THE INFLUENCE OF EXPERTISE AND CONTEXT 
ON THE CATEGORIZATION OF MUSIC: 
A COGNITIVE SCIENCE STUDY 
by 
JAMES S. IMHOFF 
A DISSERTATION 
Presented to the School of Music 
and the Graduate School of the University of Oregon 
in partial fulfillment of the requirements 
for the degree of 
Doctor of Musical Arts 
March 1995 
ii 
"The Influence of Expertise and Context on the 
Categorization of Music: A Cognitive Science Study," a 
dissertation prepared by James s. Imhoff in partial 
fulfillment of the requirements for the Doctor of Musical 
Arts degree in the School of Music. This dissertation has 
been approved by and accepted by: 
---- -------------
Dr. David P . Doerksen, Chair of the Examining Committee 
---
Committee in charge: Dr. David P. Doerksen, Chair 
Dr. Richard Clark 
Dr. Randall Moore 
Dr . Marian Smith 
Dr. Tom Givon 
Graduate School 
iii 
An Abstract of the Dissertation of 
James S. Imhoff for the degree of Doctor of Musical Arts 
in the School of Music to be taken March 1995 
Title: THE INFLUENCE OF EXPERTISE AND CONTEXT ON THE 
CATEGORIZATION OF MUSIC : A COGNITIVE SCIENCE STUDY 
Approved: 
Dr. David P. Doerksen, Chair 
Recent developments in Cognitive Science have 
demonstrated that, contrary to traditional thinking, 
categories are not rigid, feature-defined phenomena. Rather , 
they are influenced by human experience and by the context in 
which the categorization takes place. The labels people use 
to describe music reflect the way they categorize it. In 
this study, 32 music experts and 32 novices labeled short 
selections of recorded music . In each group, 16 subjects 
heard all Western Art selections (Context 1) , and 16 heard a 
mixture of Rock, Blues, Jazz , and Western Art music (Context 
2). All subjects used style terms (Classical, Renaissance, 
Baroque) as labels significantly more often than genre, 
instrument, or national origin. The results indicate that 
experts used more specific labels than novices, but context 
did not have a significant influence on the kinds of labels 
iv 
used by either group. This implies that musical categories 
are more stable than suggested by the current Cognitive 
Science literature. 
V 
CURRICULUM VITA 
NAME OF AUTHOR: James S. Imhoff 
PLACE OF BIRTH: Riverside, New Jersey 
DATE OF BIRTH: January 10, 1947 
GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED: 
University of Oregon 
Western Oregon State College 
University of Delaware 
DEGREES AWARDED: 
Doctor of Musical Arts, 1995, University of Oregon 
Master of Music Education, 1989, Western Oregon 
State College 
Master of Education in Special Education, 1972, 
University of Delaware 
Bachelor of Arts in Music Education, 1969, 
University of Delaware 
PROFESSIONAL EXPERIENCE: 
Assistant Professor, Music Education and 
Choral Conducting, Crane School of Music, 
SUNY Potsdam, 1993-
Graduate Teaching Fellow, School of Music, 
University of Oregon, Eugene, 1989-1993 
Choral Director and Music Teacher, 
North Albany Middle School, 
Albany , Oregon, 1979-1989 
vi 
Special Education Teacher, New Castle County School 
District, Wilmington, Delaware, 1972-1979 
High School and Middle School Choral Director, Tower 
Hill School, Wilmington, Delaware, 1970-1971 
Elementary Music Teacher, Hart Elementary, 
Warminster, Pennsylvania, 1969-1970 
Teaching Fellowship, University of Oregon, 
Eugene, 1989-1993 
Federal Fellowship Grant in Special Education, 
University of Delaware, Newark, 1970-1971 
vii 
ACKNOWLEDGMENTS 
The author expresses sincere appreciation to the 
Committee members and the many people who offered support 
in pursuing this research and in preparing this document. 
David Doerksen, as chair, was a source of constant 
encouragement, inspiration, and practical editorial 
advice. Richard Clark provided a questioning mind and the 
curiosity of a musician not immersed in cognitive science 
literature but nevertheless insisting on clarity and 
accuracy . Marian Smith kept the study focused on genuine 
musical questions and challenged me to rethink any 
intellectual or esthetic shortcuts I had made for the 
sake of experimental design. 
Tom Givon helped me focus on the conceptual basis 
for this study, and kept me from drifting away from this 
base . We spent all too few hours fiddling and picking 
old time music, and I still have the dollar tossed by a 
satisfied listener as we played in front of the Fall 
Creek Bakery. 
Randy Moore, although he joined the committee late in 
the process, was of great help throughout on questions of 
design and analysis and provided me with the facilities to 
viii 
run the study. Two people were not available to serve on 
the final committee, but were vital to the early stages of 
the study. Randi L'Hommedieu established rigorous 
scholarl y standards for writing and searching the 
literature. Marjorie Taylor, Professor of Psychology, 
applied Occam's Razor to my muddled ideas and supplied the 
elegance and clarity of experimental design. 
This document embodies the culmination of a dream 
that began many years ago. It was inspired and nurtured 
by two teachers who will always have a tremendous 
influence on my musical endeavors. Joseph Huszti and 
Calvin Bourgeault started me down a path--a tao--that was 
much more than a career or a profession. I did not so 
much follow it as I kept stumbling back on to it. The 
highest t ribute I can pay them is to pass along to the 
next generation the inspiration they gave me. 
Any pursuit such as this takes place over many years, 
mingled with the distractions, joys, and pains of everyday 
life. I have been blessed during those years to have the 
company and support of my wi fe, Faith. Through her, I 
learned of so many things that were every bit as important 
as my work, and every bit as much fun. 
ix 
TABLE OF CONTENTS 
Chapter Page 
I. INTRODUCTION. • 1 
Statement of the problem • • • • • 4 
Justification of the Study. • • • • • • • 7 
Theoretical Assumptions .... . .19 
Definition of Terms . . . . .24 
Design and Methodology . . ... . .29 
Analysis • • . . . . . . . . . . • .33 
Conclusions ..... .35 
II . LITERATURE REVIEW .37 
The Classical View of Categories . .••....3 8 
Cognitive Categories: A Modern Alternative •.• 40 
Expertise and Categorization. • . • ....5 4 
Current Theoretical Issues •...••.•... 66 
Categorization by Musical Style . • • . . .72 
Pilot Study . . . . . . . . . . . . • . . . . . 7 4 
Arguments Against Musical Categories. . .75 
Musical Style and Cognitive Research. . . .76 
Contributions of the Present Study .......8 5 
Summary . . . . . . . . . . . . . 8 9 
III. METHODOLOGY . . . . . . . • . . . . . 91 
The Pilot Study . . . . • .•••... 91 
The Larger Study: Expertis e and Context 92 
IV. RESULTS AND ANALYS IS . 108 
Research Questions and Results .. 108 
Conclusions . . . . . ... . . 117 
V. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS •... 120 
Summary . . . . .. 120 
Conclusions . 126 
Recommendations .. .134 
Final Philosophic al Considerations .138 
X 
APPENDIX Page 
A. RESPONSE LISTS ..• • 141 
B. RECORDED EXAMPLES • 154 
C. SUBJECT RESPONSE FORMS. • 161 
BIBLIOGRAPHY .•.. • 167 
xi 
LIST OF TABLES 
Table Page 
1 . Analysis of Variance of Mean Use 
of Style Terms in Different Contexts 
(Pilot Study) ••...... 92 
2. Summary of Label Response Types 109 
3. Frequency of "Classical" as a Style Term. • 111 
4. Analysis of Variance of Mean Use 
of Basic Level Terms • . . • • . . . . • 114 
5. Analysis of Variance of Mean Use 
of Shared/Equivalent Terms . ..•.. .•. 116 
xii 
LIST OF FIGURES 
Figure Page 
1. Comparison of Label Response Types . 110 
2. Comparison of Basic Level Term Use 
by Group and Context ..• . 113 
3. No Interaction Between Group and Context ... 114 
4. Comparison of Shared/Equivalent Term Use 
by Group and Context . . . ..• .. 116 
5. No Interaction between Group and Context .117 
1 
CHAPTER I 
INTRODUCTION 
What kind of music did Mozart write? 
This question , many would argue, cannot be answered 
in a few words. Words cannot capture so broad a sweep of 
creative genius , and so rich and diverse a body of 
brilliant music. The answer to this question must embrace 
many different forms and genre , s uch as operas, 
symphonies, concer tos, piano sonatas, church music, and 
chamber music . It must also embrace the composer's output 
at various stages of his development, from piano pieces 
written as a child to the mature and elaborate operas . 
Surely, no single word or short phrase can represent all 
of this great music . 
Music teachers , however , ask and answer just such 
questions every day. While listening, performing, and 
responding to music, teachers and students must at some 
point stop the music and talk. Too much talk, of course , 
detracts from the essentially musical experience; but this 
only heightens the need for clarity and efficiency of 
language. If teachers must assign labels to music, and if 
their students are to learn these labels and associa t e 
2 
them with the appropriate music, then everyone involved 
will benefit from an understanding of what labels convey. 
In spite of any arguments to the contrary, such 
questions are asked and answered all the time . Friends 
discussing favorite recordings, radio stations promoting 
their specialized format, and graduate level musicology 
seminars all use verbal labels to describe music. There 
is a tacit understanding that, with one or two words, one 
individual can communicate to another information that 
consists of instruments, lyrics, rhythms, melodies, 
harmonies, and numerous other musical details. 
So it appears that the question can be answered . 
Most people, however , would agree that it can be answered 
many different ways. Some individuals might answer the 
question in general terms , such as "Classical" or 
"symphonic," some more specific, such as "18th Century , 
First Viennese School." Others might reflect personal 
preference, such as "beautiful" or "boring." The answer, 
in fact, can be used as an indicator of knowledge about 
Mozart. The person who answers "singspiel" probably knows 
more than the person who answers "Classical." The kind of 
answer a person gives to such a question reflects his or 
her experiences with the music of that composer. It 
reflects , at least to some degree, that person's level of 
3 
expertise. The teacher cannot expect the student to adopt 
spontaneously the same vocabular y without at least some 
experience, training, and discourse. 
Furthermore, the circumstances in which the question 
arises might make a considerable difference in the answers 
given. One answer might seem appropriate in a college 
music appreciation course for non-musicians. Another 
might be expected in an advanced graduate course on the 
Viennese Classicist School. Still another kind of 
response might come out of a casual discussion between 
friends arguing the relative merits of Mozart, Elvis, 
Coltrane, and B.B. King. Students talking to each other 
outside class will probably use different terms than those 
they use on a music test. 
It might seem that, with so many possible ways of 
answering this question, and so many potential influences 
on the answer, every response will be different. In this 
extreme scenario, there would be as many responses given 
as there were examples played and persons asked. A far 
more realistic scenario would have at least some people 
using the same label (or the same kind of label) some of 
t he time. This study is concerned with such patterns, and 
with the kinds of answers or labels people will use most 
often. 
4 
Statement of the Problem 
In practical terms, the research problem here 
concerns communication between music teachers (experts) 
and music students (novices). Such communication requires 
a vocabulary of terms and labels, and a repertory of 
examples. Suppose a teacher wants the class to remember, 
understand, or appreciate some aspect of Baroque music. 
He or she might expect the class to recognize the term 
Baroque (as opposed to the generic uclassical"), to 
associate it with certain musical characteristics, and to 
perceive these characteristics in a recorded example. 
Recent cognitive studies in categorization have 
demonstrated that this set of expectations is more complex 
than commonly supposed. For example, students and 
teachers might use the same label, such as Baroque or 
Classical, but with very different meanings. They might 
listen to the same music but perceive very differ ent 
characteristics. Students might even recite the 
supposedly correct list of characteristics--that i s, t he 
definition--but use a very different set o f 
characterist ics when they actually listen to the mus i c . 
Furthermore, there is evidence to suggest t hat 
individuals might label or categorize a gi ven i tem 
di fferent ways i n different circumstances. The context i n 
5 
which the item is presented can have an impact on t he way 
a person describes it. A listener might label a passage 
from a Bach organ fugue "Organ music" if he or she hears 
it in the midst of a cappella choral music, but label the 
same piece "Classical" or "Baroque" in the midst of Rock, 
Blues, and Jazz recordings. 
The complexity of this situation and the 
possibilities for miscommunication call for a careful look 
at how category labels are used by different people in 
different settings. A number of studies investigate these 
issues, and they have established the theoretical and 
methodological approaches employed here. Specifically, 
this study examines the kinds of labels experts and 
novices commonly use when they describe music, and whether 
they use the same kinds of labels in different settings. 
Upper division and graduate music majors represent 
the expert population. Upper division and graduate non-
music majors represent the novice population. Subjects 
listened to a tape of 40 musical examples , each 20 seconds 
long. During or immediately after each passage, they 
responded in writing to the question "What kind of music 
is this?" There were two different settings or tapes in 
which subjects heard the music. One was an "all 
Classical" setting (Context 1 or CTXTl), the other 
included Rock, Blues, and Jazz music (Context 2 or CTXT2). 
6 
The labels used by subjects were examined for several 
different characteristics including the kind of labels 
(style, genre, national origin), how specific or generic 
the labels were, and how often subjects used the same 
labels. 
Research Questions 
Research Question #lA 
Will subjects use style terms (STs) more often than 
genre, performing media, national origin, or other kinds 
of labels? 
Research Question #lB 
Will Novices (representing the general public) use 
the generic, basic level term "Classical" significantly 
more often than other style terms? 
Research Question #2A 
Will Novices use the generic , basic level term (BLT) 
"Classical" significantl y more often than Experts? 
7 
Research Question #2B 
Will musical context influence the use of basic level 
terms? That is, will any difference between the two 
groups be greater with one tape than with the other? 
Research Question #3A 
Will Experts use the same label--a shared or 
equivalent term (SET)-- for a given example significantly 
more often than Novices? 
Research Question #3B 
Will musical context influence the use of shared 
equivalent terms? That is , will any difference between 
the two groups be greater with one tape than with t he 
other? 
Justification of the Study 
Music educators at all levels present examples of 
music to their s t udents, point out salient features of the 
music, and use appropriate labels. Students are expected 
to associate t hese labels with the music they hear and 
with similar examples they might hear later . "Much of our 
ability to perceive mus ic and the commonaliti es of musical 
8 
situations is the result of categorizing and storing away 
what we hear (Edwards, 1988, p. 129). 
When a student consistently uses the correct label 
for an example, the teacher probably assumes that the 
student understands the definition, the important 
characteristics, and some subtle aspects of the music. By 
the same token, if the student persists in using a label 
that the teacher considers incorrect, the assumption is 
that such understanding is lacking. 
This process of applying categorical labels, 
according to recent developments in cognitive science, is 
more complex than traditionally assumed. It is subject to 
a number of influences including the different levels of 
expertise of the teacher and the students, the context in 
which the examples are presented, and the categorical 
associations already learned from previous experience. 
This study empirically examines this process, and the 
results will offer useful information for music educators 
who are engaged in the process on a daily basis. 
Cognitive Studies and Music 
Cognitive science has pushed back the boundaries of 
knowledge regarding human perception, thought, and 
learning. Unfortunately, only a limited portion of this 
exciting research has been applied to music education. 
9 
The growing body of music cognition literature is based on 
a rather reductionist approach, where the musical stimuli 
are usually very short patterns of computer generated 
tones (Dowling & Harwood, 1986; Deutsch, 1982). The 
perception of isolated pitches, the grouping of a few 
notes in a rhythmic context, and the expectancy of the 
next chord are dominant themes in music cognition 
research. It is generally assumed that higher order 
concepts such as style and genre cannot be controlled in 
the scientific manner in which cognition is usually 
studied. 
When a broad concept such as style is considered, the 
goal is often to reduce it to its defining 
characteristics. Cutietta (1993), for instance, argues 
that styles are differentiated by motion, energy, flow , 
fabric, and color . Eastlund (1992) uses multidimensional 
scaling to extract "historical period, complexity, and 
tempo" (p. 19) as features that differentiate styles. 
There is evidence, however, that the cognitive process of 
categorization is more complex than this implies . 
The way people label everyday objects (Rosch & Lloyd, 
1978), the way they decide that two cases are similar or 
different (Medin & Wattenmaker, 1989) , and the way they 
organize such information in memory (Anderson, 1990) have 
been the focus of numerous studies. The research proposed 
10 
here will apply to music the principles and practices 
established by categorization research . The musical 
examples will be brief excerpts, but they will be actual 
recordings of real compositions, not computer-generated 
tone patterns. 
Contributions of this Study 
This dissertation will provide useful information for 
the development of new approaches to the teaching of 
musical concepts. In the past, educators have adopted a 
paradigm promoted by Bruner (1973) known as the Concept 
Formation Model. This model makes certain assumptions 
that have not been supported by recent research. These 
assumptions include a view of categories as discrete 
entities, clearly defined by specific features; learning 
the category simply involves identifying and recognizing 
the features (Merrill & Tennyson, 1977). 
Cognitive sci ence has given us a more complex view of 
category structure (Gardner, 1985). A category, in this 
new view, is a flexible, dynamic construct that is subject 
to influences such as prior knowledge, intuitive judgments 
of similarity, and context. Furthermore, when a large 
number of people are asked to categorize natural items 
(trees, furniture) patterns emerge from their responses. 
Some category labels are used more often and more readily 
11 
than others, and some members of a category are more 
central and typical than others. 
For example, Rosch (1978} found that people would 
most frequently respond with "robin" when shown a picture 
of that bird; they were less likely to say "bird," 
"animal," or "red-breasted robin." In addition, when 
these various labels were flashed on a screen with the 
appropriate picture, subjects responded more slowly to the 
very specific label (red-breasted) and to the very general 
label (animal). They responded most rapidly to what Rosch 
called the "basic level" term, in this case "robin." 
Tanaka and Taylor (1991} found that these response 
patterns were different for bird experts, who responded 
rapidly to the more specific labels. A bird expert, for 
example, could press the "yes" button as rapidly for 
sparrow as for bird. A novice took slightly longer to 
respond to the more specific label. The present study 
will apply some of the theory and methodology from this 
literature to the categorization of music. 
Unique Features of the Study 
In addition to the justifications above, this study 
uses several elements from the literature in new ways. 
There are a number of studies that examine expert/novice 
differences, categorization in different contexts, and 
12 
musical styles as categories. The present study, however, 
pulls these elements together in a new light and employs 
methodologies not previously applied to music. 
Musical Style Categories as Cognitive Percepts 
Cognitive literature regarding music is, for the most 
part, concerned with very localized events such as pitch 
perception and rhythmic groupings (Serafine, 1988; Dowling 
& Harwood, 1986; Howell, Cross & West, 1985). Less work 
has been done with more global issues such as style. 
Dowling and Harwood briefly mention style, but describe it 
as "invariants across sets of pieces" (1986, p. 160). 
This certainly sounds like rigid, feature-based 
categorization. 
A few studies have considered style as a variable 
(Gardner, 1973; Eastlund, 1990; Brittin, 1991), but these 
studies again assume discrete categories: A given piece 
represents a given style for both researcher and subject. 
There do not appear to be any studies that treat 
perception of style as a cognitive categorization 
phenomenon. Gardner ' s 1973 study constitutes a 
breakthrough in its admission that style is too complex to 
identify in terms of specific features. Even so, it does 
not refer to any of the categor y literature that was 
available even at that time. 
13 
Cutietta (1993) points out that categorization is the 
most fundamental process in music perception. He argues 
against the reductionist approach of teaching separate 
elements of music (rhythm, melody, harmony). Rather, 
teachers should be aware of the rapid and intuitive 
categorization taking place as their students are 
listening to examples or thinking about music. Cutietta 
writes that junior high students spontaneously categorize 
music as opera, church, classical, and rock and roll. 
Such an approach allows the subjects to express their own 
category schemas rather than respond to those of the 
researcher. In this sense, Cutietta's position is 
consistent with the assumptions of this study. It is an 
important breakthrough in the study of musical 
categorization in that it establishes style as a percept, 
rather than as a composite of separate elements. 
Cutietta's research questions, however, are quite 
different from those of the present study. Having 
rejected melody, rhythm, and harmony as defining 
attributes, he replaces them with his own proposed 
defining attributes: motion, energy, flow, fabric, and 
color. In his study, he asks subjects to rate music on 
these features, in effect priming them to attend to these 
features. It must be noted that there was great 
consistency in student ratings of the proposed attributes, 
14 
and they might indeed prove to be the basis for much of 
the musical categorization that takes place. They are 
almost certainly an improvement over the traditional 
abstractions of melody, harmony, and rhythm. 
The present study uses an open-ended task, and 
subjects will be allowed to respond with their own 
spontaneous, intuitive labels. That is, they will be free 
to form their own categories and use their own bases for 
labeling the musical examples, rather than respond to 
those of the researcher. At no point will one feature be 
cued or primed over another, and subjects may attend to 
those which they consider salient. The purpose here is 
not to find a more realistic basis for categorization, but 
to observe the influence of expertise and context on 
certain aspects of categorization. 
Free Response Format 
In the present study, each subject chooses his or her 
own basis for labeling the music. In a related pilot 
study, for example, graduate music students used style 
periods, performing groups, composer names, genres, forms, 
and other bases to label musical examples. In the present 
study, the responses of music experts are compared to 
those of the general public, but the quest ion is s t ill 
open ended: "What kind of music is this? " The responses 
15 
not only reveal knowledge and vocabulary, but also aspects 
of the music on which attention is focused. A response 
such as "piano music" implies a different focus from a 
response of "cheerful" or "early Chopin." 
At no time, however, is there an attempt to extract a 
specific feature or list of features that might influence 
such response patterns. This makes the study 
fundamentally different from Cutietta (1993) and Eastlund 
(1992), where the focus was on defining features. Here, 
the focus is on whether the label used is a style term, 
how generic or specific it is, and how often individuals 
within a group use the same label. 
Expertise and Categorization 
The influence of two factors in particular, context 
and expertise, will be examined in this process. Numerous 
studies (Tanaka & Taylor, 1991; Murphy & Wright, 1984) 
point to differences in the way experts and novices 
categorize material within a given field. These authors 
suggest that experts categorize at more specific levels. 
Barsalou (1989) suggests that categorization will vary 
from one context to the next, but Brooks (1989) argues 
that some expertise will lead to more consistent results. 
This study presents recorded musical examples to 
experts and to members of the general population. The 
16 
hypotheses predict that a frequently used basis for their 
response will be some aspect of style. Furthermore, the 
hypotheses suggest that music experts will categorize 
examples from their area of expertise at a more specific 
level and with greater consistency, even in changing 
contexts. 
Many of the previous expert/novice categorization 
studies used laboratory controlled artificial categories, 
such as color and shape sets (Murphy & Wright, 1984). So-
called expertise was established by allowing some subjects 
to practice the procedure more than others. Murphy and 
Wright argue that real-world expertise carries with it not 
only practice, but theoretical information about the 
domain of expertise. They used actual clinical 
psychologists at various levels of experience in their 
study. Tanaka and Taylor (1991), similarly, drew dog and 
bird experts from members of local dog and bird watching 
organizations. Expertise was established not by short 
term artificial practice, but by years of experience and 
by the personal recommendations of other members. 
Growth in musical expertise is associated with 
lengthy study, practice, and in many cases with 
specialization. Nevertheless, it is easy to find a group 
of people that share some expertise in one field of 
musical style. Most college music programs emphasize 
17 
Western Art traditions fostered in Europe from the fall of 
the Roman Empire to the twentieth century. While 
different schools might offer programs in Jazz studies, 
ethnomusicology, or even popular music, the common 
denominator is that body of literature generally referred 
to by the public as "Classical." 
Thus there is a group of students and faculty one 
would easily classify as experts in the field of Western 
art music. Members of the general public might very well 
be experts in one style or another, but this expertise is, 
presumably, randomly distributed. If the target examples 
are Western art music, then, advanced music students may 
clearly be considered experts. 
Context and Categorization 
Barsalou (1989) suggests that categories are subject 
to substantial influence from the context in which cases 
are presented: That is to say, the subjects will classify 
the same items quite differently from one time to the 
next. Given (1989) also argues strongly that all 
categorization takes place in some context, and is duly 
influenced. This implies that different contexts will 
yield different category schemes. Thus a piece by Bach 
might be given one label on a classical music radio 
station, but a very different label if used on an MTV 
18 
video. In fact, it is conceivable that a host of 
environmental and experiential factors will lead to 
different labels on different occasions. 
Brooks (1989), on the other hand, argues that with 
some practice, subjects might be more consistent. While 
he is referring to Barsalou's informal ad hoc category 
tasks (such as "ways to hide out from the mob," and 
"things to take out of the house in a fire"), his point 
may be extrapolated : With extensive practice, subjects 
could become quite consistent. Eventually, those subjects 
who have gained some expertise might, as a group, be very 
consistent, even under changing circumstances. 
A comparison of experts and novices would almost 
surely show differences in their categorization practices. 
But when the additional factor of changing context is 
introduced, these differences might change in degree and 
character. That is, the novice responses might change 
radically under radically different circumstances, while 
the expert responses, according to Brooks, might remain 
comparatively stable. 
In this study , musical context is defined as the 
range and variety of styles that surround the target 
examples . These targets consist of twelve Renaissance, 
Baroque, and Romantic selections from the Norton Anthology 
of Western Music (1988). In one context (CTXTl), they are 
19 
preceded and mixed in with examples from the same source 
and in similar styles. In the other context (CTXT2), they 
are preceded by and mixed in with Jazz, Rock, and Blues 
examples. 
