figure 5.4. Mono-syllabified process of [that + Cl] 185 The first type involves a phonological fusion of the frame [that + Cl]. For instance, the demonstrative hit ‗that‘ has phonologically coalesced with the most general- purpose classifier e to form a new form he in Taiwanese Southern Min. The second type involves a bare demonstrative ‗that‘, as with Beijing Mandarin də and Kunming Mandarin nə. The third type involves a bare classifier, as with Cantonese ge3 and Chaozhou Southern Min kai. It should be noted that Beijing Mandarin də and Kunming Mandarin nə could also be phonological fusions from [that + Cl]. 5.2. Development of nominalization, relativization, and genitivization in Naxi As in previous research in Tibeto-Burman linguistics on Lahu ve, the purpose of Section 5.2 is to provide a possible explanation for the development of Naxi ɡə. I will claim that the development of Naxi ɡə is due to contact with the authoritative Chinese language during its Late Medieval, Pre-Modern, and Modern periods (from 960 to 1900 AD); ɡə was a loanword borrowed from the most general-purpose classifier ɡə (个) in Mandarin. 5.2.1. Nominalization, Relativization, and Genitivization in possible Loloish Many linguists working on nominalization and/or relativization in Tibeto-Burman languages have referred to the same problem with regard to a polysemous morpheme ve in Lahu (Matisoff 1972:237-258). According to Lahuassois‘s examples, the Lahu ve functions as at least three things: a genitive marker, a relativizer, and a nominalizer. These functions are shown in (12). 186 (12) Nominalization, relativization, and genitivization in Lahu a. genitive phrase in Lahu (from Lahaussois 2002) ŋà ve mí-chɔ I Gen shoulder-bag ‗my shoulder-bag‘ b. relative clause in Lahu (from Lahaussois 2002) Và? qhe chu ve Pîchɔ-pa ô te ɣâ pig as fat Rel Shan that one person ‗That Shan over there who‘s fat as a pig‘ c. nominalized clause in Lahu (from Lahaussois 2002) ɔ-ši tɔ? la ve thà? nɔ mâ ɣa mɔ lâ blood emerge come Nml Acc you Neg get see Ques ‗Didn‘t you see that blood was coming out?‘ In searching available literature on Tibeto-Burman linguistics, I find that this type of phenomenon is only reported in three languages: Hani, Lahu, and Naxi. Based on the classification of most Chinese linguists, the three languages could belong to the Loloish language branch within the Tibeto-Burman language family (see Chapter II). The rest of the possible Loloish languages, such as Jinuo, Labo Naxi, Lisu, Mosuo, and Yi, do not apply a single morpheme as a marker of mominalization, relativization, and genitivization, as shown in Table 5.4. Table 5.4. Nominalizer, genitive marker, and relativizer in possible Loloish languages Loloish languages Jinuo Mosuo Labo Naxi Naxi Lahu Hani Yi Lisu Nominalizer mɣ44 i33 i33 ɡə33 ve33 ɣə33 su33 mɑ44 Gen. marker ε33 bɯ33 nɑ31 ɡə33 ve33 ɣə33 (none) (none) Relativizer mɣ44 i33 i33~(nɑ31) ɡə33 ve33 ɣə33 su33 mɑ44 Cognate ancestor: Proto-Na Comparisons in Table 5.4 above shows that Naxi ɡə33 and Hani ɣə33, like Lahu ve 33 , can function as nominalizers, relativizers, and genitive markers, while Jinuo, Lisu, 187 Labo Naxi, Mosuo, and Yi employ a strategy of ―syncretism (terminology adopted from DeLancey (1986:3)) in which speakers use an identical morpheme as a marker of nominalization and relativization, and use the other morpheme to mark the genitive case (if applicable). It is quite obvious that the developments of Naxi ɡə33, Hani ɣə33, and Lahu ve 33 are extremely sporadic and regional among Tibeto-Burman languages. My hypothesis will claim that the Naxi ɡə33 was a loanword borrowed from the most general- purpose classifier ɡə (个) in Mandarin.78 Table 5.4 also shows that although Mosuo, Labo Naxi, and Naxi were derived from the same ancestor (called Proto-Na), they show divergent evolutions through their nominalization, genitivization, and relativization. In the following sections, I will demonstrate that my claim about the development of Naxi ɡə33 can be substantiated. I will provide a brief history of migration and events of language contact in Mosuo and Naxi in Section 5.2.2. In Section 5.2.3, I will first discuss the development of nominalization, genitivization, and relativization typologically, then try to explain why Naxi applies a unique pattern that differs from those of Labo Naxi and Mosuo. 5.2.2. Migration and language contact Lincan Li (李霖灿) has suggested that the ancestors of the Naxi people (―Proto- Na‖ in my terminology) migrated from northern Sichuan Province to the south and first settled in Muli. Then, the Proto-Na formed two branches; one branch settled down around the Lugu Lake region, and the other branch gradually migrated across the Jinsha River to the Lijiang plain, where they invented the Dongba manuscript (Mathieu 78 ɡ ə (个) in Hanyu Pinyin (汉语拼音) is Romanized as ɡ e. It is important to show the difference between /ə/ and /e/ when doing the cross-linguistic comparison and reconstruction of the proto-language; therefore in this section I will use IPA instead of Hanyu Pinyin for demonstratives and classifiers. 188 2003:152). 79 Li argues that the absence of Dongba pictographic scripts among the Mosuo proves that the Mosuo are an older branch of the Proto-Na and that they settled in Muli and Yongning a very long time ago (Mathieu 2003:154). The Mosuo and the Naxi people have been divided by the Jinsha River for centuries. 80 The Naxi people are mostly settled on the western banks of the river in Lijiang County in settlements such as Fengke, Baoshan, Yulong Snow Mountain, Lunan, and Lijiang Ancient Town. The Mosuo people live on the eastern banks of the river and mostly inhabit the Yongning basin and the Lugu Lake regions, including Muli, Yongning, and Labo. The main toponyms of the Mosuo and Naxi regions are marked in Map 5.1 below. 79 Dongba manuscript (東巴經) is an ancient written record of the Naxi. Its name refers to its use in the Dongba Religion. The Dongba manuscript consists of about 1400 characters which are known as one of the world‘s few living pictographic scripts. 80 The Jinsha River (金沙江) travels through the basic mountain chains of the Lijiang region and feeds into the ten-thousand-mile Yangtze River. 189 Map 5.1. The Mosuo and the Naxi regions The Mosuo language is more likely affected by Tibetan due to speakers‘ Tibetan Buddhist religious beliefs. Male teenage Mosuo learn Tibetan when studying Tibetan lamaism. Mosuo speakers begin to memorize Tibetan Buddhist sutras from childhood, even though they may not understand the sutras‘ meanings. However, there does not seem to be significant Tibetan influence on the Mosuo grammar. Overall, language contact with Tibetan does not induce significant linguistic changes in Mosuo. The language of Labo Naxi appeared intact until the 1950s, when the China government required the Labo children to learn Mandarin at school. Some Labo children acquire Mosuo because most of their playmates and classmates are Mosuo children. However, the Mosuo and the Labo Naxi still maintain the self-sufficient living style they had when they presided over their own territories and civilizations. Contact-induced 190 changes seem not to have occurred. Note that the Labo Naxi language is an important link for reconstructing Proto-Na, because it is a Mosuo-like Naxi language. On the other hand, the Naxi language has been affected by Han Chinese for centuries. The Ancient Town of Lijiang was built at the beginning of the Yuan Dynasty (1279AD); it was the administrative region of the Mu lineage (木氏), the hereditary governor of the Naxi. The Mu lineage reigned over the Lijiang region for about 470 years, through the Yuan, Ming and Qing Dynasties. 81 McKhann mentions that ―Han artisans and scholars were welcomed by the royal Mu lineage, whose position in Naxi society at that time was a product of their close relations with the Chinese authorities‖ (1995:50). The Naxi vocabulary used in the Ancient Town of Lijiang and its surrounding areas includes a greater proportion of Chinese loanwords (McKhann 1995:50). In summary, it is probable to postulate that the Naxi, Labo Naxi, and Mosuo have the same ancestor—Proto-Na—based on their migrational history. The Naxi language has been influenced by Chinese since the 10th century. Since the 13th century, the Naxi lexicon has included many loanwords from Chinese, particularly Southwestern Mandarin. 5.2.3. Nominalization, genitivization, and relativization in Naxi, Labo, and Mosuo There are at least two distinct pathways for developing nominalization, genitivization, and relativization in the Sino-Tibetan languages: (1) all of the nominalization, genitivization, and relativization are simply an expression of definiteness; and (2) nominalization and relativization, which are not relevant to genitivization, are formed via grammaticalization in which both of the nominalizer and the relativizer were 81 Yuan Dynasty was established by Mongols who conquered and ruled China from 1279 to 1368 A.D. Qing Dynasty was established by Manchus who conquered and ruled China from 1644 to 1911 A.D. 191 derived from an identical morpheme, usually a noun referring to a ―place‖, ―thing‖, or especially ―people‖. The former pathway is a pan-Chinese phenomenon where the majority of the Chinese-speaking people use an identical morpheme as a nominalizer, a relativizer, and a genitive marker (see Section 5.1). The latter pathway featuring ―nominalization- relativization syncretism‖ is more likely to occur in certain Tibeto-Burman languages, such as Qiang (LaPolla & Huang 2003:226), Labo Naxi, Mosuo, and Yi (Hu 2002:138), as shown in Table 5.5. 82 Table 5.5. Nominalization-relativization syncretism Tibeto-Burman languages Qiang Labo Mosuo Labo Naxi Yi Nominalizer -m i 33 i 33 su 33 Gen. marker -tɕə bɯ33 nɑ31 (none) Relativizer -m i 33 i 33 + (nɑ31) su33 Ety. of syn. people 5.2.3.1 Nominalization-relativization syncretism Syntactically, nominalization-relativization syncretism uses an identical marker for nominalization and relativization, and a different marker for genitivization. Table 5.5 above also shows that the shared etymological meaning of nominalizer and relativizer in those languages is ―people‖. The evolution of the nominalization-relativization syncretism is easy to trace because in the case of nominal combinations such as [N(P) + 82 The idea of ―nominalization-relativization syncretism‖ is first posed by DeLancey. According to his investigation of Classical Tibetan and contemporary Lhasa Tibetan, ―the nominalization function is chronologically and systemically prior to relativization, which is merely one specialized function of nominalization‖ (1986:1). 