MUSIC COMMUNICATION RESEARCH:
THE GONNOTATIVE DIMENSIONS OF MUSIC MEANING

Thesis for the Degree of Ph. D.
MICHIGAN STATE UNIVERSITY
EARL MARSHALL PALLETT _
1967

THESIS

 

TTTTTTTTTTTTTT TTT

 

 

TITTTTTI'I/ I ‘ * t

00809 2540 I R Y
MILZI State
Ulll‘v'x': Slt’

r..—

This is to certify that the

thesis entitled

MUS IC COMMUNICATION RESEARCH:

THE CONNOTATIVE DIMENSIONS OF MUSIC MEANING

presented by .5

Earl Marshall Pallett

has been accepted tbwards fulfillment
of the requirements for

Ph . D 0 degree in communication

(/2 (rk 7% L) [7‘\Ll-tt‘ (Iii/y! _/ (A :1 (IL-Q
Major professor

 

Date February 2“, 1967

 

 

(F

   

MAL ~4 ISGBE Iv L

JAN-39*) fi- wentg
/9L0W

WWv-i’ fifth

W*‘$IW

. i 5‘
pH LI) U
\ 8L}—

    

57046

ABSTRACT

MUSIC COMMUNICATION RESEARCH:
THE CONNOTATIVE DIMENSIONS OF MUSIC MEANING

by Earl Marshall Pallett

Little is known about the meaning of music. To put it another way,
it is difficult to describe verbally the various components of music
meaning and to explain or to predict the occurrence of certain components
when listening to music. Both sociocultural and scientific conditions
are such at the present time to warrant an attempt at a verbal description
of the meaning of music.

Sociocultural conditions are: l) availability of many different
sstyles of art and music in particular. Many new styles are complex
combinations of parts of traditional styles. Thus, more understanding
of the basic aspects of music meaning (Western music of last four
centuries) is necessary in order to more accurately and precisely
interpret the Specific meaning of different styles of music. 2) Not
(enough of an attempt is made to consider the nature of music referents,
whether they be abstract properties of many observable objects and events
in the environment or unobservable emotional events. 3) Music education
in public schools and universities needs improvement. They do not
make sufficient use of verbal descriptive language in relation to music.

Several scientific theories and methodological techniques are:

1) Interest in communication theories, such as human c0gnitive processes,

Earl Marshall Pallett

the most general connotative meaning of human languages, and the influence
of verbal language on various cognitive processes. 2) New methodological
techniques to analyze multivariate data.

The purpose of this study is to describe the connotative meaning
of music. That is, the underlying dimensions of music meaning that are
common to most people and most music. Also, a description of Specific
music pattern - connotative meaning associations will be begun.

Connotative meaning is to be defined generally in terms of verbal
language. More Specifically in terms of bi-polar adjective scales and
finally as dimensions made up of the scales.

The task involves first defining the music code as clearly as possible
in terms of the range of available elements and relations and music
Styles that have been constructed from it. Then to select a set of music
patterns which will elicit a broad range of connotative meaning.

Second, a survey of theories about basic emotions, connotation,
and more Specific theories of the meaning of non-verbal codes and music
lead to the development of hypotheses about the dimensional nature of
music connotation in terms of which bipolar adjective scale selections
Tare made.

In this research, 18 melodic patterns and 26 scales were selected.
These were presented to 90 music students in June, 1966.

The methodological technique selected as most appropriate to analyze
the data is factor analysis. It was applied to a correlation matrix
of adjective scales to generate a set of independent factors. From

these factors, inferences were made about psychological dimensions of

Earl Marshall Pallett

connotation.

The results were rather clear. Four relatively independent
climensions appeared.
.l) aesthetic evaluation: beautiful-ugly, graceful-awkward, etc.
I?) mood-emotion: happy-sad, friendly-ominous, etc.
13) stability-tautness: deliberate-impulsive, stiff—elastic, etc.

1+) dynamism: strong-weak, active-passive, etc.

MUSIC COMMUNICATION RESEARCH:

THE CONNOTATIVE DIMENSIONS OF MUSIC MEANING

By

Earl Marshall Pallett

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Communication

1967

GDCopyright by
Earl Marshall Pallett

1967

ACKNOWLEDGMENTS

The author wishes to acknowledge the assistance received in
the preparation of this thesis from the Department of Communication
and the Department of Music at Michigan State University. More
Specifically, Dr. Bettinghaus, Dr. Harrison, Dr. Kumata, and Dr. Unkefer
each offered assistance of a unique nature. Also, thanks is given for

the patience and skill of Shirley Sherman in typing the final manuscript.

ii

TABLE OF

CHAPTER
I INTRODUCTION . . . . . . . .
Music as Communication . .
Music Communication Today

Music Meaning . . . . . .

II RELATED THEORY AND RESEARCH
Introduction . . . . . . .
Syntactic Description . .
Semantic Description . . .

III MESSAGE STRUCTURE . . . . .

Introduction . . . . . . .

CONTENTS

Iv’

VI

VII

Perceptually Relevant Structure .
Connotatively Relevant Structure . .
Melodic Component Structure . . . .

RESPONSE STRUCTURE O O O O O O O O O 0

Introduction . . . . . . . . . . . .
General Code for Theories of Emotion

and

Connotation

Theories of the Meaning of Non-verbal Code . . . . .

Description of the Meaning of Music

Hypotheses about Dimension of Music Connotation

and Scale Selection . . . . . . .

METHODOLOGY AND PROCEDURE . . . . . .

Introduction . . . . . . . . . . . .
Procedure . . . . . . . . . . . .
Methods of Analysis . . . . . . . .

RESULTS 0 O O O O O O O O O O O O O O

The Dimensions of Melodic Connotation

Melodic Pattern Connotations . . . .

DISCUSSION 0 O O O O O O O O O O O O 0

Summary of Findings . . . . . . . .
Future Research Needs . . . . . . .

BIBLIOGRAPHY . O O O O O O O O O O O O O O O O

APPENDICES O O O O O O O O O O O O O O O 0

iii

Page

10
19

3M
3M
3M
'43
an
58
58
59
61+
66
72
76
76
76
79
83

83
95

113

113
116

127
133

TABLE

10

LIST OF TABLES

Hypothesis About Music Connotation Dimensions
Aesthetic Evaluation . . . . . . . . . . .
MOOd "' emotion o o o o o o o o o o o o o o 0

Stability - tautness o o o o o o o o o o o 0

Factor - Structure for Each Level . . . . .
Overall Relevancy and Direction Consensus . .
Melodic Pattern Meanings . . . . . . . . . .

Pattern - Scale Associations . . . . . . . .

iv

egg:

73

86

88

89

90

90

92

96
103-105

107

LIST OF FIGURES

FIGURE Page
1 MUSiC Space 0 O O O O O O O O O O O O O O O O O O O 1+6
2 Overall Data Block . . . . . . . . . . . . . . . . 80

CHAPTER I

INTRODUCTION

Music as Communication

 

Communication is the most typical and prevelant of all human
activity. A typical communication model (8) includes a source, message,
channel and receiver. The communication process involves purposive
source transmission and receiver interpretation of messages.

Generally, the source and receiver can be a society, institution,
group, role or individual. In music, many sources and receivers have
labels such as composer, performer, critic, listener, etc.

Both the source and receiver have certain important capabilities
to facilitate communication: 1) skill in encoding or decoding the Sign
system code; 2) attitudes toward each other and the message; 3) knowledge
about the message subject matter; M) similar socio-cultural heritage.

The channel can refer to several things in relation to messages.

It can be the message "vehicle", such as sound waves or, in terms of new
technological developments, radio, television, etc. Besides this it can be
the vehicle carrier, such as air. Finally, it can be the sensory decoding
channels such as hearing.

Messages are the means by which human communication is attained.

A message has several different aspects: 1) code -- system of signs and
relations between them. The study of language codes is called
"syntactics" (57). This includes the identification of Sign elements,

such as phonemes and morphemes, and the relations between Signs, such as

morphology and syntax(grammar).

2) content -- meaning or subject matter of the message. The study
of language content is called "semantics”. Relations between Signs and
referents outside the language are described. It also can include a
description of the internal relations between the code referents or
meanings.

3) treatment -- method of message construction. The study of
language treatment is called "style". Regularities in the selection and
combination of message code elements and relations of which the source has
a choice is described.

Thus, message code and style description and analysis does not
directly unearth message meaning. We must define the communication
purpose in terms of the source or receiver and not in terms of a message
description and analysis in isolation from a communication situation.

The major task for a communication researcher is to obtain an
analysis of the meanings intended.by the source and interpreted by the
receivers plus determining who is the intended audience. This task can be
discussed usefully in terms of two sets of concepts: Specificity of intent
(23) and instrumental, consummatory, expressive and coping behavior (8,
ch. 1).

Generally, intentions can vary from highly Specific to very non-
specific in terms of audience and meaning. Expressive refers to very
non-Specific source oriented intentions, such as a yell when kicked.

Here the communicator is more concerned with expressing himself than with
possible effects on certain others. "Coping" refers to Specific intentions.

Highly Specific or "planned" communications are usually receiver oriented.

Any communication Situation_may be some of both. The music
communication Situation primarily is rather non-Specific in terms of
both audience and meaning Specificity. Of course, commercial artists
are more Specific in intention than are fine artists. That is, the
commercial artist may definitely want the receiver to buy the music but he
is not too sure about the nature of the meaning. The fine artist uses music
as an expression guided by internal, personal patterns of understanding
and feeling. Thus, here rather low Specificity of intent exists about
communicating any particular effect to certain others.
Next, "instrumental" refers to using the message as a cue or tool
to generate other behavior which will be rewarded. "Consumatory" refers
to the interpretation of the message itself offering immediate reward.
Again, any communication situation may contain some of both. Music
communication Situations are primarily consummatory, where the effect may be
dancing or an emotional reaction, etc.
Thus, it appears that music communication is not a very accurate
and precise affair. A more detailed description of music communication

in society might clarify the problem.

Music Communication Today_

 

In western society the church initially controlled the codification
of music. Consequently, music was limited in its social purposes to
maintaining the stability of religious work and other group rituals.
However, during the Renaissance the individual began to free himself from
antiquated political, economic and social restrictions. Music became a

means to express more personal artistic needs. By the time of the

industrial revolution, with its increasing leisure time for the emerging
urban, middle class music became "mass entertainment."

Generally, the social structure influenced the development of
different kinds of music and different listener roles. The most traditional
division is between classical or serious music for the elite vs. popular
music for the masses. Today a trend exists in Ameridnlpopular music to
assimilate much of classical music, with various modifications. Con-
sequently, the division between these categories tends to dissolve as
s tyles mature.

Nowadays, there is vivid evidence of a tremendous increaszin the
amount of cultural information that is accumulating and available for
human use in advanced societies. This situation is due to the growth
of scientific knowledge and technological developments such as mass
communication media. Music from many centuries and many countries is
being made available to more and more people. Many new art styles are
appearing, some of which are combinations of older Styles, such as "pop"
art. This situation is caused equally both by the availability of
examples of many past art styles and by the growing demand for art for
all the people. After three centuries of being class-bound and nation-
bound, art has again become a property of all mankind (25). For the
first time in history, music of one style or another has penetrated the
everyday life of all members of society. As Jaqaue Barzun states:

Music nowadays, whether we like it or not,

is interwoven with the texture of our lives from
morning till night. (6, p. 13)

The tremendous expansion of cultural life in America has

influenced the emergence of the cultural consumer as part of a "comfort"

class (81). But Kaplan (41) feels there is an urgent need to improve

the nature and role of cultural or aesthetic knowledge. Both artists and
consumers need more accurate and precise knowledge of the meaning of the

arts.

With the Spread of art and entertainment via mass communication has
come attempts by the media to instruct, criticize and evaluate these
messages for the consumer. But, according to Deihl:

....compounding the confusion has been the
birth of the worst thing ever to happen to

the artist and his confused public -- the
newSpaper critic. These critics, along with
sundry art historians, well meaning culture
hounds, philanthropists, quiz Show hosts, has
created a garbled and profuse vocabulary and a
distorted attitude toward art from which it may
never recover. (20, p. 9)

Art education, of course, is given in the public schools and
universities. This could be the best way for most people to obtain a
basic, general understanding of the arts. But these courses are presently
inadequate to prepare a person to deal intelligently with the vast amount
of information about the arts in magazines, newspapers and television and
with the many available art styles.

There seems to be a need, in particular, to improve music education
in America. A movement is growing presently (#8)(73) to broaden education
in general music courses in secondary schools and university music

appreciation courses emphasizing primarily a performance conservatory orien-

tation and only classical music to include emphasis on the listener-

consumer-aesthetic orientation and many music Styles.