Theoretical Assumptions 
This study requires several assumptions regarding 
categorization, some of which might be contrary to 
popularly held notions. Chief among these assumptions is 
the degree to which category labels are stable, unique to 
an example, and unique to a subject. At one extreme is 
the position that a given item will be categorized the 
same way by every subject; the rigid Platonic feature-
defined model of categories suggests something along these 
lines (Givon, 1989). At the other extreme, every 
individual will categorize a given item according to his 
or her own special ideas about the item. This would 
emphasize the influence of context (Barsalou, 1989) and 
individual intuitive theories (Medin & Wattenmaker, 1989). 
Neither extreme has been supported in the literature, 
and various issues have been raised regardi ng the rigidity 
versus the flexibility of categories. Medin and 
Wattenmaker (1989) question the basis for making judgments 
of similarity . Barsalou (1983) points out that some 
categories are traditional and learned , while others are 
20 
evoked for very specific purposes (such as "things to take 
out of the house in a fire"). Murphy and Wright (1984) 
examine differences between experienced professionals and 
beginners. 
There are consistent patterns in the ways people 
categorize everyday objects. For example, when shown a 
particular type of chair, such as a high-backed rocker or 
a captain's chair, most people call it a chair. They do 
not call it by its more specialized name, nor do they 
refer to it as a piece of furniture or a thing (Rosch, 
1978). Similarly, a Vermont sugar maple or Douglas fir 
are more likely to be called a tree than their more 
specific label or than the more generic label plant. 
Rosch called this most common and useful kind of label the 
basic level. Moreover, she found that subjects could 
respond to these labels more rapidly than to specific or 
generic labels. The basic level term (BLT) for musical 
style is one of the measures used in this study. The 
present study seeks to establish a BLT by observing which 
term--if any-- is used most frequently. 
Categorization and Music 
There is a limited body of research on the 
categorization of music. This literature, unfortunately, 
does not make full use of or reference to another even 
21 
greater body of research from the cognitive sciences. 
These new and sometimes surprising results must be 
reckoned with if musical research is to stay current and 
meaningful. They are an important factor in the 
theoretical assumptions of this study. 
The present study differs from previous musical 
category studies in several important ways. First, it 
uses the factors of expertise and context as variables, 
with the prediction of interaction between them. Second, 
it avoids the issue of defining features. The research 
questions have to do with the kinds of labels subjects 
use, not what specific features they attend to. The 
results might indeed suggest certain features, but to 
focus on them would run counter to the literature on which 
this study is based. 
Philosophical Considerations 
It must be stressed that this study is intended to 
provide useful i nformation for music teachers. While i t 
employs the terminology and methodology of cognitive 
science, it is not necessarily intended to break new 
ground in these fields. Rather, it should take the ground 
provided by nearl y three decades of empirical research and 
build upon it. 
22 
In this light, it is fitting to keep in mind that all 
of the questions considered above are still being asked. 
The cognitive literature is infected with the viruses of 
oversimplification and misunderstanding. For example, a 
number of reports state such misconceptions as "concepts 
are coded into memory as prototypes" (Jonassen, 1988, p. 
2). Rosch herself insisted that there was no such thing 
as a prototype; rather, one should speak of the 
"prototypicality" of some exemplars. 
Psychology and linguistics give us insights into the 
mind, and how the mind deals with information. But none 
of this information is direct or literal: Objective data 
can be little more than a basis for inference about actual 
mental process. The issue of links between the real world 
and our conception of it is the concern of such monumental 
philosophers as Plato (1991), Aristotle , (1991), Kant 
(1965), Wittgenstein (1958), and Lao Tse (1972). 
Verbal Knowledge and Musical Knowledge 
Even if the focus is concentrated on the meaning of 
such words as music, kind, and style, epistemological 
arguments arise that cannot be fully answered here. In 
fact, "Our facility as language users prevents us from 
recognizing the complexity of the representations and 
processes that underlie our mastery of word meanings . 
23 
There may therefore be some truth in the ... claim that 
a complete theory of meaning is impossible" (Garnham, 
1985, p. 132). In view of these seemingly unanswerable 
questions, can a study such as this be useful, or even 
justifiable? 
Three sources provide an affirmative answer. First, 
Booth and Cutietta (1991) suggest that memories about 
music are associated with verbal category labels. ''The 
results also suggest that, in subject's minds, these 
verbal stylistic labels may override musical attributes" 
(p. 130). In other words, subjects perceived two examples 
as being in the same category even though certain musical 
elements were quite different. 
Second, the First National Assessment of Musical 
Progress (1974) and the Second Assessment of Music (1980) 
use verbal responses--indeed, style category labels--as a 
measure of musical knowledge. In a format similar to that 
of the present study, the Second Assessment asks students 
"Are there any kinds of music that you like to listen to?" 
The range and specificity of the responses are an 
important indicator of musical knowledge. 
Third, renowned authorities in the field remind their 
colleagues of a fact so obvious it is often overlooked: 
"Apart from making music or listening to it, talking about 
it becomes the next best means for acquiring a musical 
24 
education. Talking about music in conjunction with making 
it or listening to it provides the most potentially 
powerful educative process" (Tait & Haack, 1984--Crane 
Symposium p. 54) Bennett Reimer puts it even more 
succinctly: "In order to teach music, you have to talk 
about it" (Reimer, 19 9 4 ) • 
Clearly, when teachers and students talk to each 
other (about music or any subject) they use words, and 
these words do have meanings, legion and ephemeral as they 
might be. The critical points for teachers are that 
labels do not necessarily represent a discretely defined 
class of music, and that understanding is not necessarily 
demonstrated by the use of a correct label. In fact, a 
teacher and a student using the same descriptive label 
might mean very different things: 
In principle, the choice of what to include and 
what to exclude in descriptions depends on one's 
pragmatic framing, i.e. on one's judgments of 
saliency, relevance, importance. None of the 
judgments are "objective"; nor can they be 
arrived at deductively or inductively; they are 
a matter of point of view; of context. (Givon , 
1989, p. 89) 
Definitions of Terms 
This study is an attempt to treat several broad 
concepts as scientific variables. These include musical 
style, expertise in music history, and musical context. 
25 
It is essential to clarify how these variables will be 
defined and measured. For example, if one subject labels 
an example "Classical," another labels it "piano music," 
and a third labels it "early 19th century chamber music," 
how will these answers be classified and tabulated? 
Style Terms 
The details of measurement and statistical analysis 
will be discussed in Chapters 3 and 4, but a brief 
explanation is in order to establish the terminology and 
direction of the study. "Classical," in the case above, 
would be classified as a style term (ST). It is an idiom 
that can be associated with a specific historical period 
(17th-19th centuries) within a particular geographical 
region (Europe). It is not confined to these, but is 
associated with them; it can be performed by different 
people on different instruments, and has its own 
collection of substyles. In this sense, Classical is a 
generic rather than specific style label . 
"Piano music , " on the other hand, is· not a style 
term, but a performing medium. Ludwig van Beethoven, 
Scott Joplin, and Jelly Roll Morton all wrote piano music 
in very different idioms or styles. Similarly, a choir 
(even a cappella) might sing gospel, doo-wop, or 16th 
century polyphony, so "choir music" would not be a style 
26 
term. References to instruments or voices are classified 
in this study as "performing media" terms. Other response 
types include national origin (French, German), and genre 
(opera, symphony, song). 
The third label, "early 19th century chamber music," 
refers to a far more specific style period, and specifies 
a genre (chamber) as well. This response, like 
"Classical," is a style term (ST), but suggests attention 
at a more detailed level. It draws finer distinctions 
between the example in question and another example. In 
the data analysis, these answers will be tabulated 
separately. 
For purposes of reliability and validity, the 
researcher and two faculty members with music history 
backgrounds independently examined the total response 
pool. Each term was judged as a Style Term, Genre, 
Performing Media, National Origin, or Other. 
Basic Level Terms 
It has been noted that examples might be labeled at 
different levels of specificity. A piece of furniture on 
which people sit is usually referred to as a "chair." In 
normal conversation , a person would never say "Sit over 
there on that piece of furniture," even though this higher 
level term would be correct English. On occasion , a 
27 
person might use a more specific label such as "armchair" 
or "high-backed rocker." The label used most easily and 
most often is referred to as the basic level term (BLT). 
Numerous studies have shown that people can answer yes or 
no to BLTs more rapidly than to more specific or more 
general labels (Rosch, 1978). 
This phenomenon, logically, should apply to music as 
well as other domains. For instance, one subject might 
call a particular passage "New Orleans ragtime" music 
while another calls it jazz. "New Orleans ragtime" is 
more specific than "Jazz;" it is indeed a subcategory of 
Jazz. Studies have shown that experts tend to use labels 
at more specific levels than novices (Tanaka & Taylor 
1991). Thus jazz might be considered a basic level term 
(BLT), whereas New Orleans ragtime is a more subordinate 
level term. 
The target examples for this study are from the 
Norton Anthology of Western Music, a collection intended 
for use in college music history and appreciation courses. 
They represent a broad spectrum of styles, genres, and 
performing media, but for the most part would probably be 
covered under the generic term Classical. Novice subjects 
are unlikely to have the same specialized vocabulary as 
the experts. They nevertheless have many options 
available, such as commonly us ed genre terms (opera, 
28 
symphony), identifying instruments, or guessing at the 
nationality. One of the hypotheses on which this study is 
based states that people will use style related terms more 
often than other kinds of labels, and that the term used 
most often by the novices will be "classical." 
Shared Equivalent Terms 
Finally, this study will examine category stability. 
Stability here refers to the extent to which subjects use 
the same label for a given example. Some of the 
literature suggests that the categories of experts will be 
more stable than those of the novices (Brooks, 1989). 
Murphy and Wright (1984) found that experts as a group 
used fewer category labels than novices . In other words, 
the experts used the same labels more often. The 
expectation here is that the experts as a group will agree 
with each other more often than novices, even when the 
context changes. 
To capture this effect, a count will be made of any 
terms used for a given musical example by more than one 
subject . These will be called shared equivalent terms 
(SETs) because some terms might be equivalent, such as 
"chorus" and "choir," or "religious music" and "sacred 
music." Such terms, while there might be subtle 
29 
distinctions, would be counted along with exact 
duplicates. 
Design and Methodology 
Subjects listened to a tape of forty excerpts, twenty 
seconds each, of varied musical selections. After each 
excerpt, they wrote the answer to the simple question: 
What kind of music is this? In studies by Rosch (1978) 
and Tanaka & Taylor, 1991, subjects were presented with 
pictures or visual cues and asked t o respond. The brief 
excerpts are the musical equivalent. The music will play 
for about 20 seconds; there will be a five-second period 
between examples during which the number of the next 
example is given. The pilot study demonstrated that this 
was enough time to listen, decide, and write, but still 
avoid a lengthy analysis and consideration of secondary or 
alternative labels. First impressions were desired; in 
fact, only the first word written was considered as data. 
Examples were presented in two different contexts. 
Cont ext 1 (CTXTl) was exclusively tracks from t he Norton 
Anthology of Western Music (NAWM) (1988), a CD collection 
of art music from antiquity to the twentieth century. 
CTXT2 was twelve of the same NAWM tracks--the t arget 
cases--mixed in with selections from rock, blues , and 
30 
jazz. The data consist of responses to the twelve 
examples that occur in both contexts. 
One part of the subject sample represents members of 
the general public; while any given individual might be 
quite knowledgeable about a particular favorite style or 
genre, this expertise should be randomly distributed 
throughout the population. In effect, this class of 
subjects can be considered novices. These subjects were 
upper division or graduate non-music majors at the 
University of Oregon. 
The other part of the sample includes advanced music 
students. They are considered experts in the field of 
serious music from the Western art tradition. Again, 
individuals might very well be experts on country music or 
jazz, but as a group the common knowledge base is most 
likely to be that which is part of their advanced study. 
Experimental Design 
Independent Variables 
The independent variables are expertise and context. 
Expertise has two levels, expert (advanced music student) 
and novice (non-music major or faculty). Context also has 
two levels. In context 1 (CTXTl) all of the examples are 
from the Norton Anthology CDs, and exemplify Renaissance, 
31 
Baroque, and Romantic styles. Classical (18th Century) 
style is not included because use of the generic label 
"Classical music" was predicted to serve as a basic level 
term (BLT). The intent was to avoid having to determine 
whether a subject was using the term in its generic 
("Classical music") sense, or in its specific ("18th 
Century Classicism") sense. 
Dependent Variables 
The dependent variable is the response pool of 768 
words; only the first word written by each subject in each 
cell was used. In one sense, the response pool is the one 
and only dependent variable. But the three research 
questions focus on different aspects of the response pool. 
Moreover, each of the three questions focuses on a 
different portion of the response pool. 
Research question 1 concerns all 768 words, 
regardless of group or context, asking whether a 
significant portion of them will be style terms. 
Question 2 concerns only those words deemed style terms, 
asking how specific or generic they are within each cell. 
Question 3 concerns all the words--style or otherwise--
used by members of each cell, asking how often members 
agreed with their peers. Thus it will be more convenient 
32 
to treat the three different concerns as three different 
variables. 
In this view, the dependent variables are the kinds 
of labels used by subjects to describe 12 target examples 
from the Norton Anthology of Western Music. These labels 
were first classified as style, genre, performing media, 
national origin, and other (based on the pilot study and 
on the actual response pool of this study). Question 1 
asks whether the style terms (STs) will predominate. 
Question 2 asks whether expertise and context will 
have an influence on the use of a generic label, the basic 
level term (BLT). Although this is a separate question 
from the first, it draws on question 1 in that a basic 
level term for style would appear as the style term most 
frequently used by the novices (because they represent the 
general population). If there are no effects, this BLT 
will appear just as often in the expert response pool, and 
in both contexts . 
Next, shared equivalent terms (SETs) were tallied 
within each of the four cells (experts and novices in both 
contexts). Question 3 asks whether experts will use more 
SETs than novices, and whether context will influence that 
difference . SETs were any term or its exact equivalent 
used by more than one subject to label a given example. 
Thus if two novices in CTXT2 called example number 3 
33 
"opera," it was considered a SET. If one subject used the 
term "choir" and another used the term "choral," it was 
also considered a SET because the two words have almost 
exactly the same meaning. 
Analysis 
Research questions lA and 1B are a matter of 
frequency count. Specifically, does the Novice group use 
one kind of label--Style Terms--significantly more often 
than other kinds (#lA); and, among these STs, is one term 
("Classical") used more than any other? A Chi Square 
tested for significance in these questions. 
Research questions 2 and 3, which concern the 
influences of expertise and context, call for a 2-by-2 
Analysis of Variance (ANOVA). First and foremost , ANOVA 
captures any interaction between expertise and context , 
and t hat is a central issue in this study. Second, it 
compares the variance within each group , and an important 
assumption is that t he experts will vary in their 
responses less than the novices. Finally, ANOVA compares 
group means; in this case, those will be the mean number 
of times each subject in each cell gives uses a BLT or an 
SET. 
34 
Variables and Measurement 
Style Terms (STs) are any labels or responses that 
specify a historical period, such as the Middle Ages, 16th 
Century, or Late Baroque. Although the target cases are 
all Norton Anthology Renaissance, Baroque, and Romantic 
examples, if a subject uses a stylistic idiom label such 
as Folk or Jazz, these would also be considered STs. 
The Basic Level Term (BLT) for style represents a 
generic as opposed to a more _specific term. For example, 
record stores typically have a section labeled 
"Classical," which might inc lude music from the 
Renaissance to the 20th Century. Classical would be a 
generic or basic level term for all these style period 
idioms. It should be noted here that a considerable body 
of research exists on the question of category levels. 
This research is examined in Chapter 2 . 
Shared or Equivalent Terms (SETs) are any labels used 
by different subjects for the same example. This would 
include obvious duplicates and equivalent terms such as 
16th Century and Renaissance, or chorus and choir. A 
panel of three experts judged the equivalence of any terms 
that were not obvious dupl icates. 
Context (CTXT ) refers to the setting in which 
categorization takes place. Half of the subjects will 
35 
hear an all-Norton tape; that is , they along with 12 
target examples t hey will hear other Renaissance, Baroque, 
and Romantic music (CTXTl ). The other half wi ll hear the 
same Norton target examples, but mixed in with example s of 
Rock, Blues, and Jazz (CTXT2 ). Studies will be examined 
that suggest the different contexts will l ead to different 
kinds of labels being used. 
Expertise refers to the group to which a subject 
belongs; there are two such groups: Experts are upper 
division and graduate music students, novices are 
equivalent non-majors. Expertise and context are nominal 
variables, both with two levels (CTXTl , CTXT2 ; expert , 
novice). This calls for a Two-way ANOVA, with possible 
interaction between the two factors. 
Conclusions 
Musical categories have characteristics similar to 
those in other fields. They are not the rigid, exclusive 
structures often found in textbooks and on tests. They 
are flexible and dynamic, but wit hin the constra ints o f 
actual features and c onsistent j udgments of similari ty. 
Barsalou (1989 ) argues t hat category structures depend 
very heavily on context, almost to the point of total 
i nstabil ity . Other research (Brooks, 1989; Diekho f f , 
1985 ) suggest s t hat practic e , understanding, and expert i se 
36 
will lend stability to category structures. This study 
measures the effects of context on categorizing mus ical 
styles, and it measures this influence at different levels 
of expertise. 
What kind of music did Mozart write? 
The answer can be found in textbooks at various 
levels. A teacher can tell a student the answer to this 
question, and demand it back on a test. But the accepted 
label might not mean the same thing to both people, and 
they might actually perceive different characteristics in 
the music. Categorization always occurs in some context, 
and the same context might create different cognitive 
references for the teacher and student. If we give 
serious consideration to this phenomenon, our teaching 
will more effective, and the music we play will be better 
understood. 
37 
CHAPTER II 
LITERATURE REVIEW 
The literature germane to this study falls under 
three general headings. The first consists of research 
into categorization as a cognitive and linguistic process. 
This body of l i terature grew out of early studies 
concerned with the way people assign labels, classify 
objects, and perceive similarities and differences. It 
established a view of categories fundamentally different 
from that established by Plato. 
The second consists of expert-novice studies in 
categorization. A number of studies have focused on the 
way naive beginners and skilled specialists categorize 
items in a given domain. These studies have focused on 
how fast subjects respond, how specific their labels are, 
and how they treat atypical examples. 
The third consists of studies that use music as the 
domain in which categories are formed. These will be 
examined for insights and results that might have an 
impact on the present study , but, for the most part, they 
will be critiqued in terms of the evidence from the two 
38 
previous sections. A superficial survey of music 
categorization literature revealed some deficiencies in 
connection with cognitive studies in other domains. In 
general, the music studies did not draw on the cognitive 
material, and the design and methodology was sometimes 
inconsistent with that of cognitive research and its 
findings. 
The Classical View of Categories 
Over two thousand years ago, Plato (Ed. and trans. 
1991) established a view of categories that still greatly 
influences category thinking today. A category was 
defined by a set of characteristics: All of the members 
had those characteristics, and any item that did not was a 
non-member . 
In Phaedo, Plato (speaking as Socrates) points out 
that tallness and shortness are mutually exclusive; "nor 
can any other opposite ... simultaneously become or be 
its own opposite" (1991, p.188). He draws similar 
conclusions regarding hot and cold, even and odd. While 
this might appear a simple explanation of the meaning of 
opposites, it sets up a rather rigid model of categories 
in that a given object in one category cannot be in the 
other . It further implies that a person labeled as tall 
39 
cannot be also labeled short; this person must be in one 
category or the other . 
Plato also discusses the concept of a characteristic 
that runs through a category (such as "large things") 
giving it definition. "When some plurality of things 
seems to you to be large, there perhaps seems to be . 
one characteristic that is the same when you look over 
them all" (1991, p. 264). This has been taken by thinkers 
though the ages to be the very basis for the existence of 
a category: one or more defining characteristics shared by 
all the members. "Both rationalists and empiricists, from 
Plato and Aristotle down, have subscribed to the view that 
mental categories are discrete and absolute" (Givon, 1989, 
p. 36) • 
Modern Applications of the Classical View 
This established an all-or-none, rule-based approach 
that still appears in educational literature (Merrill & 
Tennyson , 1977) and in concept formation studies (Bruner, 
1973). It assumes that categories function as little more 
than a bundle of specific and identifiable features. 
In this view, a given musical example will have 
certain discernible attributes that are perceived by the 
listener . When listeners are asked what ki nd of music a 
particular exampl e is, their response would be based on 
40 
those characteristics as well as their own musical 
knowledge. If the response involves style (as this study 
hypothesizes it will), then the style label used will be a 
function of such attributes as well. A review of 
literature will give evidence that the situation is not so 
clear-cut, and that musical style, along with other 
category labels, is not so easily defined. 
Cognitive Categories: A Modern Alternative 
Modern research in category theory began with Eleanor 
Rosch. She was the first to vigorously challenge the 
Platonic model of discrete, feature-defined categories. 
According to Gardner, she is "possibly the cognitivist 
whose critique did most to undermine the classical view" 
(1985, p. 342). Her work and that of her colleagues 
established t he issues, methodologies, and theoretical 
basis for nearly all subsequent category research. Rosch 
(1973) pointed out that in the laboratory, researchers 
could arbitrarily manipulate features. Large blue circles 
were just as probable as small red squares, and a subject 
entered with little notion about what to expect. In the 
real world, however, one did not often encounter animals 
that had wings, feathers, and three legs. Real world 
categories were not arbitrary like those of Bruner : Wings 
41 
and feathers were always matched with two legs. Outside 
the laboratory, people formed categories that were rich in 
inference and based on fairly consistent prior experience: 
They had never seen an animal with feathers and three 
legs; they knew from personal experience that some 
features were more likely to appear in conjunction with 
others. 
Thus, with Rosch, began a new body of category 
literature which focused on the conceptions and subjective 
experiences of the observer rather than objective features 
of the observed. Attention shifted away from objects in 
the environment and toward the judgments and experience of 
the observer. Category features were not abandoned, but 
put in the context of human cognition. 
Category Levels 
Rosch found patterns in the response times of people 
to computerized categorization tasks . When a general cue 
(tree, chair) was flashed on a screen, it was easier for 
subjects to decide whether the exemplar that followed was 
in or out (YES/NO) of a category. If the cue was too 
specific (black locust tree , captain's chair) it took 
longer for the YES-NO decision. Moreover, when subjects 
were shown common objects such as plants, animals, and 
furniture, they would most often respond at a certain 
42 
level. Rarely would the more general term (animal, piece 
of furniture) be given; usually a more specific response 
(dog, table) was given. 
The basic level was that at which most subjects 
responded: "maple tree" for example, rather than "Norway 
maple" or "plant." This level elicited the fastest 
response times (RT) in a high-speed decision task. It was 
also the term that children learned first (Rosch, Mervis, 
Gray, Johnson, & Boyes-Braem, 1976). The authors pointed 
to the cognitive economy of this phenomenon: The basic 
level term was most useful in immediately discriminating 
and classifying the environment. In terms of recognizing 
objects that had similar properties, it was more useful to 
say "a chair" than "a piece of furniture." 
Rosch and her associates (1976) recognized that 
categories reflected real-world structures. They did not 
immediately reject the objective feature lists posited by 
Plato, but they did suggest that human cognition was also 
an important factor. Category levels were not just an 
effect of an objective environment, but of human thought 
and behavior. 
Because of their hierarchical nature, these levels 
were referred to as the vertical dimension in category 
structure. Rosch called broad, general labels (plant; 
animal) superordinate; very specific labels (black locust, 
43 
red breasted robin) she called subordinate; and the more 
central, accessible, and frequently used labels she called 
basic (1978). "Object names at the basic level of 
abstraction should be the names by which objects are most 
generally designated by adult speakers of the language" 
(1976, p. 422). 
Categorization of Natural Objects 
Rosch and her associates, in 12 separate experiments, 
established methods that became the mainstay of category 
research, such as feature lists and response time. The 
first 4 experiments served to support the concept of basic 
level objects in a linguistic sense. It examined the 
words people used most often to describe or label familiar 
objects. The second set of 8 explored the cognitive, 
perceptual, and linguistic implications. Here, subjects 
formed images, matched pictures, and named objects. All 
subsequent category literature refers to this study, so it 
deserves some scrutiny here. 
In experiment 1, subjects were given an object name 
and asked to list attributes of that object. The objects 
used came from taxonomies such as "musical instrument, 
guitar, folk/classical guitar;" and "tool, hammer, ball-
peen/claw hammer" (Rosch, 1976, p . 388). The attributes 
listed were then judged for trut h by a separate set of 
44 
subjects as a reliabi lity check. The number of attribut es 
listed at the basic level was higher than at the 
superordinate (t = 6.43, p < .001). Additional 
attributes were listed at the more specific subordinate 
level, as might be expected: Having described an apple, a 
subject would need additional terms to describe a 
particular kind of apple. But the size of increase at 
this level was not as great as the increase going from 
superordinate to basic (t = 6.43, p < .001). In other 
words, more new descriptive terms were introduced at the 
basic level than at other levels in the vertical 
structure. 
The basic level was thus established as the one at 
which most information was immediately available . The 
categories were general objects such as fruits and tools, 
and the subjects were students in psychology seminars. 