192 N] in Sino-Tibetan languages, the preceding noun or noun phrase always modifies the following noun. This ―nominal modifying relationship‖ is a well-established function in Proto-Sino-Tibetan. In Mosuo or Labo Naxi, for example, the constituent of [V(P)- Nominalizer + N] in relativization is equivalent to the nominal constituent of [N(P) + N]. Because of the inherent modifying relationship between the preceding element [V(P)- Nominalizer] and the following noun [N], the nominalizer in the patten [V(P)- Nominalizer + N] can be structurally defined as a relativizer since it requires an attributive head noun. It is clear that çi 33 ‗people‘ in Naxi and xĩ33 ‗people‘ in Labo Naxi can be used as nominalizers for phrases referring to occupations, as shown in (13a-b). (13) a. çi 33 ‗people‘ in Naxi lɯ33 tu31 çi 33 rice field plant(v) people ‗farmer‘ b. xĩ33 ‗people‘ in Labo Naxi lɯ33 tu31 xĩ33 rice field plant(v) people ‗farmer‘ Furthermore, çi 33 ‗people‘ in Naxi usually corresponds to the nominalizer i33 in Labo Naxi, as shown in (14). The occupationally nominal phrase ‗vegetable planter‘ in Naxi is xo 33 phe 55 tu 31 çi 33 , as shown in (14a). 83 In Labo Naxi, ‗vegetable planter‘ is expressed as a 33 tshe 13 tu 31 i 33 , as shown in (14b). The position of the nominalization is shared by çi 33 ‗people‘ in Naxi and i33 in Labo Naxi. Thus it is possible to deduce that the nominalizer i 33 is derived from a noun meaning ‗people‘. 83 Note that [ne31 gə33] ‗that (one)‘ in (14a) was borrowed from Southwestern Mandarin, and it must be removed when reconstructing the proto-language. 193 (14) a. çi 33 ‗people‘ in Naxi tʂhɯ33 xo33phe55 tu31 [ne31 gə33] çi 33 mə31. 3S vegetable plant(v) that Cl people Cop ‗He is a vegetable planter.‘ b. i 33 ‗nominalizer‘ in Labo Naxi tʂhɯ33 a33tshe13 tu31/phu31 i33 ŋə31. 3S vegetable plant(v) Nml Cop ‗He is a vegetable planter.‘ Regarding nominalization-relativization syncretism, I would like to begin by looking at some examples from Labo Mosuo, then go on to examine equivalent examples from Labo Naxi. Labo Mosuo applies ―nominalization-relativization syncretism‖ in that it uses an identical marker i 33 to function as a nominalizer, as shown in (15), and as a relativizer, as shown in (16). Nominalization is a grammatical process in which a verb, a verb phrase, or an adjectival followed by the nominalizer i 33 can function as a noun phrase. In (15a) and (15b), the nominalized verb phrases are activities tʂɯ33 mə33 kɯ33 ―not giving money‖ and xə33 ‗leaving‘, respectively. In (15c) and (15d), the adjectivals nɑ 1 khɑ33tʂɯ33 ‗black‘ and dər33 ‗wrong‘ followed by the marker i33 are nominalized, respectively. (15) Nominalizer i 33 in Labo Mosuo a. xĩ33 dɯ33tɑ13 u33 tʂɯ33 mə33 kɯ33 i33 mə33 dʝu31. people all Adv money Neg give Nml Neg have ‗Among all people, there is no person who does not give money.‘ b. ɲɑ33 tʂhɯ33 xə33 i33 mu33. 1S 3S go Nml hear ‗I hear his leaving.‘ c. tʂhɯ33 nɑ1khɑ33tʂɯ33 i33 mə33 ɲi33 tsɯ31. 3S black Nml Neg want Evid ‗He doesn‘t want the black (one).‘ 194 d. tʂhɯ33 tʂhɯ33 dər33 i33 nər33. 3S 3S wrong Nml know ‗He knows he is wrong.‘ A relative clause is simply an attributive nominalization in which the nominalized phrase serves as a modifier to modify the following head noun. The difference between the nominalized phrase and the relative clause in Labo Mosuo is that the relative clause requires an attributive head noun, but the nominalized phrase does not. For example, the head nouns mu 13 dzɯ13 ‗girl‘, xĩ33 ‗people‘, and ɑ33bu33ɑ33mi33 ‗parents‘ need to occur right after the relativizer in (16a), (16b), and (16c), respectively. (16) Relativizer i 33 in Labo Mosuo a. de 31 dʝɑ31tsho33 i33 mu13dzɯ13 tʂhu13 u33 ɡu33mu33 dʝa13 ʂuɑ33. that dance Rel girl that Cl body very tall ‗That girl who is dancing is very tall.‘ b. no 33 ɑ33ɲi33 pi33 i33 xĩ33 tʂhɯ33 u33 tʂhɯ33ɲi31 mə33 tshɯ31. 2S yesterday tell Rel people this Cl today Neg come ‗The people who you mentioned yesterday didn‘t come today.‘ c. tʂhɯ33 õ13 ʐɯ31 i33 ɑ33bu33ɑ33mi33 lɯ31 mə33 li31. 3S self bear Rel parents even Neg look ‗He doesn‘t even look at his parents who bore him.‘ Labo Naxi uses the same marker i 33 as a nominalizer, as shown in (17a-d). Examples (17a-d) are equivalent to Mosuo examples (15a-d) above. Also, unlike the relativizer which requires an attributive head noun, the nominalizer i 33 cannot be followed by any modified noun. (17) Nominalizer i 33 in Labo Naxi a. xĩ33 dɯ33tɑ31 u33 tço33 mə33 iɑ31 i33 mə33 ndʝu31. people all Adv money Neg give Nml Neg have ‗Among all people, there is no person who does not give money.‘ b. tʂhɯ33 xə33 i33 ŋə33 mi33. 3S go Nml 1S hear ‗I hear his leaving.‘ 195 c. tʂhɯ33 nɑ31 i33 mə33 ɲi33 tsə33. 3S black Nml Neg want Evid ‗He doesn‘t want black (one).‘ d. tʂhɯ33 mə33 xo31 i33 nər33 tsə33. 3S Neg right Nml know Evid ‗He knows he is not right.‘ Labo Naxi also uses i 33 as a relativizer. In addition, it uses an extended form i 33 + nɑ31 as a relitivizer, as shown in (18). Examples (18a-c) are equivalent to Mosuo examples (16a-c) above. The construction of [i 33 + nɑ31] is a combination of the nominalizer i 33 and the genitive marker nɑ31. It should be noticed that the absence of the genitive marker nɑ31 from (18) is acceptable. (18) Relativizer i 33 nɑ31 in Labo Naxi a. de 31 dʝɑ31tsho33 i33 nɑ31 mi33dzɯ13 tʂhɑ13 ɡu33mu33 dʝa13 ʂuɑ31 iɑ55. that dance Nml Gen girl one+Cl body very tall Prt ‗That girl who is dancing is very tall.‘ b. nu 33 ɑ33ɲi33 pe55 i33 nɑ31 xĩ33 tʂhɑ13 tʂhɯ33ɲi33 mə33 tshɯ33. 2S yesterday tell Nml Gen people one+Cl today Neg come ‗The people you mentioned yesterday didn‘t come today.‘ c. tʂhɯ33 u13 ʐɯ33 i33 nɑ31 ɑ33di33ɑ33me33 lɯ31 mə33 ly31. 3S self bear Rel Gen parents even Neg look ‗He doesn‘t even look at his parents who bore him.‘ The nominalizer-plus-genitive combination also occurs in Gurung (Noonan 1997:382) and Tibetan (DeLancey 1982). DeLancey (1986:2) argues that in Lhasa Tibetan ―there is a straightforward syntactic explanation for the use of the genitive postposition as a relative marker; since the modifying clause is in fact syntactically an NP, it is marked with the same morpheme as any other dependent NP;‖ however, the relative clause of Classical Tibetan is simply a nominalization without the genitive marker. The 196 diachronic pattern found in Lhasa Tibetan and Classical Tibetan is correspondent to that in Labo Naxi and Mosuo. Beyer (1992:316) mentions that the genitive in Classical Tibetan ―is used with nominalizations functioning as attributives when the nominalization precedes the head; when it follows the head, the genitive is not used.‖84 Indeed, the use of the genitive marker in the relative clause is a structural analogy in that the genitive marker always requires an apparently modified noun. This would mean that the genitive phrase [N + Genitive + N] and the relative clause [V(P)-Nominalizer + Genitive + N] share an identical construction. So far numerous examples have been provided where the nominalizer and relitivizer in Labo Naxi and Mosuo share the same form i 33 . In what follows, I will demonstrate some sentences that involve the genitive marker in Labo Naxi and Mosuo. In the case of the possessive construction in Labo Naxi, a genitive marker nɑ31 is used, as shown in (19a). The marker nɑ31 can be omitted because of the inherent modifying relationship between two nominals mentioned above. However, the associative phrase usually requires an overt genitive marker nɑ31, as shown in (19b). It is not surprising that the possessive and associative phrase involve the same marker, because the semantic connection in the associative phrase is derived from the concept of possession via conceptual metaphor. 84 Beyer‘s examples adopted from Noonan (1997:382) are shown as follows: a. bla-ma-s btul-ba-i bgegs lama-Erg tame-Nom-Gen demon ‗the demon which the lama tamed‘ b. bgegs bla-ma-s btul-ba demon lama-Erg tame-Nom ‗the demon which the lama tamed‘ 197 (19) a. genitive marker in possessive of Labo Naxi ŋə33zɯ31 tʂhɯ33 tʂhɯ33 (nɑ31) ɑ33di33ɑ33me33 ŋə31. 1D Subj 3S Gen parents Cop ‗We two are his parents.‘ b. genitive marker in associative of Labo Naxi tʂhɯ33be33 nɑ31 li31tsɯ33 ɑ31be33 u33to33 ndɯ33. this year Gen plum last year Cmpr big ‗This year‘s plums are bigger than last year‘s.‘ Labo Mosuo uses bɯ33 as the genitive and associative marker, as shown in (20). Examples (20a-b) are equivalent to the Labo Naxi examples (19a-b) above. (20) a. genitive marker in possessive of Mosuo ɲɑ33zɯ31 tʂhɯ33 (bɯ33) ɑ33bu33ɑ33mi33 ɲi31. 1D 3S Gen parents Cop ‗We two are his parents.‘ b. genitive marker in associative of Mosuo tʂhɯ33i31 bɯ33 li31tsɯ31 ɑ33i33 u33to31 dɯ31. this year Gen plum last year Cmpr big ‗This year‘s plums are bigger than last year‘s.‘ With examples (15-20) in mind, it is possible to explain the case of Naxi ɡə33. The main puzzle concerning Naxi ɡə33 in Table 5.4 is how it could have a genitive function along with its relativization and nominalization functions. It is more difficult to believe that genitivization was derived from nominalization or relativization. Because Labo Naxi and Mosuo apply syncretism in nominalizers and relativizers (see Section 5.2.3.1), I claim that Proto-Na should employ nominalization-relativization syncretism in which it uses an identical marker i 33 to function as both a nominalizer and a relativizer; the relative clause requires an attributive head noun, but the nominalized 198 phrase does not. In addition, Proto-Na might not have a genitive marker because the genitive markers used in Loloish languages are barely cognates. 85 Labo Naxi and Mosuo developed their own genitive markers after the Naxi people migrated across the Jisha River. Their individual genitive markers are optional in the possessive construction but remain overt in the associative phrase. Labo Naxi further applies the genitive marker to form an extended nominalizer-plus-genitive combination in the relative clause via structural analogization. 5.2.3.2 Contact-induced change in Naxi ɡə33 Evidently, Naxi ɡə33 is a loanword borrowed from one of the elements in Southwestern Mandarin nei 51ɡə (那个) ‗that (one)‘ in which ɡə with a neutral tone is the most general-purpose classifier. 86 Here, I would like to claim that Naxi ɡə33 was formed through the following steps: (1) Naxi was in contact with Chinese languages, and it was especially influenced by Southwestern Mandarin for a long time; (2) Naxi adopted nei 51ɡə (那个) ‗that (one)‘ from Southwestern Mandarin into its relativization and pronounced it as nei 51ɡə33; its original relativizer i 33 was gradually replaced by nei 51ɡə33 ‗that (one)‘ through the ages; (3) meanwhile, the phrase nei 51ɡə33 in relativization was functionalized into a monosyllable marker ɡə33; and (4) the relativizer ɡə33 could have easily extended into nominalization and genitivization via analogy with nə33 in Southwestern Mandarin. 