A major difficulty with art education, and eSpecially music
"education" courses could be defined as a communication problem. Few
teachers can lecture about the arts in a persuasive manner. Most music
teachers go to the opposite extreme of the critic and use only dry,
technical language. Leonard Bernstein on his TV series is a rare example
of someone who is able to intermingle just the right kind and amount of
verbal information to help involve the listener in the work of art.

More Specifically, the problem is: what kind of verbal information
is most significant? This leads to the underlying question about the
meaning of music. A need exists for scientific research to empirically
establish new knowledge about music communication.

One contribution can be made by a scientific study of music
meaning in terms of a verbal description. This implies that the focus
will primarily be on the receiver. The resulting description will be
more understandable and of immediate use to educators and critics who can

then incorporate it into programs of instruction for all music consumers.

Music Meaning_

 

Meaning commonly refers to the idea, response or interpretation of
a Sign. Thus it intervenes as the psychological conception of the ex-
ternal or internal referents, such as environmental objects or events,
verbal behavior and emotional reactions.

According to Berlo (8, ch. 8), it is useful to distinguish between

three kinds of meaning: structural, denotative and connotative.

First, a language code contains many signs which exhibit

regularities of usage, such as word formation and sentence formation.

When messages are constructed many more stylistic regularities are

added. Structural meaning refers to the comprehension of these regular
relationships between signs and other Signs. For instance, one aspect of
structural meaning would be the comprehension of redundancy or the ability
to predict one sign from knowledge of another.

Second, denotative meaning refers primarily to qualities of objects
and events outside the language named by language signs. Further, the
best denotative meanings imply relationships to rather Specific and
concrete objects and events that are observable, such as "John F. Kennedy."
Signs such as "teacher" are more abstract and relate to large sets of
objects and also to large sets of possible behaviors. This Sign has
denotative meaning to the extent that there is agreement about the
referent objects and properties between Sources and receivers.

Third, connotative meaning refers primarily to personal eXperienceS
and less to qualities of external objects and events. Social experiences
such as attitudes and emotions are communicated. When connotative signs
such as "majestic" refer to the environment they usually are related to
large sets of objects and in terms of abstract properties. Thus,
because connotative communication about both persons and the environment
is so difficult to trace to referents, traditionally it is assumed
that less agreement exists among sources and receivers as to the referents.

What is the meaning of music? Or, of what is music a Sign? First,

an even more basic question is asked by Arron Copland:

IS there meaning to music? My answer would be
"yes". And can you state in so many words what
the meaning is? My answer to that would be "no".
Therein lies the difficulty. (18, p. 19)

Music communication is unique in that the message consists of
abstract temporal curved lines with little Specification by a source of
the nature of the referents.

This may lead to the conclusion that: a) no sign-referent relations
exist; b) only structural meaning is available; c) music is unique in
being a set of sound patterns which may have any possible object or event
as a referent.

This situation should not be interpreted as a reason fOr concluding

that a communication approach is fruitless. On the contrary, this is

just what is most needed.

CHAPTER II

RELATED THEORY AND RESEARCH

Introduction

 

To guide this study theory and research from the behavioral
sciences will be drawn upon. Since all languages have a two sided
nature (Sign system and significant system), the division between syntactic
and semantic descriptions will be most important in this chapter.

The syntactic description is usually developed prior to the semantic
description. This is because a) the Sign system is directly observable
and b) the description of its structure should be developed as independently
as possible of knowledge of the significant system to avoid bias. In
this chapter, linguistic syntactic description is discussed followed by
a discussion of music syntactic descriptive work.

Semantic description is more difficult because the significants
are more complex and meaning is unobservable. An observable semantic study
is provided by using verbal language. A new approach to the study of
connotative meaning will be adopted. Theories of cognition and language
are interrelated, culminating in a model of cognitive connotation. Based
on verbal data, inferences can be made about the nature of meaning.
Research evidence suggests that connotation has several dimensions.

Finally, comments are made in this chapter about what this research

suggests about the nature of music connotation. A statement of Specific

lO

objectives for this study closes the chapter.

Syntactic Description

 

It may be useful to first consider the approach to description in
modern science. In scientific activity during the past fifty years or so
a trend has appeared in the methods of describing phenomena. According
to L. L. Whyte (87, p. 20), "science" more apprOpriately refers to the
study of ordered complexity rather than to the study of simplicity. The
use of concepts such as "organism" and "system" indicate this.

Any object or event can be conceptualized as a System. "System"
refers to an interacting set of components, where each component is
described in detail in terms of its structure.

The modern approach to the notion of structure includes traditional
analytical "atomism" and intuitive "form-gestalt" approaches integrated
within a new conception. Rather than emphasizing properties of basic
elements in isolation (atomism) or on the qualities of phenomena as a
whole (gestalt), theoretical emphasis is on overall structure.

Generally, all phenomena are characterized as consisting of many
hierarchically organized levels of structure. Each level is defined
in terms of a set of elements and relations. Elements have a potential
for a set of relational or collective properties, and certain of these
appear at each level to produce "functional wholes". In this study,
relations will include either relational property values or statistical
correlations. The present notion of structure implies an emphasis on

the relations or arrangements between elements, within and between levels.

11

Generally, relations are of an interdependent nature rather than one-way
dependence.

Basically, this research will be descriptive in substance. That
is, attention will be on a "static" description of stable elements and
relations holding among them. Later, these descriptive statements can
be used for eXplanatory and predictive purposes within the framework

of dynamic models and causal theory.

Linguistic Description

 

The most well-developed syntactic descriptive work has taken place
in linguistics. Linguistics has developed descriptive elements and
relations which may be basic to all language descriptions. Thus, it will
be profitable to discuss linguistics at some length.

The science of linguistics is divided into descriptive (a view of
all the elements and relations available to a set of language users),
historical (changes over time in terms of groups of language users) and
comparative (changes across Space in terms of groups of language users).
The present consideration Shall only be on the descriptive branch. Further,
language only as sound and not as written symbol will be of interest.

Some of the most general aSpects of the method of linguistic
syntactic description should be mentioned. Traditionally, a taxonomic,
unordered category approach has been used, based on same-different
decisions. It is also deterministic rather than probabilistic because
the goal is to Specify the boundaries of all possible categories available
for use, independent of any particular user communication behavior.

That is, linguistics offers a picture of the elements and relations from

which any receiver selects in order to generate messages. The overall
structure consists of a hierarchically organized set of levels.

The bulk of syntactic description is on the lowest levels. Here
the sets of elements are of a maneagable Size, less dependent on semantic
variation and all elements are used with sufficient frequency so that
the actual process of data collection and descriptive work is most
productive.

Generally, the descriptive task for the linguist is one of reduction,
going from the near infinity of encoded sounds used in language behavior
to a small number which have a few common properties. Element and
relation classes are identified on each level. Two important levels of
structure must be distinguished: the lowest level includes element classes
which have no semantic meaning in isolation (phonemes), while the next
higher level element classes (morphemes, or more broadly and loosely the
vocabulary of word) do have meaning in isolation.

Given the specification of these two levels of elements, a grammar
of the language is developed. This includes classes of relations between
elements, such as morphOIOgy and syntax. That is, a structural description
of all permissibly sequences of elements is given. From this it is
possible to distinguish between prOperly and imprOperly formed sequences,
such as correct and incorrect sentences.

There are many different verbal languages. A goal of linguistics
is to describe the category system (levels of elements and relations)
from which all verbal languages draw upon to some degree. Each language

is usually distinguished in terms of nationality or country divisions.

13

Whatever elements and relations are used by most all language communities
can be considered as the "universals" of verbal language Sign system.

According to Miller (53, p. llg), the notion of verbal "style"
is used in relation to differences that exist between people who have
been exposed to nearly the same culture and language background. In
' general, "style" refers to regularities in language user selection
behaviors where choice exists. The description of styles can be non-
quantitative or quantitative. Probabilistic analysis is used. Both the
elements and relations identified in work on phonology, vocabulary and
_ grammar plus new elements and relations identified from studies of com-
position practices are used.

For instance, "dialect" refers to a subgroup of people who use the
sat: subset of phonemes. "Ideolect" refers to each individual difference
in phoneme selection. However, most style analysis uses elements associated
with significant variation in semantic meaning.

The most frequent element used is the word. In the area of content
analysis (68), studies are made of pr0paganda, tabulating the use of
different words which have the most significant intended meaning. Some-
times, the phrase or sentence unit is used. Also, type/token ratios are
tabulated, where the interest is in how much a person repeats (token)
an element (type). A related variable sometimes studied is that of
vocabulary size variation across individuals. Sometimes the syllable unit
is used in order to study the length and complexity of words selected.
Similarly, the sentence length in terms of number of words is studied.

In the area of Readability (an), word length and sentence length along

1”

with personal words and personal sentences make up the four variables
used in the Flesch Reading Ease formula (26). Besides all the above
elements, sometimes differences in punctuation, such as amount of usage
of periods or commas are studied.

According to Miner (53, p. 128), few linguistic studies of style
have used more than a Single variable at a time. He mentions only one
study where the verbal style of two individuals were measured and described
primarily with the word, phrase, clause and sentence elements, plus some
syntactical relations and various other variables, such as tempo or
number of words per minute.

In conclusion, many different descriptive elements have been used
as style variables and only a few of them have been mentioned here. Most
of these variables had only a few different values and usually were
unordered category systems. It can be concluded that there are many ways
to study verbal style. Perhaps a more relevant conclusion in terms of the
needs of the present study is that with verbal language there are many
ways to define variables which are significantly related to semantic meaning.
How is this similar to work in music syntactics and how can it be a guide

to help accomplish the present task?

Music Description

 

First, other non-verbal language description which potentially
may be more relevant than linguistics to some features of music will not
be considered. (Examples are Kinesics (9), painting (no), facial

expression (3n), etc.) Primarily, they do not yet have as well deveIOped

15

a syntactic description as does music itself. Some potentially relevant
non-verbal language work will be in the development of classification
systems of complex abstract forms.

Some aSpects of the methods of music syntactics are similar to
those of linguistic syntactics. A taxonomic approach with both unordered
and ordered element categories is used. It is deterministic in relation
to establishing the overall picture of possible elements and relations
from which to choose. Also, music structure consists of a hierarchically
organized set of levels. Finally, like verbal language music features
discrete elements and sequential organization is dominant.

However, there are some significant uniquenesses to music research.
Most important, the central task in musicology is not the description
of overall structure of music in terms of permissible elements and
relations, but is the historical study of music style (35). Thus, music
studies never are very far removed from particular sources, in this case
composers. This is probably characteristic of languages which serve
primarily an "art" function. It is also appropriate to the methodology
of humanistic scholarship, that of description of the authenticated
document. Further, musicological studies are usually of written, graphic
documents rather than of sounded documents. Within this context,
communication questions about the nature of meaning are not studied because
they require empirical testing of receiver reSponses. Linguistics is much
more closely linked to psychology and other sciences than is musicology.
Consequently, little work is done in musicology to improve methods of

teaching courses in music literature and appreciation. The discipline of

16

Ethnomusicology (sometimes referred to as comparative musicology) is
closely related to Anthropology and therefore does include both humanistic
and scientific activity.

With reSpect to the development of a broad description elements
and relations of music structure musicology has depended upon the work
done in the sub-discipline of music theory. Within music theory has been
deve10ped many basic structural features which have not been questioned for
many years. A brief description of the "pre-compositional" organization
will further clarify why much of music research is that of style analysis.

The element set identified on the lowest level is very large and
much larger than that which makes up the phoneme set in verbal language.
This means that a large number of levels of structure can be identified.
Also, music usually features more than a single sequence of tones. Thus,
besides sequential structure the description must encompass a complex,
simultaneous (vertical) structure.

No clear cut distinction is made that is analogous to the non-
meaning level (phoneme) and meaning level (morpheme) and so no distinction
is also made between morphology and syntax. Therefore, the linguistic
concepts of phonology and grammar do not apply very accurately at the present
time. (Further empirical research may decide otherwise, if better notions
of music meaning are developed.) It turns out in music that rather than
just deveIOp one set of meaning elements ("vocabulary") and relations
("grammar") from the lowest level element set ("phonology"), as was done
in verbal language, several independent element sets and relation sets have
deveIOped from three subsets from within the lowest level set: 1) pitch has

generated harmony, counterpoint and melody; 2) duration has generated rhythm;

17

3) timbre has generated orchestration.

The nature of each of these components of organization shall be
described in Chapter III. In general, these components are relevant to
different levels throughout the entire range of levels, as opposed to
having them concentrated on the lowest levels as in verbal language.
Actually, even the notion of levels is not clearly specified in music.
Elaboration of this will also appear in Chapter III.