There was no cons i deration of special knowledge, only 
general experience and common language. The subjects were 
more or less equal , and expertise was not an issue in 
these early studies . Experiment s 2, 3, and 4 involved 
similar procedures wit h physical activities, and outli ne s 
of common shapes. I n each case, there was a basic leve l 
at which the maxi mum amount of information seemed to be 
availa bl e. 
45 
The second part of the study explored the 
psychological effects of category levels. For example, 
subjects were asked to visualize an i mage of furniture. 
It was found that t hey would do so by imaging a specific , 
basic level exemplar such as a chair. There seemed to be 
no image that represented the higher order category term. 
Another experiment had subjects deciding if two pictures 
were the same, similar, or different. Before seeing the 
pictures, some subjects were given a superordinate level 
cue (tool, vegetable) and others were given a basic level 
cue (hammer, carrot). RT with a basic level cue was 
faster (p < .05). 
Still another experiment compared a sorting and 
classifying task by preschoolers, elementary school 
students, fifth graders, and adults. The basic level 
effect was dramatic . "At all age levels, basic l evel 
sorts were virtually perfect" ( 1976 , p. 417 ). 
Basic Level as Frequency 
In Rosch et al. experiment number 10, a picture 
naming task , "subjects overwhelmingly used the basic level 
name in this free-naming situation" (p. 423). This 
supported Rosch's assumption (and that of the present 
study) that "Object names at the basic level of 
abstraction should be the names by which objects are most 
46 
generally designated by adult speakers of the language" 
(p. 422). 
Again and again, it appeared that category level and 
vertical structure was a given constraint on human 
information processing. Rosch et al. summarized the 
results of their landmark study: 
The categorizations that humans make of the 
concrete world are not arbitrary , but rather are 
highly determined. They are determined in the 
first place, because the perceived world is not 
an unstructured total set of ..• attributes .• 
. categories are determined, in the second place, 
because, in so far as categorization occurs to 
reduce the infinite differences between stimuli 
to . . • useful proportions, the basic category 
cuts should be those which yield the most 
information for the least cognitive load. (1976, 
p. 428) 
It might appear circuitous to say that subjects used 
the BLT most often, and then define the BLT as the term 
used most often. The present study, however, differs from 
those of Rosch in two ways. First, Rosch also tested for 
the basic level by response time (RT). This methodology 
calls for measurements in milliseconds and is 
inappropriate for someone listening to several seconds of 
mus ic. The frequent use of these terms was an additional 
aspect she noted . Second, the present study predicts that 
the BLT will be a particular label: "Classical." That is, 
it predicts that this very label will be used most 
frequently by the novices. 
47 
Category Boundaries and Linear Separability 
As pointed out above, Plato {1991) established a 
model of category structure that included strict 
boundaries between the categories. This boundary was 
determined by the presence or absence of critical, 
defining features. Eleanor Rosch (1978) challenged this 
conceptualization, arguing that categories had a 
horizontal as well as a vertical structure. 
It creates some intuitive problems to state that 
there are no boundaries to a category. The term category 
is used regularly to describe a sharply defined group of 
ideas, objects , or people. Research designs, for example, 
call for mutual exclusivity in their nominal categories. 
Children learn their first words as mutually exclusive 
category names {Markman, 1989). A young child is puzzled 
when a "doll" is referred to as a "toy." Medin and 
Wattenmaker {1989) refer to this quality as linear 
separability, because it implies that a line can be drawn 
to separate one category from another. 
Contrary to this notion of discrete divisions, Rosch 
and her associates (1976) found a graded structure in the 
categories they investigated. The graded structure of 
categories refers to the phenomenon of some members being 
more typical of a category t han others. A robin was found 
48 
to be a prime example or prototype of the category birds. 
Penguins , c hickens, and ostriches were certainly birds, 
but subj ects rated them as less typical. These atypical 
exemplars occupied a peripheral position in t he category, 
while robins, sparrows, and c anari es occupied a 
prototypical or central position. 
Thus the horizontal dimension of a category resembles 
a bell-shaped curve, with maximum typicality in the 
center, and diminishing typicality as one moves away from 
the center. Eventually, some of the more atypical 
exemplars in one category became potential exemplars in a 
related category. In other words, there is no linear 
separability, but a gradual shift from one category to the 
next. 
Rosch and her colleagues (1976) suggested that only 
some categories are continuous while others are discrete. 
But s ubsequent studies have found graded effects in many 
different kinds of categories. Bar salou (1983) found them 
in what he termed ad hoc goal driven categories such as 
"things to take out of the house in a fire" and "ways to 
avoid getting killed by the Mafia." Subjects compared 
such options as "c hange your identity and move to the 
mountains of South America" and "become a drunk in 
Detroit " (p. 215 ). Barsalou used a Kendall's coefficient 
to test reliability among different subjects ranking the 
49 
choices in typicality. He found an average agreement of 
.87, indicating that subjects agreed on their ratings of 
the different options. Even these ad hoc categories had 
graded characteristics. 
The Graded Structure of Categories 
Armstrong, Gleitman, and Gleitman (1983) found that 
people rate the number 9 as more typically odd than 259, 
even though such a comparison contradicts the definitional 
basis for the category of odd numbers. The graded 
structure of categories seems to be an effect of human 
cognition rather than a real world phenomenon . As Plato 
would have pointed out, all odd numbers share a common 
trait (indivisibility by 2) and thus all are equally odd. 
In the perception of humans, however, some numbers seem to 
be more odd than others, or at least better examples of 
oddness. 
Graded structures do not mandate overlapping 
categories. Odd numbers are clearly linearly separable 
from even numbers, and there can be no member in both 
categories, no matter how atypical it might be judged . 
But if a number is rated as "less odd," there is some 
implication that it might be at the outer edge of the 
category, making it at least a little closer to the other 
category, "even." In this light, 259 might be construed 
50 
as "more even" than the protot ypically odd 9. The t wo 
categories, though discrete by definition , seem to have a 
somewhat graduated boundary. 
Murphy and Wright (1984) found actual overlap in 
category features used by clinical psychologist s. For 
example, children classified as "depressed" shared the 
feature "feels sad" with children classified as 
"aggressive" (p. 152). In a sense, a child exhibiting 
shared features might be marginally categorized as either 
depressed or aggressive, depending on other factors. The 
line separating the two becomes blurred. 
Barsalou characterized t he graded structure as a 
"continuum of category representativeness, beginning with 
the most typical members of a category and continuing 
t hrough its atypical members to those nonmembers least 
similar to category members" (1989, p . 102). He goes on 
to describe category continuity as an effect of long term 
memory storage of concepts. "Knowledge in long-term 
memory from which concepts are constructed for a 
particular category may generally not have clear 
boundaries. Instead, knowledge for a particular category 
may s hare much structure wit h knowledge for other 
categories" (p. 121 ). 
While subjects are able to classify i tems in 
different categories, the imaginary lines between those 
51 
categories become blurred when the memories of many 
individuals are called into action. Barsalou points out 
that this is an observed, behavioral phenomenon that 
appears when people actively categorize things. He denies 
any claim of it being a "cognitive s tructure ," and 
attributes it to an effect of the memory base of different 
individuals. Categories might have linear separability, 
he suggests, but human memory does not . 
Prototypes 
Prototypes are a somewhat controversial concept in 
cognitive research. Rosch, for example, insisted that a 
prototype was not a "mental trace" or a single exemplar 
(1978). She suggested that it was better t o write of the 
"prototypicality" of some exemplars. Hintzman (1986) 
argues that prototypes do not exist in memory, but are 
formed in recall. Other writers, ignoring Rosch, went so 
far as to claim that concepts were stored in memory as 
prototypes (Jonassen , 1988). 
Nevertheless , prototypes are a useful construct when 
considering graded and continuous categories . Whether 
they are convenient abstractions or real-world prime 
examples , they represent a central tendency in each 
category. That is , they might be thought of as a somewhat 
52 
"ill defined center" that complements the "ill defined 
boundaries." 
Clearly, there are some problems with the issue of 
discrete versus continuous categories. Lakoff (1989) 
agrees that to a mathematician, odd numbers are a 
discrete, separable category with clear limits. But the 
study that showed prototypical effects in odd numbers 
(Armstrong, Gleitman, & Gleitman, 1983) did not 
distinguish between numbers as mathematical entities and 
numbers as a linguistic phenomenon. That is, to most 
people, numbers are words that stand for the mathematical 
entities. As such, they are subject to the same 
linguistic effects as Rosch's basic objects and natural 
categories. These effects include fuzzy boundaries and 
continuity; they do not include linear separability. It 
remains to be seen if recorded musical excerpts are 
subject to the same effects. 
Hintzman (1986) has pointed out that a schema-like 
effect or prototype can be produced by a system using 
nothing but discrete memory traces of prior cases. His 
MINERVA II artificial intelligence model stores memories 
of specialized shapes; the computer can then produce a 
composite image that appears to be an "averaged" prototype 
of the whole set of shapes. The prototype is never 
53 
actually stored in memory, according to Hintzman, but is a 
secondary effect of retrieval. 
This effect , however, does nothing to discredit 
Rosch. It must be remembered that, in her later writings, 
she cautioned against thinking of prototypes as actual 
images or neurological traces (1978). Some real-world 
cases seem to exhibit a higher degree of prototypicality 
than others, but it is wrong, according to Rosch, to speak 
of any single case as an actual prototype. 
How Humans Categorize 
Hintzman's MINERVA II draws on a memory base of all 
available cases. A less structured human equivalent might 
be the use of a few specific familiar cases as a guide to 
classifying new exemplars. Rather than comparing the new 
case to a cognitive model, people sometimes simply compare 
it with a well-known exemplar. This model, as proposed by 
Brooks (1989), is somewhat closer to the classical 
category view, where new cases are compared to a feature 
list. The difference, however, is that no actual list 
need be articulated, and no specific features need be 
consciously identified. 
54 
Expertise and Categorization 
In their classic study, Rosch and her associates 
(1976) encountered one subject, an airplane mechanic, who 
produced a lengthy list of attributes for the category 
airplane. They characterized the effects of his expertise 
as the ability to "make use of attributes that are ignored 
by the average person" (p. 430). 
There were, however, other, more subtle differences, 
between the airplane expert's categories and those of the 
other subjects. For example, most subjects used a top and 
side view when imagining parts of an airplane, but the 
mechanic used the underside and engine. He was 
nevertheless able, they wrote, to "take the role of the 
average person and list attributes common t o all 
airplanes, and could imagine an average airplane shape 
from the outside" (p. 430). He could, depending on the 
context , consciously function as a novice or as an expert. 
This might be analogous to a teacher (expert) discussing 
music with a music class (novices ). 
This study, it should be remembered , represented a 
turning point in category theory. Rather t han Platonic 
feature lists, Rosch and her associates argued that "the 
basic category cuts should be t hose which yield the most 
information for the least cogniti ve l oad" (1976, p . 428). 
55 
It could be argued that, with greater in-depth 
understanding of a subject, it is easier to make those 
cuts at more specific levels. In other words, an expert 
would automatically label or categorize the same items at 
a more specific level than a novice. With a change in 
context, however, the expert might revert to the basic 
level as did Rosch's airplane mechanic. 
Expert-Novice Categorization Research 
Expertise is indeed a well-researched topic in 
cognitive science, but is usually associated with 
decision-making and problem-solving, not categorization 
(Anderson, 1990). While there is a wealth of empirical 
literature on categorization, only a few studies consider 
expertise as a factor. Those few do suggest noticeable 
differences in categorization by experts when contrasted 
with non-experts or novices. 
Chi , Feltovich , and Glaser (1981) found that experts 
in physics approached problems differently from novice 
students. This was a problem-solving study , but 
categorization of the problems was stressed. When asked 
to organize the various problems into related categories, 
experts put most problems in three large inclusive 
categories. Novices, on the other hand, had only one 
large cat egory and many smaller, more exclusive 
56 
categories. This might imply that the expert categories 
were wider in breadth and more inclusive . Novices 
apparently saw each problem as unique . 
Furthermore, novices used superficial, literal 
details as a basis for categorization, such as the 
presence of an inclined plane or pulley. Experts, in 
contrast, used general principles such as Newton's Third 
Law as a basis for classifying. Attending to the surface 
information, novices stressed the differences between 
problems rather than the underlying similarities. Experts 
saw more similarities overall, and therefore generated 
fewer categories, each of which included more cases. 
This, incidentally, should lead to more agreement among 
experts, more disagreement among novices. 
The authors interviewed their subjects and drew 
protocols of their procedures. That is, they wrote out 
step-by-step statements that followed the thoughts of the 
subjects. Analysis of the protocols suggested a 
structured knowledge set or schema on which the subjects 
drew to approach the problem. 
Both experts and novices included the general 
principle of conservation of energy, but at different 
levels . When the researchers graphically diagrammed the 
schemas, conservation of energy appears in the lower 
middle section of the novice diagram, and at the very top 
57 
of the expert diagram. The novices paid initial attention 
to specific details, while the experts gave initial 
attention to overall principles. 
Schemas and Expertise 
Schemas!, according to Anderson (1990) "facilitate 
making inferences about the concepts. If we know 
something is a house, we can use the schema definition to 
infer that it is probably made of wood or brick, and that 
it has walls, windows, and the like" (p. 135). A schema 
is somewhat like the default settings on a computer, and 
like these settings, can be easily changed if a situation 
calls for different settings. Chi, Feltovich, and Glaser 
(1981) found that physics experts were able to construct 
useful, accurate schemas from previous experience and 
knowledge, while the novices went on a case by case basis. 
It was also found that the experts included a possible 
solution in their schemas, while the novices did not. 
In contrast , different results were found in a study 
that compared expert probation officers to a non-expert 
panel (Lurigio, 1983). In this study, subjects classified 
probation cases from a description of each case . As in 
Chi, Feltovich, and Glaser (1981) the schemas used by each 
subject were compared , but in this case, experts generated 
1 Sometimes pluralized as schemata 
58 
more category schemas than novices (t [38] = 2 . 08 , p < 
.05). While this is inconsistent with the earlier 
studies, other aspects were consistent. Novice schemas 
were "simple and impoverished" (p. 139) while experts used 
a rich, interrelated ne twork of traits and strategies. 
The experts were described as "not just stringing together 
facts' (Lurigio , 1983 , p. 139). Like the physics experts, 
they seemed to use information beyond that which was 
given. 
In a card sorting task comparing expert and novice 
probation officers, Lurigio's experts had a clear 
conceptualization before placing cards. That is, they 
appeared to follow a plan that included groups of cases, 
while the novices seemed to follow a case-by-case pattern. 
In fact, experts took more time to sort the cards, once 
again in conflict with the earlier studies (Chi, Feltovich 
and Glaser , 1981; Rosch et al., 1976). Lurigio 
interpreted this to mean that the experts were invoking 
complex schemas rather than making isolated decisions. 
These response times, it should be noted, were measured in 
minutes , and they involved a conscious, elaborate 
procedure. They are not to be confused with the high 
speed YES-NO RTs of Rosch and her associates. 
59 
Category Levels and Expertise 
Rosch and her associates (1976) found what they 
termed a vertical dimension in their category structures. 
At the core of this structure was the basic level, for 
example "tree" as opposed to a specific variety or a more 
general taxonomic "plant." This phenomenon was more than 
just an average, popular name that most subjects used. It 
was the level of cue that led to the fastest RTs, the 
label that generated the most descriptive terms or 
features, and the level at which children and adults were 
best able to sort cards representing different familiar 
categories. 
It appeared that the basic level was not just a 
superficial effect among familiar objects . Rosch and her 
associates suggested that it might reflect actual 
structures in the environment as well as a convenient and 
economic aspect of cognition. 
"Basic objects for an individual, subculture, or 
culture must result from an interaction between 
the potential structure provided by the world and 
the particular emphases and state of knowledge of 
the people who are doing the categorizing. 
However, the environment places constraints on 
categorizations. Human knowledge cannot provide 
correlational structure where there is none. 
Humans can only ignore or exaggerate 
correlational structures." (1976, p.430) 
60 
Tanaka and Taylor (1991) argue that structure in the 
environment does not necessarily determine the basic 
level, but that human perception, interaction, and culture 
have a substantial influence. They point to children of 
the Tzeltal Mayan tribe, who learn the equivalent of our 
subordinate (more specific) labels early in life. These 
subordinate labels help the children discriminate plants 
as important sources of food. Only later do they learn 
the so-called basic level. In effect, their basic level 
has been shifted downward for adaptive purposes. 
Tanaka and Taylor (1991) replicated some of the 
methods used in Rosch et al. (1976) but with two distinct 
groups of subjects. One group consisted of expert dog 
handlers and members of a local American Kennel Club 
chapter. The other group consisted of experienced bird 
watchers. The pictures and cues used in the 
categorization response time task were dogs, birds, and a 
few other assorted filler items. Thus the dog experts 
served as a non-expert control group for bird cases, and 
the bird experts served as novices in the dog cases. 
Since the present study will adopt similar methodology, a 
close examination of Tanaka and Taylor is in order. 
Their method used a computer program similar to Rosch 
et al. A row of plus signs appeared on screen to focus 
attention; after one second this was replaced by a 
61 
category label. Two and a half seconds later, a picture 
appeared on a projection screen. If the picture and label 
matched, the subjects pressed true; if not, they pressed 
false. Subjects used the index finger of their dominant 
hand and were instructed to respond as fast as possible. 
They were given a two-minute rest midway through the 
experiment. 
There were a total of 128 trials: eight dogs, eight 
birds, each shown (on a slide projector)l6 times. Each 
slide was shown as a TRUE case one time at each level 
(superordinate, subordinate, basic) and once as a FALSE 
case at each level; also, 32 fillers (trees & rocks) were 
used to prevent automatic FALSE responses for non-dog and 
non-bird cases. Plant was used as a superordinate level 
FALSE case. That is, the subjects saw a plant cue and 
then a dog; this match was false, but at the superordinate 
level (plant, animal) rather than the basic level. There 
were plants, so subjects had to pay attention and consider 
the possibility of a TRUE case. Half of the fillers were 
TRUE, half FALSE. 
Tanaka and Taylor analyzed their results using a 3 x 
2 ANOVA with category level (superordinate, basic, 
subordinate) and knowledge domain (expert, non-expert) as 
independent variables. Interaction between category level 
and knowledge domain on TRUE trials was sign i ficant: 
62 
F(2,46) = 12.43, p < .001. For FALSE, results were also 
significant: F (2,46) = 14.53; p < .001. As predicted in 
their hypothesis, experts were as fast to categorize at 
the subordinate level as at the basic level. In non-expert 
domains (that is, dog experts categorizing birds and vice 
versa) RTs were faster at basic level. 
Experts were faster to respond FALSE at the 
subordinate level. For example, bird experts were faster 
when given a "sparrow" cue and robin picture than when 
given a "dog" cue and robin picture. But outside of 
their domains of expertise, RT at the subordinate level 
was slower. Novices, that is, took longer to decide on 
more specific labels. 
Four general differences were found between expert 
and novice categorization performance. In the expertise 
domain, subordinate level categories were as 
differentiated as basic level categories; that is, they 
were as richly described and as rapidly identified. 
Experts used subordinate level names as frequently as 
basic level names to identify objects. A dog expert, for 
example , was as likely to call a Doberman pinscher by its 
specific breed as to call it a dog . Expert RTs at the 
subordinate level were as fast as at the basic level. 
And, finally , "subordinate word primes produced greater 
facilitation in a physical matc hing task than basic level 
63 
word primes" (p. iv). In other words, confirming that two 
pictures matched was easier when a subordinate prime 
(rather than a basic level prime) was given before the two 
pictures. 
Category Boundaries and Expertise 
From the very beginning, Rosch argued against 
discrete Platonic categories. But if there were not 
strict boundaries between one category and the next, then 
how was one category recognizable from another? This 
study will not answer that question in any ultimate terms, 
but it might shed some light on how these elusive 
boundaries are effected by in-depth knowledge of a 
specialized field. 
The Tanaka and Taylor study (1991) is concerned 
almost exclusively with levels of categorization and 
expertise; boundaries are not discussed in any detail. In 
fact, Tanaka and Taylor specifically used exemplars that 
were central to the category, exemplars that the experts 
would recognize with no confusion. Category boundaries 
and atypical exemplars were avoided. 
Murphy and Wright (1984), on the other hand, were 
quite concerned with the breadth of each group's 
categories. They wanted to see if the outer limits of the 
category were different for the more experienced clinical 
64 
psychologists when compared to the trainees. Their test 
for this was the extent to which subjects used the same 
descriptors for two or more categories. Their categories 
included depressed, disorganized, and aggressive children, 
and their descriptive labels included such traits as 
"throws tantrums," "feels angry," and "feels sad." 
A descriptor used in two different categories implied 
some overlap. Novices associated descriptors with one 
category, and therefore, not with another. Experts, in 
contrast, used the same term to describe cases in several 
categories. With experts, children classified as 
depressed were described as "feeling sad" (p. 152) . But 
the same description was used for aggressive children . A 
child exhibiting this particular trait, therefore, might 
be classified as an instance of both categories. For 
experts, at least, the categories might be said to 
overlap . 
Murphy and Wright (1984) had four groups ranging from 
novice through experienced student to expert. The greater 
part of their results, however , were significant (p < .OS) 
only when the expert group was compared with the others . 
No significant differences (p > .10) were found between 
any two of the lower level pre-expert groups. There did 
not seem to be evidence of a gradual accumulation of 
65 
category changes, but an all-or-none shift at the highest 
level of expertise. 
More important to this study were Murphy and Wright 's 
results regarding within group agreement. In listing 
descriptive terms, the experts agreed with each other 41% 
of the time, while novices agreed with each other 22% of 
the time. In other words, given a clinical category, 
experts used shared equivalent terms to describe the 
category more often than novices, F(3, 284) = 30.9, p < 
.001. According to Murphy and Wright, uThere is clear 
evidence that ••. the level of interrater agreement 
increases with expertise" (p. 147). This leads to the 
expectation that music experts might agree more often than 
novices when applying descriptive labels to music. 
Categorization Based on Intuitive Theories 
In order to make a judgment of similarity, an 
individual must decide which properties to ignore . This 
decis ion , according to Medin and Wattenmaker (1989), is 
made on t he basis of the individual's ideas about the 
nature of the category and about the relationships between 
various properties . These theories can be purely 
intuitive , naive, heuristic, even wrong. In any case, 
they are based on the individual's experience with members 
of the category . 
66 
The implications for the present study are clear. 
The novices will have a variety of intuitive theories 
about the musical category labels they generate based on 
divergent musical taste and experiences. The experts, on 
the other hand, will have at least somewhat more 
consistent theories based on their advanced study. "The 
need to share conceptual knowledge in communication. 
places constraints on concepts ••. People attempt to 
reach a common ground or consensus by tacitly agreeing on 
certain common values or dimensions for organizing the 
concepts involved" (Keil, 1989, p. 193). 
Current Theoretical Issues 
Medin and Barsalou (1990) argue that similarity to a 
prototype or to a known exemplar is not a satisfactory 
explanation of category structure. An element of salience 
has to be introduced: Some features of an object are 
ignored while others are deemed more important. Plums 
and lawnmowers, they point out, have certain qualities in 
common (both weigh less than a ton, both were not found on 
earth 100,000,000 years ago), but these qualities do not 
seem to constitute similarity. The unanswered question is 
what separates the important features from the 
unimportant? Once again, a simplistic feature-defined 
model is not sufficient. 
67 
Stylistic similarities between Mozart and Haydn might 
be quite obvious to the musical scholar, as are 
differences between these 18th Century Viennese 
Classicists and Vivaldi, an Italian Baroque figure. But 
to the novice, Vivaldi, Mozart, and Haydn might fit very 
neatly into a category of composers who used orchestra, 
wore powdered wigs, and whose music is played in concert 
halls. Can the music expert merely dismiss these common 
features as unimportant? 
Developmental Influences on Categorization 
Keil (1989a) points out that preschoolers' notions of 
a clearly defined category such as "uncle" are not those 
found in the dictionary. His young subjects, for example, 
stated that an uncle was an adult friend who was handsome 
and brought presents. Uncles could not be seven years old 
and could not be ugly. Only later do children conceive of 
an uncle as the brother of a parent. The situation, Keil 
seems to argue, is not merely a matter of accurate 
information, but a developmental change in category 
structure. 
Keil (1989b) describes this maturation of category 
formation as a characteristic-to-defining shift, where the 
child first identifies an item based on salient 
characteristics (doggie: shaggy tail, four legs, friendly) 
68 
but later learns a more formal definition for what 
constitutes a dog, and what differentiates it from a cat. 
Social function has an impact in Keil's view: As the 
child grows and must communicate with others regarding 
different concepts, there is a tendency to use a more 
analytic definition. The attention, however, is still on 
the defining features of a category; the ability to 
perceive, understand, and communicate these features is 
the developmental issue. 
Neisser (1989) argues that this characteristic-to-
defining shift is also found in adults learning about new 
categories: "something similar occurs when we move from 
ignorance to expertise in any new domain. Novices have no 
choice but to judge by appearance, sticking as close to 
the basic level as they can" (p. 20). This once again 
implies significant differences between categorical 
responses by experts and novices. It clearly suggests, 
for example , that novices will use basic level terms more 
often than experts. 