85 Most Tibeto-Burman linguists (including DeLancey (1984) and Matisoff (2003)) believe that there should be a genitive particle *ki or *gi in Proto-Tibeto-Burman. 86 The Chinese characters 那个 in Southwest Mandarin pronounced as [nə31 ɡ ə33], whether [nei51 ɡ e] or [nə31 ɡ ə33], must be pronounced as [ne31 ɡ ə33] when borrowed by Naxi. 199 Therefore, Naxi and Chinese finally share the same structural pattern in nominalization, genitivization, and relativization. In the following sections, I will provide examples to demonstrate the contact-induced change of Naxi ɡə33. Unlike Labo Naxi and Mosuo, Naxi uses a single morpheme ɡə33 as a nominalizer, a relativizer, and a genitive marker, as shown in examples (21-23), respectively. Examples (21a & b) are equivalent to Labo Mosuo examples (15a-b) and Labo Naxi examples (17a-b) above. Comparing these correspondent sentences, I find that they share all of the grammar except the nominalizer. Naxi has changed the original nominalizer from i 33 into ɡə33, as shown in (21). (21) Nominalizer in Naxi a. çi 33 ndɯ33xə33 mbe33 tçə55 mə33 iə55 ɡə33 mə33 ndʝy31. people all Adv money Neg give Nml Neg have ‗Among all people, there is no person who does not give money.‘ b. tʂhɯ33 nɑ31 ɡə33 mə33 ɲi33. 3S black Nml Neg want ‗He doesn‘t want the black (one).‘ In addition, Naxi has changed the marker of the relative clause from i 33 or i 33 nɑ31 to ɡə33, as shown in (22). Examples (22a-b) are equivalent to Labo Mosuo examples (16a-b) and Labo Naxi examples (18a-b) above. (22) Relativizer in Naxi a. ndzər33tsho33 ne31 ɡə33 mi33 tʂhɯ33 ku55 ŋɡu33mu33 dʝa13 ʂuɑ31. dance Prog Rel girl that Cl body very tall ‗That girl who is dancing is very tall.‘ b. nɯ33 nɯ33 ɑ31ɲi33 ʂə55 ɡə33 ɑ33tʂhɯ33 ku55 tʂhɯ33ɲi33 mə33 tshɯ31. 2S Subj yesterday say Rel that Cl today Neg come ‗The people you mentioned yesterday didn‘t come today.‘ Dongba characters prove that until the 11th century, the Naxi people still used nɑ31 as the genitive marker. For example, in an associative phrase like tu31 ʐɯ33 nɑ31 khɑ33 200 dʐɯ33 ‗a big city of thousands of streets‘ in Dongba, nɑ31 is used as the genitive marker to connect tu 31 ʐɯ33 ‗thousands of streets and khɑ33 dʐɯ33 ‗big city‘.87 However, Naxi has changed the genitive marker from nɑ31 into ɡə33, as shown in (23). Examples (23a-b) are equivalent to Labo Naxi examples (19a-b) and Labo Mosuo examples (20a-b) above. (23) a. genitive marker in possessive of Naxi ŋɑ33ɲi33ku55 tʂhɯ33 tʂhɯ33 ɡə33 ɑ33bɑ33me33 uɑ31. 1D Subj 3S Gen parents Cop ‗We two are his parents.‘ b. genitive marker in associative of Naxi tʂhɯ33mbe33 ɡə33 se33li31 ɑ33mbe33 tçɑ31 ndɯ31 mu33sɯ33. this year Gen plum last year Cmpr big rather ‗This year‘s plums are bigger than last year‘s.‘ The Chinese language with which Naxi has been in contact is Southwestern Mandarin, such as Kunming Mandarin. Kunming Mandarin uses ―that + classifier‖ (nə + ɡə) in its nominalization, relativization, and genitivization, as shown in (4) in Section 5.1.2.3 above. In addition, nə ɡə ‗that (one)‘ can be replaced by a monosyllable marker nə33 without changing the meaning. Naxi borrowed this ―that + classifier‖ pattern in its relative clauses, as shown in (24). In examples (24 a-c), ne 31 ɡə33 ‗that (one)‘ is used as an expression of definiteness for the modified head noun çi 33 ‗people‘. (24) ne 31ɡə33 in Naxi a. tʂhɯ33 tʂhɯ33 lɯ33 tu31 ne31 ɡə33 çi 33. 3S Subj rice field plough that Cl people ‗He is a farmer.‘ b. tʂhɯ33 ɲi33 ndʐə31 ne31 ɡə33 çi 33 (uɑ31). 3S fish catch that Cl people copula ‗He is a fishman.‘ 87 This example is based on a dictionary entitled Naxizu Xiangxing Biaoyinwenzi Zdiani (纳西族象形标音 文字字典) edited by Lincan Li (李霖灿) and published in 2001. 201 c. tʂhɯ33 the33ɣɯ33 so31 ne31 ɡə33 çi 33 tɑ55 mə31 sɯ33. 3S book learn that Cl people only Neg still ‗He is still a student.‘ The loaned classifier ɡə33 finally functionalized as a marker of definiteness without accompanying the demonstrative ne 31 ‗that‘, as shown in (25 a-c). This mono- syllabified process of the marker is very common in nominalization, relativization, and genitivization in Chinese dialects; for example, Taiwanese he, Kunming Mandarin nə, Beijing Mandarin də, Chaozhou Southern Min kai, and Cantonese ge3 (see Section 5.1). (25) ɡə33 in Naxi a. tʂhɯ33 tʂhɯ33 lɯ33 tu31 ɡə33 çi 33. 3S Subj rice field plough Rel people ‗He is a person who works in the rice field.‘ b. tʂhɯ33 ɲi33 ndʐə31 ɡə33 çi 33 (uɑ31). 3S fish catch Rel people copula ‗He is a person who catches fishes.‘ c. tʂhɯ33 the33ɣɯ33 so31 ɡə33 çi 33 tɑ55 mə31 sɯ33. 3S book learn Rel people only Neg still ‗He is still a person who is studying.‘ It is reasonable to believe that Naxi relativizer ɡə33 has extended to its nominalization and genitivization via analogy with nə33 from Southwestern Mandarin. 5.2.4. Discussion The argument concerning a contact-induced change in nominalization, relativization, and genitivization of Naxi can be established because evidence exists showing that the development of Naxi ɡə is due to language contact with Chinese, particularly Southwestern Mandarin, from the Late Medieval Chinese period (from 960 A.D.). 202 Typologically speaking, only Chinese languages involve a single morpheme that functions as a genitive marker, a relativizer, and a nominalizer. The ancestor of Naxi, Proto-Na, applies the ―nominalization-relativization syncretism‖ which uses an identical marker i 33 as a relativizer and a nominalizer. Naxi ɡə was a loanword borrowed from the most general-purpose classifier ɡə (个) in Mandarin. 203 CHAPTER VI CONCLUSION This dissertation primarily exists to provide new perspectives from which to solve six puzzles regarding linguistic changes on the basis of typological and diachronic evidence. These linguistic puzzles have arisen due to diachronic developments. Tackling such challenging puzzles requires expertise in the field of historical linguistics. Chapter II aims to investigate the relationship between ―tonal splits of checked syllables in *L‖ and ―subgrouping of the Loloish language branch‖. The current work has attempted to compare tonal patterns cross-linguistically in different dialects of Chinese, different dialects of Yi (Lolo), and possible Lolo-Burmese languages based on the assumption that they had a single *checked syllable in earlier times, and that they later underwent tonal splits conditioned by the rule of Matisoff‘s two-way tonal splits. That is, the voiced initial consonant of *checked syllables tend to develop lower-pitched tones in modern languages, while the voiceless initial consonant of *checked syllables tend to develop higher-pitched tones. After several comparative studies in different dialects of Chinese, different dialects of Yi (Lolo), and possible Lolo-Burmese languages, the typological evidence shows that the Loloish tonal split corresponding to *L checked syllables did not simply fall into the two-class contrast that Matisoff has characterized as ―low-checked‖ and ―high-checked‖. Most of the patterns of tonal splits, such as ―regularity‖, ―flip-flop‖, and ―irregularity‖, occur in different dialects of Chinese and Yi (Lolo). This fact shows that a language with an irregular pattern of tonal splits in *checked syllables can be categorized with patterns of ―regularity‖ and ―flip-flop‖ in the same language group. In addition, the 204 behavior of the tonal split in checked syllables of the proto-language has nothing to do with the subgrouping of the modern languages because it often fails to predicate a more reliable genetic relationship between languages. Since Matisoff‘s methodology is weak when used in subgroupings of Loloish, it is not necessary or prudent to accept the genetic position of Naxi Matisoff proposed in 1972, even though it may be correct. Concerning the puzzle of the genetic position of Naxi, the current problem that faces us is that Naxi and Mosuo show two different patterns of tonal splits corresponding to *checked syllables, as shown in Figure 2.3. Based on my investigation, Naxi belongs to the subgroup of flip-flop, while Matisoff put it in the subgroup of irregularity. Moreover, Mosuo, which was not included in Matisoff‘s study, belongs to the subgroup of irregularity. I have suggested that Naxi and Mosuo have a common ancestor called Proto-Na; it is assumed that at an earlier period, the two contemporary languages shared the same tonal system, but that variation developed at a later stage. But what is the tonal system of Proto-Na? Reconstructions of tonal changes in the period from Proto-Na to Naxi and Mosuo are necessary before it is possible to locate the genetic position of Naxi. In order to successfully reconstruct the tonal system of Proto-Na, all of the subgroups of Naxi (including Dayanzhen, Lijiang, and Baoshan) and Mosuo (including Yongning, Beiqu, and Guabie) must be studied further. Concerning the comparative, cross-linguistic studies, the phenomenon of tonal splits conditioned by the phonetic nature of the syllable-initial consonants seems to be scattered throughout many countries of Southeast Asia; languages spoken in these areas, such as Thai and Vietnamese, may be affected. What exactly are the patterns of tonal splits corresponding to *checked syllables in Thai, Vietnamese, etc.? Do all of these 205 languages, like Chinese and Lolo-Burmese, demonstrate different patterns of tonal splits in *checked syllables, such as ‗regularity‘, ‗flip-flop‘, ‗irregularity‘, and ‗un- split‘? The phonology of the proto-languages of Thai and Vietnamese have been reconstructed via the comparative method. 88 I believe these unsolved questions would be good topics for future research. Concerning the biggest problem about the conditioning features of tonal splits, the languages that involve the regular pattern in *checked syllable represent a very small group among all of the Sino-Tibetan languages. The issues about the patterns of ‗flip- flop‘, ‗irregularity‘ and ‗un-split‘, which are used by the most speakers of the Sino- Tibetan languages, are less-discussed. In addition, the mysteries surrounding the conditioning features of tonal splits that brought about the pattern of ‗flip-flop‘ remain unresolved. 89 Furthermore, the research on the diachronic development of the pattern of ‗irregularity‘ is currently in a state of chaos. Chapter III proposes a cognitive principle, ‗schematization‘, to explain semantic development of the motion verb RETURN—还 and the action verb TAKE—把 in Chinese. The image schemata of RETURN and TAKE are used to demonstrate how Chinese speakers use and understand RETURN and TAKE diachronically. In Section 3.1, I investigate the motion verb meaning ―to return‖ in Chinese. I re- examine the development of RETURN huan/hai (还) from a diachronic viewpoint. RETURN is not a simple concept because it requires both a retraversal path and an unmarked path-traversed portion. According to the etymological evidence from ancient 88 Please see the related references: Fang-kuei Li (李方桂) (1977), Pittayaporn (2009), and Thompson (1976:1113-1203). 