It must be made clear that each component with its elements and
relations also is the location of the development of restrictions
(permissible and non-permissible sequences). However, only the harmony
and rhythm components feature an appreciable number of restrictions.
Further, in music within a Single component there may be several
principles of practice, such as tonal or atonal harmony.. However, as is
characteristic of an art, a composer does not have to pay attention to
any of the existing organization. (At an extreme, all of any piece of
music is stylistic.) In sum, only a few "pre-compositional" constraints
appear on many levels, so the constraints are never very "oppressive."

Given that music is an abstract art form having primarily
connotative meaning, it is necessary to permit much freedom of selection
on all levels and components of structure. This compensates for there
being "less" depth or different kinds of semantic meaning for any given
music pattern than exists fer any given verbal pattern.

All of these components along with the lowest level element set

clefine the nature of the music Sign system that is available to a composer.

18

Music styles are defined by which and how many components are
the object of selection and the uniqueness of their selection behavior,
which consequently implies that "compositional" constraints are added to
each component by each style. Given that there are few "pre-compositional"
constraints and many components with many elements and relations, many
different styles can be developed. Thus, it can be seen why much of
musicology is style analysis.

Briefly, style divisions have been in terms of primitive - complex
and Eastern - Western civilizations. A further breakdown by
Palisca (66)is primitive music, folk music, European, Asiatic, Oceanic,
African and North and South American music. Within the European schools
or periods of style, such as romantic and classic, some universals in
selection behavior have existed: pitch scales, triadic chords, fixed
tempo, symmetrical measure groupings, etc. During the 20th century,
most older styles are used plus many new styles have been developed.

Some comments Should be made at this point about the use of
musiCOIOgical research to meet the needs of the present study. Generally,
music style research has placed no Special emphasis on the study of
those variables which have the most semantic meaning as was done in
linguistics. Also, no new variables have been added, probably because of
the presence of so many variables within the many components. Due to the
"humanistic" bias, elements and relations, such as tonality, chord
progressions, motives, themes, and form sections, or overall features such
as loudness, tempo, meter, etc., have been studied, without consideration

for differences in semantic meaning. Some considerations have been made

in psychology (21) about the semantic significance of these variables,

19

but further tests are needed. For instance, the phoneme-morpheme levels
distinction needs to be tested.

Generally, Chapter III will demonstrate that the best approach in
this study is to first slice into overall music structure in terms of
lowest level elements and components.: Some components, such as melody
overlap other components considerably? Only one component, that of melody,
will be studied. It will be Shown to be sufficient for two reasons:

1) within this component, a broad Slice across many music styles can be
made. Thus, a rather broad range of meaning appears to be attached,
possibly at least a small portion of all aspects of music meaning.

2) It permits a more careful selection of certain familiar, relatively

well-defined variables on a hierarchically organized set of levels.

Semantic Description

 

Music Sign-Significant relations

 

An important distinction between verbal and non-verbal languages
is related to the structural nature of the sign-Significant relation.
Generally, relevant structural relationscnn either be iconic (or resemble)
or be arbitrary and not depend upon structural resemblances. Most
language sign-significant relations include some of both. However, non-
verbal languages, including music, primarily feature iconic relations.

Many different styles of music exist, from random music to more
determinate program music. Theorists from Aesthetics, Psychology and
Anthr0pology have ideas about the semantic meaning of music. "Causal"
theories about the influence of music on various responses must wait

until a more complete description of Sign-significant relations is

2O

accomplished. Theories which primarily describe iconic relations shall

be considered first.

Iconic Relations

Formalists (69, part 10) stress only music sign-sign (as
significant) relations, or structural meaning. Different formalists focus
on different aSpects of music structural relations. Although formalists
do not refer explicitly to the notion of iconicity, most music Sign -

Sign relations vary in degrees of resemblance. For example, a motive will
regularly be followed by a repetition or transposition of it.

The formalist theory of music meaning is too limited. First, in an
absolutist fashion, it is implied that music meaning is isolated from
meaning obtained from other sources. Second, although structural meaning
is important in relation to all music styles, it is not a sufficient
description of the meaning of music. Most all music is probably associated
with emotions and feelings, too. Most theorists suggest relation to
unobservable components of human emotional behavior. Actually, the
simplest theory about the relation of emotion is that of hedonism or
sensual enjoyment (69, part 2). Although the structural nature of the
sign-significant relation is not made explicit, the basic materials of
music, such as harmonic sounds and instrumental-vocal timbres, are felt
to be related to emotional reactions.

‘Langer (46, ch. 8) Spells out the relation to emotion more
elaborately. 'She states that music, more than any other Sign system,
emphasizes the single purpose of enabling man to express abstractly the

patterns of his most basic uniqueness: emotion. It is felt that man has

21

a need to symbolize emotional experience just as other experiences, and
that the abstract forms of music Seem to be most suitable. Further,
music motion patterns are felt to be Similar to patterns of emotion.
Music is like the growth, deviation and resolution of emotions. Roger
Sessions observes:

....in embodying movement, in the most subtle

wand most delicate manner possible, it communicates

the attitudes inherent in, and implied by, that

movement; its Speed, its energy, its elan or impulse,

its tenseness or relaxation, its agitation or its

tranquility, its decisiveness or its hesitation. It

communicates in a marvelously vivid and exact way the

dynamics and abstract qualities of emotion. (78, p. 23)

Meyer (51) offers a more precise theory about the resemblance
between music emotion. Meyer suggests that music patterns are related
iconically to emotion arousal in terms of its probabilistic structural
information.

The previous theories Suggest that music is similar in important
ways to internal emotional patterns. Other theories, such as Referentialism
(69, part 3), posit that music communicates Specific human emotions like
happiness, anger, majestic, etc. Here the resemblance is based on
relations to a more observable emotional behavior, such as skeletal move-
ments. For instance, a piece of music might be bouncy which may only be
Similar to human motor movements occurring in "happy" events. With an
event such as majestic or heroic, the relations of Similarity become more
tenuous. The significant cues of heroism depart from the immediate
surface proPerties of the event and the music patterns that appear to be

relevant (based on some research by this writer) are tonal chords, tonal

sequential pitch intervals, brass instruments and medium or slow tempo, etc.

22

Copland begins to suggest the boundaries of iconic music meaning relations
to emotional events:
How close Should the intelligent music lover wish
to come to pinning a definite meaning to any
particular work? No closer than a general concept,
I Should say. Music expresses at different moments
serenity or exuberance, regret or triumph, fury or
delight .....take such a theme as the first main
one of the Ninth Symphony (Beethoven) ...a feeling
of strength, a feeling of power.... but one Should
never try to boil it down to "the fateful hammer
of life", etc. That is where the difficulty begins.
(18, pp. 20-21)

The above iconic relations also begin to suggest a shift from
internal to external significants, some of which may not be primarily
emotional events. Music may have iconic relations to objects and events,
which themselves have either iconic or non-iconic emotion reSponses attached
to them.

A few music Styles, such as "prOgram" music, may be practically
identical to the external event and thus be Similar to a large number
of relevant cues in an event such as a bird call. Probably, there is a
somewhat fewer number of Similarities to important cuesin events such as
human motor movements or dancing. Even fewer may exist with temporal
phenomena such as wind and water. Also, as was suggested by Sessions in
relation to the nature of the music relations to aspects of emotion
patterns, music is similar to conceptualized classes of abstract temporal
forms, such as "stability", "elasticity", etc., which have a very large
number of objects and events as members. Further, psychoanalytic

theories (39) about music partially symbolizing basic but usually un-

conscious needs for feminity, sex, etc., would fit here. Finally,

23

Anthropologists feel that music is an integral part of culture and thus

inevitably reflects its general structure and values:
.... the incorporation in sound, of dynamic tensions,
tonicities and detensions that parallel, reflect or
reconstitute those experiences in the individual
biological-social life and in the collective social
continuum characteristics of a culture. (77)

Generally, although much descriptive research is obviously needed, it
seems that music is iconic to less properties besides rather abstract pro-
perties of most objects and events in the environment than is usually the
case with most other human languages that have iconic relations, such as
scupture, painting, etc. Further, the relations that exist may only be
vague resemblances. However, music patterns may be Similar to very
frequently appearing abstract properties of most objects and events, such
as stability, movement, tension, conflict, etc. Further, music patterns
define a set of abstract forms which are highly regarded emotionally.

Music expresses these patterns, making them observable. Thus, research into

the nature of music Sign-significant relations is certainly warranted.

Arbitrary Relations

Most clearly arbitrary kinds of sign-significant meaning relations
are of little interest, because they vary considerably across individuals
and music Style. For instance, a music pattern may be used in a film to
<ienote a particular character or a certain piece of music may be someone's
wedding music. However, a non-iconic relation of great generality is that
of verbal language. It can be used to deal with some of the complexities

of semantic description that are important to this study.

2“

More specifically, most different kinds of music meaning are

interconnected (cognitive organization) and also connected to emotional

‘W"-‘ ,

experience. In relation to the afore-mentioned aesthetic theories
(formalism, kinetic-syntactic and referentialism), Meyer states:
Any account of music communication that pretends
to completeness must find a place for all three.
The aspects of musical experience designated by these
positions are inextricably linked. (52, p. 258)
Some of the sign-Significant relations overlap, are redundant.
This is especially true in the case of verbal significants. Most other
significants also have verbal labels for their various states. (This
has made it possible to write about these other significants in this paper.)
It is with the use of verbal labels that the requirements set by Meyer
can be met. It further turns out that the most music significants are
labelled within the domain of connotative descriptive adjectives or verbs.
Whereas most semantic studies of human languages using verbal
connotation are usually inadequate because more components of meaning
(denotation), with music a connotative description appears to be sufficient.
Verbal connotations will be used for several reasons: 1) They are
the most easily measured, because they are available to every receiver and
they are observable; 2) They will reveal the broadest picture of overall
music meaning, because they are used as labels for most significants;
3) The organization of this verbal translation may have some functional
resemblance to the cognitive organization of music meaning; A) They offer

insight into the nature of individual and group abilities to communicate

verbally about music.

25

The main goal in this study is to describe the meaning of music
in an economical fashion. Afterward, a study can be made of the present
methods of verbal instruction used in music education courses to supplement

the presentation of music and suggestions made about developing more effective

overall programs of instruction.

Connotation

 

There are several sources of relevant research to serve as a guide
for the develoPment of a connotative description of music. This includes
philosoPhers of aesthetics, linguists, psycholinguists, and social

psychologists. Again, we Shall begin by considering linguistics.

Problems in Semantic Description

 

The semantics of verbal language has received less attention
than syntactics (In). Three major difficulties are suggested: 1) The
meaning of a particular Sign, such as a word, usually cannot be studied
to much of an extent in isolation, independent of a particular context of
its use, for it can have different meanings depending on the context.
That is, the context provides information leading to the definition, not
;providing the definition itself. 2) Given that there are many aSpects
to the semantic meaning of verbal items, a comprehensive definition would
require the development of a classifitory scheme of all the objects and
events in reality. 3) In terms of methodology, linguistics has never
been prone to use the inferential method, which would be necessary in

relation to unobservable referents.

26

Briefly, definitional work in linguistics has led to the use of the
word unit and only in terms of denotative meaning. Based on the use of
satisfactory grammatical and lexical meanings it is relatively easy to
determine the meaning of larger sequences. Therefore, it has not been
necessary to develop dictionaries fer sentences. Actually, dictionaries
do exist for some types of larger sequences (idioms).

Denotative meanings are the properties of verbal symbols which
are essential for socially approved use in the Speech community. Denotation
includes grammatical and lexical components. This implies a division
between structured and referential language significants. For instance,

a grammatical definition may include reference to sentence structure
(subject-predicate) and to form class (e.g., noun), while the lexical
definition utilizes synonyms and Specifies the property that this refers
to in particular referent contexts (e.g., "strike" means such and such in
the context of baseball).

The traditional definition of connotation used in linguistics which
complements the traditional definition of denotation raises another
problem: It has not been considered to be used regularly by most members
of a speech community. It is considered an individual, personal matter,
based on vague, uniquely defined external and internal referent contexts
such as personality, attitudes, etc. That is, connotations of specific
objects seems to vary continually within and between pe0ple. Consequently,
no work has been done on the connotative definition of verbal units in
linguistics.

In order to justify connotative definition research first requires

considerable psychOIOgical analysis of the nature of the referent contexts

27

of a representative sample of people from a Speech community. If a good
deal of interpersonal agreement is revealed, then connotative definitional
work would be useful.

It is posited here that connotation has both common and unique
components. Most people would agree that they share many personal
feelings with others, such as the "primary" emotions. Also, to some
extent emotional behavior is learned. This increases the probability
that most people share many personal behaviors, especially within a common
environment, such as a particular socio—cultural System. We might assume
then that human emotions have both a common and a unique component across

individuals. The problem is to determine the nature of this commonality.

Cognition

It seems that recent psychological research about human symbolic
processes or "c0gnition" has suggested an approach which reveals aspects
of connotation that Operate in common across groups of individuals.