Stability of Expert Categorization 
Barsalou (1989) argues against stability in category 
structures. Context is so important that the same 
subjects will respond differently on different occasions. 
Readers of the word "frogs" might not associate i t with 
69 
the category "things eaten by humans;" they are more 
likely to do so, however , if it is read in the context of 
"French restaurant." Context can have a dramatic affect 
on categorization. This suggests that neither experts nor 
novices would show reliable patterns in classifying 
musical examples under different contexts. 
Brooks ( 1989) finds some fault with Barsalou's 
methodology in justifying unstable category structures. 
He argues that Barsalou's ad hoc studies (things to take 
out in a fire, ways to avoid the Mob) do not give the 
subject a chance to practice classifying the items. 
Subjects are presented with an unusual and somewhat 
artificial task, not at all like real-life categorization 
procedures . Given time to theorize and organize, writes 
Brooks, the subjects might develop some stability in their 
response patterns. Groups of subjects given such an 
opportunity might develop within-group consistency. 
Brooks does not use the term expertise, but it 
appears that he offers it as a foil to Barsalou's changing 
contexts. In fact, Barsalou admits that "subjects from 
the same population correlate around .50 with each other. 
although agreement is generally low, it exists and 
must be accounted for" (p. 123). Thus some consistency is 
found using general knowledge, general populations, and 
somewhat bizarre ad hoc categories . Might not even more 
70 
be found in the highly-organized categories of a music 
history expert? 
Indeed, studies using knowledge organization as a 
test of understanding (Diekhoff, 1983) indicated that 
expert instructors were more consistent and reliable in 
their judgments of similarity than the novice students. 
Diekhoff suggests using these judgments of similarity as a 
basis for evaluation and grading. He found a high 
correlation between the judgments of his top students and 
a panel of experts. 
Murphy and Wright (1984), in comparing category 
description among clinical psychology professionals, found 
clear expert/novice differences. The experts frequently 
used the same or similar terms to describe the category, 
while novices did not. With regard to the issues taken 
between Barsalou and Brooks, Murphy and Wright seem to 
support substantial category stability among experts, but 
not among novices . Again, this suggests music experts 
will agree more often than novices. 
Areas of Disagreement 
Homa, Rhoads, and Chambliss (1979) found that experts 
had tighter category clusters than novices. That is, 
members of a category were rated as more similar to one 
another and as more different from members of other 
71 
categories. This contradicts Murphy and Wright ( 1984) , 
who found broader, more inclusive, and less distinctive 
c ategories with experts . The l atter authors explain tha t 
the results are reconcilable because the methodologies 
were different; but further investigation is needed to 
settle the issue. 
Subjects in Homa et al. (1979) rated the similarity 
of objects within categories, while those in Murphy and 
Wright used feature lists. The earlier study involved 
classification skill, and the focus was on 
distinctiveness , while the more recent study used a more 
open-ended procedure. It is easier to use a descriptive 
label in more than one category, if one feels so inclined, 
t han it is to classify an object in t wo different 
categories. 
In the study by Lurigio ( 198 3 ), expert probation 
officers took more time than novices to categorize 
criminal suspects. Initially, t his seems to contradict 
the hypothesis o f faster RTs for experts . Lurigio 
suggests that t he time delay was caused when experts 
called up rich schemas from memory, whereas novices made 
case-by-case decisions based on relatively short feature 
lists . Furthe rmore, in this study sub jects used a 
conscious procedure and protocols, not immediate r esponse 
72 
patterns. The times in Lurigio were measured in minutes, 
not milliseconds. 
Categorization by Musical Style 
The first thesis in this study is that the preferred 
domain of response to a recorded musical example will 
involve style. That is to say, subjects will call the 
various examples names such as "Jazz " , "Rock, " or 
"Classical" more often than "vocal, " "piano," or 
"cheerful . " The pilot study gave some evidence to support 
this position, but most of the evidence is anecdotal and 
informal. The same kind of evidence supports the second 
thesis , that music experts will use more specific terms in 
their basic level response, such as "Renaissance," 
"Baroque," and "Romantic." 
This evidence, as discussed in Chapter I, includes 
the following informal observations: (a) When asked what 
kind of music they like, people usually answer with a 
style term such as country, rock, or classical; (b) 
record stores commonly organize their shelves under 
similar style labels; (c) radio stations announce their 
specialized programming in style terms; and (d) music 
history courses and texts are organized around such style 
terminology . 
73 
In addition, support for the categorization o f music 
by style comes from Booth and Cutietta (1991). Subjects 
in this study listened to a tape of 12 songs which, 
according to the authors, fell into four style categories: 
Christmas songs, rock songs, musical show tunes, and 
children's songs. After two subsequent hearings, 48 hours 
apart, subjects tried to recall the titles. Some subjects 
recalled titles in the approximate order they originally 
heard them (serial recall), others seemed to group titles 
by style category, and a third group seemed to use no 
particular strategy. Of the three groups, those using a 
categorical strategy had the highest recall. As with the 
present study, the researchers did not give the subjects 
any cues for the categories; categorization, they 
maintain, was spontaneous . 
Further evidence for spontaneous categorization by 
styl e comes f r om "subjects who, when they could not 
remember a specif ic song, wrote responses such as 'another 
Christmas song' or 'another Broadway song' (p. 130). In 
their conclusions, Booth and Cutietta called for "mor e 
research ... t hat uses free recall as a tool t o explore 
more thoroughly t he relationship between verbal labels and 
musical recall, as well as the cognitive processes 
i nvolved with both" (p. 130). 
74 
The present study, while it uses labeling rather than 
recall, is an attempt to provide such research. It 
follows the path of categorization by the subjects (rather 
than the researcher) , and it draws on methodology normally 
associated with verbal rather than musical information. 
It is distinctly different, however, in its focus on 
category levels (generic versus specific), on expert-
novice differences, and on the influence of musical 
context. 
Pilot Study 
In preparation for the expert/novice study, a pilot 
study was conducted in the summer of 1992 at the 
University of Oregon. In this study, subjects were all 
experts--that is, they were all graduate and upper 
division music students. As in the larger study, these 
subjects listened to brief excerpts of recorded music: 
half all Western art music, and half mixed with Rock, 
Blues and Jazz. The priming question was "What style of 
music is this?" 
In the pilot, context had a significant effect on the 
kinds of labels used. That is, subjects who listened to a 
narrow context--all "Classical"--used style terms 36.25% 
of the time , while those who listened to an extended 
context--Rock , Blues, Jazz, and Classical--used STs 63.25% 
75 
of the time. This difference proved significant (Fl, 14; 
p = .01). 
The results clearly supported context as a source of 
variation in the kinds of labels used by subjects in the 
pilot study. These results, together wi th the work of 
Barsalou (1989), and of Booth and Cutietta (1991), are the 
basis for the hypothesized context effects in the larger 
study. 
Surprisingly, even though the question was "What 
style of music is this?" subjects answered a total of 
50.25% of the time with labels that were not, by the 
definition adopted for this study, style terms at all. 
Such labels included genre (opera, lieder), performing 
media (orchestra, choir), and national origin (Italian, 
German)2. Since the larger study gave no such cue for 
style (uwhat kind of music is this?") any preference for 
STs might well be attribut ed t o t he l ist ener rather than 
the study design. 
Arguments Agai nst Musical Categories 
Arguments are often heard that mus i c cannot be 
categorized. Grout (1973), a standard text in many 
2 Some subjects also tried t o identify (that is, to name) 
the work or the composer. This option occurred so 
infrequently i n the main study that it i s i ncluded i n the 
"other" response category. 
76 
college music history courses, states that "Historical 
labels are a matter of convenience, and most of them are 
necessarily inaccurate" (p. 159) . Robinson and Winold 
(1976), in a text designed for choral conducting students, 
adopt a similar view: "We have been stressing those 
aspects which make one period of choral literature 
different from other periods. It is, however, one of the 
sublime glories and mysteries of music that it ultimately 
resists assignment to any closed chronological or 
geographical category" (p. 498). 
If one assumes the rigid Platonic category model, as 
these statements seem to, then categorization is indeed 
difficult if not impossible. But if the more intuitive 
category model is used, a strong case can be made for 
using generally accepted style terms. If categorization 
is such an automated process, the listener might be 
labeling the music from the first few notes. That indeed 
is the central phenomenon of this study. 
Musical Style and Cognitive Research 
While there appears to be little experimental 
literature treating musical style in terms of cognitive 
categorization, there are a number of studies that do use 
it as a variable . Many of these studies treat style as a 
nominal, discrete variable, defined in terms of historical 
77 
perspective: A given selection is or is not an example of 
a given style. The implication is that style is a matter 
of certain features, and students attend to these 
features. Cutietta (1993) proposed a set of defining 
features (motion, energy, flow, fabric) and tested their 
reliability. Eastlund (1992) used multidimensional 
scaling to extract the "dominant dimensions" of musical 
style. These two studies have been mentioned earlier, so 
a careful look at others in the field is in order here. 
Brittin (1991) examines the effects of overtly 
categorizing popular music on preference in college non-
music majors. The study is concerned with correlation 
between preference and category labels such as Pop, Rock, 
and Jazz. The distinction between groups is whether they 
were given category labels to use in classifying examples, 
or whether they were allowed to make up their own. 
While preference was measured on a continuous 
Lickert-type scale, the style categories were discrete. 
The process by which examples were categorized, whether by 
the researcher or by the subjects, was not an issue. This 
implies an assumption of a rigid, rule-based category 
structure: Any given example is simply in one category or 
another. The question of subjects class ifying the same 
example differently was not raised , nor was the question 
78 
of subjects rejecting the labels assumed by the 
researcher. 
An earlier study by Eastlund (1990) measured the 
influence of cognitive style on discriminating musical 
style. While "cognitive style" refers to a body of 
research not relevant here, it should be mentioned that it 
refers to such parameters as leveling-sharpening (the 
ability to discern fine size differences among geometric 
figures) and conceptual differentiation (the number of 
groups formed in a sorting task) . In simple terms, it 
claims to measure whether a student considers things as a 
group or as individual cases. 
Eastlund pays considerable attention to the non-
discrete aspects of style categories: "Listening for 
style is a complex process ... Ambiguity surrounds 
category boundaries ... Members of a stylistic category 
appear not to be related by a single rule, but seem to be 
related by commonalities that are sometimes difficult to 
verbalize" (p. 51-52). This shows healthy respect for the 
non-discrete Roschian categories mentioned above, although 
there are no references to such cognitive literature ; 
Eastlund does cite Gardner's study (1973 ) extensively. 
Nevertheless, Eastlund's actual experimental 
methodology treated style as a discrete variable. 
Subjects judged two passages, some from the same 
79 
composition, some from different compositions as same or 
different, then rated the degree of similarity on a 
Lickert scale . If a subject rated two passages as same 
when they were from the same composition, that answer was 
considered correct. This approach seems reasonable enough 
until one considers the assumptions on which it is based. 
If membership in a style category is indeed so 
complex and difficult to verbalize, can it be assumed that 
two passages from a given composition will be the same 
style, while a passage from some other composition will 
not? It seems quite possible, for instance, that a given 
piece might change along several dimensions such as 
instrumentation, tempo, or dynamics within a few measures. 
If students consider these features important to style, 
they will give a response that Eastlund might label 
incorrect. 
Eastlund's study, it must be noted, shows greater 
understanding of the nature of categories than most music 
education literature . The assumptions made are quite 
reasonable within the experimental framework adopted. But 
it refl ects the fact that music educators, when they deal 
with categories, deal with them in rigid terms. Eastlund 
writes: "Stylistic analysis concerns discovery of the 
attributes held in common by a group of compositions" 
(1990, p. 51). This implies a simplistic category 
80 
structure based on the presence or absence of specific 
attributes. It stands in contrast with the statements 
regarding the ambiguous, complex nature of listening for 
style. It reflects what Gardner (1985) terms the 
classical view, where categories are discretely defined by 
such attributes. 
Music Education and Style Categories 
The idea of using such questions and answers to 
measure musical knowledge is nothing new. The First 
National Assessment of Musical Progress (1974) and The 
Second Assessment of Music (1980) include a section which 
begins with the question "Are there any kinds of music 
that you like to listen to?" Students were then directed 
through a series of similar questions and more and more 
specific levels. Responses were categorized under such 
general headings as popular or classical, then more 
specific headings such as jazz, vocal and instrumental 
art, folk, rock , soul, and country-western . 
Students were also asked to classify recorded 
examples as similar or different in style. Again, moving 
to more specific levels, they were asked to classify 
different sub-styles of jazz such as ragtime, boogie-
woogie, and Brubeck . The results of this activity wer e 
81 
used to assess increases or declines in the musical 
knowledge among different age groups. 
Barrett (1989) identifies classifying as one of the 
core thinking skills in music: 
Classifying involves grouping items into categories 
based on attributes of the items ..• As the learner 
becomes more sophisticated, classification systems 
become more subtle, requiring finer distinctions. 
Students might be requested to classify jazz 
recordings by school or traditions of playing. A 
crucial step in classifying is the labeling process, 
which aids the student to recall the item at a later 
time, along with items belonging to the same 
category. (p. 51) 
Gardner (1973) found that children were able to 
classify different musical styles at a surprisingly early 
age. As part of a developmental study, Gardner had 
children at five age levels (between 6 and 19) listen to 
recordings of western symphonic music. Gardner was 
surprised at "the overall excellence of the Ss at the 
task" (p. 74 ) despite the grave reservations of colleagues 
who had listened to the recorded examples he used. 
With the insight that has made him a leading voice in 
the application of cognitive science to education, Gardner 
makes a critical comment that seems overlooked in many 
other studies: 
What are the cues in the musical stimulus? . 
Though rhythm , melody, instrumentation, and 
volume are probably the most prominent cues, the 
82 
listener may also take into account details of 
ornamentation or interaction among instruments 
or any other perceptible element. For this 
reason one can not state with confidence on what 
basis a stylistic judgment is made, nor can one 
insure that a certain aspect is or is not a cue. 
(p. 74) 
One question addressed here is that of automatic or 
preferred ways of categorizing music. That is to say, 
when an individual hears music, what are the first 
observations and associations made, and what are the first 
labels consciously applied? The hypothesis states that 
style--Classical, Jazz, Rock, Blues--will be the most 
frequently used domain of labels. 
There is no empirical evidence that style is the most 
common or preferred categorical response to music; the 
present study seeks such evidence. There are, however, 
some informal indications. Consider the commonly asked 
question, "What kind of music do you like?" A typical 
response does not refer to ensemble, nation, or mood, but 
to style. Radio stations announce themselves as rock, 
country, and classical. Record stores organize their 
shelves similarly. Bars and clubs tend to specialize in 
various rock idioms, jazz, country, or blues. 
Within the realm of academia, music courses are, for 
the most part, organized and labeled in terms of 
historical style periods: Music of the Renaissance, Music 
83 
of the Baroque, and Music of the 20th Century for example. 
Similarly, music history texts bear such labels as Music 
in the Middle Ages (Reese, 1968), Music in the Renaissance 
(Brown, 1976), and Mysic in the Romantic Era (Einstein, 
1947). 
It should be noted that informal real-world evidence 
about labeling was the basis for the groundbreaking 
research of Rosch and her associates nearly a generation 
ago. Eschewing the laboratory-controlled features of 
randomized colors, circles and squares, Rosch used tools, 
animals, and furniture. These were objects with which 
people interact on a daily basis, objects which, when 
people talk to each other, they instinctively and 
intuitively label and categorize. Surely music might be 
considered in very similar terms. 
Styles in Music History Literature 
Palisca (1981) asks a question that is at the heart 
of this study: "Is it justifiable to lump together into 
one stylistic period and under one label such diverse 
modes of musical expression?" (p. 2). Cognitive science 
offers an answer, but not a simple one. It does indeed 
seem justifiable, even essential to organize information 
and label one collection of items differently from 
another. But it must be remembered that such 
84 
organizations and labels are not rigid or discrete, and 
that any two items so lumped might seem quite different 
when viewed from various perspectives and by various 
observers. 
Seaton (1991) points out that "learning theory has 
clearly shown that information is only absorbed and 
retained when it is incorporated into some coherent 
pattern" (p. vi). It could well be argued that 
categories, style-based or otherwise, provide such 
coherent patterns. Surprisingly, Seaton addresses the 
issue of categories as more than feature-bundles without 
explicitly using cognitive terminology: 
"Western musical tradition is best regarded not 
as one of changing traits of style per se but as 
one of changing models for musical expression. 
These models in turn justify the articulation of 
music history into periods ... They also account 
for historical divisions and connections at 
different levels." (1991, p. vi) 
Seaton's "changing models" seem closely related to 
prototypes, with style periods being centered on specific 
exemplars rather than feature-bundles or rules. Far from 
negating the existence of style categories, these models 
actually "justify" them. An example should clarify this. 
In musical terms, it might be argued that Mozart's 
Eine Kleine Nachtmusik (K. 525) is a clear (prototypical) 
case of 18th century classical style, while his Fantasia 
85 
in C minor (K. 475) is less typical. Nachtmusik 
exemplifies clear , tonal, harmonies and melodies, with 
classical tonic-dominant key relationships, articulated 
phrases and sections, and the use of sonata and rondo 
forms (Seaton, 1991). The Fantasia is more dramatic and 
chromatic, exemplifying a style more akin to 19th century 
Romanticism (Grout, 1973). 
If one were teaching the 18th century style, Eine 
Kleine Nachtmusik would be the better model. It 
exemplifies those features and forms that characterize the 
First Viennese School. Keeping in mind the graded nature 
of the category, however, it would be misleading to ignore 
the Fantasia; it is an atypical member, but a member 
nonetheless. Presented with this atypical example, the 
students are likely to experience some confusion, but in 
the end they will have a more realistic idea of what was 
going on in 18th-Century Vienna. 
Contributions of the Present Study 
The first aspect of this study that sets it apart 
from existing literature is that musical style will be the 
domain of knowledge. No such study is available in 
cognitive expert-novice categorization literature. This 
time, a highly-structured body of information is the 
subject matter, not trees and birds. Similar studies have 
86 
been done with organized domains such as clinical 
psychiatry (Murphy & Wright, 1984) and Physics (Chi, 
Feltovich, & Glaser, 1981), but for the most part the 
research uses general knowledge domains such as dogs and 
birds (Tanaka & Taylor, 1989) or trees and furniture 
(Rosch et al. 1976). Since expertise is nonnally 
associated with highly-organized and structured domains, 
it seems appropriate to use music history here. 
There are methodological problems raised by the use 
of actual recorded music. The response time device would 
be difficult to use because the musical example requires a 
number of seconds to hear and categorize. Gardner (1973), 
Seaton (1991) and others point out the complexity of 
musical style and the difficulty of verbalizing its 
features. But music teachers and their students must cope 
with this complexity on a regular basis. This study will 
offer some empirical information regarding the different 
ways students {novices) and teachers (experts) might 
perceive the very same recorded example. 
Unique Features of the Present Study 
Rosch, Mervis, Gray, and Boyes-Braem {1976) claimed 
to have discovered universal principles of categorization. 
These universals were not, they stated, the content of any 
category. Rather , t he principles of category formation 
87 
were the universals. Thus one culture or one individual, 
while following the same universal principles, might come 
up with a different set of category members from another 
culture or individual. 
With regard to the present study, this means that the 
canonical taxonomy of Baroque, Classical, and Romantic 
music is not a fixed list as it might appear from 
traditional texts. There certainly are commonalities and 
differences, and there certainly are categories. But the 
structure and contents of those categories need further 
investigation. Rosch 's airplane mechanic, when asked to 
envision an airplane, did so from below, while the other 
subjects used a side view. Perhaps music experts, like 
the mechanic, hear music from a totally different angle. 
Contributions of this Study to Music Education 
Musical categories are obviously different in content 
from those of natural objects such as dogs and birds, but 
are they different in structure? Vertical levels and 
graded typicality have been demonstrated in natural object 
categories (Rosch et al., 1976) . It would be wrong, 
however, to assign these characteristics to only bird, 
dog, a nd furniture categories . They have been tested and 
supported in such diverse fields as corrections (Lurigio, 
1983), psychiatry (Murphy & Wright , 1984), and medical 
88 
diagnosis (Medin, Altom, Edelson, & Freko, 1982). 
Barsalou (1989) found these characteristics in what he 
tenned goal directed ad hoc categories. 
Murphy and Wright (1984) consider this issue. 
uAlthough some people do attain expertise in natural 
object domains usually investigated by psychologists 
(e.g., birds, flowers tools), many are experts in more 
abstract domains, like chess positions, musical styles, 
diseases, foreign policy, and so on. One might argue that 
these abstract domains are very different from object 
categories, but there seems to be little reason to believe 
that the effect of expertise should be qualitatively 
different in the two domains" (p.146, this writer's 
emphasis). In other words, the expertise effects found in 
object categories might well be applicable to more 
abstract categories such as musical style and genre. 
There are indeed differences between dogs and 
composers, and in the way people become experts in the 
different fields. Much of the knowledge about famous 
composers has come from academic study, not direct 
experience . Tanaka and Taylor (1991) theorize that 
physical interaction gives their dog experts an advantage 
over bird experts, who must view from a distance. Lakoff 
(1987) argues that real-world physical interaction and 
bodily experience might be the basis for very early 
89 
categorization of objects. Could musical expertise be of 
a more abstract and less physical nature? 
The Special Case of Music Expertise 
Musicians do have an opportunity to interact with the 
music of famous composers, but in a very different way 
from experts i n other fields. They play, sing, or conduct 
this music, all quite physical activities; they analyze it 
using traditional harmonic and rhythmic approaches. 
Experts, it seems reasonable to assume, have had more 
interaction with music than novices, more years to 
practice, play, conduct, and listen. 
Summary 
With context and expertise as the independent 
variables, this study should be of interest to music 
teachers at every level. Although some teachers are 
obviously more experienced and more knowledgeable than 
others, it still seems appropriate to think of them as 
more expert than their students. It also seems 
appropriate to assume that, in earning their degrees and 
credentials , they have been exposed to music at higher 
levels and in greater depth than the vast majority of 
their students. 
90 
As a result of these different backgrounds , the 
teacher hears the music in a different context from that 
of the student. The teacher , for example, considers the 
lesson plan, the important concepts to be learned, and 
chooses a musical selection that demonstrates those 
concepts. This choice is made by an infonned person who 
has spent considerable time listening to and studying 
music, including the example to be played. 
Students, on the ot her hand, come from a culture 
dominated by electronically reproduced commercial music. 
They enter a room known as the music room, where certai n 
kinds of music are heard quite unlike that with which they 
are familiar. They very probably never heard the musical 
selection chosen by the teacher. Teacher and students 
listen to the same recording , but with different knowledge 
bases and in different contexts . There is a serious 
question as to whet her they hear the same things. 
If musical categories are not the discrete, feature-
defined structures described by Plato , then an alternative 
model for them must be found. The organization and 
presentation of musical information would benefit from an 
empirical cognitive base. This study could serve as a 
beginning of an ever widening and ever deepening 
understanding of musical concepts, and a fresh approach to 
their teaching. 
91 
CHAPTER III 
METHODOLOGY 
The Pilot Study 
In the pilot study, all subjects were music experts 
(N=l6), and the independent variable was context. Half of 
the subjects listened to a narrow range of all "Classical" 
excerpts; this was analogous to the all-Norton Anthology 
tape (CTXTl) in the larger study . The other half listened 
to a mixture of popular styles as well as Classical, 
analogous to the Rock, Blues, and Jazz tape (CTXT2). Thus 
context was a variable, while expertise was not a factor . 
Even though subjects were asked to write down the 
style of music they heard, their responses included 
genres, national origins, and other kinds of labels. The 
use of style terms was greater in the mixed context: 
F(l,14)=8.8, p=0.0102. That is, the musical context 
appeared to have an effect on the kind of labels people 
used. These results a re summarized in Table 1. 
92 
TABLE 1. Analysis of Variance of Mean Use 
of Style Terms in Different Contexts 
(Pilot Study) 
Source of Sum of df Mean F p F 
Variation Squares Square Critical 
Between 0.2916 1 0.29 16 8.8 12 0.0 10 4.6001 
Within 0.4633 14 0.033 1 
Total 0.7549 15 
The Larger Study: Expertise and Context 
The pilot study showed some influence of context on 
the l abels used by experts. The main premise of the 
larger study is that experts and novices will react 
differently to the change in musical context. In the 
larger study, novices were expected to show an even 
greater change in the kinds of label s they used in the two 
different contexts. That is, the kinds of labels used by 
novices should be quite different when all the examples 
are Cl assical as opposed to a mixed sampling of styles. 
93 
Subjects 
All subjects were upper division or graduate students 
at the University of Oregon. There were 32 music majors 
(experts) and 32 non-majors (novices), and each subject 
listened to a series of 40 brief recorded examples, 12 of 
these being the target examples for the study. This 
produced a pool of 768 responses, a reasonable pool from 
which to test the hypotheses. 
Subjects were recruited by advertisements posted in 
the Music building, on the Psychology "paid studies " 
bulletin board, at several locations around a graduate 
housing complex, and several other central bulletin 
boards. Subjects were paid $5 for participation, or given 
credit through the psychology department subjects program. 