89 Since the pattern of flip-flop violates the universal phonetic mechanism of tonogenesis. Wang (1967) and Yue-Hashimoto (1986) argue that the pattern of flip-flop derived from the pattern of regularity. 206 Chinese and typological evidence from Tibetan and Labo Mosuo, the image schema of RETURN huan/hai (还) is similar to the refresh symbol . Regarding synchronically versatile uses of huan/hai (还), the prototypical image schema of RETURN is shown to consist of four conceptual senses: (1) repetitious sense, (2) opposite sense, (3) sequential sense, and (4) continuant sense. All of the versatile uses of RETURN are conceptually abstracted from the individual senses of the image schema of RETURN. The meaning of ―again‖ is derived from its repetitious sense; the meaning of ―unexpectedly‖ is derived from its opposite sense; the meaning of ―also‖, ―or‖ or ―more‖ is derived from its sequential sense; and the meaning of ―still‖ is derived from its continuant sense. In Section 3.2, I investigate the Chinese action verb meaning ―to take‖. The constructions with TAKE in Chinese are normally associated with the most-studied construction—the disposal construction. The disposal construction in Mandarin is known as the BA construction. Most linguists disagree with Li Wang‘s opinion because the ‗disposal‘ notion he poses is too specific, without much room for extension to the other types of disposal sentences. Thus, the majority of articles related to issues of the BA construction in Mandarin tend to adopt a theory that could explain all of the examples of the BA construction, even thought their goals are never put into practice completely. 90 The puzzle of the disposal construction arises from its complicated diachronic evolution in semantics as well as syntax. Therefore, before creating a theory to explain the disposal construction, it is necessary to understand its diachronic development. It is impossible to adopt only one theory in order to analyze all types of the constructions with 90 Please see the related references: Chappell (1992); Hopper & Thompson (1980); Sun (1995); Tsao (1987); and Ziegeler (2000). 207 TAKE. Based on the historical evidence from ancient Chinese and typological evidence from dialects of modern Chinese, the representation of image schemata involving TAKE in both ancient and modern Chinese was composed of four independent serial events: ―thing-transferred event‖, the ―thing-located event‖, the ―thing-transformed event‖, and the ―instrumental causative event‖. All of these serial events consist of three arguments in their image schema. Removing the THING argument from their image schema in the former three events will cause semantic insufficiency and syntactic ungrammaticality. However, the THING object, or INSTRUMENT in the instrumental causative event, is allowed to be omitted from the image schema. This is probably because the disposal verb used in this event conveys the sense of both ―manipulativeness‖ and ―affectedness‖. After the omission of the INSTRUMENT from the instrumental causative event, a disposal event that involves two arguments in a single event was formed. The image schema of the disposal event represents a transfer of causative force from an ‗energy initiator‘—causer—to an ‗energy receiver‘—causee. The disposal construction is a new form that was derived from the instrumental causative event by the Tang Dynasty due to the omission of the instrument. The semantic relationship between the disposal event and the instrument causative event is very close because both events contain the senses of ―manipulativeness‖ and ―affectedness‖ in their image schema, event though the serial event has become the single event. The other new form is the ―agent-patient event‖, which was derived from the disposal event when the nominal argument in the event is not a tangible object. The agent-patient event and the disposal event share the same syntactic structure, but they differ completely in semantics 208 because the grammatical TAKE is used to mark the PATIENT in the former but marks the CAUSEE in the latter. In comparative cross-linguistic studies, most Sino-Tibetan linguists seem to agree that the BA construction or ‗disposal‘ construction is unique to Chinese since no similar construction has been found in any other language. However, I found a pu-construction in Labo Naxi which may correspond to the Mandarin BA-construction to some extent. In addition, Liberty A. Lidz (2010:809), the author of the dissertation entitled A Descriptive Grammar of Yongning Na (Mosuo), has glossed the word pɔ as a verb meaning ‗to take‘ or ‗to use‘ and as an instrumental marker of disposal construction.91 Since Mandarin ba, Mosuo pɔ , and Labo Naxi pu have an identical etymological meaning (TAKE), it would be very interesting to investigate the similarity and dissimilarity of their functions in every respect. Chapter IV focuses on a long-standing puzzle concerning the agentive passive marker in certain dialects of Chinese, such as Mandarin and Xiang. In these dialects, speakers use a single morpheme, such as gei (给) ‗to give‘, na (拿) ‗to take‘, jiao (教/叫) ‗to call‘, and rang (让) ‗to let‘, to function as both the passive/agent marker and the disposal/patient marker. Based on the typological similaries, I claim that these agentive passive markers arise from the causative construction, theconfiguration of which can be demonstrated as [NP1 primary causer + gei/ na/ jiao/ rang + NP2 secondary causer + V + NP3 causee]. Under this configuration, gei ‗to give‘ and rang ‗to let‘ occur in the permissive 91 The function and gloss of pɔ ‗to take‘ in Mosuo provided by Lidz (2010:809) is misleading. As mentioned in Section 3.2 in Chapter III, TAKE in Chinese is associated with an INSTRUMENT in the instrumental causative construction, and to mark a CAUSEE in the disposal construction. Based on my understanding of Labo Mosuo and Labo Naxi, Mosuo pɔ and Labo Naxi pu, like Mandarin ba, are used to mark a CAUSEE in the disposal construction. 209 causative construction; na ‗to take‘ occurs in the instrumental causative construction; and jiao ‗to call‘ occurs in the manipulative causative construction. The passive meaning was diachronically derived from the permissive causative construction, the instrumental causative construction, and the manipulative causative construction under the same process. This ―causative-to-passive‖ process involves three steps: ―topicalization‖, ―omission of primary causer‖, and ―structural reanalysis‖. In the first step, the NP3 causee is topicalized, placed in the sentence-initial position; the constituent order after topicalization can be represented as [NP3 causee, NP1 primary causer + gei/ na/ jiao/ rang + NP2 secondary causer + V]. In the second step, the NP1 primary causer is omitted by the speaker; the constituent order after the omission of the NP1 primary causer can be represented as [NP3 causee, + gei/ na/ jiao/ rang + NP2 secondary causer + V]. In the final step, the pause particle after the NP3 causee is neglected; the final configuration [NP3 causee + gei/ na/ jiao/ rang + NP2 secondary causer + V] is reanalyzed as a passive construction by the listener. When the process is completed, the passive meaning starts to be conventionalized as a fixed passive construction in certain dialects of Chinese. The passive and disposal markers employ the same morpheme because of a mismatch of interpretation between the speaker and the listener. The speaker wants to express a causative meaning, but the sentence is understood as a passive construction by the listener. However, it should be noticed that this mismatch does not bring about any misunderstanding in their conversation. Both the speaker and the listener focus on the affected causee. The notion of ‗causee is affected by something/someone‘ as understood by the listener is very similar 210 to the notion of ‗a person causes the causee being affected by something/someone‘ as expressed by the speaker when the primary causer ‗a person‘ is omitted. In comparative cross-linguistic studies, there are at least four types of lexical verbs that can serve as a single source of both the passive and the disposal marker in certain dialects of Chinese (Shi & Wang 2009 and Wu 1999). They are GIVE-type, TAKE-type, CALL-type, and ALLOW-type. The TAKE-type disposal, such as ba (把) in Beijing Mandarin, was discussed in the second part of Chapter III. Moreover, the GIVE-type passive, TAKE-type passive, CALL-type passive, and ALLOW-type passive were discussed in Chapter IV. The likely parallel independent developments of the GIVE-type disposal, CALL-type disposal, and ALLOW-type disposal in these dialects of Chinese are worth investigating. The GIVE-type disposal, as mentioned at the beginning of Chapter IV, is derived via a metaphorical process. Newman (1996:172) mentions that the literal GIVE is often ascribed to a causative expression in which ―the GIVER is said to be the entity which ‗causes‘ it that the RECIPIENT comes to have the THING‖. The CALL-type disposal and the ALLOW-type disposal are only used by a very small population of Mandarin speakers in Henan, Shanxi, and Shandong. Data collection from these speakers is necessary before conducting an investigation. Chapter V proposes a possible explanation for the development of nominalization, relativization, and genitivization in Chinese and Naxi. Most Chinese dialects and Naxi apply a single morpheme functioning as a genitive marker, a relativizer, and a nominalizer. 