In contemporary psychology, the importance of theorizing about
unobservable, "central" organism events is gradually being accepted.
Cognitive theories are primarily about intervening psychological processes
between symbolic input and output. "Cognition" traditionally referred
to the study of thinking, but now encompasses intersecting phenomena
distinguished with labels such as "meaning", "perception", "learning",
"creativity" and "memory". The unifying theme is the individual's
construction of an internal model of the universe around and within him

by using symbol systems.

28

Two basic kinds of questions are asked about c0gnitive behavior:
1) How are the symbols decoded? That is, how is meaning learned? For
instance, neo-behaviorists offer a "representational mediation process"
(63, ch. 1) which features several stages of stimulus-reSponse conditioning.
Cognitive theorists (5n) offer a cybernetic oriented approach featuring
hypothesis testing. The present study assumes that connotative meanings
are learned rather rapidly. Only adults have been studied and the
assumption is that learning has reached a plateau and is relatively stable.

2) What do symbols mean? This is a question about cognitive
structure or content and the concern of this study. Currently, psychological
research is studying aspects of the internal organization of meaning which
are relatively similar across individuals and Sign systems. That is,
underlying categories or dimensions of structure are suggested.

Generally, many music sign-significant meaning relations or con-
cepts are developed and organized by most peOple. Following the model
of cognitive content suggested by Bruner (12) which includes three modes
of representation: enactive (motor), iconic (imagery) and symbolic (verbal
language), most music concepts would be within the iconic representation.
This includes percepts and images of the spatial and temporal fields and
.involves the connection of abstract music patterns with Similar patterns

conceptualized in relation to other phenomena. The central problem to be

...nmw’u ' ' ' "'

”A... ...- ...a-..

_erce<1is how music patterns are grouped together and classified.
, First, music patterns can be grouped based on structural variables
developed.in music syntactic description. But most of these variables

are used in terms of structural meaning and not connotative meaning.

29

For instance, might music cognitive organization be in terms of the four
categories of pitch, duration, timbre and loudness, or in terms of
components such as melody, rhythm, and harmony? This is hardly likely when
considering the ways in which enjoyable music patterns are described:
stable, heroic, exploding. Possibly, connotative organization may lead to
larger, more general groupings. In this study, we shall correlate structural
variables to connotative variables and begin to eXpose the nature of there
influence on cognitive organization.

Second, human emotional behavior plays an eSpecially important role
in the development of music cognitive structure. Langer (47, p. 71)
suggeSts that perception is a process of Simplification from the countless
possible impinging stimuli and emotion is the act of emphasizing the
exciting features of the simplification. Aldrich (2) offers an aesthetic
mode of perception (opposed to a scientific mode) where perception moves
to the surface of objects and emotion is allowed to engulf it. Obviously,
not much is known about these interrelated phenomena, but it is likely
that emotions influence the overall selection and organization of music
cognitive structure. Also, emotion further plays an important role as an
““IBEé¥fiéi Significant (abstract pattern of response) about which concepts
can be develoPed and since it is present in the human receiver it operates
as another part of the overall response to music. The concern of this
study is only with emotions as it influences the selection and organi-
zation of cognitive structure which determines the interpretation of
music patterns.

Third, many theorists (2n) suggest and much research support exists

for the notion that the nature of cognitive structure is dependent on

30

the nature of the languages that it processes. Especially verbal
language is influential because it is the best means to facilitate
the broadest range of intrapersonal and interpersonal communication. To
quote Cofer:
"Much thought involves, or, if you will, is verbal
process and the fOrm and content of a language
probably makes some kinds of thinking and perceiving
easier..J'(15, p. 9n)

That is, verbal language is not only used externally but internally.
In psychological work, verbal elements and relations result in being used
both as parts of hypothetical models of cognitive structure and as
Operational definitions of those unobservable phenomena.

It was mentioned above that Bruner suggests a symbolic mode of
representation. It stresses methods of representation that are not
necessarily similar to the phenomena being represented. That is, words
do not usually resemble the object they are representing. When verbal
language element classes and grammatical rules are learned, Bruner then
states:

"It becomes possible for him [child] to
represent and systematically transform the
regularities of experience with far greater
flexibility and power than before." (12, p. A)

Russian research by Vygotsky (85) has supported their notion of a
second Signal system which replaces classical conditioning with an in-
ternalized linguistic system for controlling experience itself. Thus,
the symbolic representation may influence the iconic representation. For

intance, words have a structural feature of opposition (strong-weak) and

this may influence the organization of concepts about music patterns

31

and other Similar patterns. The nature and extent of influence needs

much research. It is obvious that recognizable perceptual patterns

exist far in excess of ordinary language words available to name them.

Osborne points out:

"When we need to talk about un-named shape

qualities [in totum], we are at a loss and
begin to use the language of metaphor, analogy

and emotion." (62, p. HG)

Research is needed about the nature and organization of concepts

about abstract visual and temporal forms. A useful beginning attempt

will be made by approaching the complexities of music meaning through

the use of verbal language. Fortunately, some recent work in psycho-

linguistics has completed part of the work.

The; interest has been to study the common portion of connotative
meaning for most people which is also common to most human languages.
It is felt that a study of the commonness between people should precede

a Study of difference. This work was based partially on previous studies

iDVOlving verbal descriptions of human emotions used to define many

sYlnbOls in some human languages, such as facial expressions (88). It
seems that a few emotion words are applied to many objects and events.
Further, the study of synaesthesia behavior (63, ch. 1) revealed that

stimuli from several modalities shared meanings and that most peOple seemed

to sFlare this integrated meaning System based on the use of verbal

corn-1 Otrative labels .

The notion of "dimensions" of connotation is suggested by Osgood

(
63) where the research question shifts to: "which out of all the many

32

available connotative concepts (e.g., beautiful, majestic, balanced,
floating, happy, etc.) usually do and do not covary regularly for most
peOple?" By the use of a new multivariate methodological technique
called Factor Analysis (33), it is possible to determine the amount of
independence between the clusters of concepts. This model of cognitive
structure includes a combination of quantitative mathematical and verbal
language Structural features, such as cartesian coordinate system,
correlation and bi-polar adjective scales. The dimensions and adjective
opposites may begin to offer some clues to the overall organization of
music meaning concepts.

The results of the research by OSgood revealed several regularly
appearing dimensions across different cultures and different symbol
systems. Three major dimensions are: evaluation, potency and activity.
The first refers to attitudinal judgments, the second refers to judgments
such as strong-weak and hard-soft and the third refers to active-passive
and excitable-calm judgments. Even though people may vary to some extent
in their judgments on each scale, it appears that most people use these
three dimensions. Thus, the notion of a common component of connotation
is supported.

Given that music appears to have primarily connotative meaning, this
kind of dimensional approach may turn out to represent a considerable
Portion of the music cognitive structure.

What connotative dimensions are involved in music commufication?
Possibly the above three dimensions will appear. No one has attempted

empirical research of this particular nature with music. It is likely

33

that an evaluation dimension will appear. Music taste and preference

phenomena has been of continuing interest to composers, listeners, social
scientists and the commercial music business for hundreds of years (22).
More interesting questions will be about the nature of the scales making

Up the evaluation dimension and of the nature of other dimensions and how

they relate to evaluation. Chapter IV will study in detail all previous

Uses of the dimensional approach and suggest hypotheses about the nature

of mus ic connotation .

Sununary

Finally, to summarize, we have suggested that communication re-
Search is an interdisciplinary approach: from philOSOphy we have drawn
a logical model for language analysis; from linguistics a descriptive
model; from psychology a model of cognitive structure; from musicology
descriptive work on variable identification; from aesthetics various
the(Juries about the meaning of music.

There are two goalsin this study: 1) to describe the internal
din‘lel'isional structure of the connotative meaning of music; and 2) to
establish associations between music elements and connotative elements.
On the input side, we intend to describe and manipulate music patterns
and on the output Side describe verbal connotative behavior, then in-
ferences will be made about the nature the intervening cognitive content.

Finally, to quote Rapaport:

The tracing of the pathways of the connections
between musical structures and collective emotional
needs would, I think, prove an exciting task. As
far as I know, very little has been done in this

direction. (70, p. l89)

CHAPTER II I

MESSAGE STRUCTURE
Introduction

The task in this chapter is to obtain a broad sample of music

patterns. First, with the help of psychological and musicological

theories, perceptually relevant aspects of music structure will be

identified and described. This description will be carried on in terms

Of the linguistic scheme of phonology, vocabularly,‘ grammar and style.
Secondly, it is suggested that all perceptually relevant structure

is not equally relevant connotatively. Consequently, the most

C=<>nnotatively relevant component of music structure Shall be selected.
From this component a small sample of patterns will be selected in terms

0f three stages-levels, dimensions and variable values.

Perceptually Relevant Structure

For the psychologist, perception or comprehension of the sign-Sign

relations or syntactics of music is the first stage in the decoding
Process. But most syntactics descriptions and style analyses of music
haVe been based on little empirical research about receiver decoding
abilities. In this paper there will be no measurements of the ability to
Initially,

°°mpr~ehend various patterns or features of certain patterns.

It Shall be assumed that there is a good deal of agreement in this task
0 . . . . .
f "Stimulus definition" across receivers.

31+

35

Prior research in experimental psychology has established that the
basic low level units on pitch and duration are "perceptually relevant".
Pitch range, pitch intervals and duration lengths are well within the
thresholds (absolute, differential and terminal).

But "music" usually refers to temporal motion which implies the
decoding of many sounds in sequence. Given that the input consists of a
series of quickly decaying bits of sound energy, the receiver must
"expand the present". Wertheimer (86, p. 72) states that when presented
with a series of discrete stimuli, we do not as a rule eXperience a
number of individual things. When the separate sounds are contiguous,
they will interact so that the occurrence of each affects the appearance
of the other. This interaction leads to the perception of relational
properties and larger sequential meaning units. Larger wholes are
experienced. In general, according to Boring (10, p. 168), relational
properties (pitch intervals, etc.) dominate non-relational properties
(absolute pitch values) in perceptiod.

There are many relational properties that appear to be perceptually
relevant. Kohler (us, ch. 6) states that a longer, more complex relational
preperty of "shape" is the most important property of things. In
pictorial research Gibson (30, chap. 5) and Attneave (3) feel that surface
contour is an important cue Since it is a high information region. Also,
Miller (55) has found support for the notion that human beings use longer
units ("chunks") because they are better at making Slow, complex decisions
(many alternatives) than fast, Simple decisions.

Few psychological theories are available to deal with complex Shape

(
0T .

43:!

both

SUEZ

16$:

' .
tgzgflfgi"

“a

 
 

36

Gestaltists and Associationists have both (based

or form perception.
on different assumptions) offered very general laws of perceptual

organization, such as proximity, figure-ground, closure, symmetry and
contiguity, repetition, climax, contrast, return and balance, respectively.

These can be summarized as asPects leading to a general law of simplicity:
But these concepts are not sufficiently tied to

we see what is Simplest.

stimulus variables from which to predict reSponses in complex situations
Little of the necessary empirical

such as listening to string quartets.
testing of hypotheses has been accomplished because of the different

Possible locations of independent variable (distal or proximal) and

tremendous number of stimulus variable that would have to be defined
Hochberg (38, p. 89) suggests angles and continuous lines

and controlled.
var°5.ables in relation to pictorial perception of single line drawings.

But music usually offers more complex patterns such as the following:
AS a melodic sequence gets longer, more relational preperties (harmonic,
r'Ilb'thmic, melodic) and amount of past experience becomes significant in
the perceptual selection process. Coding strategies are learned in order

to Separate relevant from less relevant elements. For instance, there is

a need to separate major accented tonal relations from minor ornamentation.

ISlocl'lberg would suggest that if presented with a sequence of tones, the
s -
lxnp lest "path" will appear which connects the relevant tones. Little

0
b no research is available to Suggest a list of most probable basic

n
paths" in music perception.
Initially, it is necessary to adopt the assumptions of the

““13 -
leologist that his syntactic descriptions containing many variable

id . .
exitifications are perceptually relevant. Actually, with few or no

37

denotative constraints on the construction of patterns, composers must be

rather sensitive to receiver pattern perception decoding abilities. For

instance, they follow rules such as not using pitch intervals greater than

a 10th.
At this point there will be an overview of the syntactic description

of music structure presented in terms of linguistic levels of analysis-

phonology, vocabulary and grammar, and style.