Upper division and graduate music students were 
desired because they would have taken the sequence of 
music history courses that would give them considerable 
familiarity with the Renaissance, Baroque, and Romantic 
styles in question. While some juniors would also have 
had these courses, the case for expertise could be made 
more securely with juniors, seniors and graduates. This 
also captured the experts without specifically mentioning 
music history or styles, a mention which might easily have 
influenced the kinds of responses they gave. In effect, 
94 
it avoided "tipping them off" to the researcher 's desired 
results. 
The novice sample was simply chosen to match the 
experts: juniors, seniors and graduates, but with no 
academic music background. Thus experts and novices could 
be assumed to be approximately equal in age, intellectual 
ability, and non-musical academic experience. Any 
differences in performance, therefore, could indeed be 
attributed to musical expertise. 
Measures 
The Instrument 
As described in the preceding chapters, the 
categorization of music was measured by a free response 
form headed with the question "What kind of music is 
this?" While much of the categorization literature uses 
response times to cues, this was not practical in the case 
of music. Milliseconds of response time might be 
meaningless when the music itself takes a significant 
amount of time to be perceived and labeled. 
Tanaka asked subjects "to say the word that names the 
object as quickly as possible" (1989, p.20). Upon hearing 
the response, Tanaka recorded whether it was a BLT or 
otherwise. For the musical study, an equivalent procedure 
95 
was to have the subject write his or her own response. 
Answering aloud might have interfered with listening to 
the music, would have been difficult to record when more 
than one subject was participating, and might have led to 
inaccurate judgments on the part of the researcher. 
Consistent with Tanaka's procedure, subjects were asked to 
write their first thoughts and not to change them; 
furthermore, only the very first word was used as data in 
the study. The inference was that this might best capture 
their first thoughts about the music. Another assumption 
was that "Italian Baroque" represented a different 
response from "Baroque--Italian. [see APPENDIX A: 
Response lists]. 
Of great importance was the exact selection of 
musical examples presented on the tape. The intent was to 
present a stable, reliable set to the subjects, but in two 
different contexts. While other studies (Gardner, 1973) 
had attempted this with different passages from the same 
composer or piece, this study required more precision. 
Also, the variable of musical context had to be 
manipulated with some validity: One could argue for 
endless factors that influence a listener ' s perceptions. 
These two problems, a reliably stable set and a valid 
change of context, were solved with a series of steps more 
stringent than those found anywhere in the literature. 
96 
Selecting and Presenting the Examples 
The target examples are from the Norton Anthology of 
Western Music set of compact discs, and are listed in the 
accompanying text as Renaissance, Baroque, and Romantic 
examples. Classical (18th century) examples were avoided 
because this label might be confused with the generic 
basic level term. The NAWM in a CD format satisfied the 
conditions of (a) high quality of performance and recorded 
sound, (b) clearly identifiable stylistic categories, and 
(c) ease in capturing the exact same 20 seconds of music 
on several different context tapes. 
Nearly any given example of music has too many 
attributes to list and control (Gardner, 1973). An ideal 
experimental situation would have been to select and 
present passages purely at random. Unfortunately, with 40 
examples total and 12 targets, early attempts at a random 
list were unsatisfactory. For instance, one randomized 
list produced nearly all group rather than solo music, 
while another was nearly all vocal with very little 
instrumental. It was feared that these attributes might 
affect the labeling of the few cases that were different. 
That is, it would unintentionally highlight the one or two 
cases of solo or instrumental music; in effect, it would 
introduce an undesired variable or a priming effect. 
97 
To accommodate all of these concerns, a constrained 
randomization was used, with the commitment to a balance 
of vocal and instrumental and of solo and ensemble music. 
All NAWM Renaissance, Baroque, and Romantic selections 
were assigned random numbers and so ordered. From this 
order, examples were placed in four separate columns: 
vocal solo, vocal ensemble, instrumental solo and 
instrumental ensemble. With selections from rotating 
columns, the resulting target set presented a more 
balanced sampling of music to the subjects than that 
resulting from pure randomization. 
One other factor played a role in the selection. 
Music of many styles and genres often begins with a few 
introductory bars. Often this passage is markedly 
different from the main body of the piece, as when a piano 
plays the opening measures of a Schubert Lied, or an 
allegro movement begins with an adagio prelude. The 
compact disk format allowed for precise timing, so that 
each selection was started at least 10 seconds into the 
piece. If this resulted in a major change (such as the 
entrance of a choir or soloist) very near the end of the 
passage, the starting point was adjusted to either include 
the event as a significant part of the 20 second passage 
or to eliminate it. 
98 
Setting the Musical Context 
Once again, a myriad of features might be used to 
define "musical context." This study, however, focused on 
style as a variable, so the two musical contexts were 
defined on the basis of style. Context 1 was a 
constrained all-NAWM set; Context 2 a mixed Blues, Rock, 
Jazz, and NAWM set. 
To ensure that some context was established by the 28 
non-target examples, the targets were shifted toward the 
end of the sample and mixed in. Starting with the final 
position (number 40), the targets were spaced with one, 
then two, then three of the context-setters between them. 
This pattern was repeated until the twelfth target was 
placed (position number 8). Consequently, subjects heard 
7 context-setting selections before hearing the first 
target. 
Once the 12 targets were chosen, the other examples 
were selected in a similar constrained random order. The 
NAWM examples were taken from the same list as the 
targets, with attention to solo/ensemble and vocal 
instrumental features. The Blues and Jazz examples were 
taken from two anthologies, somewhat equivalent to the 
NAWM. The intent was to provide an arguably matched set 
99 
that was already labeled and categorized, creating sub-
categories within the generic Jazz and Blues labels. 
The Smithsonian Classic Jazz collection (1988) 
provides high quality CD recordings of significant Jazz 
works. These works are categorized by the editors under 
such headings as Ragtime, New Orleans, Swing, and Modern 
(Williams, 1987). Since these selections were context 
setters rather than data relevant targets, less stringent 
measures were used. A sampling of three tracks from each 
of the Early Swing and Modern categories was drawn up, 
then randomized for position on the tape. 
The same process was used for the Blues selecti ons, 
using the anthology Blues Masters (1993). These examples 
came from Volume 7, Blues Revival, volume 8, Delta Blues, 
and volume 9, Postmodern Blues. Again, subjects could 
respond any way they chose, but they at least had the 
opportunity to use subsets of the category "Blues." 
For the Rock selections, no authoritative anthology 
was readily available. A great number of "best hits of .• " 
collections exist, but these tend to be commercial 
promotions by a given company rather than carefully 
constructed anthologies. A collection of CDs representing 
at least three sub-styles was needed, so examples included 
early rock from the 1950s and 1960s (Elvis, Chuck Berry), 
British groups (Wings, Dire Straits), and Southern groups 
100 
(Lynard Skynard, Allman Brothers). As with the NAWM, 
Jazz, and Blues examples, a subject had the opportunity to 
respond with the generic "rock," a more specific label, or 
in any subjective manner chosen. 
Finally, to avoid ordering effects, an additional 
tape was made in both contexts with the examples in 
reverse sequence. Targets still occupied the same 
positions (from number 8 to number 40) but in reverse 
order; Context setters also occupied the same positions 
but in reverse order. The original order was dubbed A, 
the reverse B; thus 4 tapes were used: lA, lB, 2A, and 2B. 
A complete listing of all titles, composers, and artists 
appears in Appendix B, Table 2. 
The Variables 
The two nominal independent variables were expertise 
and context. Experts were the music majors, novices the 
non-music majors. All subjects were upper division or 
graduate students, so the groups could be assumed to be 
matched on such fac t ors as maturity, intelligence, and 
general linguistic abilities. Context 1 was the all-
Classical tape , context 2 the tape using mixed styles. 
The "context setters" were the 28 examples which preceded 
and followed the 12 target examples. 
101 
As explained in Chapter 1, the dependent variable was 
the pool of 768 words used by subjects to describe the 
musical examples. This pool, however, was treated as 
three separate variables because the research questions 
examine three different portions and three different 
aspects of the 768 words. 
The first of the three hypotheses tested was that 
style terms would occur more often than random. That is, 
words referring to a historical musical idiom period such 
as Classical, Renaissance, or early 16th Century would 
constitute a higher percentage of the pool than references 
to the instrument (voice, piano), the genre (opera, 
symphony) or identification (composer or title). Relevant 
to research question 1, the total number of style terms 
was compared to the total response set. 
Basic level terms (BLTs) were by definition those 
words used most frequently by the novices. Studies 
demonstrate that experts use more specific labels than the 
general public. However, context might have some effect 
on this difference. The third hypothesis states that the 
use of BLTs will be different between the two contexts, 
and that this effect will be different for experts and 
novices. In other words, context and expertise will 
interact. This is the central issue of research question 
2. 
102 
Finally, for any given target, each subject might use 
a unique label that no other subject used. The 
hypothesis, however , stated that experts would use the 
same term more often than novices. Within each group, 
every time a subject used the same term as one or more 
other subjects, it was counted as a SET. Research 
question 3 concerns the influences of expertise and 
context on SETs. 
Validity and Reliability 
BLTs are in one sense a mathematical phenomenon, i n 
that they are the most frequently appearing words (Rosch 
et al., 1976). There are few threats to validity when one 
term appears with substant ially greater frequency. SETs 
are more problematical and some subjective judgment is 
involved. Equivalent terms include words that have, for 
all practical purposes, the same meaning. For example, 
two experts using the labels "Renaissance" and "16th 
century" might be said to agree in their categorization of 
an example . "Church" and "sacred" might also be argued as 
equivalent terms. "Mass " and "Kyrie " would not: The 
latter is more specific than the former ; similarly, "early 
Renaissance" and "Renaissance," "chamber music" and 
"string quartet," and "concert choir" and "a cappel la 
chorus" would not be considered exact equivalents. A 
103 
panel of the researcher and two music history faculty 
members made independent judgments of equivalency. 
Design and Hypotheses 
This study involves six research questions addressed 
by three hypotheses, so it might be best to represent it 
in three parts. Research question lA concerns the 
frequency with which style terms (such as Classical, 
Baroque, and Sixteenth Century) are used overall; 
Question lB concerns novices using "Classical" as the BLT. 
Research questions 2A and 2B concern the level (generic or 
specific) at which subjects label examples. Research 
questions 3A and 3B concern how often subjects use the 
same labels as others in their group. Each research 
question, of course, has an appropriate hypothesis, but it 
will be clearer if each hypothesis has a subscript based 
on its variable. The hypothesis for style terms, that is, 
will be symbolized HST, while those for basic level terms 
and shared equivalent terms will be symbolized as, 
respectively, HBLT and HSET• 
104 
Hypothesis Number One: Style Terms (Hs~l 
If style is not a preferred basis for categorizing 
music (null hypothesis) then STs should occur with the 
same frequency as other kinds of labels. If their 
frequency is higher than expected, however , then the 
alternative hypothesis is supported. STs will be compared 
with genre (opera, church, symphony), media (piano, 
orchestra, choir) , national origin (Italian, French), and 
other. The hypothesis states that subjects will use style 
terms more often than genre, media , nationality, or other 
terms. 
HsT0 (null): STs =Genre= Media= National= Other 
HsTl : STs >Genre ~ Media z National z Other 
Hypothesis Number Two: Basic Level Terms (HBLTl 
A BLT cannot be determi ned until after the pool of 
responses is carefully examined. It is operationally 
defined here as the style term used most frequently by the 
novice population. This is consistent with studies from 
Rosch et al. (1972) to Tanaka and Taylor (1991). 
Novices are expected to use the BLT significantly 
more often than experts; experts tend to use more specific 
105 
(subordinate level} terms. This is a well-documented 
phenomenon in categorization literature , but has not been 
tested in a musical setting. 
The situation is more complex than that , however , 
when context is introduced as a variable . The prediction 
is that Expert-Novice differences in BLT use will be 
greater in one context than in the other . There is no 
clear basis for predicting whether the change from CTXTl 
to CTXT2 will cause an increase or decrease in Novice 
BLTs, only that it will cause a change. The hypothesis 
states that Experts will use fewer BLTs than Novices, and 
that musical context will have an interactive effect on 
this between- group difference . 
No difference between EXP/NOV or between CTXT1/CTXT2 
Exp BLT< Nov BLT 
CTXTl BLT~ CTXT2 BLT 
Interaction 
Hypothesis Number Three : 
Shared Equivalent Terms (HSETl 
Experts should share more equivalent terms than 
novices. That is , they will agree more often and use the 
106 
same label (or equivalent labels) for a given target 
example. Murphy and Wright (1984) found this to be true 
with clinical psychologists, comparing experienced senior 
staff members with trainees. The number of SETs will be 
tallied for both groups and both contexts. Interaction is 
predicted: That is, a change of context should have a 
greater effect on the novices than on the experts. As 
with the BLT comparison, there is no clear basis for 
predicting whether the change from CTXTl to CTXT2 will 
cause an increase or decrease in Novice SETs, only that it 
will cause a change. The hypothesis states that Experts 
will use more SETs than Novices, and that context will 
have an interactive effect on the between-group 
differences. 
HsETO (null): 
No difference between EXP/NOV or between CTXT1/CTXT2 
HsET 1: EXP SETS> NOV SETs 
HsET 2: CTXTl SETs ~ CTXT2 SETs 
HsET 3: Interaction 
The study took place on the University of Oregon 
campus between April and July, 1993. All tapes and 
response sheets have been kept for possible use in further 
analysis. Detailed lists of all recorded examples appear 
107 
in APPENDIX B, and the subject response forms used appear 
in APPENDIX C. 
108 
CHAPTER IV 
RESULTS AND ANALYSIS 
Research Questions and Results 
The study is based on three main hypotheses, each 
addressing two research questions. Hypothesis one is 
concerned with how often subjects used style terms and how 
often novices used a particular term, "Classical." 
Hypothesis two is concerned with the use of this basic 
level term as influenced by differences in expertise and 
context. Hypothesis three is concerned with how often 
members of a group use the same labels, and how this is 
influenced by differences in expertise and context . The 
results will be discussed in the order of the research 
questions as stated in Chapter 1. 
Style Terms 
Research Question #lA: ST Frequency 
The first hypothesis stated that subjects would use 
style terms (STs) more often than genre, performing media, 
national origin, or other kinds of labels. STs consisted 
109 
of the established labels Renaissance, Baroque, Romantic, 
and the generic Classical. In addition, any term that 
referred to a historica l perspective (16th century, early, 
modern) or a commonly used musical idiom (folk, jazz) were 
taken as style terms. Based on the pilot study and the 
actual responses gathered in this study, responses were 
sorted into 5 nominal categories including style terms . 
The other nominal categories were : genre (opera, church, 
symphony); media of performance {vocal, choral, piano); 
national or linguistic (Italian , German), and other. As a 
test for reliability, the researcher's classification of 
label types was compared with those of two music history 
faculty members. This panel of three agreed with 78% 
reliability . 
This prediction was made independent of group or 
context, so the overall count of STs was analyzed rather 
than cell by cell . Subjects used STs 319 times out of a 
possible 768 ; other res ponse types are tallied in Table 2: 
TABLE 2. Summary of Label Response Types 
Style Genre Media Nat'l Other 
319 155 163 54 77 
110 
FIGURE 1 demonstrates even more clearly the 
predominance of style terms compared to those of genre, 
media of performance, national origin, and other kinds of 
labels. 
350 
30 
25 
w w a:: 
_J Lu 
> z~  I- :I: 
en w l-(!) o 
FIGURE 1. Comparison of Label Response Types 
Chi Square was calculated; x2 = (4, N=768, p< .001). 
Clearly, subjects used STs more than other kinds of 
labels. This supports the first hypothesis. 
Research Question #lB: Specifying the Basic Level Term 
The BLT for style was operationally defined as the 
style term used most often by the Novice group 
(representing the general population). As predicted, 
"Classical" was the most frequently used label, 78 out of 
111 
a total 129 STs. The next most frequently used label was 
nopera,n a genre rather than a style term, at 54 times . 
Among STs , the next most frequently used labels were 
nBaroque" and "Medieval" at 10 times each, not even 
approaching "Classical" at 78 . 
A chi square test proved significant, x2 (1, N = 129, 
p = . 0146), clearly establishing "classical" as the most 
frequently used style term among novice subjects . These 
results are summarized in Table 3 . 
TABLE 3 . Frequency of "Classical" as a Style Term 
"Classical" Other Total 
78 51 129 
Basic Level Term Response Patterns 
The Basic Level Term for Style has now been 
established as "Classical." The next step is to find how 
use of this BLT is influenced by expertise and context . 
Research Question #2A and 2B: 
BLTs with Expertise and Context 
The second hypothesis stated that novices would use 
BLTs more often than experts, and that this difference 
112 
would be influenced by context (i.e., interaction was 
predicted). A 2X2 ANOVA tested this prediction. 
Main Effects 
As predicted, there was a significant difference 
between the groups, with experts using the label 
"classical" 22 times and novices using it 78 times. Thus 
expertise was a significant source of variation . This is 
consistent with nearly all of the expert/novice 
literature, and with Tanaka and Taylor (1991) in 
particular. 
There was , however , no significant difference due to 
context in either group (p = .441). These results are 
somewhat at odds with Barsalou's context dependent 
categorization theories (1989). That is , a change of 
musical context did not change the use of the generic 
"classical" label. The results support category stability 
across context . Brooks (1989) suggested stability among 
the more experienced group, but the stability in the 
Novice responses is somewhat surprising. Figure 2 shows 
the clear difference between groups, with no influence 
from context . 
113 
Comparison of BLT Use 
■ CTXTl 
□ CTXT2 
EXP t-0/ 
BLT BLT 
FIGURE 2. Comparison of Basic Level Term Use 
by Group (EXP/NOV) and Context (CTXT) 
Interaction 
The prediction of interaction between expertise and 
context was not supported by the results. Expertise alone 
seems to be a factor in the use of the basic level style 
term "classical." That is, experts persisted in using 
"Renaissance , Baroque, and Romantic" while the novices 
persisted in using "classical" whether the context was all 
Norton Anthology or rock, blues, and jazz. 
114 
TABLE 4. Analysis of Variance of Mean Use 
of Basic Level Tenns 
Source of Sum of df Mean F p F 
variation Squares Square Critical 
Context 1.5625 l 1. 5625 0.574 0.451 4.0012 
Group 49.0 1 49.0 18.02 0.00 4.0012 
Interaction 0.0625 1 0.0625 0.023 0.88 4.0012 
Within 163.13 60 2.7188 
Total 213.75 63 
The lack of interaction is demonstrated more clearly 
in FIGURE 3. The lines are nearly parallel, indicating 
that while the two groups differed to some degree, context 
had no effect within each group. Interaction would be 
indicated by more divergent or crossed lines. 
--•--crxr 
1 
0+---------~ --o-crxr 
EXP t-0/ 2 
BLT BLT 
FIGURE 3. No Interaction Between Group and Context 
115 
Shared Equivalent Term Response Patterns 
Research Questions #3A and 3B : 
SETs with Expertise and Context 
SETs cons isted of any term used by two or more 
subjects on a given example , or of any term that was an 
equivalent (such as nchoir" and nchoral") . This part of 
the study tested the theory that categor ization by experts 
is more stable than categorization by novices. The 
hypothesis stated t hat a change in context will have 
little influence on the experts , but a signi ficant 
influence on the novices (i.e. , interaction between 
expertise and context). As with styl e terms, a panel of 
the researcher and two music history faculty independently 
judged for equivalency . This panel agreed with 88% 
reliability . 
The results did not support this hypothesis : neither 
expertise nor context was a source of significant 
variation. Although experts agreed more often than 
novices (289 versus 254 times) the difference was not 
significant : F(l, 60 ) = 3 . 12 , p = .08. Context had 
virtually no effect at all: F(l, 60) = .06, p = .80. 
FIGURE 4 shows a difference, though not significant, 
between experts and novices in their use of SETs, as well 
as the lack of influence of context on either group . 
116 
9.5 
9 
■ men 
8.5 
8 0 CTXT2 
7.5 
7 
0.. I- > 1-
x 1.t.1 ow 
1,1.J (I') z (f) 
FIGURE 4. Comparison of Shared/Equivalent Term Use 
by Group (EXP/NOV) and Context (CTXT) 
In greater detail, TABLE 5 shows the ANOVA for both 
groups in both contexts, and FIGURE 5 shows the lack of 
interaction between group and context. 
t 
TABLE 5. Analysis of Variance of Mean Use 
of Shared/Equivalent Terms 
Source of Sum of df Mean F p F 
Variation Squares Square Critical 
Context 0.3906 1 0 . 3906 0.064 0.802 4.00 12 
Group 19.141 1 19 .141 3 . 118 0 .08 3 4.0012 
Interaction 0.1406 1 0.1406 0.023 0.880 4 . 0012 
Within 368 . 31 60 6. 1 385 
Total 387 . 98 63 
117 
9.5 
9 --•--crxr 
8.5 1 
8 --o-- CTXT 
7.5 2 
7+-----------t 
EXP IOI 
SET SET 
FIGURE 5 . No Interaction Between Group and Context 
Once again, the categorization patterns appear more 
stable than might be e xpected, even within the novice 
group. Subjects in both groups agreed with their peers 
whether the context-setting excerpts were all Renaissance, 
Baroque, and Romantic , or Rock, Blues, and Jazz . The 
expert/novice differences suggested by Murphy and Wright 
(1984) did not appear , nor did t he context effects 
suggested by Barsalou ( 1989). 
Conclusions 
I n some cases , the results supported t he hypotheses, 
while in other cases they did not . Some of the results, 
indeed , appear rather inconsist ent with previous research 
and with the pilot study. The following discussion will 
118 
consider the possible reasons for both the expected and 
unexpected results. 
Style Terms and BLTs 
The results support the use of style terms as the 
most favored manner of categorizing music, significantly 
more so than genre, performing media, or national origin. 
Among the STs, the most frequently used term for the 
novices was "Classical," establishing it as the basic 
level term for musical style. 
Experts tended to use more specific level terms 
(Renaissance, Baroque, Medieval; Early ..• , and Late 
. ), as suggested by Tanaka and Taylor (1991). The 
difference was significant: F(l,60) = 16.2, p = .0002. 
Use of BLTs, however, was not influenced by context. Even 
when all the examples were "Classical" (actually 
Renaissance, Baroque, and Romantic , CTXTl) novices used 
that generic term 36 times . Novices used the term only 
slightly more often (42 times) when the target examples 
were preceded by Rock, Blues, and Jazz, and the difference 
was not significant. 
The implications are that musical experts, like those 
in other domains, categorize music at more specific 
levels. They rarely used the generic term "classical," 
preferring the more specific Renaissance, Baroque , or 
119 
Romantic labels . This difference persisted across the two 
contexts, suggesting tha t level of categorization is a 
function of expertise but not of context . 
Shared Equivalent Terms 
Within each of the four cells, subjects used the same 
or very similar labels (SETs) 543 times out of a response 
pool of 768. There were no differences between groups or 
between contexts. The number of times subjects agreed 
overall far exceeded that of disagreements. That is, 
subjects used the same label f or the same example more 
often than they used different labels. This difference 
was significant when tested wit h a chi square: x2 (1, N = 
768), p < .001. In other words, independent of group or 
context, all subjects agreed with their peers 
significantly more often than they disagreed. 
The results from each part of the study clearly 
support category stability within a group. The only 
differences that appeared were expert/novice differences 
in the BLTs. The way individuals label or categorize a 
passage o f music seems to be a relatively fixed response. 
The intensive study of music in the western art tradition 
leads to more specific levels of categorizing, but does 
not lead to more consistent agreement with one's peers. 
120 
CHAPTER V 
SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS 
Summary 
Thi s study began with a question. That question was 
simple and direct : "What kind of music did Mozart write?" 
The answer has proven considerably more complex and 
lengthy than one might expect . This complexity is not 
merely a function of Mozart's music , but of human 
cognition as well. To say that every person hears 
something different in Mozart's music is as great an 
oversimplification as it would be to say we all hear the 
same thing. The truth lies somewhere between these 
extremes. 
There was a rich diversity in the labels used to 
describe the musical examples. Some subjects used highly 
subjective and imaginative language, including "On Golden 
Classical Pond" and "Medieval Monks bemoaning their 
celibacy." It is perhaps regrettable that such creative 
descriptions might be marked "wrong" on a test using 
teacher-generated labels. A fundamental premise of this 
study is that such labels are just as valid and meaningful 
121 
as any textbook terminology. The fact that they are 
neither style terms nor shared equivalent terms--as 
defined here--has no bearing on their usefulness as an 
indicator of the subject's knowledge and perceptions of 
the music. 
Nevertheless, for the most part, both expert and 
novices used a surprisingly stable set of labels, and this 
stability was not significantly affected by context . 
Members of both groups tended to agree with their peers 
more often than they disagreed, and a relatively few terms 
(Classical, opera) dominated the pool of 768 responses . 
Research Questions and Results 
Research Question #lA 
Will subjects use style terms (STs) more often than 
genre, performing media, national origin, or other kinds 
of labels? 
The results of the study supported the hypothesis 
regarding use of style terms: Subjects overall used STs 
significantly more than any other kind of label . That is , 
they used labels such as Classical, Baroque, and 
Renaissance more often than labels such as piano, 
symphony, and Italian. 
122 
Research Question #lB 
Will Novices (representing the general public) use 
the generic, basic level term "Classical" significantly 
more often than other style terms? 