211 Section 5.1 has shown that the use of nominalization, genitivization, and relativization applies only one robust function in terms of ―definiteness‖ in all of the dialects of Chinese. There are four types of constituents used for the expression of definiteness in nominalization, genitivization, and relativization: (1) ―that + classifier‖ (pan-Chinese), (2) phonological fusion of [that + classifier] in Taiwanese, (3) a bare ‗that‘ in Beijing Mandarin and Kunming Mandarin, and (4) a bare classifier in Cantonese and Chaozhou Min. Concerning the four constituent types of expressions of definiteness, I find that different dialects of Chinese may apply different types. A dialect may also apply more than one type (e.g. literary Cantonese applies the bare ‗that‘ and colloquial Cantonese applies the bare classifier). This common value of definiteness is the primary explanation for why Chinese speakers use an identical morpheme as a nominalizer, a relativizer, and a genitive marker. In addition, it also shows that the Mandarin conditional subordinator (de hua/的话), cleft (shi~de), and sentence final marker (de) are semantically related to the sense of definiteness. All of the contiguous and diachronic developments of de are based on the semantic nature of definiteness. Section 5.2 points out the two distinct pathways in the development of nominalization, genitivization, and relativization in the Sino-Tibetan languages. In the first pathway, all of the Chinese dialects use an identical morpheme as a nominalizer, a relativizer, and a genitive marker; all of those markers in Chinese are simply expressions of definiteness. In the second pathway, many of the Tibetan-Burman languages apply nominalization-relativization syncretism in which they use one marker for nominalization and relativization, and another different marker for genitivization. 212 Within this typological classification, the idea that Labo Naxi, Mosuo and Naxi derived from the same ancestor, Proto-Na, is problematic. However, these contemporary languages show divergent evolution in their nominalization, genitivization, and relativization. Naxi ɡə is used as a genitive marker, a relativizer, and a nominalizer, but Labo Naxi and Mosuo use a cognate marker i for nominalization and relativization and use other markers na and bɯ33 for genitivization, respectively. This shows that the development of Naxi ɡə is due to contact with the authoritative Chinese language in its Late Medieval, Pre-Modern, and Modern Chinese forms (from 960 to 1900 A.D.); ɡə was a loanword borrowed from the most general-purpose classifier ɡə (个) in Mandarin. Another puzzle that arose earlier than that of Naxi ɡə is the polysemous morpheme ve in Lahu (Matisoff 1972:237-258). Like Naxi ɡə, Lahu ve is also a single morpheme used as a genitive marker, a relativizer, and a nominalizer. In searching available literature on Tibeto-Burman linguistics, I found that this ―3 in 1‖ (three functions in one morpheme) phenomenon only occurs in Hani, Lahu, and Naxi. Is there any plausible explanation for the development of Lahu ve and Hani ɣə? The puzzle of Lahu ve has probably remaining unsolved for several decades because scholars have overlooked the influence of external factors, especially contact- induced change. In addition, the etymology of Lahu ve is still arguable. DeLancey (to appear) mentions that ―the first serious study of a final particle construction in Tibeto- Burman is Matisoff‘s (1972) description of ve in the Loloish language Lahu. Matisoff (1985) identifies ve, as well as the Jinghpaw final particle ai, and others, as reflexes of the same Proto-Tibeto-Burman copula *way‖. 213 However, Lahu ve 33 could be a nominalizer rather than a copula. DeLancey (to appear) mentions that the general structure of a nominalized clause construction in a verb-final language is ―verb stem—nominalizer—(copula)‖. According to Chang‘s (1986) Lahuyu Jianzhi (拉祜语简志 (A grammar of Lahu)), Lahu ve33 used as a particle in the final position of the construction usually occurs in the pattern: ―verb stem—ve33—(zu31)‖; the optional zu 31 is a common copula in Lahu; ve 33 could be a nominalizer. Data collection from various dialects of Lahu speakers is necessary before conducting a further investigation. 214 APPENDIX A LIST OF ABBREVIATIONS 1S first singular 1D first dual 2S second singular 3S third singular Acc accusative Adv adverb Asp aspect Cl classifier Cmpr comparative Comp complement Cont continuous aspect Cop copula Def definite Erg ergative Evid evidential Gen genitive Impfv imperfective Neg negative Nml nominalizer NP noun phrase Perf perfective Pl plural Prog progressive Prt particle Ques question particle Rel relativizer Subj subject V verb 215 APPENDIX B REFERENCES & RECONSTRUCTIONS FOR *VOICED INITIALS IN CHINESE Karlgren‘s (高本汉) Reconstruction Han zi gu jin yin hui (page) Hanyu fangyan cihui (page) Jiangsusheng he Shanghaishi fangyan gaikuang (page) 1 thin (薄) *buɑk 292 352 650 2 pull out (拔) *buɑt 113 270 576 3 butterfly (蝶) *diɛp 302 56 604 4 wax (腊) *lɑp 273 152 579 5 deer (鹿) *luk 421 34 635 6 stone (石) *dziæk 223 30 619 7 honey (蜜) *miɪt 300 106 621 8 wheat (麦) *muək 423 71 647 9 hot (热) *niæt 186 379 589 10 sun; day (日) *niɪt 132 14 619 216 APPENDIX C REFERENCES & RECONSTRUCTIONS FOR *VOICELESS INITIALS IN CHINESE Karlgren‘s (高本汉) Reconstruction Han zi gu jin yin hui (page) Hanyu fangyan cihui (page) Jiangsusheng he Shanghaishi fangyan gaikuang (page) 1 pen (笔) *piuet 238 183 619 2 peel (剥) *pok 24 277 593 3 pick (摘) *thiek 120 262 648 4 bamboo(竹) *tiuk 237 169 637 5 color (色) *siɪk 277 249 587 6 snow (雪) *siuæt 383 5 612 7 rule (尺) *tshiæk 75 158 616 8 holiday (节) *tsiɛt 240 23-24 606 9 pigeon (鸽) *kəp 417 39 590 10 horn (角) *kok 317 242 651 217 APPENDIX D IPA DESCRIPTIONS FOR WU DIALECTS (*VOICED INITIALS) Karlgren (高本汉) Suzhou Wuxi Shanghai Wenzhou 1 thin (薄) *buɑk boʔ23 bɔʔ2 boʔ2 bo12 2 pull out (拔) *buɑt baʔ23 baʔ2 bɑʔ2 bo12 3 butterfly (蝶) *diɛp diɪʔ23 laʔ2 laʔ2 di12 4 wax (腊) *lɑp laʔ23 lɔʔ2 loʔ2 la12 5 deer (鹿) *luk loʔ23 lɔʔ2 loʔ2 ləu12 6 stone (石) *dziæk zɒʔ23 zɑʔ2 zɑʔ2 zei12 7 honey (蜜) *miɪt miɪʔ23 miʔ2 miʔ2 mi13 8 wheat (麦) *muək mɒʔ23 mɑʔ2 mɑʔ2 ma12 9 hot (热) *niæt ɲiɪʔ23 ɲiəʔ2 ɲiəʔ2 ɲi12 10 sun; day (日) *niɪt ɲiɪʔ23 ɲiəʔ2 ɲiəʔ2 nE12 218 APPENDIX E IPA DESCRIPTIONS FOR WU DIALECTS (*VOICELESS INITIALS) Karlgren (高本汉) Suzhou Wuxi Shanghai Wenzhou 1 pen (笔) *piuet piɪʔ4 piəʔ5 piəʔ5 pi23 2 peel (剥) *pok poʔ4 pɔʔ5 poʔ5 po23 3 pick (摘) *thiek tsɒʔ4 tsɑʔ5 tiəʔ5 tsa23 4 bamboo(竹) *tiuk tsoʔ4 tsɔʔ5 tsoʔ5 tɕiu23 5 color (色) *siɪk sɤʔ4 səʔ5 səʔ5 sE23 6 snow (雪) *siuæt siɪʔ4 siəʔ5 siəʔ5 ɕy23 7 rule (尺) *tshiæk tshɒʔ4 tshɑʔ5 tshɑʔ5 tshei23 8 holiday (节) *tsiɛt tsiəʔ5 tsiəʔ5 9 pigeon (鸽) *kəp kɤʔ4 kəʔ5 kəʔ5 ky23 10 horn (角) *kok koʔ4 kɔʔ5 koʔ5 ko23 219 APPENDIX F IPA DESCRIPTIONS FOR HAKKA & MIN DIALECTS (*VOICED INITIALS) Karlgren (高本汉) Meixian Xiamen Chaozhou Fuzhou 1 thin (薄) *buɑk phɔk4 poʔ5 poʔ4 pɔʔ4 2 pull out (拔) *buɑt puiʔ5 poiʔ4 peiʔ4 3 butterfly (蝶) *diɛp iaʔ5 iaʔ4 tieʔ4 4 wax (腊) *lɑp lap4 laʔ5 laʔ4 laʔ4 5 deer (鹿) *luk luk4 lɔk5 tek4 løyʔ4 6 stone (石) *dziæk sak4 tshioʔ5 tsieʔ4 suɔʔ4 7 honey (蜜) *miɪt mɛt4 bit5 bik4 miʔ4 8 wheat (麦) *muək mak4 beʔ5 beʔ4 maʔ4 9 hot (热) *niæt ɲiat4 liat5 ieʔ4 10 sun; day (日) *niɪt ɲit21 lit5 zik4 niʔ4 220 APPENDIX G IPA DESCRIPTIONS FOR HAKKA & MIN DIALECTS (*VOICELESS INITIALS) Karlgren (高本汉) Meixian Xiamen Chaozhou Fuzhou 1 pen (笔) *piuet pit21 pit32 pik21 peiʔ23 2 peel (剥) *pok pɔk21 pak32 pak21 puɔʔ23 3 pick (摘) *thiek tsak21 tiaʔ32 tiaʔ21 tieʔ23 4 bamboo(竹) *tiuk tsuk21 tik32 tek21 tøyʔ23 5 color (色) *siɪk sət21 sik32 sek21 6 snow (雪) *siuæt siɛt21 seʔ32 soʔ21 suɔʔ23 7 rule (尺) *tshiæk tshak21 tshioʔ32 tshieʔ21 tshuɔʔ23 8 holiday (节) *tsiɛt tsiɛt21 tsueʔ32 tsoiʔ21 9 pigeon (鸽) *kəp kap21 kap32 kaʔ23 10 horn (角) *kok kɔk21 kak32 khak21 kɛyʔ23 221 APPENDIX H IPA DESCRIPTIONS FOR MANDARIN DIALECTS (*VOICED INITIALS) Beijing Shenyang Xian Xuzhou Pixian Jinan 1 thin (薄) pau35 pau35 pho24 pə55 pə55 pə42 2 pull out (拔) pa35 pa35 phɑ24 pɑ55 pɑ55 pa42 3 butterfly(蝶) tier35 tier213 tiə55 tie55 tie42 4 wax (腊) la51 la41 lɑ21 lɑ313 lɑ313 la21 5 deer (鹿) lu51 lu41 lou21 lu313 lu313 lu21 6 stone (石) ʂɩ35 sɿ35 ʂɩ24 ʂɩ55 ʂɩ55 ʂɩ42 7 honey (蜜) mi51 mi41 mi313 mi313 mi21 8 wheat (麦) mai51 mai41 mei21 me313 mei313 mei21 9 hot (热) ʐə51 ie41 ʐə21 ʐə313 ʐə313 ʐə21 10 sun; day (日) ʐɩ51 i41 ɚ21 ʐɩ313 ʐɩ313 ʐɩ21 222 APPENDIX I IPA DESCRIPTIONS FOR MANDARIN DIALECTS (*VOICELESS INITIALS) Beijing Shenyang Xian Xuzhou Pixian Jinan 1 pen (笔) pi214 pi213 pi21 pe313 pei313 pei213 2 peel (剥) pau55 pau33 po21 puə313 puo313 pə213 3 pick (摘) tʂai55 tsai35 tsei21 tʂe313 tʂei313 tʂei213 4 bamboo(竹) tʂu35 tsu35 tsou21 tʂu313 tʂu313 tʂu213 5 color (色) ʂair51 sai213 sei21 se313 ʂei313 ʂei213 6 snow (雪) ɕye214 ɕye213 ɕye21 ɕye313 ɕye313 ɕye213 7 rule (尺) tʂhɩ214 tʂhɩ213 tʂhɩ21 tʂhɩ313 tʂhɩ313 tʂhɩ213 8 holiday (节) tɕie35 tɕie213 tɕie21 tɕiə313 tɕie313 tɕie213 9 pigeon (鸽) kə55 kə33 kə21 kə313 kə313 10 horn (角) tɕiau214 ka33 tɕyə313 tɕyə313 ka213 223 APPENDIX J IPA DESCRIPTIONS FOR CANTONESE DIALECTS (*VOICED INITIALS) Karlgren (高本汉) Yangjiang Guangzhou 1 thin (薄) *buɑk pɔk454 pɔk33 2 pull out (拔) *buɑt pɐt2 3 butterfly (蝶) *diɛp tip454 thie22 4 wax (腊) *lɑp lap454 lap22 5 deer (鹿) *luk lʊk454 lʊk2 6 stone (石) *dziæk ʃɪk454 ʃɛk22 7 honey (蜜) *miɪt mɐt454 mɐt2 8 wheat (麦) *muək mɐk454 mɐk2 9 hot (热) *niæt jit454 jit22 10 sun; day (日) *niɪt jɐt454 jit22 224 APPENDIX K IPA DESCRIPTIONS FOR CANTONESE DIALECTS (*VOICELESS INITIALS) Karlgren (高本汉) Yangjiang Guangzhou 1 pen (笔) *piuet pɐt24 pɐt5 2 peel (剥) *pok mɐk24 mɔk33 3 pick (摘) *thiek tʃak33 4 bamboo(竹) *tiuk tʃʊk24 tʃʊk5 5 color (色) *siɪk ʃɪk24 ʃɪk5 6 snow (雪) *siuæt ʃyt33 7 rule (尺) *tshiæk tʃhɪk24 tʃhɛk33 8 holiday (节) *tsiɛt tʃit21 tʃit33 9 pigeon (鸽) *kəp kɐp21 kap33 10 horn (角) *kok kɔk21 kɔk33 225 APPENDIX L IPA DESCRIPTIONS FOR MANDARIN DIALECTS, XIANG, AND GAN (*VOICED INITIALS) Yangzhou Nanjing Chengdu Kunming Changsha Nanchang 1 thin (薄) pɑʔ4 poʔ5 po31 po31 po24 phɔk5 2 Pull out (拔) pæʔ4 pɑʔ5 pha31 pa31 pa24 phat5 3 butterfly(蝶) thiəʔ4 theʔ5 tie31 tie31 thie24 thiɛt5 4 wax (腊) æʔ4 lɑʔ5 na31 la31 la24 lat5 5 deer (鹿) lɔʔ4 luʔ5 nu31 lu31 ləu24 luk5 6 stone (石) səʔ4 ʂɩʔ5 sɿ31 ʂɩ31 sɿ24 sak5 7 honey (蜜) miəʔ4 miʔ5 mi31 mi24 mit5 8 wheat (麦) mɔʔ4 məʔ5 me31 mə31 mə24 mat5 9 hot (热) iəʔ4 ʐəʔ5 ze31 ʐə31 ye24 lɛt5 10 sun;day(日) ləʔ4 ʐɩʔ5 zɿ31 ʐɩ31 zɿ24 ɲit5 226 APPENDIX M IPA DESCRIPTIONS FOR MANDARIN DIALECTS, XIANG, AND GAN (*VOICELESS INITIALS) Yangzhou Nanjing Chengdu Kunming Changsha Nanchang 1 pen (笔) piəʔ4 piʔ5 pi31 pi31 pi24 pit5 2 peel (剥) pɑʔ4 poʔ5 po31 po31 po24 pɔt5 3 pick (摘) tsəʔ4 tsəʔ5 tse31 tsə31 tsə24 tsak5 4 bamboo(竹) tsɔʔ4 tʂuʔ5 tsu31 tʂu31 tsəu24 tsuk5 5 color (色) seʔ4 səʔ5 se31 sə31 sə24 sɛt5 6 snow (雪) ɕyeʔ4 ɕyeʔ5 ɕye31 ɕie31 ɕie24 ɕyɔt5 7 rule (尺) tshəʔ4 tʂhɩʔ5 tʂhɿ31 tʂhɩ31 tʂhɿ24 tshak5 8 holiday (节) tɕiəʔ4 tɕieʔ5 tɕie31 tɕie31 tɕie24 9 pigeon (鸽) kəʔ4 koʔ5 ko31 kuo31 ko24 kɔt5 10 horn (角) kɑʔ4 koʔ5 ko31 kuo31 ko24 kɔk5 227 APPENDIX N REFERENCES & RECONSTRUCTIONS FOR *VOICED INITIALS IN YI AND LOLOISH Bradley‘s Reconstruction (*L) & (No.) Matisoff‘s Reconstruction (*LB) & (No.) The Tibeto-Burman Lexicon Huang (ed.)