Phonolog

This term refers to non-meaningful, non-isolatable dimensions of

the lowest level unit, commonly called the "tone". The tone corre5ponds

Somewhat to the "syllable" in verbal language. When considering the notion

of phonology in relation to connotation, the criteria of lack of meaning

of low level elements does not hold, but the criteria of no wntext in-

dependent, stable meaning does hold. Separate pitch and duration values

may vary a good bit in meaning as the context grows larger and relational

Prepepties become more important. Due to the accumulation of properties,

they may increase in meaning as "parts" of larger pattern "wholes".
Actually, these notions will be tested to some extent in this study.
There are four basic pr0perties or dimensions to consider on the lowest
1e v81 ‘-' pitch, duration, loudness and timbre. A rough estimate of the
total number of combined values on these dimensions is 225. With this

' mat a number at the lowest level of structure, the possible structural

c
on‘Dllexity in terms of the variety of hierarchical develoPment in a

“he - ,
1c composrtion 18 enormous.

38

Grammar
In one sense there is no vocabulary and grammar at all in music

as in linguistics where all the users of the language accept a "common

practice" set of "words" and syntax. To quote Myhill:

The classification of possible tonal material
into good and bad or (better) permissible and
non-permissible is thoroughly out of place in
a music theory for the present time. (58, p. 197)

Fewer general practices exist in music than in verbal language.
Most of verbal language, grammar resides at the lowest levels of structure,

While the common practices in music appear at all levels but to a lesser

eX‘tent. In music there is wider acceptance and even expectations of new

Vocabulary and grammatical rules which contradict old rules being
deVeIOped and co-existing together which leads to a reduction in the range

of usage of the old rules. For instance, atonal music was deve10ped and

co“exists with traditional tonal music and has achieved a wide range of
alcc"eptance in a manner that never happened, say, with the introduction of
James Joyce's Finnegan's Wake in verbal language.

At this point a brief description of the overall available classes

0 O 0 O I I 0 O O
f e.‘Lements and relations will be discussed. Within this organization on

man .
y levels, composers can construct muSic messages.

In Western music, the interest has focused primarily on developing
Q
c”T‘D<>nents of organization derived from the basic pitch, duration, loudness
d timbre dimensions. Actually, the most effort has been devoted to

p -
1": Q11 and the least to loudness.

Important aSpects of loudness, such as accents and dynamics are

39

incorporated and elaborated to a large extent as aspects of other
components and so will be mentioned in these later contexts. General
theories about the use of timbre have grown around the problems of
orchestration and scoring music patterns for different numbers and
combinations of voices and instruments. In the past several hundred
years there have been many new instruments and the possibilities for
different kinds and sizes of groups, from unaccompanied solos to
symphonies, has increased. General practices have standardized the use
of many combinations of timbres, such as string quartet, dance band,
symphony, band, etc. (#3). Other common practices are that percussion
instruments are associated with rhythmic aSpects of music, while strings,
reeds and brass are assigned to harmony and melody.

Theories of temporal organization or "rhythm" as it is most commonly
called have been neglected in western music, eSpecially in proportion
to its complexity and importance. There are three basic general practices
that should be mentioned here: 1) pulse- one of a series of regularly
recurring precisely equivalent stimuli, marking off equal units in the
'temporal continuum. 2) Meter- the measurement of the number of pulses
between more or less regularly occurring accents. There are strong
and weak accents and the most frequent accent patterns are duple and
triple. 3) Rhythm- another level of accented-unaccented pulse or beat
patterns contained within the melodic sequence of tones. That is, in
music there are different kinds of temporal organization operating
simultaneously, where the meter is one and the melodic rhythm is another.

(17)

1+0

Grammatical theories about pitch organization are of central interest
in western music. There are three somewhat separable pitch centered
components: melody and harmony, plus a counterpoint component based on an
(integration of some eSpects of both melody and harmony.

Harmony has received the most grammatical attention in western
music. Harmony refers to the vertical organization of many simultaneously
occurring tones and to low level aSpects of the sequential relations
between these vertical arrays. The theory is based on the concept of
tonality and has generated two general practices: tonal and atonal.

Tonality approach is based on the diatonic scale where different
scale positions, such as the tonic and dominant, are assigned different
significances. With the notion of the "triad" (root, third and fifth)
as central a class of possible chords are available. Class relations or
chord progressions follow the principles of root dominancy and dissonance
reduction.

The atonal approach devises practices where the traditional tonal
key centered dominance is neutralized. The basic feature is to use the
12 tone chromatic scale instead of the diatonic scale. From this has been
developed a 12 tone row technique which determines the order of a sequence
of pitches. In this way sequences are composed that tive no continual
preference to any tone or interval. 80 the intervals and pitch shapes
contribute to overall unity what tonal centers did in traditional practices.
Actually, it turns out that the music is composed in such a way that the
tonal center shifts rapidly, since it is perceptually impossible to

completely eliminate tonality with the present phonology.

#1

Generally, melody refers to shaped single tone sequences. Not
much in the way of common practices have developed within the melodic
component. Several large vocabulary classes are continually distinguished
in musicological analysis, such as intervals, motives, phrases and themes.
Rules for establishing relationships between these elements are the
,general aesthetic "laws", such as repetition, contrast, balance, continuity,
etc., and various transformations, such as inversion, augmentation, etc.

The counterpoint component might more correctly be referred to as
instrumental harmonic counterpoint in relation to polyphonic music (72).
This deals with the means for developing musical structure with two or
more simultaneously occurring sequential lines. It uses certain melodic
vocabularies, such as interval and motive, in conjunction with most of
the harmony elements and relations.

Briefly, higher level principles of organization have develoPed
over the centuries. According to Copland (18, p. 76), given the goal
of balance, the most basic principle to be applied on the higher levels
is that of repetition. Various "forms" have appeared and apply to two
broad levels: 1) based on the identification of thematic material, the
divisions within a section of a piece are suggested. A familiar form at
this level is ABA or AABA. Higher forms deal with entire movements,

such as symphony, sonata and suite.

Style

Since no pre-compositional organization is necessary, virtually all
of the music vocabularies and grammars can contribute to the development

of musical styles. Thus, style basically implies a regularized process of

42

element and relation selection or invention in order to generate messages.
Since music vocabularies are very large, many different styles can
emerge.

The first problem in style analysis is to determine what bodies
of music are units of style. It might be a cultural area, such as EurOpe,
or a school or individual, etc. In western music, certain selection
combinations have obtained recognition and have been regularized. They
have been labelled as "classic", "romantic," "jazz," etc.

The second problem in style analysis is to determine the nature of
the homogeneity within a set of pieces. Nettl states:

The main purpose of describing bodies of music is, after
all, to distinguish them from each other in ways which
are significant in the sense that they reflect
differences in cultural and historical tradition.
Another purpose is to tell the listener what makes the
music... have the particular effect on him which it

has. (59, p. 187)

All the elements and relations of music organization are available
to the musicologist. Traditionally, pitch aspects of music are used, such
as tonality, scales, intervals and harmonic chords. Further, a statistical
approach is needed to find out which variable values occur most frequently.
The goal is to describe the behavior of many different variables using
many orders of statistical analysis to find out not only the frequency

of occurrence but the varied nature of the musical context in which these

occurrences take place.

1+3

Connotatively Relevant Structure

 

We have seen that there are many components of music-timbre-
orchestration, rhythm, harmony, melody counterpoint and forms. Each
one of these components includes an extremely large number of pattern
possibilities. However, it may be that neither all components nor levels
of structure contained within them are equally relevant to connotative
meaning. The interest is in those aSpects of music which have connotative
significance. According to Cooke (16, ch. 3) and Meyer (51, p. 7), the lower
levels of structure seem to be most relevant to connotative meaning. That
is, variation in connotative meaning is influenced by variation in
patterns around the "attention span". This limits it to components such as
melody, harmony, rhythms and timbre.

But are all these components equally relevant? Recently, Henkin (36)
did an impirical study using factor analysis, of variations in preference
of a wide variety of music excerpts from different styles. This study
may help to answer the question. The results of the factor analysis of
preferences showed two strong factors, easily labelled in terms of
components. They are melody and rhythm. A weaker factor, called timbre-
texture-dynamic appeared also. It appears that these three are inter-
related. No harmonic factor was pulled out. This agrees with COpland
(18, p. 22) who states that melody, rhythm and timbre obtain the
attention and concentration of the listener while harmony is taken for
[granted as an underlying aSpect revealing itself primarily as a part of
melody. Consequently, we might argue that harmony is closely related to

melody in terms of connotative relevance since they both come from the

in»

pitch dimension. Thus, the major contributors to expressive differences
are the melodic and rhythmic features. That is, they have the most
connotatively significant vocabularies.

This points up the fact that music vocabularies and grammars have
been developed without enough empirical study of the nature of various
aspects of human decoding such as perception and connotation. That is,
harmony features the most well-develOped grammatical theory of music,
but it appears that our sampling does not necessarily have to concentrate
on that region of the space which offers the most structural theory.

Are melody and rhythm equally important in terms of expression?

Since melody intersects with most other components and thus contains rhythm,
sufficient attention for the present can be given to rhythm just by
sampling from melodic structure. Further, it contains the timbre, loudness
and monophonic texture parts of the third factor. In general, this suggests
that within melody appears most of the significant features of most kinds

of music styles.

In conclusion, by selecting melody component or subpopulation of
patterns, it will be possible to draw a sample which has a reasonable
combination of sc0pe and depth for the initial study of the connotation

of music.

Melodic Component Structure

 

Because of the complexity and role of melodic phenomena in music,
there is only a minimal suggestion of elements and relations. These
various elements and relations have been mentioned previously. More

elaboration on definitions of melody would be appropriate. Melody is

45

defined as:
l) a sequence of two or more tones in which pitch is varied.

2) an upper (on the surface) "wavy" line among several simultaneously
occurring lines.

3) a sequence with pr0perties of symmetry, continuity, closure and
singleness of tonality.

'u) a sequence of period length (around 8 measures).

While most of these definitions consider melody in terms of high
level wholistic units, the present research will focus on melodic
structure where melody is the highest level of structure.

In order to begin to sample, an operational definition of the
population is needed. Given that melodic phenomena is a part of a larger
music space, the overall model will be based on a "matrix" method. Since
many of the boundaries of the music Space are undefined, first an n x m
matrix of cells shall be considered, with melodic phenomena operationally
defined as the first row of cells. To enable the use of one sampling
unit on all levels the basic unit which differentiates each cell will be
a tone. The symbols in each cell of Figure 1 refer to the simultaneous
and sequential asPects of music structure. The first subscript refers
to sequential music patterns of length from 1 to n tones and the second
subscript refers to vertical, simultaneous music patterns of length from
1 to m tones.

The criteria of dealing with many levels of structure is central.
Although there is no agreement in music theory as to the definition of
levels of melodic structure each successive sequence in tone length is

considered to be the next higher level.

46

MUSIC SPACE

 

C11 C12 C13 Ciu C15 015 017 C13 C19 ... ... cluo ....

 

C21

 

C31

 

#1

 

Figure 1

Not only does melodic structure contain many levels but the tone
contains pitch, loudness, duration and timbre dimensions where each of
which consists of a cluster of variables with many values. It can be
seen that several stages of sampling will be required: A) levels,

B) dimensions, in terms of variables, C) variable values. These stages
of sampling are interdependent. Since the melodic component contains a
very large number of patterns and since the final sample of patterns
must be rather small in this initial research, quite a number of
limitations or constraints must be placed on sampling at every stage.

So the sampling criteria must be selected with care.

'47

Level Sampling

 

Given the tone unit, approximately 50 cells or levels of the first

row are within the melodic sampling domain. In terms of Figure 1,

C11 - C150. Generally, the goal is to select enough melodic patterns to
find the range of meaning across melodic structure. Using linguistic
terms, it can be suggested that each cell contains a vocabularly. Since
in temporal sequences the parts appear somewhat in isolation, it may be
common to attach meaning to all cells. All cells but Cll have a set of
sequential syntactical relations. At present the focus shall only be

on a vocabulary sampling. Whether each cell contains a set of words
similar in connotative meaning is to be determined.

As a first criteria, the sample must include a good Spread of
levels, so the lowest and highest levels shall be selected. To the extent
that the lowest single tone unit, Cll’ is phonemic in character the structure
of syntactic theories of codes related to denotative use are similar to
syntactic theories of codes related to connotative use. If there is
consensus about the meaning of some properties of the single tone,
then it will be interesting to find out the nature of its influence on
meaning formation at higher levels. As for the highest level, melodies
vary a good deal in length in terms of angle tones. Initially, melodies
or themes which are around no tones, Ciao, in length shall be selected.

Since the rate of growth of connotative meaning probably is
greatest at the lowest levels, a more complete sample should be taken.

30 besides C11, 012 and C13 shall be selected. For some music styles,

148

such as contemporary styles, even a three tone sequence may be used as
a basic melodic unit upon which to generate longer sequences and unify a
whole composition. More extremely, the frequent use of a two tone

sequence, for instance, Stravinsky's "Rite of Spring? which features the

 

continual use of the pitch interval of a 4th, attachs a specific
character to the piece. At these lowest levels there should appear the
most rapid eXpansion in kinds and amounts of connotation. Possibly, on
level two intersubject agreement as to nature of meaning may increase
markedly.