Novices did use "Classical" significantly more than 
any other ST, supporting the hypothesis. This establishes 
it as the basic level term (BLT) for style. For the 
purposes of this study, and in keeping with the research 
literature, the BLT has been operationally defined as the 
term used most often by the Novice group. 
Research Question #2A 
Will Novices use the generic, basic level term 
"Classical" significantly more often than Experts? 
The results supported the hypothesis: Novices did 
indeed use the generic BLT "Classical" significantly more 
often than Experts. This was consistent with the findings 
of nearly all of the literature. It is hardly surprising 
that Experts would use more specialized terms in 
describing the musical examples, but there was some 
evidence that the between-group difference might be 
influenced by context. 
123 
Research Question #2B 
Will musical context influence the use of BLTs? That 
is, will any difference between the two groups be greater 
with one tape than with the other? 
Context had no significant effect on BLT use. Novices 
used more BLTs than Experts whether the music was all 
classical (CTXTl) or mixed styles (CTXT2). These results 
failed to support the hypothesis, and run somewhat counter 
to the arguments of Barsalou (1989). They also appear 
inconsistent with the pilot study, wherein a similar 
manipulation of context led to a significant change in the 
kinds of labels used. 
Research Question #3A 
Will Experts use the same label for a given example 
significantly more often than Novices? 
The difference between the two groups was not 
significant, although experts agreed with each other 
slightly more often than novices. Again, this runs 
counter to some of the literature (Brooks, 1989) , which 
had suggested that experts would be more consistent in 
their labeling, especially when the context was changed. 
It appears that the Novices used a few terms ("Classical" 
124 
and uopera" in particular) so often that it led to a high 
level of agreement. 
This high level of agreement explains the difference 
between these results and those of Murphy and Wright 
(1984). In that study, the setting was a children' s 
clinical psychology center. The experts, like those in 
the music study, were veterans who tended to agree with 
each other in their labeling of the cases presented. The 
psychology novices, however, were beginning trainees who, 
while not yet experienced, had some motivation for 
appearing astute and perceptive in their descriptions of 
cases that they would indeed soon be treating. The music 
novices, in contrast, had no motivation to use anything 
other than the most familiar and superficial vocabulary. 
Furthermore, they were almost certainly influenced by the 
forces mentioned early in this study: the marketing 
practices of record companies and radio stations and the 
colloquial practice of everyday informal discourse about 
music. Not surprisingly, the very responses used most 
often by the novices are those used by junior high 
students in Cutietta (1993): Classical and opera. This 
even suggests that the supposed novices were indeed 
experts in the sense that they had learned and practiced 
these informal categories years earlier. 
125 
It should be noted that , consistent with Tanaka and 
Taylor (1991), the expert SETs were at more specific 
levels , and carried more detailed information than the 
novice SETS. This was also in keeping with other expert-
novice studies including Chi, Feltovich, and Glaser 
(1981) , and Lurigio (1983). Nevertheless, novices used 
shared or equivalent terms just as often as the experts. 
Research Question #3B 
Will musical context influence the use of SETs? That 
is, will any difference between the two groups be greater 
with one tape than with the other? 
There was no significant difference between the two 
groups or between the two contexts, and there was no 
interaction . Although the different groups used different 
terms, they tended to agree with members of their own 
groups. This is, perhaps, the most surprising result of 
the study. Experts, the literature suggested, might be 
fairly consistent and stable in their use of learned, 
specific labels. But there was much evidence that 
novices, with limited training and vocabulary, would use 
different kinds of terms in different contexts. That was 
not the case. 
Barsalou (1989), more than any other author , argues 
that categorization is not a stable process. Depending on 
126 
the circumstances, he proposes, the same person might 
categorize the same item different ways at different 
times. But even with completely different "context-
setting" musical excerpts (all Classical versus Rock, 
Blues, and Jazz) categorization patterns remained stable. 
Conclusions 
People--experts and novices--agreed significantly 
more often than they disagreed on what kind of music they 
were listening to . In this sense, some might argue that 
they "heard the same thing." But the overwhelming 
evidence from categorization research on which this study 
is based does not allow such a simplistic assumption. 
Individual theories about music, heuristics , experience, 
training, and an ever changing environment all have an 
impact on the choice of a label. More importantly, they 
have an impact on the individual's concept of what that 
label means. Two people using the same label, in other 
words, do not necessarily mean the same thing . The 
obvious case, returning to the original question, would be 
the expert and novice agreeing tha t Mozart wrote 
"Classical" music, but meaning different things. 
The expert probably uses the term to designate the 
18th Century Viennese Classicist School, associated with 
Haydn, Mozart, and early Beethoven. The novice probably 
127 
means music for an orchestra, played in concert halls by 
people in formal outfits. This case is so obvious, in 
fact, that music of the Viennese Classicists was 
intentionally eliminated from the Norton Ant hology 
selections of Context 1. It is quite possible that, in 
the few cases where experts used the term "Classical," 
they were mistaking some late 17th-Century piece for an 
early 18th-Century piece . Such a j udgment is entirely 
consistent with the concept of non-discrete, non-Platonic 
categori es. 
A label, then, does not carry with it the traditional 
Platonic discretely defined category; agreement on a label 
does not necessarily imply agreement on a list of defining 
characteristics. This is the fundamental implication 
found in Rosch et al . (1976), Barsalou (1989 ), Medin and 
Wattenmaker (1989), Murphy and Wright (1984), Tanaka and 
Taylor (1991), and nearly all of the literature on this 
complex subject. It is also the thought that must be held 
in mind while evaluating the results of this study. 
Use of Style Terms 
As predicted, both groups used style related terms 
more often than other kinds of terms . The present study 
makes no claims a s to whet her this is an effect of 
conditioning by radio stations, record companies , and 
128 
music courses, or an effect of some innate aspect of human 
cognition. Such questions might be the focus of future 
studies. 
Cutietta (1993) assumed style to be the preferred 
basis for categorization and searched for defining 
features. He adopted his own hypothetical feature list 
(motion, energy, flow, fabric, and color) and asked 
students to rate the music on the these factors. This 
approach primes the subjects to attend to their own 
perceptions of these factors. 
In contrast, the present study merely asked "What 
kind of music ••. ;" had the question been "What style?" 
or "What kind of instruments?" or "When do you think this 
was written?" subjects would have been primed to respond 
in a particular manner. In effect, the wording of the 
question avoided the imposition of a schema and allowed 
each subject to construct his or her own basis for 
responding. Cutietta's assumptions about flow and fabric 
might very well be accurate, but a design free of such 
priming cues would lend them greater validity. 
Basic Level Terms 
Gardner (1973) surprised himself and his musical 
colleagues when he examined young children's ability to 
consistently relate passages from the same composer. 
129 
Musical style categories seem to develop well before any 
overt t raining or study takes place . This, in fact, might 
be one reason for the unexpected results of the present 
study. In a sense, t he novices had nearly as much 
experience as the experts at categorizing music, but at in 
a less formal and rigorous atmosphere--thus the 
differences in level but not in consistency. 
Further explanation comes from a more precise 
examination of the two studies most closely related to 
this one: Tanaka and Taylor (1991) and Murphy and Wright 
(1984). These two studies supplied elements of 
methodology and design, but obvious differences were 
necessitated. For example, the use of real-time musical 
examples made it difficult to adopt a meaningful timed 
response format, which was an important aspect of Tanaka 
and Taylor and much of the categorization literature . 
Subjects did, at one point, write labels under pictures, 
which suggested the format used here. 
More important is a theoretical--or perhaps semantic-
-difference. Tanaka and Taylor suggested that for the 
experts, the more specific labels were the basic level 
terms. That is, for a bird expert, "robin , " "crow," 
"jay," and "cardinal" are as accessible and easily 
identifiable terms as "bird" is for the general 
population. The musical analog s uggests that, for music 
130 
experts, "Renaissance," ''Baroque," and "Romantic" actually 
are t he basic level terms. 
It might seem merely two different ways of saying the 
same thing: experts automatically categorize at more 
specific levels, or the basic level for experts is more 
specific. Tanaka and Taylor clearly demonstrated 
differences in the level at which experts and novices make 
their most immediate categorical response. Once that fact 
has been established, the distinction has more to do with 
defining BLT than with further inferences about human 
cognition. This approach might take "Renaissance", 
"Baroque", and "Romantic" as a composite BLT for the 
experts. 
Even with this alternative approach, context had 
little or no effect. Experts used the three STs 57 times 
in CTXTl, 54 times in CTXT2. Novices used their BLT 
{"Classical") 36 times in CTXTl, 42 times in CTXT2. A Chi 
Square test showed that these counts do not differ 
significantly: x2 (3, N=l89), p = . 102. Using Tanaka and 
Taylor's working definition of BLT, then, it appears that 
there is little difference even between groups. Once 
again, category stability overc omes individual 
differences. 
131 
Shared Equivalent Terms 
Neither expertise nor context had any significant 
effect on SETs. On the surface, it appears that experts 
and novices agreed with members of their own group at 
about the same rate. Any explanation of these unexpected 
results is, at this point, conjecture. Such conjecture 
might very well be the basis for further study, as new 
hypotheses arise to be tested. 
The central question is still one of category 
stability versus changes due to expertise and context. 
Much of the literature as well as the pilot study 
suggested that response patterns--SETs in particular--
would be influenced by these two variables. Since the 
results did not follow the predictions, a more careful 
scrutiny of the literature might yield an explanation. 
As discussed above, the novices might have become 
quasi-experts in that they had learned and practiced 
describing music in basic level terms early in their 
development. The very terms they used most often were no 
different from those used by Cutietta's junior high 
students: Classical and opera (1993). This logic is 
somewhat analogous to Tanaka and Taylor's view of the 
basic level itself changing with expertise. 
132 
Tanaka and Taylor's (1991) view assumes that both 
experts and novices were categorizing at their own basic 
level: Basic level, they argued, is not inherent in the 
material, but in the eyes of the beholder. There was no 
fundamental difference in the way the two groups 
categorized, only in the specificity of the labels. It 
can be argued similarly that the music novices were 
"experts," but at a basic level of terminology. That is, 
they were just as experienced at classifying music as 
"Classical" and "opera" as the experts were at classifying 
"Renaissance" and "Baroque." 
Another possibility is that the two groups agreed for 
different reasons, reasons that seem consistent with the 
relevant literature. Experts might agree because they 
rely on a similar base of training and study, using 
discrimination skills and specific labels beyond the 
capabilities of the novice group. Novices, on the other 
hand , quite possibly agree because they rely on stock, 
stereotypical labels that they have relied on for years. 
Much of the research suggests that experts form rich, 
elaborate schema for a category, while novices rely on 
specific items or rules (Lurigio, 1983; Chi , Feltovich, & 
Glover , 1981; Murphy & Wright, 1984). It is possible 
that, following some rule for classifying music, the 
novices used similar terms to label the examples. This, 
133 
in fact, runs counter to the argument that novice 
categories will be less consistent. Perhaps the tendency 
to use more rigid rule-based schema overcame any influence 
of context. 
Such rigid categorization, however, belies nearly all 
of the literature since Rosch. It seems to take us back 
to the Platonic, all-or-none kinds of categories that have 
been so discredited by empirical study. This apparent 
conundrum is the very problem that makes the teaching of 
musical style (and other abstract concepts) so difficult. 
And it is here that this study will make its most 
important contribution to those who would teach such 
concepts. 
Music educators must take responsibility not only for 
the kinds of verbal labels their students associate with 
music, but also for the meaning of the labels and the 
processes they entail. Although context did not prove a 
significant factor in this study, it might be an important 
factor in understanding the real world situation. Among 
themselves, music experts readily recognize the broad, 
rich, and diverse nature of a category such as 
"Renaissance music." They know that it includes the 
esoteric contrapuntal devices of Okeghem, the intensely 
emotional and chromatic madrigals of Gesualdo, the light-
hearted dance music of the English Virginal School. 
134 
When these same experts turn to their students, 
however, they sometime adopt a more rigid Platonic 
approach. In order to simplify things for the less 
experienced young listeners, they use definitions based on 
short, easily memorized feature lists. They point out 
salient features in a few examples, and they proclaim 
"That is Renaissance music." Attentive students can 
recite the feature list definition, and with very little 
practice, correctly label recorded examples. 
Cognitive category literature and at least some of 
the results of this study point out the errors inherent in 
such a simplistic approach. Listening for a specific set 
of defining features is not necessarily listening to the 
music. Reciting the definition or using the approved 
label does not necessarily indicate in-depth musical 
knowledge. 
Recommendations 
Recommendations for Further Study 
This study took theoretical and methodological 
approaches from categorization research and applied them--
for the first time--to musical styles. In this sense, it 
was a first step, and, as with any new study, leaves 
considerable work to be done by further research. There 
135 
are many possible directions this research might follow, 
but it is appropriate to suggest a few here. 
The range of target examples was quite limited. The 
idea was to present a reasonable number of context-setting 
excerpts, plus the twelve targets, in a reasonable amount 
of time. The hypotheses being tested were focused on the 
twelve targets because these were the only examples heard 
by every subject. A follow-up study might examine the 
entire response set, a much larger pool of labels. That 
is, one might examine every word written by every subject 
for every example. This might lend some insights into the 
results. 
Another direction for future research might be subtle 
changes in the methodology. The data here consisted of 
the very first word written by each subject for the twelve 
targets. It is possible, with more elaborate equipment, 
to record the spoken comments of s ubjects, as Tanaka and 
Taylor (1991) did. Perhaps the very act of writing led to 
a less spontaneous, more considered response pool. All 
the subjects were college student s, and all were 
accustomed to taking written tests . They may have 
experienced some interference from a "testing" mind set, 
even though they were assured that t here was no " right or 
wrong" answer, and that all results were anonymous. 
136 
A third direction is perhaps the most important and 
the most obvious. If the results are ultimately to be 
considered useful to classroom music teachers, the study 
should be replicated with real teachers and real classroom 
students (i.e., children). Serafine (1991) found some 
important developmental trends in the perception of brief 
musical patterns, and these might very well have their 
counterpart in perceptions of style in recordings of 
actual music. Cutietta (1993) states that junior high 
students categorize Western art music as "classical," 
"opera," and "church" music. These category schemas seem 
very close to those of the University of Oregon Novice 
group, but no extensive generalizations should be made 
without more empirical support. 
Recommendations for Application 
What kind of music did Mozart write? 
If students and teachers continue to think of 
categories as rigid rule-defined concepts, then they will 
have to sort out the contradictions when they encounter 
music that does not follow these rules. If "Classical" 
must mean music composed by the 18th Century Viennese 
School , then every student who uses the word in a generic 
sense--as well as record stores and radio stations--is 
wrong. They have failed the music test. 
137 
The teacher's task , under the Platonic view, is to 
erase the student's defining rules and replace them with 
those of the authorities . The student is clearly wrong, 
and the teacher is clearly right: "nor can any other 
opposite . •. simultaneously become or be its own 
opposite" (Plato, 1991, p.188). This either/or situation 
is not a function of musical style, student acumen, or 
teacher skill. It is a function of an entrenched view of 
category structure . 
If the strong case for individualized, theory-driven 
categories is applied to the same situation, we have not 
only a more realistic view but also a chance for better 
communication. Both student and teacher categorization 
systems can coexist; students can indeed learn much about 
music without necessarily adopting a textbook list of 
defining features . The goal should not be to call 
Mozart's music "Classical" and Bach's music "Baroque." It 
should be to understand and appreciate fundamental 
differences and fundamental similarities between these two 
great composers . The mere use of a category label does 
not constitute such an understanding. 
Style terms as category labels are unavoidable. It 
is possible to have fuzzy, heuristically defined concepts , 
but it is impossible to have no concepts at all. Whether 
concepts are "stored in our memory as prototypes" 
138 
(Jonassen, 1988) or whether individual memory traces form 
a concept in recall (Hintzrnan, 1986), the concepts are 
there. We need words to communicate these concepts. 
Knowing that the same word means different things to 
different people should, at least, lead to less 
misunderstanding. At best, it might lead to more 
effective communication and a better sense of mutual 
respect among teachers and students. 
Final Philosophical Considerations 
The student who complains that "all classical music 
sounds alike" and the teacher who believes in a rigid 
definition are, in a sense, victims of the same 
misconception. Both have concepts that are more concerned 
with rigid similarities than with rich diversity. Both 
are concerned with (arguably) superficial features rather 
than musical substance. 
Category labels, or words for concepts , serve a vital 
function in our thought and in our communication. But 
they must be used with the knowledge that their meaning is 
flexible and subjective. Rosch , Barsalou, Medin, and 
others have struggled with the psychological and 
linguistic aspects of this issue, and have been duly cited 
throughout this study . But in the end, the final question 
is philosophical. Plato (1991), Aristotle , (1991), Kant 
139 
(1965), and Wittgenstein (1958) struggled with the meaning 
of categories and names. In examining this idea and those 
of other profound thinkers, one figure seemed to most 
clearly articulate the position and advice that emerged 
here. 
The facts of the world in their sensible 
diversity are always before us, but our theoretic 
need is that they should be conceived in a way 
that reduces their manifoldness to simplicity. 
Our pleasure at finding that a chaos of facts is 
the expression of a single underlying fact is 
like the relief of the musician at resolving a 
confused mass of sound into melodic or harmonic 
order .• . But alongside this passion for 
simplification there exists a sister passion, 
which in some minds--though they perhaps form the 
minority--is its rival. This is the passion for 
distinguishing; it is the impulse to be 
acquainted with the parts rather than to 
comprehend the whole ... It prefers any amount 
of incoherence, abruptness, and fragmentariness 
(so long as the literal details of the separate 
facts are saved) to an abstract way of conceiving 
things that, while it simplifies them, dissolves 
away at the same time their concrete fullness. 
Clearness and simplicity thus set up rival 
claims, and make a real dilemma for the thinker. 
(William James, 1952, p. 4-5) 
James' "chaos of facts" is too often associated with 
the student, while the "passion for distinguishing" is too 
often associated with the teacher . On the contrary, James 
holds that we must all face the chaos, and we must all 
satisf y our pass ion f or distinguishing. The rival claims 
140 
of clearness and simplicity set up a dilemma for thinking 
students and thinking teachers. We must face the dilemma 
together. 
141 
APPENDIX A 
RESPONSE LISTS 
(Exact replication of Subjects' written responses 
with first word, fragments of following words, 
crossed-out words, and occasional 
misspelled words) 
142 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
101 1 CHORAL MUSIC 110 1 CHORAL 
101 2 MID 1900-OPE 110 2 BAROQUE ARIA 
101 3 RENAISSANCE 110 3 RENAISSANCE D 
101 4 CONTEMPORAR 110 4 ROMANTIC LIE 
101 5 SYMPHONIC M 110 5 BAROQUE DOUB 
101 6 CLASSICAL MA 110 6 SOLO GUITAR 
101 7 PATRIOTIC 110 7 ROMANTIC 
101 8 ORGAN MUSIC 110 8 BAROQUE ORGA 
101 9 EARLY MOTET 110 9 EARLY COUNTE 
101 10 PIANO SONATA 110 10 ROMANTIC SOL 
101 11 MOTET-I SH 110 11 CHORAL - COUN 
101 12 REN. FRENCH TU 110 12 RENAISSANCES 
108 1 CLASSICAL - CH 111 1 BAROQUE 
108 2 BAROQUE OPER 111 2 EARLY CLASSIC 
108 3 INSTRUMENTA 111 3 LATE MEDIEVAL 
108 4 GERMAN ART SO 111 4 ROMANTIC LIE 
108 5 SYMPHONY OR 111 5 BAROQUE CONC 
108 6 GUITAR ._ SPAN 111 6 BAROQUE 
108 7 NATIONALISTI 111 7 MID/LATE ROM 
108 8 BAROQUE ORGA 111 8 LATE BAROQUE 
108 9 MELISMATIC/ 111 9 MEDIEVAL 
108 10 ROMANTIC PIA 111 10 MID ROMANTI 
108 11 RENAISSANCE - 111 11 BAROQUE 
108 12 LATE MIDDLE A 111 12 LATE MEDIEVAL 
109 1 CHORAL 117 1 CLASSICAL CHO 
109 2 ITALIAN OPERA 117 2 BAROQUE VOCA 
109 3 ENGLISH DANC 117 3 RENAISSANCE/ 
109 4 GERMAN LIEDE 117 4 ROMANTIC W/ 
109 5 SYMPHONY - V 117 5 CLASSICAL ORC 
109 6 xx 117 6 CLASSICAL GUI 
1 09 7 P IANO & STRIN 117 7 PIANO CONCER 
109 8 ORGAN FANTAS 117 8 BAROQUE ORGA 
109 9 EARLY LATIN M 117 9 MELISMATIC 0 
109 10 PIANO SONATA 117 10 ROMANTIC PIA 
109 11 CHORAL FUGUE 117 11 CHOIR 
10 9 12 15TH CENT CHA 117 12 MEDIEVAL MOT 
143 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
118 1 A CAPELLA CHO 124 1 BRUCKNER VIR 
118 2 SOPRANO W/IN 124 2 ITALIAN ORATO 
118 3 RENAISSANCE 124 3 RENAISSANCE C 
118 4 GERMAN ROMA 124 4 MARIAN ANDER 
118 5 ORCH, CLASSIC 124 5 VIVALDI! OR C 
118 6 SOLO GUITAR FA 124 6 EARLY SOLO PI 
118 7 PNO TRIO, ROM 124 7 STRING QUARTE 
118 8 ORGAN MUSIC. 124 8 ORGAN/BAROQ 
118 9 "KYRIE ELEISO 124 9 CATHOLIC MAS 
118 10 PIANO, ROMAN 124 10 ROMANTIC SOL 
118 11 CHORUS IMITA 124 11 EARLY CHORAL 
118 12 FR TROUVERE SO 124 12 FRENCH - EARL 
119 1 CHORAL 125 1 xx 
119 2 OPERA ARIA - E 125 2 xx 
119 3 EARLY MEDIEV 125 3 CONSORT /REN 
119 4 ART SONG 125 4 xx 
119 5 STRING QUARET 125 5 BAROQUE/ORCH 
119 6 EARLY MUSIC 0 125 6 PIANO/QUINTE 
119 7 STRING/PIANO 125 7 ORGAN/BAROQ 
119 8 ORGAN - CHURC 125 8 xx 
119 9 CHANT STYLE C 125 9 PIANO/ROMAN 
119 10 ROMANTIC PER 125 10 xx 
119 11 CHORAL, CANO 125 11 xx 
119 12 A TROUBADOUR 125 12 xx 
120 1 CHORAL, SACRE 126 1 CHORAL 
120 2 L ' ORFEO, GLUCK 126 2 BAROQUE 
120 3 CHAMBER MUS 126 3 BAROQUE 
120 4 GERY.lY1  FRENC 126 4 VOCAL SOLO 
120 5 SYMPHONIC, C 126 5 CONCERTO GRO 
120 6 GUITAR OR LUT 126 6 xx 
120 7 SYMPHONIC, P 126 7 QUNTET 
120 8 BUXTEHUDE, OR 126 8 BAROQUE 
120 9 MALE CHORUS, 126 9 CHORAL 
120 10 CHOPIN, RUBAT 126 10 PIANO SONATA 
120 11 CHORAL, SACRE 126 11 BAROQUE VOC.A 
120 12 CHANSON, REN 126 12 FRENCH IMPRO 
144 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
128 1 FULL CHOIR, BA 131 1 CLASSICAL - CH 
128 2 BAROQUE(?) AR 131 2 BAROQUE ARIA 
128 3 REN. DANCE MU 131 3 EARLY BAROQU 
128 4 (ALTO) SONATA 131 4 ROMANTIC LIE 
128 5 BAROQUE CONC 131 5 BAROQUE CONC 
128 6 GUITAR SONATA 131 6 RENAISSANCE C 
128 7 STRING QUARTE 131 7 ROMANTIC TRI 
128 8 ORGAN SONATA 131 8 BAROQUE ORGA 
128 9 MEN'S CHOIR - 131 9 EARLY RENAISS 
128 10 PIANO SONATA 131 10 CLASSICAL PIA 
128 11 BAROQUE CANT 131 11 LATE RENAISSA 
128 12 RENAISSANCE D 131 12 RENAISSANCE F 
129 1 CHORAL 
129 2 BAROQUE ARIA 
129 3 RENAISSANCE -
129 4 LIEDER 
129 5 BAROQUE CONC 
129 6 CLASSICAL GUI 
129 7 ROMANTIC - PI 
129 8 BAROQUE - ORG 
129 9 GREGORIAN CH 
129 10 CLASSICAL - PI 
129 11 xx 
129 12 xx 
130 1 RENAISSANCE 
130 2 BAROQUE OPER 
130 3 RENAISSANCE, 
130 4 ROMANTIC M-E 
130 5 BAROQUE, CONC 
130 6 GUITAR/LUTE 
130 7 ROMANTIC PIA 
130 8 BAROQUE, ORG 
130 9 MEDIEVAL MOT 
130 10 PIANO, ROMAN 
130 11 MOTET 
130 12 MEDIEVAL CHA 
145 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
102 1 MONOPHONIC 105 1 CHORAL OCTAV 
102 2 SOP. ARIA WIT 105 2 BAROQUE (ITAL 
102 3 X 105 3 RENAISSANCE D 
102 4 FRENCH ART SO 105 4 GERMAN ART SO 
102 5 ORCH - STRING 105 5 BAROQUE CHAM 
102 6 BAROQUE SUITE 105 6 CLASSICAL GUI 
102 7 MARCH - STRIN 105 7 TRASHY CLASSI 
102 8 ORGAN - POLYP 105 8 ORGAN PRELUD 
102 9 CHORAL POLYP 105 9 MEDIEVAL/EAR 
102 10 SOLO PIANO ET 105 10 CHOPIN PRELU 
102 11 CHORAL MONOP 105 11 CHORAL - RENA 
102 12 CELTIC SONG W 105 12 RENAISSANCE B 
103 1 A CAPELLA CHO 106 1 LATE BAROQUE 
103 2 ITALIAN - BAR 106 2 EARLY BAROQU 
103 3 EARLY INSTRU 106 3 RENAISSANCE E 
103 4 SOP. SOLO GERM 106 4 LATE ROMANTI 
103 5 BAROQUE CONC 106 5 CLASSICAL (OR 
103 6 GUITAR ARRAN 106 6 BAROQUE GUITA 
103 7 xx 106 7 CLASSICAL QUI 
103 8 ORGAN SOLO co 106 8 LATE BAROQUE 
103 9 ORGANUM 106 9 EARLY BAROQU 
103 10 SOLO PIANO - M 106 10 ROMANTIC PIA 
103 11 A CAPELLA - MA 106 11 LATE BAROQUE 
103 12 ITALIAN 106 12 RENAISSANCE C 
104 1 SACRED MOTET 107 1 SACRED CHORA 
104 2 EARLY BAROQU 107 2 EARLY OPERA M 
104 3 RENAISSANCE 107 3 RENAISSANCE I 
104 4 LIEDER 107 4 ROMANTIC voe 
104 5 VIVALDI CONC 107 5 BAROQUE ORCH 
104 6 LUTE - RENAISS 1 07 6 RENAISSANCE G 
104 7 BRAHMS QUINT 107 7 ROMANTIC ERA 
104 8 BACH ORGAN FU 107 8 BAROQUE ORGA 
104 9 MEDIEVAL 107 9 MEDIEVAL CHU 
104 10 CHOPIN NOCTU 107 10 ROMANTIC ERA 
104 11 BAROQUE CHOR 107 11 RENAISSANCES 
104 12 RENAISSANCE 107 12 EARLY RENAISS 
146 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
112 1 CONCERT CHOR 115 1 BAROQUE 
112 2 BAROQUE ITAL. 115 2 CLASSICAL voe 
112 3 REN/BAR LUTE 115 3 RENAISSANCE 
112 4 ART SONG - FRE 115 4 ROMANTIC voe 
112 5 SYMPHONIC - 115 5 CLASSICAL 
112 6 BAR/REN. LUTE 115 6 CLASSICAL GUI 
112 7 STRING QT & PN 115 7 CLASSICAL 
112 8 BAROQUE ORGA 115 8 BAROQUE 
112 9 EARLY REN CHO 115 9 RENAISSANCE 
112 10 PIANO SOLO - L 115 10 ROMANTIC 
112 11 CHORAL - GENE 115 11 RENAISSANCE 
112 12 14TH C. - ARS 115 12 RENAISSANCE 
113 1 BAROQUE CHOR 116 1 CHORAL, A MAS 
113 2 BAROQUE ARIA 116 2 ARIA FROM BAR 
113 3 RENAISSANCE C 116 3 BAROQUE 
113 4 GERMAN LIED ( 116 4 LIEDER (A GOOD 
113 5 BAROQUE CONC 116 5 BAROQUE 
113 6 CLASSICAL GUI 116 6 GUITAR INSTRU 
113 7 CLASSICAL PNO 116 7 STRING QUARTE 
113 8 BAROQUE ORGA 116 8 ORGAN - BACH? 