/(No.) 1 hand *lak/L (111) *lak/L (166) 107 2 pig *wak/L (21) *wak/L (168) 284 3 year *C-kok/L (477B) *C-kok/L (34) 777 4 enough *lok/L (560) *lok/L (164) 1349 5 six *C-krok/L (483) *C-krok/L (35) 802 6 waist *ap/L (118) *ɡyok/L (6) *džok/L 98 7 lick *m-lyak/L (630) *ʔlyak/L (179) *m-lyak/L 1674 8 vomit *C-pat/L (577) *C-pat/L (38) 1688 9 needle *ɡ-rap/L (382) *rap/L (191) 578 10 sleep *yip/L (735) *(ʔ)yip/L (180) 1646 11 goat *(K)-cit/L (4) *V-cit/L (27) 276 12 bite *C-tsat/L (634A) *C-tsat/L (24) 1753 13 hungry *C-mwat/L (637) *mwat/L (132) 1298 14 brain *(C)-nok/L (140) *(ʔ)nok/L (156) 132 15 stand *ʔ-ryap/L (687) *ʔrap/L (175) 1784 16 thirsty *C-sip/L (638) *C-sip/L (129) 1478 17 eight *C-yet/L (485) *ʔrit/L (171) 804 18 new *C-ʃik/L (536) 1050 19 boil *C-sak/L (644) *(ʔ)ɡyak/L (61) 1804 20 kill *C-sat/L (706) *C-sat/L (124) 1602 21 sew *ɡyup/L (680) *(ʔ)ɡrup/L (63) 1324 Note that *C-prefix is a cover-symbol which indicates *b, *d, *g, and *l prefixes in *TB. 228 APPENDIX O REFERENCES & RECONSTRUCTIONS FOR *VOICELESS INITIALS IN YI AND LOLOISH Bradley‘s Reconstruction (*H) Matisoff‘s Reconstruction (*LB) & (No.) The Tibeto-Burman Lexicon Huang (ed.)/(No.) 1 chicken *k-rak/H (50) *k-rak/H (184) 293 2 black *c-nak/H (503) *s-nak/H (142) 1005 3 bean *s-nok/H (276A) *s-nok/H (140) 430 4 come out *ʔ-dwak/H (656) *ʔtwak/H (102) *Ntwak/H 1207/1208 5 be afraid *(sə)-ɡrok/H (689) *ʔkrok/H (104) *Nkrok/H 1366 6 shoot *(ʔ)-m-pok/H (337) *Npök/H (108) 1611/1612 7 eye *c-myak/H (92) *s-myak/H (145) 79 8 blow *s-mut/H (690) *s-mut/H (143) 1219 9 ascend/ go up *ʔ-dak/H (652) *ʔtak/H (98) *Ntak/H 1609 10 cold *c-grak/H (514) *ʔkrak/H (99) *Nkrak 1063 11 sharp *tak/H (543) *tak/H (41) 1020 12 tree *sik/H (303) *sik/H (118) 372 13 bird *s-ŋyak/H (48) *s-ŋak/H (141) 326 229 APPENDIX P IPA DESCRIPTIONS FOR YI DIALECTS (*VOICED INITIALS) Matisoff‘s Reconstruction (*LOW) Yi 1 hand *lak/L Weishan: lɪ21; Mile: le2; Lunan: le2; Xide: lo55; Wuding: lɑ55; Luquan: la55; Nanhua: le21; Sani: le2 2 pig *wak/L Weishan: vɪ21; Mile: ve2; Lunan: ve2; Xide: vo55; Wuding: vɑ55; Luquan: va55; Nanhua: ve21; Sani: ve2 3 year *C-kok/L Weishan: kho21; Mile: khu2; Lunan: qho2; Xide: khu55; Wuding: khu55; Luquan: kho55; Nanhua: khu55; Sani: qhu3 4 enough *lok/L Weishan: ʔlo21; Mile: lu2; Lunan: lo2; Xide: lu55; Wuding: lu55; Luquan: lo55; Nanhua: lo21; Sani: lu2 5 six *C-krok/L Weishan: kho21; Mile: tʂhu2; Lunan: kho2; Xide: fu55; Wuding: tɕhu55; Luquan: tɕho55; Nanhua: tɕho55; Sani: khu2 6 waist *ɡyok/L Weishan: dʐo21; Mile: dʐu21; Lunan: dʐo2; Xide: dʑu55; Wuding: dʐ u55; Luquan: dʐ o55; Nanhua: dʐ u21; Sani: dʐ u2 7 lick *m-lyak/L Weishan: la21; Mile: ɬa2; Lunan: ɬa2; Xide: ʑo55; Wuding: lɚ55; Luquan: lɚ55; Nanhua: læ21; Sani: ɬɑ2 8 vomit *C-pat/L Weishan: tshɿ21; Mile: phi2; Lunan: phɪ2; Xide: pɑ55; Wuding: ti55; Luquan: phi55; Nanhua: pɑ55; Sani: phɪ2 9 needle *rap/L Weishan: y21; Mile: ɣe2; Lunan: ɣɤ2; Xide: ʑi55; Wuding: ɣɤ55; Luquan: ɣɤ55; Nanhua: ɣə21; Sani: ɣɤ2 10 sleep *(ʔ)yip/L Weishan: ʔɪ21; Xide: i55; Wuding: ji55; Nanhua: ʔi21; Sani: ji2 11 goat *V-cit/L Weishan: tʂhɿ21; Xide: tʂhɿ55; Wuding: tʂhɿ55; Luquan: tʂhi55; Nanhua: tʂhɿ55; Sani: tɕhi2 12 bite *C-tsat/L Weishan: kho21; Xide: ɕi55; Wuding: tɕhɤ55; Luquan: tɕhɤ55; Nanhua: khɑ55; Sani: qhɯ2 13 hungry *mwat/L Weishan: mɯ21; Xide: mi55; Wuding: ɲi55; Luquan: ɲi55; Nanhua: me21; Sani: n2 230 14 brain *(ʔ)nok/L Weishan: ʔno21; Mile: nu2; Lunan: no2; Xide: no55; Wuding: nu55; Luquan: no2; Nanhua: nu21; Sani: nu44 15 stand *ʔrap/L Weishan: hy21; Mile: xo2; Lunan: hɤ2; Xide: hi55; Wuding: he55; Luquan: he55; Nanhua: xə21; Sani: hɤ2 16 thirsty *C-sip/L Weishan: sɿ21; Xide: sɿ55; Wuding: si55; Luquan: si55; Nanhua: ɕɿ55; Sani: sz2 17 eight *ʔrit/L Weishan: hɪ21; Xide: hi55; Wuding: hi55; Luquan: hi55; Nanhua: xe21; Sani: he2 18 new *C-ʃik/L Weishan: xɯ21; Xide: ʂɿ55; Wuding: ɕi55; Nanhua: ɕi55; Sani: ɕi2 19 boil *(ʔ)ɡyak/L Weishan: tɕɪ21; Xide: tɕo55; Wuding: tʂɑ55; Nanhua: tɕe55; Sani: tɕe55 20 kill *C-sat/L Weishan: ɕɪ21; Xide: si55; Nanhua: se55 21 sew *(ʔ)ɡrup/L Weishan: ɡu21; Xide: ɡu55; Wuding: nɚ55; Nanhua: næ21; Sani: nɑ2 231 APPENDIX Q IPA DESCRIPTIONS FOR YI DIALECTS (*VOICELESS INITIALS) Matisoff‘s Reconstruction (*HIGH) Yi 1 chicken *k-rak/H Weishan: ʑɪ33; Mile: ʑe33; Lunan: ʑe4; Xide: vɑ33; Wuding: ɣɑ2; Luquan: ɣa2; Nanhua: ʑi33; Sani: je44 2 black *s-nak/H Weishan: nɪ33; Mile: ne33; Lunan: ne4; Xide: nɔ33; Wuding: nɑ2; Luquan: na2; Nanhua: ne33; Sani: ne44 3 bean *s-nok/H Weishan: no33; Mile: nu33; Lunan: no4; Xide: nu33; Wuding: nu2; Luquan: no2; Nanhua: no33; Sani: nu44 4 come out *ʔtwak/H Weishan: do33; Mile: du33; Lunan: do4; Xide: du33; Wuding: du2; Luquan: do2; Nanhua: du33; Sani: du44 5 be afraid *ʔkrok/H Weishan: ɡo33; Mile: dʐ u33; Lunan: ɡo4; Xide: dʑu33; Wuding: tɕi2; Luquan: dʑo2; Nanhua: tɕi55; Sani: ɡu44 6 shoot *Npök/H Weishan: ba33; Mile: ba33; Lunan: ba4; Xide: mbe33; Wuding: mphɚ2; Luquan: mbhɚ2; Nanhua: bæ33; Sani: bɑ44 7 eye *s-myak/H Weishan: ʔmɪ33; Xide: ɲɔ33; Wuding: nɑ2; Nanhua: me33; Sani: ne44 8 blow *s-mut/H Weishan: ʔm33; Xide: mo33; Wuding: mɯ2; Nanhua: mɯ33; Sani: m44 9 ascend *ʔtak/H Weishan: dɪ33; Mile: de33; Lunan: de4; Xide: li33; Wuding: dɑ2; Luquan: da2; Nanhua: de33; Sani: de44 10 cold *ʔkrak/H Weishan: gu33; Xide: ŋɡo33; Wuding: dʑɑ2; Nanhua: ɲæ33; Sani: ŋe33 11 sharp *tak/H Weishan: ʔnɑ55; Mile: the33; Lunan: the4; Xide: thɔ33; Wuding: thɑ2; Luquan: tha2; Nanhua: the33; Sani: the44 12 tree *sik/H Weishan: sɿ33; Mile: si33; Lunan: si4; Xide: sɿ33; Wuding: si2; Luquan: si2; Nanhua: ɕi33; Sani: sz44 13 bird *s-ŋak/H Weishan: n33; Xide: tsɿ33; Wuding: ŋɑ2; Nanhua: ɲæ33; Sani: ŋe33 232 APPENDIX R REFERENCES FOR *VOICED INITIALS IN NAXI AND MOSUO Naxi Naxiyujianzhi (纳西语简志) Mosuo Naxiyujianzhi (纳西语简志) & A Descriptive Grammar of Yongning Na (Mosuo) 1 hand He (1985:143) He (1985:143) 2 pig He (1985:139) He (1985:139) 3 year He (1985:137) He (1985:137) 4 enough Huang (1992, no.1349) Lidz (2010:795) 5 six He (1985:170) He (1985:170) 6 waist He (1985:144) He (1985:144) 7 lick 8 vomit He (1985:155) He (1985:155) 9 needle He (1985:150) He (1985:150) 10 sleep He (1985:157) He (1985:157) 11 goat He (1985:139) He (1985:139) 12 bite Huang (1992, no.1753) Lidz (2010:821) 13 hungry 14 brain Lidz (2010:796) 15 stand He (1985:156) He (1985:156) 16 thirsty 17 eight He (1985:170) He (1985:170) 18 new He (1985:168) Lidz (2010:818) 19 boil He (1985:158) He (1985:158) 20 kill He (1985:159) He (1985:159) 21 sew 233 APPENDIX S REFERENCES FOR *VOICELESS INITIALS IN NAXI AND MOSUO Naxi Naxiyujianzhi (纳西语简志) Mosuo Naxiyujianzhi (纳西语简志) & A Descriptive Grammar of Yongning Na (Mosuo) 1 chicken He (1985:139) He (1985:139) 2 black He (1985:167) He (1985:167) 3 bean He (1985:141) He (1985:141) 4 come out Huang (1992, no.656) 5 be afraid He (1985:162) Lidz (2010:776) 6 shoot 7 eye He (1985:143) He (1985:143) 8 blow He (1985:155) He (1985:155) 9 ascend/ go up He (1985:163) He (1985:163) 10 cold 11 sharp Huang (1992, no.543) Lidz (2010:820) 12 tree Lidz (2010:815) 13 bird 234 APPENDIX T IPA DESCRIPTIONS FOR POSSIBLE LOLOISH & BURMISH (*VOICED INITIALS) Matisoff‘s Reconstruction (*Low) Loloish & Burmish 1 hand *lak/L Lahu: lɑ21; Hani (Luchun): la31; Hani (Mojiang): la31; Lisu: lɛ31; Jinou: lɑ55; Naxi: lɑ31; Mosuo: lɑ31; Written Burmese: lɑk4; Burmese: lɛʔ4; Achang: lɔʔ55 2 pig *wak/L Lahu: vɑ21; Hani (Luchun): ɣa31; Hani (Mojiang): ja31; Lisu: vɛ31; Jinou: va55; Naxi: mbu31; Mosuo: bu13; Written Burmese: wɑk4; Burmese: wɛʔ4; Achang: oʔ55 3 year *C-kok/L Lahu: qhɔ21; Hani (Luchun): xu31; Hani (Mojiang): xv31; Lisu: kho31; Naxi: khu55; Mosuo: khu13; Written Burmese: hnɑs4; Burmese: niʔ4; Achang: nək55 4 enough *lok/L Lahu: lɔ21; Hani (Luchun): lu31; Hani (Mojiang): lv31; Lisu: lo31; Jinou: lo55; Naxi: lu31; Mosuo: lu31; Written Burmese: lɔk4; Burmese: lɑuʔ4 5 six *C-krok/L Lahu: khɔ21; Hani (Luchun): ku31; Hani (Mojiang): khv31; Lisu: tɕho31; Jinou: tɕho44; Naxi: tʂho55; Mosuo: khər13 Written Burmese: khrɔk4; Burmese: tɕhɑuʔ4; Achang: xʐoʔ55 6 waist *ɡyok/L Lahu: tɕɔ21; Hani (Luchun): tsɿ31; Hani (Mojiang): tshɿ31; Lisu: dʑo31; Jinou: tʃo55; Naxi: thɯ55; Mosuo: tʂa13 7 lick *m-lyak/L Lahu: lɛ21; Hani (Luchun): mje31; Hani (Mojiang): mɤ31; Lisu: lɯ31; Jinou: mrə55; Written Burmese: jɑk4 ; Burmese: jɛʔ4; Achang: liap55 8 vomit *C-pat/L Lahu: phe21; Jinou: phi55; Naxi: phy55; Mosuo: phi13 9 needle *rap/L Lahu: ɣo21; Hani (Luchun): ɣo31; Hani (Mojiang): ɣo31; Jinou: kø55; Naxi: ko31; Mosuo: ɣo13; Written Burmese: ɑp4; Burmese: ɑʔ4; Achang: ap55 10 sleep *(ʔ)yip/L Lahu: zi21; Hani (Luchun): ju31; Hani (Mojiang): ʒv31; Jinou: ji55; Naxi: i55; Mosuo: ʐɯ13; Written Burmese: ip4; Burmese: eʔ4 235 11 goat *V-cit/L Lahu: tɕhe21; Hani (Luchun): tsi31; Hani (Mojiang): tʃhɿ31; Lisu: tʃhi31; Jinou: tshɿ55; Naxi: tshɯ55; Mosuo: tshɯ13; Written Burmese: shit4; Burmese: sheʔ4; Achang: paʔ55 12 bite *C-tsat/L Lahu: tɕhe21; Hani (Luchun): ko31; Hani (Mojiang): kho31; Lisu: kho31; Jinou: thə55; Naxi: thər55; Mosuo: thæ13; Written Burmese: kok4; Burmese: kɑiʔ4 13 hungry *mwat/L Lahu: mɤ21; Hani (Luchun): mi31; Hani (Mojiang): me31; Lisu: mɯ31; Jinou: mæ55; Achang: ʂut55 14 brain *(ʔ)nok/L Lahu: nɔ21; Hani (Luchun): nɔ31; Hani (Mojiang): nɤ31; Lisu: nɯ55; Jinou: no55; Mosuo: ɬu13; Written Burmese: hnɔk4; Burmese: nɑuʔ4; Achang: nuʔ31 15 stand *ʔrap/L Lahu: xu35; Hani (Luchun): ɕo31; Hani (Mojiang): ʃo31; Lisu: he31; Jinou: xɛ55; Naxi: ɕy55; Mosuo: xĩ13; Written Burmese: rɑp4; Burmese: jɑʔ4; Achang: ʑap55 16 thirsty *C-sip/L Lahu: ɕi35; Hani (Mojiang): ʃv31; Lisu: se31; Jinou: ɕi55; Written Burmese: ŋɑt4; Burmese: ŋɑʔ4; Achang: ʂut55 17 eight *ʔrit/L Lahu: xi35; Hani (Luchun): ɕe31; Hani (Mojiang): xɛ31; Lisu: he31; Jinou: xɛ44; Naxi: χo55; Mosuo: xo13; Written Burmese: hrɑs4; Burmese: ɕiʔ4; Achang: ɕet55 18 new *C-ʃik/L Lahu: si35; Hani (Luchun): ʃɤ31; Hani (Mojiang): sɿ31; Lisu: ʃi31; Jinou: ɕi55; Naxi: ʂɿ55; Mosuo: ʂɯ13; Written Burmese: tθɑs4; Burmese: tθiʔ4; Achang: ʂək55 19 boil *(ʔ)ɡyak/L Lahu: tɕɑ35; Hani (Luchun): tʃha31; Hani (Mojiang): tɕa31; Lisu: tɕɑ55; Jinou: tʃha55; Naxi: tɕiə55; Mosuo: tɕɑ13; Written Burmese: khjɑk4; Burmese: tɕhɛʔ4 20 kill *C-sat/L Hani (Luchun): si31; Hani (Mojiang): se31; Lisu: se31; Jinou: sɛ55; Naxi: sər55; Mosuo: sɯ13 Written Burmese: tθɑt4; Burmese: tθɑʔ4; Achang: sat55 21 sew *(ʔ)ɡrup/L Hani (Luchun): kv31; Hani (Mojiang): gu31; Lisu: dʒi31; Jinou: tɕu55 Written Burmese: khjup4; Burmese: tɕhoʔ4; Achang: xʐop55 236 APPENDIX U IPA DESCRIPTIONS FOR POSSIBLE LOLOISH & BURMISH (*VOICELESS INITIALS) Matisoff‘s Reconstruction (*HIGH) Loloish & Burmish 1 chicken *k-rak/H Lahu: ɣɑ54; Hani (Luchun): xa33; Hani (Mojiang): xa33; Lisu: ɣɑ33; Jinou: ja42; Naxi: æ31; Mosuo: a13; Written Burmese: krɑk4; Burmese: tɕɛʔ4 Achang: kʐuaʔ55 2 black *s-nak/H Lahu: nɑ54; Hani (Luchun): na33; Hani (Mojiang): na33; Lisu: nɛ33; Jinou: na42; Naxi: nɑ31; Mosuo: nɑ13; Written Burmese: nɑk4; Burmese: nɛʔ4; Achang: lɔk55 3 bean *s-nok/H Lahu: nɔ54; Hani (Luchun): nɤ33; Hani (Mojiang): nɯ33; Lisu: no33; Naxi: nu31; Mosuo: nu31; Achang: tsheʔ31 4 come out *ʔtwak/H Lahu: tɔ54; Hani (Luchun): tv33; Hani (Mojiang): du33; Lisu: do33; Jinou: to42; Naxi: thv33; Written Burmese: thwɑk4; Burmese: thwɛʔ4; Achang: thoʔ55 5 be afraid *ʔkrok/H Lahu: kɔ54; Hani (Luchun): kv33; Hani (Mojiang): gu33; Lisu: dʑo33; Jinou: khø33; Naxi: ʐər33; Mosuo: dwæ13; Written Burmese: krɔk4; Burmese: tɕɑuʔ4; Achang: ʐoʔ55 6 shoot *Npök/H Lahu: bɔ54; Hani (Luchun): pɤ33; Hani (Mojiang): bɤ33; Lisu: bɯ33; Jinou: pə42; Written Burmese: pɑs4; Burmese: piʔ4; Achang: pək55 7 eye *s-myak/H Lahu: mɛ54; Hani (Luchun): ma33; Hani (Mojiang): mja33; Lisu: miɛ33; Jinou: mja42; Naxi: miə31; Mosuo: ɲɑ31; Written Burmese: mjɑk4; Burmese: mjɛʔ4; Achang: ɲɔʔ55 237 8 blow *s-mut/H Lahu: mɤ54; Hani (Luchun): mɤ33; Lisu: mu33; Naxi: mu31; Mosuo: mu13; Written Burmese: hmut4; Burmese: moʔ4; Achang: mut55 9 ascend *ʔtak/H Lahu: tɑ54; Hani (Luchun): ta33; Hani (Mojiang): da33; Lisu: dɛ33; Jinou: ta42; Naxi: ndo33; Mosuo: do33; Written Burmese: tɑk4; Burmese: tɛʔ4; Achang: tɔʔ35 10 cold *ʔkrak/H Lahu: kɑ54; Hani (Luchun): ka33; Lisu: dʑɛ33; Achang: kʐuat55; Achang: kʐuat55 11 sharp *tak/H Lahu: tsi54; Hani (Luchun): tha33; Hani (Mojiang): ta33; Lisu: tshɛ35; Jinou: tha42; Naxi: tha55; Mosuo: thɑ13; Written Burmese: thɑk4; Burmese: thɛʔ4; Achang: thɔʔ55 12 tree *sik/H Lahu: si54; Lisu: si35; Jinou: sɯ44; Mosuo: sɯ33; Written Burmese: tθɑs4; Burmese: tθiʔ4 13 bird *s-ŋak/H Lahu: ŋɑ54; Lisu: ɲiɛ35; Jinou: ŋa42; Written Burmese: hŋɑk4; Burmese: ŋɛʔ4; Achang: mɔʔ55 238 REFERENCES CITED Bennett, Paul A. 1981. The evolution of passive and disposal sentences. Journal of Chinese Linguistics 9.61-90. Beyer, Stephan V. 1992. The Classical Tibetan Language. Albany: State University of New York Press. Bradley, David. 1975. Nahsi and Proto-Burmese-Lolo. Linguistics of the Tibeto-Burman Area 2.1.93-150. Bradley, David. 1978. Proto-Loloish (Scandinavian Institute of Asian Studies Monograph No. 39). London: Curzon Press. Bradley, David. 1979. Lahu dialects (Issue 23 of Oriental monograph series). Canberra: Australian National University Press. Chang, Gongen (常肱恩). 1986. Lahuyu Jian Zhi [拉祜语简志 (A grammar of Lahu)]. Beijing: Minzu Publisher Press. Chao, Y. R. 1930. A system of tone letters. Le Maitre Phonetique 45.24-7. Chappell, Hilary (曹茜蕾). 1992. Causativity and the ba construction in Chinese. Partizipation: Das sprachliche Erfassen von Sachverhalten, ed. by Hansjakob Seiler and Walfried Premper, 509-30. Tübingen: Gunter Narr Press. Chappell, Hilary (曹茜蕾). 2000. Dialect grammar in two Early Modern Southern Min texts: A comparative Study of dative kit, comitative cang and diminutive -guia. Journal of Chinese Linguistics 28.2.247–302. Chappell, Hilary (曹茜蕾). 2006. From Eurocentrism to Sinocentrism: the case of disposal constructions in Sinitic Languages. Catching language: the standing challenge of grammar writing, ed. by Felix Ameka, Alan Dench and Nicholas Evans, 441-86. Berlin: Mouton de Gruyter Press. Chen, Chuseng (陈初生). 1983. Zaoqi chuzhishi luelun [早期处置式略论 (A discussion of disposal construction in the early stage)]. Zhongguo Yuwen (中国语文) 3.191-206. Chen, Kang (陈康). 1986. Yiyu de sengdiao duiying [彝语的声调对应 (Tonal correspondences in Loloish)]. Minzuyuwen (民族语文) 5.30-9. Chen, Kang (陈康). 1988. Yiyu de jindiaolei [彝语的紧调类 (Tense tones in Loloish)]. Minzuyuwen (民族语文) 1.18-26. 239 Chen, Kang (陈康). 1991. Yiyuzhi diaolei quanshi [彝语支调类诠释 (An explanation of tones in Loloish)]. Minzuyuwen (民族语文) 3.19-27. Chen, Kang (陈康). 1993. Yimianyu seyin yunwei yanbian guiji [彝缅语塞音韵尾演变 轨迹 (Development of syllable-final stops in Lolo-Burmese)]. Minzuyuwen (民族语 文) 1.22-6. Chen, Kang (陈康). 1997. Lun yiyuzhi shengdiao xitong de fasheng yu liebian [论彝语 支声调系统的发生与裂变 (The origin of tones and tonal splits in Loloish)]. Minzuyuwen (民族语文) 1.8-17. Chen, Matthew Y. 2000. Tone sandhi: patterns across Chinese dialects (Cambridge. Studies in Linguistics Vol. 92). Cambridge: Cambridge University Press. Dai, Qingxia (戴庆厦). 1981. Yiyuzhi yuyan de qing zhuo shengmu [彝语支语言的清浊 声母 (Voiceless and voiced initial consonants in Loloish)]. Academic Journal of the Central University for Nationality (中央民族学院学报) 65-85. Dai, Qingxia (戴庆厦). 1992. Yimianyu biguan shengmu de laiyuan ji fazhan [彝缅语鼻 冠声母的来源及发展 (The origin and development of pre-nasalized initial consonants in Lolo-Burmese)]. Minzuyuwen (民族语文). 1.42-51. Dai, Qingxia (戴庆厦), and Zhichao Cui (崔志超). 1985. A Chang Yu Jian Zhi [阿昌语 简志 (A grammar of Achang)]. Beijing: Minzu Publisher Press. DeLancey, Scott. 1984. Etymological notes on Tibeto-Burman. Linguistics of the Tibeto- Burman area, ed. by Graham Thurgood 8.1.59-77. DeLancey, Scott. 1986. Relativization and nominalization in Tibetan and Newari. Presented at the 19th annual meeting of the International Conference on Sino-Tibetan Languages and Linguistics. DeLancey, Scott. 2004. Grammaticalization. Morphology: An International Handbook on Inflection and Word Formation, ed. by G. Booij, C. Lehmann, J. Mugdan, and S. Skopeteas, 1590-9. Berlin and New York: De Gruyter Press. DeLancey, Scott. 2011. Finite structures from clausal nominalization in Tibeto-Burman languages. Nominalization in Asian Languages: Diachronic and Typological. Volume I: Sino-Tibetan and Iranian Languages (Typological Studies in Language Series), ed. by F.H. Yap & J. Wrona. Amsterdam/Philadelphia: John Benjamins Press; to appear. Ding, Sizhi Picus. 2001. Semantic change vs. categorial change: A study of the development of ba in Mandarin. Journal of Chinese Linguistics 29.1.102-28. 240 Fang, guoyu (方国瑜), and Zhiwu He (和志武). 1979. Naxizu de yuanyun qianxi he fenbu [纳西族的渊源,迁徙和分布 (The origin, migration and distribution of Naxi)]. Minzuyanjiu (民族研究) 1.33-41. Fangyan Diaocha (ed.) 1960. Jiangsusheng he Shanghaishi fangyan gaikuang [江苏省和 上海市方言槪况 (A dialectal survey in province of Jiangsu and Shanghai City)]. Jiangsurenmin Press. Feng, Chuntian (冯春田). 1990. Shilun jiegou zhuci ―di‖ de yixie wenti [試論結構助詞 ‗底/(的)‘的一些問題 (A few questions about structural particle—―di‖)]. Zhongguo Yuwen (中国语文) 219.6.448-52. Feng, Chuntian (冯春田). 1992. Tang wudai mouxie yufa xianxiang tanxi [唐五代某些語 法現象探析 (A preliminary analysis of certain language phenomena during the Tang period)]. 232-67. Jinan: Shandong Education Publishing (山東教育出版社) Press. Ge, Xingzhi (盖兴之). 1982. Mianyi yuzhi chutan [緬彝語支初探 (Studies in the Mian- Yi language branch)]. Minzuxuebao (民族學報) 2.207-31. Ge, Xingzhi (盖兴之). 1986. Ji Nuo Yu Jian Zhi [基诺语简志 (A grammar of Jinuo)]. Beijing: Minzu Publisher Press. Ge, Xingzhi (盖兴之), and Zhuyi Jiang (姜竹仪). 1990. Naxiyu zai zangmianyuyan zhong de diwei [纳西语在藏缅语言中的地位 (The genetic position of Naxi in Tibeto-Burman)]. Minzuyuwen (民族语文) 1.63-73. Haudricourt, André-Georges. 1961. Bipartition et tripartition des systèms de tons dans quelques langues d'Extrême-Orient (Two-way and three-way splitting of tonal systems in some Far-Eastern languages) translated by Christopher Court. Bulletin de la Société de Linguistique de Paris 56.163-80. Hopper, Paul, and Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Language 56.2.251-99. He, Jiren (和即仁) and Zhuyi Jiang (姜竹仪). 1985. Na Xi Yu Jian Zhi [纳西语简志 (A grammar of Naxi)]. Beijing: Minzu Publisher Press. Her, One-Soon. 1990. Historical development of ba and jiang in the Tang Dynasty. Language Variation and Change 2.277-94. Cambridge: Cambridge University Press. Hu, Suha (胡素华). 2002. Yiyu Jiegou Zhuci Yanjiu [彝语结构助词研究 (Research on Structural Particles in Yi)]. Beijing: Minzu Publisher Press. 241 Huang, Bufan (黄布凡) (ed.) 1992. Zang Mian Yu Zu Yu Yan Ci Hui [藏缅语族语言词 汇 (A Tibeto-Burman Lexicon)]. Beijing: Central University for Nationality Press. Huang, Shuanfan (黄宣范). 1986. The history of the disposal construction revisted— evidence from Zen Dialogues in the Tang dynasty. Journal of Chinese Linguistics 14.1.43-51. Jiang, Lansheng (江蓝生). 1999. Chusuoci de lingge youngfa yu jiegou zhuci ―di‖ de youlai [处所词的领格用法与结构助词―底‖的由来 (The genitive use of place words and the origin of the structural particle di3)]. Zhongguo Yuwen (中国语文) 269.2.83- 93. Jiang, Lansheng (江蓝生). 2004. Kua cen fei duanyu jiegou ‗de hua‘ de cihuihua [跨层非 短语结构―的话‖的词汇话 (About the lexicalization of the cross-structural combination ―de hua‖)]. Zhongguo Yuwen (中国语文) 302.5.387-99. Jiang, Shaoyu (蒋绍愚). 1997. Baziju luelun—jian lun gongneng kuozhan [把字句略论, 兼论功能扩展 (A discussion of ba-sentence and its functional extensions)]. Zhongguo Yuwen (中国语文) 281.4.298-304. Jiang, Shaoyu (蒋绍愚). 2002. Geiziju jiaoziju biao beidong de laiyuan—jian lun yufahua, leitui he gongneng kuozhan [给字句, 教字句表被动的来源, 兼论语法化, 类推, 和功能扩展 (The origin of gei-sentence and jiao-sentence: grammaticalization, analogy, and functional extensions)]. Yuyangxue Luncong (语言学论丛) 26.159-77. Jiang, Shaoyu (蒋绍愚). 2009. Jindai hanyu de jizhong beidongshi [近代汉语的几种被 动式 (Several passive patterns in Modern Chinese)]. Academic Journal of Shanxi Normal University (山西师范大学学报) 38.6.87-92. Jiang, Zhuyi (姜竹仪). 1993. Naxiyu dongbu he xibu fangyan yufa yitong gaizhu [纳西 语东部和西部方言语法异同概述 (Differences between eastern and western dialects in Naxi)]. Minzuyuwen (民族语文) 4.43-8. Lahaussois, Aimée. 2002. Nominalization, relativization, genitivization in Thulung Rai. Proceedings of the 18th Annual Meeting of the Berkeley Linguistics Society. University of California-Berkeley Press. Lakoff, George. 1987. Women, fire, and dangerous things: What categories reveal about the mind. Chicago: University of Chicago Press. LaPolla, Randy, and Chenglong Huang. 2003. A Grammar of Qiang. Berlin: Mouton de Gruyter Press. 242 Li, Charles N. (李讷) and Sandra A. Thompson. 1981. Mandarin Chinese—A Functional Reference Grammar. University of California-Berkeley Press. Li, Fang-kuei (李方桂). 1977. Handbook of Comparative Tai. Honolulu, HI: University of Hawai‘i Press. Li, Lincan (李霖燦). 2001. Naxizu Xiangxing Biaoyinwenzi Zdiani (納西族象形標音文 字字典). Yunnan Minzu Press. Li, Yongsui (李永燧). 1992. Mianyi yuyan shengdiao bijiao yanjiu [缅彝语言声调比较 研究 (A comparative study of tones in Lolo-Burmese)]. Minzuyuwen (民族语文) 6.11-24. Li, Yongsui (李永燧). 1996. Mianyyu diaolei—lishi bijiaofa de yunyong [缅彝语调类: 历史比较法的运用 (A comparative reconstruction of tones in Lolo-Burmese)]. Minzuyuwen (民族语文) 5.1-16. Li, Yongsui (李永燧). 1999. Lun mianyiyu [论缅彝语 (A discussion of Lolo-Burmese)]. Minzuyuwen (民族语文) 2.18-26. Li, Yongsui (李永燧). 2000. Gongtong mianyiyu yunlei chulun [共同缅彝语韵类刍论 (A discussion of tones in Lolo-Burmese)]. Minzuyuwen (民族语文) 4.1-12. Li, Yongsui (李永燧). 2008. Mianyiyu—yizhong shengdiao zuyu [缅彝语:一种声调祖 语 (The origin of tones in Proto-Lolo-Burmese)]. Minzuyuwen (民族语文) 3.44-54. Li, Ying-che. 1974. What does ―disposal‖ mean? Features of the verb and noun in Chinese. Journal of Chinese Linguistics 2.200-18. Lidz, Liberty A. 2010. A Descriptive Grammar of Yongning Na (Mosuo). Ph.D. Dissertation. Austin: University of Taxes Press. Liu, Minzhi (刘敏芝). 