With reference to the middle levels or cells, no statistical data
is available about the most frequently used cells. However, two broad
classes of patterns have traditionally been posited: motive and phrase.
80 the task will be to sample from each of these classes.

A motive is usually no less than three tones in length and no
more than six-eight tonesin length. This is a typical expressive unit
in baroque, romantic and modern styles. It is a label for the smallest
(actually melodically incomplete) ideas or kernels with some pitch shape
or rhythmic pattern of distinction used as the basic unifying unit in a
composition. It probably has relatively context independent connotative

meaning. So initially in this research, cell C 6 shall be selected.

1
The phrase is a longer sequence which permits more internal

structural relations such as tonality and meter. Thus, it is probably more

context independent and melodically complete. In a sense, it is

"one—half" of a melody or theme (a measures long). It is the typical

basic unit of classic, romantic and popular music. It may contain several

different motives and seems to vary in length roughly from eight to twenty

1+9

tones. Actually, the variation in pattern lengths gets high within this
higher level class of patterns. Consequently, phrases varying around
cell C112 shall be selected.

In making the above decisions, help has come from musicologists such
as Reti (7n), Newman (60) and Grout (31). In summary, six levels of

structure have been sampled: C11, C12, €13, C16, C112 and ClflO’

Dimension Sampling

 

All the basic dimensions (pitch, loudness, duration and timbre)
will be used as they are non-isolatable. Actually, all possible melodic
properties will be available and the question becomes one of determining
the nature of their appearance.

Each of these dimensions is to be conceptualized as being made up

of one or more variables. Each has only one variable in C11 but within

the succeeding levels the number of variables increase due to the
appearance of relational properties. Also, across dimension variables
appear, such as contour and symmetry. The large number of variables
within dimensions on higher levels are probably correlated in terms of their
influence on meaning. So only a few variables on any level.will need to be
identified.

Dimensions and their variables can either be manipulated, held
constant or permitted to vary in an uncontrolled fashion.

First, performance variables such.as.rubato, phrasing, articulation,
and tone quality (differentiated from timbre) could not be controlled

very well in this study. These variables are very difficult to measure.

50

They might have been held constant if one performer had been found who
played all the instruments to be selected. This was not possible in the
present circumstances. However, by using "school" musicians tended to
reduce the expression of extreme individualism and approximated holding
these variables constant. In the future, these variables should definitely
be mahipulated.

Secondly, overall loudness level was held only approximately
constant. It was found to be too difficult to vary when only using single
instruments. It is much easier to vary when the group size varies.

Other aSpects of loudness such as stress and accent are varied as a part
of other dimensions.

Many pitch, duration and timbre dimension variables are manipulated
on a wide range of values. Actually, timbre refers to only one variable
which reappears at each level. .

Variables on the two lowest levels are few in.number and are
identified clearly and precisely. They are listed below.

Variable selection is not so easy on the higher levels. More
possible variables appear on succeeding levels. Various musicologists
have suggested significant variables on these levels. For instance,
Cooper and Meyer (17) and Nettl(59, ch. 5) suggested most of the duration
variables. Dallin (19, ch. 2) and Cooke (16, ch. 3) suggested the pitch
contour variable and narrow-wide pitch range variable. Grout (31) among
others used the style category variable. Forte (28) among others was
referred to for the tonality variable.

Few of these variables are identified very clearly. Especially

the more commonly used but gross style category system (by period) and

51

melodic contour variables offer only vague clues as the nature and number

of more objective variables to use as principles of classification.

Many

objective variables such as narrow-wide range, tonality, conjunct-

disjunct, etc. are available. Both style system and objective variables

are used but they are somewhat redundant.

The following is a cell by cell list of selected, manipulated

variables organized in terms of dimensions.

cellll - PitCh:
Duration:
Timbre:

Cell12 - Pitch:
Duration:
Timbre:

- Pitch:
Duration:
Timbre:

Cell13

cellls - PitCh:
Duration:

Timbre:

Cellllz Pitch:
Duration:

Timbre:

- P' :
iuo 1tCh

Duration:

Cell

Timbre:

absolute pitch levels.
duration lengths.
instrumental music.

intervals and direction.
distance pairs, Speed.
instrumental music.

interval pairs and direction angle change.
metric-non-metric lengths, Speed.
instrumental music.

shape, major tonal-minor tonal-atonal, conjunct-disjunct,
and melodic style category by period.

staccato-legato, metrical-non-metrical lengths,
repetitions-varied series of lengths, meter, Speed.
instrumental music.

contour complexity, major tonal- minor tonal-atonal,
narrow-wide range, melodic styles by period.
staccato-legato, simple-complex length series, metric-
non—metric lengths, syncopated-nonsyncopated, meter, Speed.
instrumental music.

contour complexity, major tonal-minor tonal-atonal,
narrow-wide range, melodic styles by period.
staccato-legato, Simple-complex length series, metric-
non-metric lengths, meter, Speed.-

instrumental music.

52

Variable Value Sampling

 

In the final sampling stage the actual music patterns are obtained
and can be exibited. The previously selected variables all have two or
more values or states. In general, these variables vary in terms of
the clarity of Specification of different values. This shall not be too
much of a handicap in the present research because only a few values can
be selected from each variable. Further, due to the gross differences
between values to be sampled, the "cultural" definition rather than the
"physical" definition of variable values shall be used. For instance,
instead of defining a particular pitch value as 256 cycles it will be
defined as middle C.

The main reasons why only a few variable values can be used are
subject centered: availability of subjects in terms of class hours and
subject fatigue. Given the maximum of one class hour, this results in
being able to use a maximum of only 20 patterns. With Six levels of
Structure, approximately three patterns can be used. Thus only three
variables values at most can be drawn from each variable on each level.

The range of values of all variables on the lowest level are
reasonably small in number, are defined clearly and are used with relatively
equal frequency. That is, there is a subset of pitch values and duration
lengths within which the usage is approximately equal, in relation to
the typical melodic structure. Further, the values on all variables
(pitch, duration) but one (timbre) are related quantitatively. When
variable values are related quantitatively rather than nominally there

are more interrelationships between the values and this helps to make the

53

basic sampling criteria of obtaining a wide number of values much easier
to meet.

Four main categories of timbre values exist (percussion, string,
reed, brass) and all are presumed to be used a Significant number of times
with melodic phenomena. The single most frequently melodic instrument
iS selected from each category in the initial research and sampling
from all levels is only among these four instruments.

On level two and three the values of the variables are defined
clearly but the quantitative nature is more complex. Further some
information about frequency of usage has been collected here Ortmann (61),
Fuchs (29).

The values on each variable will be sampled separately and then
combined. To minimize bias from personal experiences with music, several
sampling procedures were developed. The author along with several
musicologists each selected values for each variable. These sets of
‘values were compared and a set was selected. If there were several
equally applicable patterns, then the selection was made by a random
procedure. A more detailed Specification of the value selections and
combinations follows.

The following 18 patterns Shall define the Six levels of melodic

structure to be used in this research:

._I\ .
Cellll

 

 

 

 

 

 

 

 

 

 

 

51+

pitch- a wide range selected within the frequent "melodic range"
duration - a wide range of duration lengths; speed = 100.

timbre - percussion (piano), string (violin) and reed (clarinet).

 

 

 

 

 

 

 

 

 

Cell
12 n
I
fir
4-
pitch - intervals of M2nd, PSth and octave;
two down direction and one up direction.
duration - pair of long durations, pair of Short durations
and one of varied lengths; speed 50, 90, 120.
timbre - brass (trumpet) and clarinet and violin.
Cell ‘
13

 

 

 

pitch- repetition of "zero" interval, varied pair (M2nd-M6th),
and repetition of 2ndS (M2nd-m2nd);

zero angle or direction change, upward direction with

slight direction change, down direction with no direction
change.

duration - repeated metric series; combination non-metric-
metric, repeated submetric series; Speed of 80, 100, luO.

timbre - piano, violin and trumpet.

The complexity of melodic structure on the succeeding three levels
requires the addition of more involved sampling procedures. Problems are

exposed when dealing with the relatively complex high level variables.

55

First, the variable values or, in the case of nominal relations,
categories are oversimplified. A study of the schemes used by Dallin
(19, ch. 2), Marquis (ug, ch. 1), Cooper and Meyer (17), Copland
(18, Ch. 5) and others reveals that many possible values on each of
many variables are collapsed into a Single category and the boundaries be-
tween categories are not clearly delineated. For instance, upward
direction-downward direction-arch countour categories and repetitious-
varied duration length series categories reveal these weaknesses. AS a
rough check on the exhaustiveness and to counter the possible weaknesses of
oversimplification the Dictionary of Music Themes (5) was used. It
contains themes from over 10,000 of the most well-known music works composed
in western civilization during the past four centuries. This was used
to see if there were many melodic patterns that could be found which were
not easily categorized by the above schemes.

Next, Since complex higher level patterns include a tremendous number
of possible combinations of variable values and the present interest is to
find music meaning in terms of patterns that are used regularly, it seems
best to sample "real" patterns that are familiar. It was possible to
(unstruct patterns on lower levels because all possible combinations are
used.

To use real patterns produces sampling prdblems, as the variable
valies are already combined and so selection must be simultaneous. This
Inakes it harder to meet the criteria of a wide range of values. So the
sampling criteria becomes one of maximizing the overall differences be-

tween the three patterns to be selected on each of the three higher levels.

56

The sampling procedure continued as the patterns for each level were
selected by the author and by the above mentioned musicologists. Given
these patterns, a Single set of three was chosen for each level so that
no pattern contained a variable value which appeared.with some other
pattern such as a particular rhythm variable value being repeated across
two patterns. The sampling was completed.when maximum differences existed
between patterns. In the case of equally adequate patterns, a random
selection method was used.

A more detailed Specification of value selections and combinations

follows.

C
ell16

 

pitch - upward direction, downward direction, upward arch;
atonal, major mode tonal and minor mode tonal;
very disjunct, conjunct, and somewhat disjunct;
modern, baroque and romantic styles.

duration - moderately staccato, staccato and legato;
triple submetrical pattern, duple submetrical
pattern and varied around metric length; Speefih of
70, 100, 130; two duple and one triple meter.

timbre - piano, clarinet and violin.

C
e11112

 

 

 

 

 

 

 

 

 

pitch - Simple arch contour, complex contour with many
direction changes, and oscillation contour;
medium pitch range, very wide range, narrow range;
major mode tonal, atonal, minor mode tonal;
romantic, modern and felk song styles.

‘9.

57

duration - legato, staccato and moderately staccato;
Simple repetition of submetric, complex intermixing
of fragmented duple and triple patterns, and repeated
syncopation; Speed of 100, 80, 120; duple meter.

timbre - violin, piano and clarinet.

Cell
luO lili III

I’_----Z‘-------

‘-"'-uOI'-- '-

   
 

 

1 ' --. '6

 

 

 

 

 

 

 

 

 

pitch - Simple oscillation contour, wide upward arch contour,
and complex mixture of many downward and upward directions;
narrow range, wide range with several climax points,
lower position but wide range with many ordered,
building climax points;
simple diatonic major mode tonal, chromatic major mode tonal,
atonal; folk song, romantic and modern.

duration - staccato, legato, intermediate;
repeated metric length, repeated submetric, and
alternately augmented pattern around metric length;
Speed of iuo, 100, 80; duple meter.

timbre - clarinet, violin and piano.

CHAPTER IV

RESPONSE STRUCTURE

Introduction

 

Just as we have described music structure in terms of various
levels, dimensions and variables, we shall offer a description of music
connotation structure. The main goal in this chapter is to describe
music connotation in terms of dimensions through the study of the
organization of verbal adjective variables which will lead to inferences
about cognitive connotative structure.

By finding dimensions the definition of connotation can be reduced
to only a small number of adjectives. Thus the attempt includes a re-
duction of the great amount of connotative conceptual redundancy that has
occurred in communication about the meaning of music.

It is hoped that this definition will be relevant to most music,
and most people most of the time. Therefore, a "structured response"
approach will be used.where the adjective variables are selected as a
part of the hypothesized dimensional description and this fixes the
overall domain of choice for the subjects. This makes it all the more
important to study carefully a wide range of relevant research.

Chapter IImaintained that research with connotative behavior
developed jointly in many areas such as linguistics, psychology and
aesthetics. This chapter will discuss the somewhat narrow descriptive
theories of emotion, and the broader descriptive theories of connotation

which utilize operational definitions in terms of verbal language.

58

59

These descriptive theories will be divided into three categories:
1) general code-free descriptions. 2) descriptions of the meaning of
spatial and time oriented non-verbal codes. 3) descriptions of music
meaning.