113 9 RENAISSANCE C 116 9 CHORAL, MEN'S 
113 10 CHOPIN PIANO 116 10 AH! CHOPIN, P 
113 11 BAROQUE CHOR 116 11 CHORAL, A CAP 
113 12 FRENCH RENAI 116 12 BAROQUE? VOC 
114 1 HOMOPHONY 121 l CHORUS MASS -
114 2 ARIA (CLASSIC 121 2 VOCAL - OPERA 
114 3 IMITATIVE POL 121 3 BAROQUE 
114 4 ARIA-LIKE (W/ 12 1 4 VOCAL SOLOW/ 
114 5 CLASSICAL (FRO 121 5 CLASSICAL SYM 
114 6 SPANISH GUITA 121 6 CLASICAL SOLO 
114 7 FRENCH JAZZ 121 7 STRING QUARTE 
114 8 EARLY ORGAN 121 8 BAROQUE FUGU 
114 9 GREGORIAN CH 121 9 MASS/RENAISS 
114 10 ROMANTIC (CH 121 10 OMANTIC PIAN 
114 11 POLYPHONY (I 121 11 CHORUS/CLASS 
114 12 MELISMATIC P 121 12 VOCAL/OPERA 
147 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
122 1 BAROQUE CHOR 132 1 CHORAL 
122 2 OPERA 132 2 ITALIAN ARIA-
122 3 MEDIEVAL (LAT 132 3 RENAISSANCE-
122 4 FRENCH CABAR 132 4 GERMAN LIEDE 
122 5 BAROQUE ORCH 132 5 BAROQUE CONC 
122 6 RENAISSANCE L 132 6 RENAISSANCE-
122 7 ORCHESTRAL 132 7 LATE CLASSICA 
122 8 VERY EARLY OR 132 8 BAROQUE-ORGA 
122 9 GREGOIHAN CH 132 9 MASS 
122 10 ROMANTIC 132 10 CHOPIN PIANO 
122 11 CHORAL 132 11 MADRIGAL 
122 12 MEDIEVAL 132 12 TROUVERE SONG 
123 1 ORATORIO/ CHU 
123 2 BAROQUE ITALI 
123 3 RENAISSANCE I 
123 4 FRENCH SONG 
123 5 CLASSICAL SYM 
123 6 SPANISH GUITA 
123 7 PIANO QUINTE 
123 8 BAROQUE ORGA 
123 9 CHANT 
123 10 PIANO IMPRO 
123 11 ORATORIO/ANT 
123 12 FRENCH MEDIE 
127 1 MEDIEVAL CHA 
127 2 EARLY BAROQU 
127 3 MEDIEVAL CON 
127 4 GERY.. .' \H FRENC 
127 5 CONCERTO GRO 
127 6 MEDIEVAL LUTE 
127 7 PIANO QUINTE 
127 8 MEDIEVAL ORG 
127 9 MEDIEVAL CHA 
127 10 CHOPIN 
127 11 MEDIEVAL CHO 
127 12 MEDIEVAL SON 
148 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
201 1 OPERA CHORUS 206 1 CLASSICAL - w 
20 1 2 DIGNIFIED, CH 206 2 OPERA 
201 3 BAROQUE INSTR 206 3 THE KIND WE W 
201 4 VOCAL, ALTO, G 206 4 GERMAN OPERA 
201 5 CLASSICAL, VIO 206 5 18TH C. BAVAR 
201 6 FLAMENCO PIE 206 6 SPANISH GUITA 
201 7 DRAMATIC PAR 206 7 PIANO W/STRI 
201 8 CHURCH "FILLE 206 8 CLASSICAL - BA 
201 9 xx 206 9 MUSIC FROM A 
201 10 WALTZ 206 10 SOFT, PIANO 
201 11 CHRISTIAN CHO 206 11 CHOIR, CLASS! 
201 12 ORIENTAL 206 12 BALLAD, IRISH 
202 1 xx 207 1 CHOIR/CHORAL 
202 2 OPERA 207 2 SOLO PIECE/CH 
202 3 RENAISSANCE 207 3 SYMPHONY/FL 
202 4 ACCOMPANIM 207 4 SOLO OPERA PI 
202 5 CLASSICAL 207 5 SYMPHONY/BE 
202 6 SPANISH GUITA 207 6 FOREIGN/SPAN 
202 7 DRAMATIC 207 7 SYMPHONY 
202 8 ORGAN 207 8 CHURCH/ORGA 
202 9 MONASTIC CHO 207 9 MEN'S CHORAL 
202 10 PIANO 207 10 SOLO PANO 
202 11 CHOIR 207 11 CHORAL 
202 12 MINSTREL 207 12 FRENCH SOLO 
205 1 CHORAL - CLAS 213 1 CLASSICAL 
205 2 ITALIAN CLASS 213 2 OPERA 
205 3 MEDIEVAL INS 213 3 CHAMBER 
205 4 GERMAN CLASS 213 4 OPERA 
205 5 CLASSICAL ORC 213 5 CLASSICAL ORC 
205 6 CLASSICAL GUI 213 6 CLASSICAL 
205 7 CLASSICAL INS 213 7 G:f.ASS fGAI:. - CH 
205 8 CLASSICAL ORG 213 8 CHAMBER 
205 9 MEDIEVAL voe 213 9 OPERA 
205 10 CLASSICAL PIA 213 0 CLASSICAL (PIA 
205 11 MEDIEVAL CHO 213 11 CHIOR 
205 12 FRENCH VOCAL 213 12 CHAMBER MUS 
149 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
216 1 CHORAL 223 1 CHOIR 
216 2 OPERA 223 2 OPERA 
216 3 xx 223 3 xx 
216 4 OPERA 223 4 GARBAGE 
216 5 CLASSICAL 223 5 SYMPHONY 
216 6 CLASSICAL GUI 223 6 GUITAR SOLO 
216 7 CLASSICAL 223 7 FUNERAL PROC 
216 8 CHAMBER MUS 223 8 HAUNTED HOUS 
216 9 CHANT 223 9 CHURCH CHOIR 
216 10 CONCERTO 223 10 LOUNGE MUSIC 
216 11 OPERA 223 11 CHOIR 
216 12 xx 223 12 FOREIGN 
218 1 MODERN CLASS 224 1 CHOIR 
218 2 OPERA 224 2 OPERA 
218 3 BAROQUE - REN 224 3 MIDDLE AGES F 
218 4 MODERN 224 4 OPERA 
218 5 ROMANTIC CLA 224 5 CLASSICAL 
218 6 MEDIEVAL 224 6 GUITAR FOLK 
218 7 MODERN CLASS 224 7 CLASSICAL 
218 8 BAROQUE 224 8 CHURCH ORGAN 
218 9 GREGORIAN 224 9 CHURCH 
218 10 PIANO ROMAN 224 10 CLASSICAL 
218 11 CLASSICAL 224 11 CHOIR 
218 12 BAROQUE 224 12 xx 
222 1 CHURCH CHANT 225 1 BALLET MUSIC 
222 2 MIDDLE AGES C 225 2 OPERA 
222 3 RENAISSANCE F 225 3 ORCHESTRA OLD 
222 4 CLASSICAL OPE 225 4 FOREIGN 
222 5 CLASSICAL EUR 225 5 CLASSICAL 
222 6 MIDDLE AGES L 225 6 FOLKLORE 
222 7 CLASSICAL EUR 225 7 CLASSICAL 
222 8 MIDDLE AGES C 225 8 BIBLE 
222 9 GREGORIAN CH 225 9 MONK CHOIR 
222 10 CLASSICAL PIA 225 10 SAD 
222 11 CHURCH CHANT 225 11 CHOIR W/ ORCH 
222 12 FOLK - MIDDLE 225 12 OPERA OR PLAY 
150 
_fil EXAMPLE LAB L _fil EXAMPL L 
226 1 CHOIR 230 1 CHOIR 
226 2 OPERA/BAROQU 230 2 OP.ERA FEMA 
226 3 BAROQUE 230 3 CHAMBER M 
226 4 OPERA 230 4 FEMALE GER 
226 5 CLASSICAL 230 5 CLASS I CAL S 
226 6 CLASSICAL GU I 230 6 CLASSICAL G 
226 7 CHAMBER 230 7 TRIO CLASSI 
226 8 CLASSICAL MED 230 8 BAROQUE OR 
226 9 MONK CHANTI 230 9 GREGORIAN 
226 10 LOUNGE PIANO 230 10 CLASSICAL P 
226 11 CHOIR 230 11 CHOIR 
226 12 ROYAL CLASSIC 230 12 TRIO CLASSIC 
227 1 SYMPHONIC CH 
227 2 OPERA 
227 3 QUINTET 
227 4 SOLO CONCERT 
227 5 SYMPHONY OR 
227 6 ~ SOLO 
227 7 ORCHESTRA 
227 8 PARLOUR MUSI 
227 9 OPERA 
227 10 SOLO CONCERT 
227 11 SYMPHONY OR 
227 12 OPERA 
228 1 CHOIR 
228 2 OPERA 
228 3 FOLK MUSIC 
228 4 OPERA/ CHEESY 
228 5 CLASSICAL / SY 
228 6 CLASSICAL MU 
228 7 xx 
228 8 CARNIVALESQU 
228 9 CHANTING 
228 10 CLASSICAL PIA 
228 11 CHOIR 
228 12 OPERA/FOLK 
151 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
203 1 QUIRE [ESL] MU 209 1 CHOIR 
203 2 OPERA 209 2 OPERA 
203 3 (CLASSICAL MU 209 3 CLASSICAL 
203 4 OPERA 209 4 OPERA 
203 5 CLASSICAL MU 209 5 CLASSICAL 
203 6 CLASSIC GUITA 209 6 CITAR MUSIC 
203 7 CLASSICAL MU 209 7 CLASSICAL 
203 8 ORGAN IN A CH 209 8 ORGAN MUSIC 
203 9 QUIRE [ESL] IN 209 9 CHOIR 
203 10 LIGHT MUSIC (P 209 10 CLASICAL PIAN 
203 11 QUIRE [ESL] 209 11 CHOIR MUSIC 
203 12 OPERA (FLUTE) 209 12 OPERA 
204 1 CLASSICAL 210 1 CLASSICAL CHO 
204 2 OPERA 210 2 OPERA 
204 3 MEDIEVAL CHA 210 3 EARLY CLASSIC 
204 4 CLASSICAL 210 4 OPERA 
204 5 BAROQUE CLASS 210 5 CLASSICAL ORC 
204 6 CLASSICAL GUI 210 6 CLASSICAL GUI 
204 7 BAROQUE CLASS 210 7 CLASSICAL 
204 8 FUGUE 210 8 CLASSICAL/HA 
204 9 GREGORIAN CH 210 9 CLASSICAL-BA 
204 10 CLASSICAL 210 10 CLASSICAL 
204 11 CLASSICAL 210 11 CLASSICAL/CH 
204 12 MEDIEVAL CHA 210 12 OPERA 
208 1 CHORAL - I ALW 211 1 CHURCH 
208 2 ITALIAN OPERA 211 2 OPERA 
208 3 CHAMBER QUAR 211 3 CHAMBER 
208 4 GERMAN W/TH 211 4 OPERA 
208 5 17TH CENT/CLA 211 5 CLASSICAL 
208 6 GUITAR, SATUR 211 6 FLAMENCO 
208 7 MODERN - RETR 211 7 CHAMBER 
208 8 CHUCK EVANS A 211 8 CHAMBER 
208 9 MONKS BEMOA 211 9 GREGORIAN 
208 10 NOCTURNE - ES 211 10 CLASSICAL 
208 11 VISIONS OF PA 211 11 CLASSICAL 
208 12 ITALIAN BEAUT 211 12 CHAMBER 
152 
SUBJ EXAMPLE LABEL SUBJ EXAMPLE LABEL 
212 1 CHOIR 217 1 SOARING - MOS 
212 2 CHOIR 217 2 OPERATIC, BUS 
212 3 FOLK 217 3 CLASSICAL, NI 
212 4 FOLK 217 4 NOT SOMETHIN 
212 5 JAZZ 217 5 SYMPHONIC -
212 6 FOLK 217 6 SPANISH - CLA 
212 7 JAZZ 217 7 ROYAL - POMP 
212 8 CHOIR 217 8 TINNY MUSIC 
212 9 CHOIR 217 9 GREGORIAN CH 
212 10 xx 217 10 PIANO 
212 11 CHOIR 217 11 UPLIFTING 
212 12 CHOIR 217 12 A PERFORMANC 
214 1 CHORAL 219 1 BIG CHOIR CLA 
214 2 OPERA 219 2 19TH CENTURY 
214 3 CLASSICAL 219 3 COURT MUSIC 
214 4 OPERA 219 4 DEPRESSING W 
214 5 SYMPHONY CL 219 5 CLASSICAL ORC 
214 6 CLASSICAL GUI 219 6 SEGOVIAN SPA 
214 7 CLASSICAL 219 7 CLASSICALY CO 
214 8 ORGAN HYMNA 219 8 HAPPY PIPE OR 
214 9 CHOIR BOYS - G 219 9 MONESTARY 
214 0 PIANO CONCER 219 10 ON GOLDEN CLA 
214 11 OPERA 219 11 CHEERY OPERA 
214 12 CLASSICAL 219 12 ELECTRIC "SOU 
215 1 CATHOLIC CHUR 220 1 GOSPEL/CHURC 
215 2 OPRA (MAYBE I 220 2 OPRA 
215 3 EUROPEAN REN 220 3 ORCHESTRA 
215 4 GERMAN 220 4 MUSICAL PLAY 
215 5 CLASSICAL 220 5 CLASSICAL 
215 6 SPANISH GUITA 220 6 ITALIAN MUSI 
215 7 VICTORIAN CLA 220 7 SILENT MOVIE 
215 8 DARK AGES CAT 220 8 CHURCH ORGAN 
215 9 MONKS IN MOU 220 9 OLD RELIGIOUS 
215 0 CLASSIC PIANO 220 10 WALTZ TYPE M 
215 11 OPRA - CLASSIC 220 11 OLD ENGLISH M 
215 12 EUROPEAN REN 220 12 OPRA 
153 
SUBJ AMPLE LAB L UB.I EXAMPL LABEL 
221 1 CHRISTMAS MU 232 1 RELIGIOUS 
221 2 OPERA 232 2 OPERA 
221 3 OLD MID-AGE T 232 3 CLASSICAL 
221 4 OPERA 232 4 OPERA 
221 5 CLASSICAL 232 5 CLASSICAL (LOV 
221 6 CLASSICAL 232 6 CLASSICAL GUI 
221 7 ORCHESTRA - C 232 7 CLASSICAL 
221 8 MID. CENTURY 232 8 RELIGIOUS 
221 9 CHURCH MUSIC 232 9 RELIGIOUS 
221 10 CLASSICAL 232 10 CLASSICAL 
221 11 FROM PLAYS? 232 11 RELIGIOUS 
221 12 OPERA 232 12 OPERA 
229 1 CHIOR 
229 2 OPERA 
229 3 ORCRASTRA 
229 4 OPERA 
229 5 SYMPHONY OR 
229 6 SPANISH GUITA 
229 7 ORCRASTRA (SP 
229 8 ORGAN 
229 9 CHOIR 
229 10 PIANO 
229 11 OPERA 
229 12 OPERA 
231 1 1850'S ORCHES 
231 2 GERMAN OPRA 
231 3 BAROQUE 16 5 0 
231 4 GERMAN CHURC 
231 5 1850'S ORCHES 
231 6 SPANISH GUITA 
231 7 FULL ORCHESTR 
231 8 PIPE, GERMAN 
231 9 PASTORAL, 150 
231 10 PIANO, 1700'S 
231 11 CHOIR, GERMA 
231 12 BAROQUE, ITAL 
154 
APPENDIX B 
RECORDED EXAMPLES 
(Target examples indicated 
with an asterisk*) 
155 
CTXTI 
E ·mp NAWM Composer Title Vol/CDffra k Timing 
# # 
l 96 Bach Wenn. wir l 7 19 0:10 
2 69 Monteverdj Poooea l 5 8 0:10 
3 140 Weber die Frieschut~ II 3 38 0:40 
4 42 Dufay Adieux l 3 44 0:10 
5 88 Bach Nun Komm I 6 13 0:10 
6 127 Liszt Maz.epfXl rJ 2 14 0:00 
7 56 Weelkes OCare I 4 40 0:10 
8* 143 Bruckner Virga Jesse JI 4 45 0:10 
9 100 Frescobaldi Toccara I 7 34 0:10 
IO l30 Mendelssohn Midsummer's ND (12 26 0:10 
11* 76 Scarlatti Griseu:ia l 5 27 0:20 
12 45b Narvaez Mille Rewerz J 3 48 0:10 
13* 58 Attaignant Danseries I 4 44 0:10 
14 91 CorelJi Trio Sona1a 1 6 32 0: 10 
15 136 Mahler Kindertolenlied II 3 17 0:20 
16 59 Morton L'homme arme I 4 46 0:10 
17* 135 Wolf Kenn.st du Il 3 9 0:20 
18 74 Purcell Fairy Queen [ 5 22 0:10 
19 139 Meyerbeer Les Hugen.ots [( 3 29 0:00 
20* 92 VivaJru Concerto Grosso £ 7 2 0:10 
21 50 Cara lo non compro r 4 12 0:10 
22* 60 Milan Fantasia XI I 4 48 0:00 
23 83 Carissimi Jephthe I 6 2 0: lO 
24 128 Liszt Nuages gris II 2 19 0:10 
25 40 Palestrina Missa Papa Marcelli I 3 35 0:10 
26* 132 Brahms Piano Quintet II 2 34 0:00 
27 101 Gautier Gigue I 7 35 0:10 
28 129 Berlioz Symphfantas (12 20 0:20 
29* 94 Buxtehude Danker I 7 15 0:10 
30 98 Dowland Lachrymae Pavone I 7 24 0:10 
31* 38 Taverner Missa Gloria I 3 29 0:10 
32 73 Lully Armide I 5 15 0:10 
33 133 Schubert Kennst du JI 3 I O:JO 
34 61 Tye In nomine I 4 53 0:10 
35* 126 Chopin Nocturne II 2 12 0:10 
36 78 Gay BeJ?.J?.ar's Opera I 5 36 0:10 
37 125 Field Nocrurne I[ 2 11 0:10 
38* 80 Gabrieli Hodie I 5 40 0:10 
39 39 Arcadelt Missa Noe I 3 41 0:10 
40* 41 Dufay Re ·veillies vou.\ I 3 41 0:10 
156 
CTXTIB 
Ex NAWM Composer Title CDffrack Timing 
# # 
1 39 Arcadelt Missa Noe I 3 41 0:10 
2 125 Field Nocturne II 2 11 0:10 
3 78 Gay Befutar 's Opera I 5 36 0:10 
4 61 Tye In nomine I 4 53 0: l0 
5 133 Schubert Kennst du II 3 I 0:10 
6 73 Lully Armide I 5 15 0:10 
7 98 Dowland Lachrymae Pavane I 7 24 0:10 
8* 41 Dufay Resveillies vous I 3 41 0:10 
9 129 Berlioz Symphfantas II 2 20 0:20 
lO 101 Gautier GiRUe I 7 35 0:10 
11* 80 Gabneli Hodie I 5 40 0:10 
12 40 Palestrina Missa Papa Marcelli I 3 35 0:10 
13* 126 Chopin Nocturne II 2 12 0:10 
14 128 Liszt NuaReS Rris ll 2 19 0:10 
15 83 Carissimi Jephthe I 6 2 0:10 
16 50 Cara Jo non compro I 4 12 0:10 
17* 38 Taverner Misj·a Gloria I 3 29 0:10 
18 139 Meyerbeer Les HuRenot IJ3 29 0:00 
19 74 Purcell Fairy Queen I 5 22 0:10 
20* 94 Buxtehude Danket I 7 15 0:10 
21 59 Morton l'homme arme I 4 46 0:10 
22* l32 Brahms Pian.o Quintet II 2 34 0:00 
23 136 Mahler K indertotenLied II 3 17 0:20 
24 91 Corelli Trio Sonata I 6 32 0:10 
25 45b Narvaez Mille Regref"z, I 3 48 0:10 
26* 60 Milan Fan1asiaXI I 4 48 0:00 
27 130 Mendelssohn Midsummer ' ND II 2 26 0:10 
28 100 Frescobaldi Tocccua l 7 34 0: .10 
29* 92 Vivaldi Concerto Grosso 172 0:10 
30 56 Weelkcs O Care I 4 40 0:10 
31* 135 Wolf Kennst du 113 9 0:20 
32 127 Liszt Mazeppa 112 14 0:00 
33 88 Bach Nun komm I 6 13 0:10 
34 42 Dufay Adieux l 3 44 0: LO 
35* 58 Attaignant Danseries I 4 44 0:10 
36 140 Weber die Freischut7, 113 38 0:40 
37 69 Monteverdi Poppea l 5 8 0:10 
38* 76 Scarlatti Griselda I 5 27 0:20 
39 % Bach Wenn wir l 7 l.9 0: 10 
40* 143 Bruckn r Virga Jesse II 4 45 0:10 
157 
CTXT2A 
Ex NAWM Artist/ Album I Title Vol/Track Time 
# # Composer 
I Wings Wings Greatest/ 4 :30 
Junior's Farm 
2 L. Johnson Blues Masters/ VIII 5 :30 
Onzhe Wall 
3 Di re Straits Money for Nothing/ 5 :40 
Tunnel of love 
4 Can Heat Blues Masters/ VII 15 :40 
On the Road Again 
5 LSkynard (iOk/ & Plalinuml 5 :20 
You Got Thal RiRht 
6 JR Morton Classic]az:d J 9 :10 
KinR Porter Stomp 
7 R Johnson Blues Masters/ YIU 8 :10 
Crossroad Blues 
8* 143 Bruckner VirRa Jesse ll 4 45 0:10 
9 Sara Vaughn C/a,;sic Jazz/ rv 4 :20 
All Alone 
10 Band Best of/ 5 :30 
life Is a Carnival 
I I * 76 Scarlalti Griselda I 5 27 0:20 
12 Howl Wolf Blues Masters/ VIII 13 :30 
Sittin on Top ojrhe World 
13* 58 Attaignant Danseries I 4 44 0: 10 
14 Paker Classic Jazz! (It 14 :30 
Embraceable You 
15 Fitzgerald Classic Jaz:J ll 12 :IO 
You'd Be So Nice 
16 Armstrong Classic Jazz/ I 13 : 10 
Bi£ Butter & E22 Man 
17* 135 Wolf kennst du II 3 9 0:20 
18 Taylor Blues Masters/ lX 8 :30 
I'm a Woman 
19 Oliver ClassicJQ::2/ I 5 : 10 
Dippermouth Blues 
20* 92 Vivaldi Concerto Grosso r 7 2 0:10 
21 Coltrane Classic Jaw V 6 : 10 
Alabama 
22* 60 Milan Fantasia XI I 4 48 0:00 
23 Cray Blues Masters/ IX 10 : 10 
Phone Booth 
158 
24 El.vis Top / 8 Hits/ l3 :10 
RinR Around Your eek 
2~ Waters Blues Masters/ VII 8 :20 
Got m_ymojo 
26* l32 Brahms Piano Quin1et [I 2 34 0:00 
£/ Mayall Blue Masters/ vu 14 : 10 
Death of JB Lenoir 
28 Hender on Clas 'ic Jazz/ lI .1 :20 
Stampede 
29* 94 Buxtehude Danket I 7 15 0:10 
30 Domino Bel of Fats/ 8 :20 
Be my guest 
31 * 38 Taverner Mi.i;sa Gloria I 3 29 0:10 
32 LRichard 18 Greatest Hits/ 6 :10 
Heebie-Jeebies 
33 Basie Classic Jaw II 20 :10 
Dol!.Rin Around 
34 Allman Filmore East/ ll 1 :30 
Hot'Lanta 
35* 126 Chopin Nocturne II 2 12 0:10 
36 Berry R&R Rarities/ 5 :10 
Come on 
37 Beatles Rubber Soul/ 6 : 10 
The Word 
38* 80 Gabriela Hodie I 5 40 0:10 
39 St RVaughn Blue Mater I IX 13 :30 
Pride & Joy 
40* 41 Dufay Resveillies vous I 3 41 0:10 
159 
CTXT28 
Ex NAWM Artist/ Album I Title Vol/frack Time 
# # Composer 
I St R Vaughn Blues Masters/ IX 13 :30 
Pride & Joy 
2 Beatles Rubber Soul/ 6 : IO 
The Word 
3 Berry R&R Rarities/ 5 : IO 
Come on 
4 Allman Filmore East/ IT 1 :30 
Hor'Lanta 
5 Basie Classic l aw II 20 :10 
D022in Around 
6 L Richard 18 Greaiesr Hits/ 6 :10 
Heebie-Jeebies 
7 Domino Best of Faist 8 :20 
Be my guesl 
8* 41 Dufay Resveillies vous I 3 4l 0:10 
9 Henderson Classic Jaw lJ 1 :20 
Stampede 
10 Mayall Blues Masters/ vn 14 :10 
Death ofJ B Lenoir 
11* 80 Gabrielli Hodie I 5 40 0:10 
12 Waters Blues Masters/ VII 8 :20 
Got my mojo 
13* 126 Chopin Nocturne II 2 12 0:10 
14 Elvis Top 18 Hits/ 13 :10 
Rini!, around your neck 
15 Cray Blue.'i Masters/ IX 10 :10 
Phone Booth 
16 Coltrane C/,assic l ad V 6 :JO 
Alabama 
17* 38 Taverner Missa Gloria I 3 29 0: lO 
18 Oliver Classic }~ I 5 : 10 
Diooerrn.ourh Blues 
19 Taylor Blues Masters/ IX 8 :30 
I'm a Woman 
20* 94 Buxtehude Danket I 7 15 0:10 
21 Armstrong Classic la-::::/ I 13 :10 
Bil!. Butter & EJ!f! Man 
22* 132 Brahms Piano Quintet er 2 34 0:00 
23 Fitzgerald Classic Ja::;:J JI 12 : 10 
You 'd he so nice 
24 Parker Classic l a::::/ m 14 :30 
Embracahle you 
160 
25 Howl W If Blues Masters/ VHI 13 :30 
Sitt in on top of the world 
26* 60 Milan Fantas·ia XI I 4 48 0:00 
27 Band Best of! 5 :30 
Life is a carnival 
28 Sara Vaughn Clas ic l azz/ rv 4 :20 
All Alone 
29* 
30 R Johnson Blues Masters/ VIII 8 :10 
Crossroad Blues 
3 1* 135 Wolf kennsl du 11 3 9 0:20 
32 JR Morton Classi Jaw I 9 :10 
KinR Porter Stomp 
33 LSkynard Gold & Platinum! 5 :20 
You Got Thal RiRht 
34 Can Heat Blue Masters/ Vll 15 :40 
On the road aRain 
35* 58 Atta.ignant Danseries I 4 44 O:JO 
36 Di re Straits Mone , f or Nothing/ 5 :40 
Tunnel of Love 
37 L. John on Blue Masters/ vm s :30 
On the Wall 
38* 76 Scarlatti Griselda I 5 Tl 0:20 
39 Wi ng Wings Greatest/ 4 :30 
Junior's Farm 
40* 143 Bruckner Virr<a Jesse 114 45 0:10 
161 
APPENDIX C 
SUBJECT RESPONSE FORMS 
162 
Subject Profile 
Age ___  Gend r (circle M F 
Major or Department: 
Please circle: undergraduat graduat GTF 
faculty 
Have you taken any music history courses in college? YES NO 
Have you had any musical training while in college? 