2006. Songdai jiegou zhuci ‗di‘ de xinxing yongfa ji qi laiyuan [宋 代结构助词“底”的新兴用法及其来源 (Newer uses of the structural particle di in the Song Dynasty and their origin)]. Zhongguo Yuwen (中国语文) 310.1.59-64. Lu, Jianming (陆俭明). 1980. ―Huan‖ and ―geng‖ [―还‖和―更‖]. Yuyanxue luncong (语 言学论丛) 6.191-209. Beijing: Commercial Press. Lu, Shuxiang (吕叔湘). 1984. Lun di3, di4 zhi bian ji di3 zi de youlai [论底地之辨及底字 的由来 (A discussion on the controversy of di3 and di4 and the origin of di3)]. Han Yu Yu Fa Lun Wen Ji (汉语语法文集). Beijing: Commercial Press. 243 Maddieson, Ian, and Peter Ladefoged. 1985. Tense and lax in four minority languages of China. Journal of Phonetics 13.433-54. Mathieu, Christine. 2003. A history and anthropological study of the ancient kingdoms of the Sino-Tibetan borderland Naxi and Mosuo. Edwin Mellen Press. Matisoff, James. 1970. Glottal dissimilation and the Lahu high-rising tone: a tonogenetic case-study. Journal of the American Oriental Society 90.1.13-44. Matisoff, James. 1971. The tonal split in Loloish checked syllables. Occasional Papers of the Wolfenden Society on Tibeto-Burman Linguistics. University of Illinois-Urbana Press. Matisoff, James. 1972. The Loloish Tonal Split Revisited (Research Monograph No. 7, Center for South and Southeast Asia Studies). University of California-Berkeley Press. Matisoff, James. 1972. Lahu nominalization, relativization, and genitivization. Syntax and Semantics (Volume 1), ed. by John Kimball, 237-58. NY: Academic Press. Matisoff, James. 1973. Tonogenesis in Southeast Asia. Consonant Types and Tone. Southern California Occasional Papers in Linguistics, ed. by Larry M. Hyman, 1.71- 95. University of California-Los Angeles Press. Matisoff, James. 1985. God and the Sino-Tibetan copula, with some good news concerning selected Tibeto-Burman rhymes. Journal of Asian and African Studies (Tokyo Foreign Lamguages University) 29.1-81. Matisoff, James. 2003. Handbook of Proto-Tibeto-Burman: system and philosophy of Sino-Tibetan reconstruction. University of California Press. McKhann, Charles F. 1995. The Naxi and the nationalities question. Cultural Encounters on China’s Ethnic Frontiers, ed. by Stevan Harrell, 39-62. Seattle: University of Washington Press. Mei, Tsu-lin (梅祖麟). 1990. Tang song chuzhishi de laiyuan [唐宋处置式的来源 (The origin of the disposal construction in Tang and Song dynasties)]. Zhongguo Yuwen (中国语文) 3.191-206. Mei, Tsu-lin (梅祖麟). 1992. Hanzangyu de ―sui‖, ―yue‖, ―huan (xuan)‖ ―huan‖ ji qi xiangguan wenti [汉藏语的―歲、越‖, ―還(旋)、圜‖及其相关问题]. Zhongguo Yuwen (中国语文) 254.325-38. Mei, Tsu-lin (梅祖麟). 1988. Ciwei ‗di‘, ‗de‘ de laiyuan [词尾―底‖,―的‖的来源 (The origin of the particle di and de)]. Bulletin of the Institute of History and Philology Academia Sinica (中研院史语所集刊) 59.1.141-72. 244 Mei, Tsu-lin. 2000b (梅祖麟). Tang song chuzhishi de laiyuan [唐宋处置式的来源 (The origin of the disposal construction in Tang and Song Dynasties)]. Mei Tsu-lin yuyanxue lunwenji [梅祖麟语言学论文集] 188-221. Beijing: Commercial Press. Meng, Qinghui (孟庆惠). 1997. Hefeihua de zhe, na he sheme [合肥话的―这‖―那‖和―什 么‖ (―This‖, ―that‖ and ―what‖ in Hefei Mandarin)]. Zhongguo Yuwen (中国语文) 259.4.297-304. Newman, John. 1996. Give: A Cognitive Linguistic Study [Cognitive Linguistics Research 7]. Berlin and New York: Mouton de Gruyter Press. Noonan, Michael P. 1997. Versatile nominalization. Essays on language function and language type (in honor of Talmy Givón), ed. by Joan Bybee et al., 373-94. Philadelphia: John Benjamins Press. Peyraube, Alain. 1988. Syntactic change in Chinese: On grammaticalization. Bulletin of the Institute of History and Philology 59.3.617-52. Taipei: Academia Sinica Press. Peyraube, Alain. 1996. Recent issues in Chinese historical Syntax. New Horizons in Chinese Linguistics, ed. by C.-T. James Huang and Y.-H. Audrey Li, 161-214. Boston: Kluwer Academic Publishers Press. Pittayaporn, Pittayawat. 2009. The Phonology of Proto-Tai. Ph.D. dissertation. Department of Linguistics, Cornell University Press. Shen, Jiaxuan (沈家煊). 2001. Two constructions with Hai in Mandarin Chinese [跟副词 ‗还‘有关的两个句式]. In Zhongguo Yuwen (中国语文) 285.483-93. Shi, Feng (石锋), and Decai Zhou (周德才). 2005. Nanbu yiyu song jin yuanyin de shengxue biaoxian [南部彝语松紧元音的声学表现 (An acoustic analysis of tense and lax vowels in Southern Yi)]. Studies in Language and Linguistics (语言研究) 25.60-5. Shi, Yuzhi (石毓智), and Charles N. Li (李讷). 1998. Hanyu fazhan shi shang jiegou zhuci de xingti—lun DE de yufahua licheng [汉语发展史上结构助词的兴替—论―的‖的语 法化历程 (On the historical replacements among the three particles: zhi, zhe, and de)]. Zhongguo Shehui kexue (中国社会科学) 6.165-80. Shi, Yuzhi (石毓智), and Charles N. Li (李讷). 2002. The establishment of the classifier system and the grammaticalization of the morphosyntactic particle de in Chinese. Language Sciences 24.1-15. Elsevier Science Press. 245 Shi, Yuzhi (石毓智), and Tongshang Wang (王统尚). 2009. Fangyan zhong chuzhishi he beidonshi yongyou gongtong biaoji de yuanyin [方言中处置式和被动式拥有共同 标记的原因 (Common markers of the disposal and passive constructions in Chinese dialects)]. Hanyu Xuebao (汉语学报) 26.2.43-53. Sio, Joanna Ut-Seong. 2009. The Cantonese ge3. Hong Kong Polytechnic University. Manuscript. Sun, Chaofen (孙朝奋). 1995. Transitivity, the ba construction and its history. Journal of Chinese Linguistics 23.1.159-95. Sweetser, Eve E. 1988. Grammaticalization and semantic bleaching. Berkeley Linguistics Society 14th Proceedings 389-405. University of California-Berkeley Press. Thompson, Laurence C. 1976. Proto-Viet-Muong Phonology. Oceanic Linguistics Special Publications 13.1113-203. University of Hawai'i Press Thurgood, Graham. 2003. A subgrouping of the Sino-Tibetan languages: The interaction between language contact, change, and inheritance. The Sino-Tibetan languages, ed. by G. Thurgood and R. LaPolla, 3-21. London: Routledge Press. Traugott, Elizabeth C. 1982. From propositional to textual and expressive meanings: some semantic-pragmatic aspects of grammaticalization. Perspectives on historical linguistics, ed. by W. P. Lehmann and Y. Malkiel, 245-71. Amsterdam: John Benjamins Press. Tsao, Feng-fu (曹逢甫). 1987. A Topic-comment approach to thr BA construction. Journal of Chinese Linguistics 15.1.1-54. Wang, Li (王力). 1947. Zhongguo Xiandai Yufa [中国现代语法 (Modern Chinese Grammar)]. Shanghai: Zhonghua Press. Wang, Li (王力). 1958. Han Yu Shi Gao [汉语史稿]. Beijing: Zhonghua Press. Wang, William S-Y. 1967. Phonological features of tone. International Journal of American Linguistics 33.2.93-105. Wu, Fuxiang (吴福祥). 2003. Zai lun chuzhishi de laiyuan [再论处置式的来源 (A further discussion regarding the origin of the disposal construction)]. Studies in Language and Linguistics (语言研究) 23.3.1-12. Wu, Yunji (伍云姬). 1999. An etymological study of disposal and passive markers in Hunan dialects. Journal of Chinese Linguistics 27.2.90-123. 246 Wu, Yicheng, and Adams Bodomo. 2009. Classifiers ≠ Determiner. Linguistic Inquiry. 40.3.487-503. MA: MIT Press. Xing, Janet Zhiqun. 2006. Mechanisms of semantic change in Chinese. Studies in Language, ed. by Bernard Comeir, 30.3.461-83. Netherlands: John Benjamins Press. Xu, Hui-ling, and Stephen Matthews. 2011. On the polyfunctionality and grammaticalization of the morpheme kai in the Chaozhou dialect. Nominalization in Asian Languages: Diachronic and Typological Perspectives. Volume 1: Sino-Tibetan and Iranian Languages, ed. by Foong-ha Yap and Janick Wrona. Amsterdam/Philadelphia: John Benjamins Press, to appear. Xu, Lietong (徐烈烔), and Danqing Liu (刘丹青). 1998. Huati de Jiegou yu Gongneng [话题的结构与功能 (The structure and function of topic)]. Shanghai: Shanghai Education Press. Xu, Lin (徐琳), Yuzhang Mu (木玉璋), and Xingzhi Ge (盖兴之). 1986. Li su Yu Jian Zhi [傈僳语简志 (A grammar of Lisu)]. Beijing: Minzu Publisher Press. Yang, Hsiu-Fang (杨秀芳). 2004. The grammaticalization of ―KU‖ (故) and its use in Min dialects‘ [论―故‖的虚化及其在闽方言中的表现 (A discussion on grammaticalization of ―KU‖ in Southern Min)]. Taida Wen Shi Zhe Xuebao (台大文 史哲学报) 199-242. Yap, Foong-ha, Pik-ling Choi, and Kam-siu Cheung. 2007. Delexicalizing di: How a Chinese noun evolves into a nominlizer, with attitude. Department of Linguistics and Modern Languages, Chinese Uinversity of Hong Kong. Manuscript. Yap, Foong Ha, and Shoichi Iwasaki. 2003. From causative to passive: A passage in some East and Southeast Asian languages. Cognitive Linguistics and Non-Indo- European Languages (Cognitive Linguistics Research 18), ed. by Eugene Casad and Gary Palmer, 419-46. Berlin: Mouton de Gruyter Press. Yap, Foong-ha, and Stephen Matthews. 2008. The development of nominalizers in East Asia and Tibeto-Burman languages. Rethinking Grammaticalization—new perspectives, ed. by Lopez-Couso and Seoane, 309-41. Philadelphia: John Benjamins Press. Yap, Foong-ha, Stephen Matthews, and Kaoru Horie. 2004. From pronominalizer to pragmatic marker: Implications for unidirectionality from a crosslinguistic perspective. Up and down the cline: The nature of grammaticalization (Typological Studies in Language 59), ed. by Olga Fischer, Muriel Norde and Harry Perridon, 137-68. Amsterdam and Philadelphia: John Benjamins Press. 247 Ye, Youwen (叶友文). 1988. Sui Tang chuzhishi neizai yuanyuan fenxi [隋唐处置式内 在渊源分析 (The origin of the disposal construction in Tang and Song dynasties)]. Journal of Chinese Linguistics 16.1.55-71. Yeh, Meng. 1998. On 还 hai in Mandarin. Journal of Chinese Linguistics 26.2.236-80. Yuan, Jiahua (袁家骅) (ed.) 1964. Han Yu Fang Yan Ci Hui [汉語方言詞汇 (Lexicon of Chinese Dialects)]. Beijing: Wenzigaige Press. Yuan, Yulin (袁毓林). 2003. Cong jiaodian lilun kan juuwei ‗de‘ de jufa yuyi gongneng (从焦点理论看句尾―的‖的句法语义功能) [On the syntactic and semantic function of ‗de‘ in the sentence-final position: from a viewpoint of the modern ‗focus theory‘]. Zhongguo Yuwen (中国语文) 292.1:3-16. Yue-Hashimoto, Anne O. 1986. Tonal flip-flop in Chinese dialects. Journal of Chinese Linguistics. 14.2.161-82. Zhang, Binglin (章炳麟). 1915. Xin Fang Yan (新方言). Zhang Shi Cong Shu (章氏丛书). Zhang, Chengcai (张成材). 2007. Shangzhou fangyan kouyu zhong de shuliang lianyin bianhua [商州方言口语中的数量连音变化 (Sound changes in Shangzhou dialects)]. Zhongguo Yuwen (中国语文) 317.2.159. Zhou, Fagao (周法高) (ed.) 1973. Han Zi Gu Jin Yin Hui [汉字古今音汇 (A Pronouncing Dictionary of Chinese Characters in Archaic & Ancient Chinese, Mandarin & Cantonese)]. Hong Kong: Chinese University of Hong Kong Press. Zhu, Minche (祝敏彻). 1957. Lun chuqi chuzhishi [论初期处置式 (A discussion of the disposal construction in the early stage)]. Yuyanxue luncong (语言学论丛) 1.17-33. Beijing: Commerical Press. Zhu, Minche (祝敏彻). 1982. Zhuzi yulei zhong ―di‖, ―di‖ de yufazzuoyong [朱子语类中 ―地‖―底‖的语法作用 (The function of di4 and di3 in Zhuzi Yulei)]. Zhongguo Yuwen (中国语文) 3.193-97. Ziegeler, Debra. 2000. A possession-based analysis of thr ba-construction in Mandarin Chinese. Lingua 111.807-42.