These theories are valuable in two ways: 1) they suggest hypotheses

about the connotative dimensions of music. 2) they guide adjective

sampling.

General Code-Free Theories of Emotion and Connotation

 

These theories provide the broadest possible picture of the basic
asPects of connotative Structure. Possibly only a portion may be relevant
to music connotation. Two ways of describing this structure have been used.
First the categorical approach will be considered, followed by early and
recent dimensional approaches.

One relatively recent categorical description of primary emotions
by Tompkins and McCarter (82) offers evidence that there are nine
primary affects: enjoyment, interest, surprise, fear, anger, disgust,
shame, distress, and neutrality.

Another categorical theory by Plutchik (67) posits a Similar set
of categories but groups them in terms of bi-polar relations, and suggests
that the variables within each bi-polar pair of categories are related in
terms of an underlying intensity dimension. This structure is derived
:from animal and human adaptive behavior theory. There are four bi-polar
categories: reproduction (joy)- deprivation (sorrow); destruction (anger) -
protection (fear); incorporation (acceptance) - rejection (disgust);

orientation (surprise) - exploration (expectation).

60

The two descriptions differ Slightly because the former does not
include an incorporation category, while the latter does not include
neutrality category.

Early dimensional descriptions of emotions for centuries suggested
a single pleasure-pain continuum. An empirical test by Woodworth (88, ch. 5)
some years ago using a variety of adjectives did not support a one
(iimensional description. In more recent research (76), two and sometimes
three relatively independent dimensions have been found: pleasure,
activation (or Sleep-tension), attention-rejection (or control, extro-
version-introversion, approach-withdrawal, etc.). Actually, much of this
latter dimensional research was not code-free in that most of the results
were based on only one important Symbol of one code: facial expressions.

It appears that the categorical and dimensional descriptions of
emotions are interrelated in that the categories describe more fully a
limited set of familiar behavioral events, while the dimensions describe
more abstract underlying properties of a larger set of possible behavioral
events. Thus, any one of the categories might be partially related to one
or all three underlying "referent" dimensions. Although it will be useful
to compare the above categorical theories to the early categorical music
meaning research at a later point in the chapter, the present descriptive
study will only use the dimensional approach.

The most recent dimensional approach and that which most in-
fluences the present study is one that does not attempt to find the
underlying dimensions of emotion. The phenomena of interest is human

communication (symbolic behavior, Sign learning theory) and the problem

61

has been to find the most generalizable aSpectS of meaning which appear to
be relevant across all human symbol systems. Osgood (63) suggests that
connotative meaning is most generalizable.

A model of overall cognitive structure is developed primarily around
two theories: 1) neo-behavioristic association learning theory- repre-
sentational mediation process, and 2) methodological technique of factor
analysis.

Briefly, a "semantic Space" is postulated as a region of some
unknown dimensionality, Euclidian in nature (a multi-dimensional cartesian
coordinate system). Factor analysis implies a mathematical classification
procedure into a series of dimensions. The dimensions (or "a axes")
are independent of each other. Further, there is some finite number of
"representational mediation reactions" available to the organism, and the
number of these alternative reactions corresponds to the number of
dimensions in the semantic Space. When a person judges some object or
event, he "places" it on a psychological continuum defined by a pair of
recriprocally antagonistic reactions which correSpond to the bi-polar
adjective scales. A rather limited number of Specific scales can be used
to operationally define the space and will be representative of the main
underlying organism dimensions of connotative meaning.

Attneave (u) states that those who think that a factor analytic
styled model is too complicated for human decoders are misguided in the
wrong direction. Such a model is too simple.

But in the initial work by OSgood there was no development of

hypotheses about the semantic (connotative) contents of these dimensions.

62

This lack of theory about the nature and boundaries of each possible
cognitive dimension makes it difficult to interpret the results and
assessing their validity, toward revealing something more about human
behavior than how adjectives are organized.

Actually, the adjectives were initially organized aS bi-polar
opposites, permitting two directions besides intensity judgments.

Kelly (42, ch. 2) offers evidence in support of this basic notion about
the content of cognitive structure. That is, bi-polar relations are
typical but not absolutely necessary.

Generally, the boundaries of connotation were and still are rather
vague. The initial criteria was to select those features of connotation
which would not vary in relevancy as the various codes and code symbols
varied. Thus, the most frequently used and most abstract descriptive
adjectives were selected. The set of scales included many summary
evaluations (good-bad) and those which seemed to be relevant to the most
abstract time and Space properties of external objects and events (active-
passive etc.).

Chapter I stated that Osgood found three main dimensions. Since
the initial research, many replications have been made across codes and
cultures and the same three dimensions continually appear. So it is likely
that this research may be revealing some general organizational features
of human cognition.

A more elaborate definition of each of the dimensions Shall be
presented now. 1) evaluation: this is the dominant first factor which
Operates somewhat like a general factor as subsets of its adjectives
adso appear on other factors. The definition comes from adjectives such as

good-bad.

63

2) potency - it is best defined by hard-soft, masculine-feminine, strong-
weak. It refers to the power, size, weight and toughness of objects. Two
subsets of scales correlated highly to both potency and evaluation.

First, hedonistic goodness- beautiful-ugly and pleasant-unpleasant. Second,
meek goodness - clean-dirty, graceful-awkward and weak-strong.

3) Activity - it is defined by active-passive and excitable-calm. So it
refers to the quickness, excitement, warmth and agitation of objects and
events. A subset of scales correlated highly to both activity and
evaluation which is labelled dynamic goodness and included successful-
unsuccessful, important-unimportant and progressive-regressive.

Two other dimensions of somewhat lesser importance in terms of ex-
plained reSponse variance appeared. First, stability was defined in terms
of stable-changeable, rational-initive and cautious-rash. Second, tautness
was defined by angular—rounded, tight-loose and stiff-elastic. Several
other factors appeared very weakly and might be defined as novelty,
receptivity and_aggressiveness.

What clues do we have to suggest hypotheses about the music
connotation dimensions at this point? First, adjective selection shall be
as broad as that by Osgood. The selection of abstract adjectives is very
appropriate in relation music structure. Adjectives will be arranged in
bi-polar fashion. By using adjectives from each of the dimensions
. gradually established by Osgood will offer a means of interpreting the
present results. Initially, then, the hypothesis will suggest the three

main dimensions of evaluation, potency, and activity.

61+

However, thehypothesis is not yet in its final form, because when
dealing with only music, it is expected to be both Similar to the general
structure and to differ in some unique way. To quote Osgood:

In every instance in which a widely varied sample of
concepts has been used.... the same three factors
have emerged in roughly the same order of magnitude.
When the sample of things being judged is restricted
in some fashion, the nature and order of magnitude of
the factors may change. (#2, p. 72)

For instance, Osgood offers a general principle ("congruity") about
code-dimension or, more Specifically, concept-scale interaction: in the
process of human judgment, all scales tend to shift in meaning toward
parallelism with the dominant attribute of the concept being judged.

For instance, the more evaluative the concept being judged, the more the
meaning of all scales Shift toward evaluative connotation. In the next

section only non-verbal codes will be considered and it will be interesting

to find out the nature of the Shift among dimensions.

Theories of the Meaning of Non-Verbal Codes

 

Non-verbal codes which are most relevant are those that evolve in
time, such as film and the dance. However, no research on the connotative
meaning of these codes has been done. Nor has it been done with
architecture or sculpture and other non-verbal codes which are primarily
oriented to aesthetic knowledge as is music. However, there is much
literature in these areas (71) which do offer many descriptive adjectives,
such as rushing-leisurely, floating-dragging, etc., that are primarily
related to the temporal asPects of objects and event.

Overall emotional organization has already been mentioned in

relation to several spatial oriented codes. Much of the early work on the

65

descriptive of emotions focused on facial eXpressions. Osgood recently
applied his approach to this code (6n) and found three dimensions.
However, some differences appeared between these dimensions and those
found when dealing with all codes. The first dimension was that of
pleasure, which correSponds to evaluation. The second dimension was
activity but the third dimension was best labelled as control and was
more Similar to stability than to potency. This suggests that activity
is more important than potency in spatial codes. Also, stability may
be a somewhat more important dimension of non-verbal code connotation.
A second piece of research focused on the connotative meaning of
representational and abstract painting (83). Both artists and non-
artists were used as judges. First, an analysis of the data based on
non-artists reSponses to representational paintings produced the three
main dimensions of evaluation, activity and potency. Activity accounted
for the most response variance. When artists were used the same three
dimensions appeared again but activity was even stronger and potency
was even weaker. When non-artists responded to abstract painting, the
semantic structure was a jumble and uninterpretable, but artists
judgments produced a single, very strong evaluative dimension. That is,
if the painting is evaluated favorable, it is also active and potent.
This also implies that for abstract art activity and potency are inter-
dependent and correSpond to what Osgood calls a dynamism dimension.
However, when Tannenbaum and Eliot (80) used a set of two-dimensional
abstract "nonsense" visual shapes, presented one at a time to air force

personnel, activity appeared again as the first dominant dimension and

66

independent of potency. Three other dimensions appeared - evaluation
plus two which were felt to be aSpects of potency (size and Shape of
objects): masculine (hard-soft) and feminine (angular~rounded). The
latter aSpect of potency is called tautness by Osgood.

What further information about the music connotative hypothesis
is suggested? The results are Somewhat conflicting. However, the weight
of evidence supports activity as the main attribute of Spatial codes and
thus may appear with music. Yet activity and potency may be interrelated
in music connotation because music involves abstract forms. So a
dynamism dimension may appear. Further, stability has appeared and Since
concepts of balance and harmony are very important in music aesthetics, it
may appear as a significant aspect of music connotation. Lastly,
tautness should be considered because music most centrally involves the

perception of shapes.

Descriptions of the Meaninngf Music

 

First, a good deal of searching for descriptive adjectives was
clone by this writer in articles on music in magazines and newspapers
(High Fidelity, New York Times, Saturday Review, Downbeat, etc.). Two
results seemed to appear: 1) Some descriptions were rather complex,
and Specific, revealing the critic's need to be creative ("flamboyant")
etc.; However, 2) most critic's descriptions were chosen from familiar
and basic research on music meaning. Consequently, reference shall only
be made to the research, because the present interest is toward defining

the broad underlying dimensions of music meaning.

67

The most well-known descriptive study of the internal organization
of music meanings was done by Hevner (37) in 1936. A categorical approach
was taken. A list of descriptive adjectives was selected from a set of
frequently used emotion labels. The subjects checked the list to
determine which adjective associated with which music. Thus the categories
were determined empirically. Eight clusters were feund and two kinds of
interrelations between these clusters appeared: 1) Similar to color
theory, a "circular" structural law operated, where as one proceeded away
:from any cluster the nature of the content became less similar and the
most distant cluster was opposite in meaning. 2) The clusters could also
be related in a bipolar fashion to form four "dimensions." No further
research was conducted to find the nature of the underlying dimensions.

In relation to the present study, several of these "dimensions" might
be a substructure within one more general dimension of music connotation.

This research by Hevner exhibited several methodological weaknesses,
such as too restricted and poor techniques of sampling, which weakened
the ability to make generalizations beyond the data. Also, there was
no broader cognitive or semantic theories for interpreting the data.

The resulting instrument turned out to be rather unreliable and of
restricted use.

Several attempts at modifying the original Hevner clusters seem to
offer some improvements. These will be described and displayed. First,
in 1942, Ivy Campbell (13) found that 12 clusters were needed to describe
music expression. Some of these categories were not obtained via emp’rical
research but are nonetheless interesting. The following table reveals

the categories although only a few adjectives are listed from a much

68

larger group in each category in the original table.

 

 

r
_ gaity joy i vigorous erotic
humorous rejoicing heroic passionate
solemn tenderness calm yearning
majestic sweet detached pensive
reverent aspiring
sorrow eerie cruelty rage
melancholy ominous satiric violent
agitated

 

 

 

 

Also, in 195”, Farnsworth (21) modified the Hevner description in
the fellowing manner. The circular relation is still in evidence but no
attempt was made to determine the nature of the possibly two underlying

dimens ions .

joy
happy
- playful
light frustrated
whimsical
‘ graceful .agitated
delicate vigorous
lyrical exciting
leisurely majestic
tender dramatic
calm triumphant
yearning sacred
pathetic reverent
sad
solemn

serious

69

There are several cluster disagreements between Campbell and
Farnsworth. Each had some categories that did not appear in the others
because of differences in adjective sampling.

Adjectives varied in difficulty of use by subjects with music,
such as yearning, tender, calm, pathetic, yielding, etc. These are
mostly very complex descriptive terms.

Generally, these descriptive attempts offer depth but not much
breadth in terms of connotation. Sampling of adjectives is still limited
to simple to complex moods, personality characteristics with some
denotational relations to Specific subjects of objects and events in the
environment.