No Yes, college courses Y s, private lessons 
Music hi tory Voice or instrument lessons 
Choir band, or rchestra 
If you are a faculty member, how many years have you been teaching in your 
ubject area? 1-3 mor than 
ther any other music related experience or trainin that you might have had? (a 
bri fans er i fine). 
163 
Instructions: 
You will hear a series of musical examples lasting about 20 seconds each. 
Listen for the first few seconds, and decide what kind of music you think this selection 
represents. You decide what this question means and how you will answer it. You 
may change your mind about how to answer after you have heard a few examples. 
That is perfectly alright, but you must not go back and change earlier answers. 
Write down your answer quickly, since the next example will start a few 
seconds after the end of the preceding one. You will hear the number each 6me so you 
will not get lost. If you cannot decide on an answer, mark an X and go on. Once you 
have begun, we cannot stop the tape. The whole procedure should take about 15-20 
minutes. 
This is not a "name that tune" test. You might recognize some of the music, 
but the question is: what kind of music is this. 
Please do not ask any questions aloud if you are taking the study with other 
people, since your questions might influence their thinking. If something is unclear, 
raise a hand and point to the area that is confusing. 
YOU decide what is meant by "kind of music," so do not ask for examples or 
explanations. 
A reminder: 
I am interested in your first decisions, so do not go back and change an 
answer. It is quite possible that you wi ll change your mind as you hear more 
examples, but please do not go back. 
WAIT TO TURN THE PAGE 
164 
Tape: IA I B 2A 2B 
What kind of music is th.is? 
1. 
2. 
3. 
4. 
5. 
6. 
7. 
8. 
9. 
JO. 
11. 
12. 
13. 
14. 
15. 
165 
What kind of music is thi s? 
16. 
17. 
18. 
19. 
20. 
21. 
22. 
23 . 
24. 
25. 
26. 
27. 
28. 
29. 
30. 
166 
What kind of mu ic is thi ? 
3 l. 
32. 
33. 
34. 
35. 
36. 
37. 
38. 
9. 
40. 
T hank you for participating in this study. I would appr ciat your not discussing 
it with anyone who might be a futur subj l, sin our an wers might influence 
theirs. If you have questions about the natur of the study or the musical 
el tions used, I will be happy to pro id you with the information. Thank you 
on aoain. 
167 
BIBLIOGRAPHY 
American Psychological Association. (1989). Publication 
Manual (3rd ed.). Washington, D. C.: American 
Psychological Association . 
Anderson, J. R. (1990). Cognitive psychology and its 
implications (3rd ed.). New York: W.H. Freeman . 
Aristotle (1991) . Categories. In R. E . Allen (Ed. and 
Trans .) Greek philosophy: Thales to Aristotle (3rd ed., 
pp. 285-291) 
Armstrong , s . L ., Gleitman, L ., & Gleitman , H. (1983). 
What some concepts might not be. Cognition, 13 , 263-
308. 
Ausubel, D. ( 1968 ). Educational psychology: A Cognitive 
view. New York: Holt , Reinhart & Winston . 
Barrett, J. R . (1989) . Core thinking skills in music. 
In E. Boardman , Ed. Dimensions of musical thinking. 
(pp. 45-56). Reston, VA MENC. 
Barsalou, L. (1983). Ad hoc categories. Memory and 
Cognition, 11, 211-227. 
Barsalou , L. (1989) . The instability of graded 
structure: Implications for the nature of concepts. 
In u. Neisser (Ed.), Concepts and conceptual 
development: Ecological and intellectual factors in 
categorization (pp . 101- 141) . Cambridge, UK: Cambridge 
University Press. 
Bartlett, F. C. (1932). Remembering : A Study in 
experimental and social psychology . London: Cambridge 
University Press. 
Blues Masters. (1993) . Rhino Records R2 71130 . 
168 
Booth,G. o., & Cutietta, R. (1991). The applicability of 
verbal processing strategies to recall of familiar 
songs. Journal of Research in Music Education, 39, 
121-131 . 
Borg, w., & Gall, M. (1989). Educational Research (3rd 
ed.). Longman, New York. 
Bradley, J. (1990). The role of cognitive structure in 
expertise . Unpublished doctoral dissertation, 
University of Texas, Arlington. 
Brittin, R. v. (1991). The effect of overtly 
categorizing music on preference for popular music 
styles. Journal of Research in Music Education 39, 
143- 151 . 
Brooks, L . R. (1989) Decentralized control of 
categorization: The role of prior processing episodes. 
In u. Neisser, (Ed.), Concepts and conceptual 
development : Ecological and intellectual factors in 
categorization (pp. 141-174). Cambridge, UK: Cambridge 
University Press. 
Brown, H. M. (1976). Music in the Renaissance. 
Englewood Cliffs, NJ: Prentice Hall . 
Bruner, J . (1973). Beyond the information given: 
Studies in the psychology of knowing. New York: w. W. 
Norton. 
Bruner, J . s., Goodnow, J . J., & Austin, G. A. (1957). a 
study of thinking. New York : Wiley & Sons. 
Chi, M., Feltovich , P., & Glaser, R. (1981). 
Categorization and representation of physics problems 
by experts and novices. Cognitive Science, 2, 121-152 . 
Cole, M., & Means, B. (1981). Comparative studies of how 
people think . Cambridge, MA: Harvard University Press. 
Conrad, C. E. (1971 ). An analysis of a hierarchical 
model of semantic memory organization. Unpublished 
masters thesis, University of Oregon, Eugene. 
169 
Corbett, A. T. (1975). Artificial scattering and 
semantic distance in a categorization task. 
Unpublished masters thesis, University of Oregon, 
Eugene. 
Corrigan, J . (1990). Effects of expertise on category 
knowledge : changes in basic level. Unpublished 
doctoral dissertation, University of Delaware , Newark. 
Cutietta, R. (1993). Musical elements: Who said they're 
right? Music Educators Jounal, li, 48- 53 . 
Deutsch, D. (Ed.) . (1982) . The psychology of music. New 
York : Academic Press. 
Diekhoff , G. M. (1983) . Testing through relationship 
judgments. Journal of Educational Psychology, 75, 
227- 233. 
Donington, R. (1973). A performer's guide to baroque 
music. London: Novello. 
Dowling, W. J., & Harwood, D. L. (1986). Music 
cognition. San Diego, CA: Academic Press. 
Eastlund, J. (1990) . Musical style discrimination as a 
function of selected cognitive style variables. 
Contributions to Music Education 12, 50- 66 . 
Eastlund, J. ( 1992 ). A Multidimensional scaling analysis 
of musical style . Journal of Research in Music 
Education , 40 , 204- 215. 
Edwards , R. (1988) . Transfer and performance instruction. 
In C. Fowler, (Ed . ) . The Crane Symposium: Toward an 
understa nding of the teaching and learning of music 
performance. (pp . 119-142). Potsdam, NY: Crane 
School of Music 
Einstein , A. (1947). Music in the romantic era. New 
York : Norton . 
First national assessment of musical progress . (1974 ) . 
Report No. 03 - MU- 40 . Denver , CO : National Assessment 
of Educational Progress, Education Commission of the 
United States . 
170 
Fowler, C. (Ed . ). (1988). The Crane Symposium: Toward an 
understanding of the teaching and learning of music 
performance . Potsdam, NY: Crane School of Music 
Frederick, w. C. , & Klausmeier, H. J. (1965) . The 
effects of instructions , concept complexity , method of 
presentation , and order of concepts upon a concept 
atta inment task . (Report No . BR-5- 0216-TR-3). 
Madison , WI : Wisconsin University Research and 
Development Center for Learning and Re-Education. 
(ERIC Document Reproduction Service No. ED 010-203) 
Gagne , E. (1985) . Cognitive psychology of school 
learning . Boston: Little , Brown, & Co. 
Gahart , M. (1987) . Toward an understanding of social 
schemata: expertise , level of categorization, and 
schema use . Unpublished doctoral dissertation, 
University of California , Los Angeles . 
Gardner, H. (1973) . Children ' s sensitivity to musical 
styles. Merrill-Palmer Quarterly 12., 67-77 . 
Gardner , H. (1985) . The mind ' s new science : A History of 
the cognitive revolution . New York : Basic Books. 
Garnham, A. (1985) . Psycholinguistics : Central topics . 
New York: Methuen . 
Givon, T. (1989). Mind , code, and context. Hillsdale, 
NJ: Lawrence Earlbaum Associates. 
Gregory , R. (Ed.). (1989) . Oxford companion to the mind. 
Oxford, UK: Oxford University Press . 
Grout, D. (1973) . A History of western music. New York: 
Norton . 
Haack , P. (1982). A study of high school music 
participants' stylistic preferences and identification 
abilities in music and the visual arts. Journal of 
Research in Music Education, 30, 213-220. 
Harnad, s. (1990) . Category induction and 
representation. Ins. Harnad (Ed.) , Categorical 
perception : The Groundwork of cognition (pp. 535-
555). Cambridge, UK: Cambridge University Press . 
171 
Harnad, s. (Ed.). (1990). Categorical perception: The 
groundwork of cognition. Cambridge, UK: Cambridge 
university press. 
Hintzman, D. (1986). Schema abstraction in a multiple-
trace memory model. Psychological Review 93, 411-428. 
Homa, D., Rhoads, D., & Chambliss, D. (1979). The 
evolution of conceptual structure. Journal of 
Experimental Psychology: Human Learning and Memory,~' 
11-23. 
Houston, w. B. (Ed.). (1990). Handbook of research on 
teacher education. New York: MacMillan. 
Howell, P., Cross, I., & West, R. (1985). Musical 
structure and cognition. London: Academic Press. 
Iuele, P. (1991). Representation and feedback in the 
formation of physical science concepts. Unpublished 
doctoral dissertation, Rutgers University, New 
Brunswick, NJ. 
Jonassen, D. H. (1988, January). Prototype development 
and acquisition in an interactive video concept lesson. 
Paper presented at the annual meeting of the 
Association for Educational Communications and 
Technology, New Orleans , LA. (ERIC Document 
Reproduction Service No. ED 295-650) 
Kant, I. (1929). Critique of pure reason (N. K. Smith, 
trans.). New York: St. Martin's Press. (Original work 
published 1781). 
Katsuyama, R. M., & Hoffarth, G.D. (1978). Salience and 
developmental effects upon the learning of rules. 
Biennial Southeastern Conference on Human Development, 
Atlanta, GA. (ERIC Document Reproduction Service No. 
ED 165-889) 
Keil , F . (1989) Conceptual development and category 
structure. In u. Neisser (Ed.), Concepts and 
conceptual development: Ecological and intellectual 
factors in categorization (pp.175-200). Cambridge, 
UK: Cambridge University Press. 
172 
Keil, F., & Kelly, M. (1990). Developmental changes in 
category structure. Ins. Harnad (Ed.), Categorical 
perception: The groundwork of cognition (pp. 491-
510). Cambridge, UK: Cambridge University Press. 
Kellogg, W. A. (1985). Knowledge and instantiation of 
meaning in reading: On-line studies of expertise. 
Unpublished doctoral dissertation, University of 
Oregon, Eugene. 
Kennedy, M. (1987). Inexact sciences: Professional 
education and the development of expertise. In E. 
Rothkopf, (Ed.), Review of research in education 
(Vol. 14). Washington, D.C.: American Educational 
Research Association. 
Lakoff, G. (1987). Women, fire, and dangerous things: 
What categories reveal about the mind. Chicago: 
University of Chicago Press. 
Lakoff, G. (1989) Cognitve models and prototype theory. 
In u. Neisser (Ed.), Concepts and conceptual 
development: Ecological and intellectual factors in 
categorization (pp. 63-100). Cambridge, UK: Cambridge 
University Press. 
Langer, s. K. (1975). Philosophy in a new key: A study 
in the symbolism of reason, rite, and art. Cambridge, 
MA: Harvard University Press. 
Lao Tzu (1973). Tao Te Ching (Ch'u Ta-Kao, trans.). 
New York: Samuel Weiser. 
Laughlin, P.R. (1973). Focusing strategy in concept 
attainment as a function of instructions and task 
complexity. Journal of Experimental Psychology, 98, 
320-327. 
Lurigio, A. J. (1983). Schematic approach to 
understanding probation officers' judgments of 
criminals. Unpublished doctoral dissertation, Loyola 
University, Chicago. 
173 
McCauley, R. (1989). The role of theories in a theory of 
concepts. In U. Neisser (Ed.), Concepts and 
conceptual development: Ecological and intellectual 
factors in categorization (pp. 288-310). Cambridge, 
UK: Cambridge University Press. 
McKinney , C. W. (1987). Effects of best examples, 
critical attributes, definitions, and practice on 
concept acquisition and prototype formation. Journal 
of Social Studies research, 11, 1-14. 
Medin, D., Altom, M., Edelson, S., & Freko, D. (1982). 
Correlated symptoms and simulated medical 
classification . Journal of Experimental Psychology: 
Learning, Memory, and Cognition,~, 37-50. 
Medin, D., & Barsalou, L. (1990). Categorization process 
and categorical perception. Ins . Harnad (Ed.), 
Categorical perception: The groundwork of cognition 
(pp. 455-490). Cambridge, UK: Cambridge University 
Press. 
Medin, D., & Wattenmaker, W. (1989). Category 
cohesiveness, theories, and cognitive archeology. In 
u. Neisser (Ed.), Concepts and conceptual development: 
Ecological and intellectual factors in categorization 
(pp. 25-62) . Cambridge, UK: Cambridge University 
Press. 
Merrill, M. D., & Tennyson, R. D. (1977). Teaching 
concepts: An instructional design guide. Englewood 
Cliffs, NJ: Educational Technology Publications. 
Merrill, M. D., & Tennyson, R. D. (1978). Concept 
classification and classification errors as a function 
of relationships between examples and nonexamples. 
Improving Human Performance Quarterly, 1, 351-364. 
Murphy, G., & Wright, J. (1984). Changes in conceptual 
structure with expertise: differences in real-world 
experts and novices . Journal of Experimental 
Psychology: Learning, Memory and Cognition , lQ, 144-
155. 
174 
Neisser, u. (1989). From direct perception to conceptual 
structure. In u. Neisser (Ed.), Concepts and 
conceptual development: Ecological and intellectual 
factors in categorization (pp. 1-10). Cambridge, UK: 
Cambridge University Press. 
Neisser, U. (Ed.). (1989). Concepts and conceptual 
development: Ecological and intellectual factors in 
categorization. Cambridge, UK: Cambridge University 
Press. 
Newell, A. (1973). Production systems: Models of control 
structures. New York: Academic press. 
Norton Anthology of Western Music. (1988). (Compact Disk 
No. AS 20523). New York: Norton. 
Palisca, C. (1981). Baroque music. Englewood Cliffs, 
NJ: Prentice-Hall. 
Parker, I. V. (1990). The reliability, sensitivity, and 
criterion-related validity of concept comparisons and 
concept mapping for assessing reading comprehension. 
Unpublished doctoral dissertation, University of 
Oregon, Eugene. 
Plato (1991). Parmenides. In R. E. Allen (Ed. and 
Trans.) Greek philosophy: Thales to Aristotle (3rd ed. 
p. 257-269). New York: Macmillan. 
Reese, G. (1940). Music in the middle ages. NY: Norton 
Reimer, B. (1994, July). A profile of the 1992 NSSE 
Yearbook: The Arts, education, and aesthetic knowing. 
Paper presented at the meeting of the International 
Society of Music Educators, Tampa, FL. 
Rosch, E. (1973). Natural categories. Cognitive 
Psychology, i, 328-350 . 
Rosch, E. H., & Lloyd, B. (1978). Cognition and 
categorization. Hillsdale, NJ: Earlbaum & Assoc. 
Rosch, E. H., Mervis, c. B., Gray, W. D., Johnson, D. M., 
& Boyes-Braem, P. (1976). Basic objects in natural 
categories. Cognitive psychology,~, 382-439. 
175 
Rosch, E. H., Mervis, C. B., Gray, w. D., Johnson, D. M., 
& Boyes-Braem, P. (1976). Basic objects in natural 
categories. Cognitive psychology,~, 382-439. 
Rosch, E. H., Mervis, C. B., Gray, w., Johnson, D., & 
Boyes-Braem, P. (1975). Basic objects in natural 
categories. (Working paper no. 43). Berkeley, CA: 
California University, Language Behavior Research 
Laboratory. (ERIC Document Reproduction Service No. ED 
142-051) 
Rothkopf, E. (1987). Review of research in education 
(Vols. 14-15). Washington, D.C.: American Educational 
Research Association. 
Russell, B. (1972). A history of western philosophy. 
New York: Simon & Schuster. 
Schmidt, H., & Boshuizen, H. (1990, April). Effects of 
activation of prior knowledge on · the recall of a 
clinical case. Paper presented at the annual 
conference of the sociation, Boston, MA. 
Seaton, D. (1991). Ideas and styles in western musical 
tradition. Mt. View, CA: Mayfield. 
Silverman, D. (1990). The effect of category structure 
on judgments of similarity. Unpublished doctoral 
dissertation, University of New Mexico, Santa Fe. 
Skinner, B. F. (1957). Verbal behavior. New York: 
Appleton-Century-Crofts. 
Stepich, D. (1991). From novice to expert: Implications 
for instructional design. Performance and Instruction, 
30, 13-17. 
Steve, M. H., & Tennyson, R. D. (1974, April). Prior 
memorization of definitions, examples and nonexamples 
with conjunctive and disjunctive concept learning 
tasks. Paper presented at the annual meeting of the 
American Educational Research Association, Chicago. 
(ERIC Document Reproduction Service No. ED 091-430) 
176 
Taebel, D. K. (1971). The effect of various 
instructional modes on children's performance of music 
concept tasks. (From Dissertation Abstracts 
International, 33, 6957A) 
Tait, M. and Haack, P. (1984). Principles and processes 
of music education: New perspectives. New York: 
Teachers College Press. 
Tanaka, J. (1989). Object categorization and expertise. 
Unpublished doctoral dissertation, University of 
Oregon, Eugene. 
Tanaka, J. & Taylor, M. (1991). Object categories and 
expertise: Is the basic level in the eye of the 
beholder? Cognitive Psychology 23, 457-482. 
Taylor, J. (1989). Linguistic categorization: Prototypes 
in linguistic theory. Oxford UK: Clarendon. 
Tennyson, R. D. (1973). The effect of negative instances 
in concept acquisition using a verbal learning task. 
Journal of Educational Psychology, 64, 247-260. 
Trabasso, T. R. (1963). Stimulus emphasis and all-or-
none learning in concept identification. Journal of 
Experimental Psychology, 65, 398-406. 
Tversky, A., & Gati, I. (1978). Studies of similarity. 
In E. H. Rosch, & B. Lloyd, Cognition and 
categorization. Hillsdale, NJ: Earlbaum. 
Tversky, A., & Kahneman, D. (1973). Availability: A 
Heuristic for judging frequency and probability. 
Cognitive Psychology,~, 207-232. 
Vessey, I. (1988). Expert-novice knowledge organization: 
An empirical investigation using computer program 
recall. Behavior and Information Technology, l, 153-
171. 
Vickers, J. (1986). Expert-novice differences in 
knowledge structures of action. Ottowa: National 
Library of Canada. 
177 
Whorf, B. (1956). Language, Thought, and Reality: 
Selected writings of Benjamin Lee Whorf, ed. J.B. 
Carroll. Cambridge, MA: MIT Press. 
Wittgenstein, L. (1958). Philosophical investigations. 
New York: Macmillan.