It might be useful to compose the code-free categorical and
dimensional research by Plutchik and Osgood to this early music meaning
research.

For instance, calm, rage and violence may fall into the activity
dimension. Rage, violence, plus cruelty, satric and ominous may fall
into the potency dimension and the destruction - protection bi-polar
categories. Possibly gaity, joy, heroic, passionate, solemn, pensive,
sorrow, heroic and majestic will fall primarily on evaluation and repro-
duction-deprivation bi-polar categories, plus to some extent on incorporation-
rejection.

From the Plutchik categories it is suggested that evaluation
(incorporation-rejection) could be independent of reproduction - deprivation
under certain conditions. Possibly, a condition such as an aesthetic mode
of decoding may be sufficient. Under this condition the pragmatic or

inauwmental meaning of surface cues do not contribute much weight to the

7O

evaluation of the object. Also, in terms of aesthetic theories, most

of these adjectives are within the referentialist position. Thus it

could be that an emotion—mood dimension will appear independent of activity,
potency, evaluation, etc. However, it is difficult to say how many of these
adjectives will define it. It probably will be made up of basic moods such
as happy and sad. Only a small sample from each of the above categories
will be sufficient to obtain some notion of what dimensions underly the
entire set of adjectives.

For the present purposes the following adjectives Shall be sampled
from the Campbell and Farnsworth clusters: gay, happy, humerous,
extroverted, friendly, rash, violent, majestic, aSpiring, delicate, pensive,
sad, serious, introverted, cautious and ominous.

The object is to find dimensions that are not Specific to a subset
of music nor a subset of denoted objects and event properties. This will
lead to a measuring instrument that has much generalizability and
possibly more reliability. In future research, the underdying dimensions
can be used to sort out the connotative components of the various complex
adjectives such as heroic, whimsical, yearning, etc.

Other aesthetic theories about the meaning of music have never been
subjected to empirical study. First, the formalists feel that music
asymbolizes nothing (of any importance) beyond itself. The understanding
and enjoyment of music depends upon the comprehension of factors such as
symmetry, perfection of proportion and other relatively Spatialized,
static aSpects of structure.

Secondly, Meyer (51) and Langer (46), feel that the most important

aspects of music patterns are the comprehension of dynamic, rapidly changing

71

growth, deviation and resolution. Further, Bearsley (7) suggests these
dynamic structural properties have verbal labels, such as stability,
complexity, abrupt, exploding, etc. These Should increase the importance
of stability and activity dimensions. Further, according to research by
Rogge (75) these Vkinetic" descriptions were easier to use in association
with music patterns than were mood descriptions.

Now we will turn to available research using a dimensional approach
to Study music connotative meaning. First, Van De Geer, (8H), presented

a small set of pitch intervals (simultaneously occurring, C21) and only

used a small set of Scales with no stability adjectives present. The
three main dimensions appeared. Interestingly, activity and tautness were
collapsed and it was the strongest factor.

Secondly, Solomon (79) had a group of Navy sonar men judge the
meaning of a set of sonar signals on 50 adjective scales. Generally, the
three dimensions of evaluation, activity and potency appeared. However,
potency accounted for more variance than did activity. Stability collapsed
onto evaluation while tautness scales appeared on both activity and
evaluation dimensions. So stability and tautness may operate in a complex
fashion with music.

Finally, Brown, et. a1. (11), studied the connotative meaning of
vocal singers. Three sopranos, tenors and baritones were used. The
selection of adjective scales was heavily Slanted toward static Spatial
(timbre) properties with few dynamic temporal descriptive scales. Thus,
the results revealed a large evaluative first factor and a second potency

factor, but the other factors were difficult to interpret.

72

These research results are somewhat contradictory, eSpecially in
relation to activity and potency. Most of the above pieces of research
featured very Short simple "patterns" (C11 and C21). The only one which
used extended patterns produced a potency dimension and no activity
cfimension, but this may have been due to scale selection bias.

Further suggestions can be obtained from some work by Osgood
(63, ch. n). For instance, judgments about the concept "time" produced
no activity factor at all. It seems that things "in time" can be active,
but time, at the extreme, operates as a referent outside the system, so
to speak. Also, judgments of "symphony" produced all the three major factors,
but activity was the weakest. Potency and stability were about equally
strong. 80 one might consider that while activity seems to be strong in
Spatial oriented arts, potency might be the stronger in temporal oriented
arts. In the particular case of music, activity and potency may turn out
to be correlated and so a dynamism dimensions will appear.

Hypothesis about Dimensions of Music
Connotation and Scale Selections

 

 

Based on the foregoing information an attempt will be made to put
the various pieces together and form a hypothesis about the meaning of
music.

Generally, a Six dimension structure is hypothesized (see Table 1).
First, the three main dimensions of connotation were selected. One of the
aims of this research is to see if they generalize across music symbol
system. Secondly, modern aesthetic theories have supported the relevance

of connotative dimensions such as activity, potency and stability to music.

73

Table l. Hypothesis about Music Connotation Dimensions

 

Evaluation

 

'good-bad
pleasant-unpleasant
beautiful-ugly
graceful-awkward
lucid-obscure

Activity

active-passive
tender-violent
aSpiring-resolving
floating-dragging

Mood

 

happy-sad
gay-pensive
humorous-serious
friendly-ominous
extrovert-introvert

Tautness

stiff—elastic
tight-loose
sharp-blunt

Potency

strong-weak
rugged-delicate
solid-hollow
plain-majestic
formal-passionate

Stability

steady-changeable
cautious-rash
leisurely-rushing
deliberate-impulsive

 

71+

Thirdly, a mood dimension has been suggested based on the early music
expression research and general theories about the structure of emotions.
Lastly, in light of various results with Simple auditory stimuli and
:3patial abstract shapes, a tautness dimension has been offered.

With reSpect to the adjective scales, 26 were selected. Due to
methodological constraints each dimension has at least three adjectives
on it. Since rather common adjectives have been used, no pre-testing of
the bi-polarity relations appeared to be necessary.

The present research deals with many levels of melodic structure.
This means that a study can be made of the stability of the dimensional
hypothesis across levels. However, some suggestions about variations
might be appropriate. Osgood's three main dimensions will hold up across
all levels as will tautness. The stability dimension may not appear until
the third level of melodic structure. Also, the mood dimension may not
appear to account for much variance until the second or third level.

The above "pure" Six dimensions hypothesis was stated in order to
have a baseline from which to consider some of the various complex
interrelations that have appeared in the previous studies between these
(imensions. Some predictions of possible dimensional complexities are the
following: 1) Evaluation and stability will collapse. 2) Activity and
tautness will collapse. 3) Mood will be interrelated more with activity
than with potency. 4) Activity and potency will collapse and a dynamism
dimensions will appear. This reduces the number of suggested dimensions

to only three or four.

75

In summary, the goal has been to suggest a set of independent
(imensions which will be relevant to most music and most people most of
the time. No prior research has attempted such a broad, general

descriptive study.

CHAPTER V

METHODOLOGY AND PROCEDURE

Introduction

 

Much has already been said about methodology in the previous
chapters because theory and methodology are so closely intertwined.
Generally, in this chapter the procedure for obtaining the data and the
methods of analyzing this data will be discussed. The procedure includes a
consideration of the message presentation, subject sample and questionnaire
construction and administration. The methods of analysis includes the
statistical design for analyzing the internal structure of connotative
scales and the statistical tests for assessing the significance of the

relations between music patterns and connotative scales.

Procedure

Message

Competent musicians from the Michigan State University Music
Department performed and recorded<flltape a sample Of 18 melodic patterns
(three from each level). The time length of the patterns varied from
1 second to 30 seconds and the time between patterns was at least 10 seconds.

Since it cannot be assumed that the order (prior context of other
patterns) of the patterns is not influential toward producing similar
responses across scales and intercorrelations between scales, an empirical
test of this hypothesis was incorporated into the research. Three tapes

were made, each with a different order of levels:

76

77

B) cell ----- cell C) cell - cell - cell

11 """ celliuo

celll2 - cell - cell The initial pattern order within levels was

112 13‘
at random (A), then they were numbered 1 - 18. Thus, for B, after the
levels were reversed the patterns within each level were also reversed.
For C, after the cells were randomly ordered, the patterns within were

randomly ordered. Consequently, three entirely different orders were

constructed by varying both levels and patterns. (See Appendix I).

Subjects

Three music education classes at Michigan State University were
used. This was the second term in a course designed to train people to
teach music in elementary schools. The three sections were assumed to
be Similar in most relevant variables. The total of 79 subjects were
almost entirely junior girls, ranging in age from 19 to 26. The sections
were of unequal n (31, 30 and 18). The class time length was 50 minutes
and the task took approximately #0 minutes, about 2.5 minutes per pattern.

Classes were assigned randomly to one of the treatments. The
three groups were used on successive days in the same room at the same
time (3:00). During the presentation of the treatments, this author
announced each level in an attempt to maintain high motivation and reduce
boredom. The writer administered all three classes. A pre-test was run
in order to aid the author in deciding what Should be included in an
opening verbal introduction and to learn to run the equipment (tape

recorder) smoothly. These data were not included among the data to be

analyzed.

luO

78

Questionnaire Construction and Administration

Basically, Since the goal is to find the common aSpectS of
connotation, a "structured response" approach has been used. The question,
"What does this pattern mean?" is not equivalent to the question, "List
all the associations you can make to this pattern?" It is the former
question that is of interest and implies that the overall domain of
possible adjective choices will be fixed in common for all subjects. Thus,
a questionnaire utilizing the semantic differential approach of Charles
Osgood (63, p. 19) was used.

The questionnaire was mimeographed on 8 1/2 by 11" paper. It
consisted of an introductory page followed by 18 pages of reSponse Sheets,
one for each pattern. The subject wrote down the announced music pattern
number on the top of each Sheet before reSponding to it.

The introduction obtained various information about the subject
and explained the use of the rating scales (See Appendix II). The author
read over the instructions and stressed that this was not a "test" with
right and wrong answers. Further, the general context of melodic patterns
to be presented was described in order to provide a similar frame of
reference for the subjects when responding to each pattern.

A five step bi-polar scale was used:

, good : : : :bad

 

Only two steps of intensity discrimination in each direction was felt to be
sufficient in this initial research. The "zero" category implies that the

Structured response approach is not equivalent to a "forced choice" method.

79

The 26 scales on each page were systematically ordered. First,
the "hard" or possibly less familiar adjectives were placed near the
top left-hand column. This might have reduced the problem of subject
fatigue by having the decisions become easier as he went along. Second,
the scales were ordered so that from response to reSponse the subject
would alternate and not repeat within the hypothesized factors. It is
assumed that the subjects all marked the left column and then the right
<nlumn. Lastly, random procedure determined the polarity position. (See

Appendix III)

Methods of Analysis

 

Major Design

 

The major hypothesis was concerned with the internal dimensions of
music connotation. A preliminary analysis was made of the influence of‘
the.particular pattern context used in this study on the internal
correlational structure by constructing three different pattern orders.
Since this context could possibly effect both the central tendency or
between scale correlations, both analysis of variance and homogeneity of
correlation Significance tests were used. Twenty-Six Simple one-way
anovas were computed, with three pattern orders as treatments and the 26
different scales as dependent variables. If any scale was significant
further anovas would be used to find out which pattern levels were the
most influential. The homogeneity of correlation tests are more closely
related to the main factor analysis. This test involves cell by cell

comparisons across three matrices, thus three cell tests. Given a 26 by

26 matrix, this required (26-25)/2 = 325 three cell tests.

80

An overview of the steps involved in using factor analysis will be
useful. Given 79 subjects responding on 26 scales for 18 messages

produces approximately 37,u00 pieces of data.

*

 

 

 

 

 

 

 

a:
I E
t =.
c .,
‘ ’:
”magnum...” ”S

 

 

afaLJL;¢,c.T:'

Figure 2

Grouping data first among the patterns for each subject and then
across subjects permits correlating all possible pairs of scales in terms
of the overall subject data for each scale. This produces a 26 by 26 matrix
of correlations. A Pearson product-moment correlation technique was
employed.

Assuming that the clustering of subsets of scales was not obvious
from observing the surface data patterns among the correlation matrix cells,
a principal axis factor analysis was applied to the correlation matrix.
Briefly, this means that the adjectives are correlated to mathematically
selected referent vectors (criteria of minimizing residual variance) and
applied successively in an orthognal relationship. Further, in order to
emphasize the simple (pure) structure of the data, a orthogonal varimax
rotation was used, along with the Kiel-Wrigley criterion for limiting the

number of factors.

81

So the adjective scales were clustered about a set of independent
vectors or "underlying" dimensions. Each dimension is operationally defined
by the scales which correlate (load) highly on it. That is, the conceptual
nature of the dimension is determined from the properties that the high
loading scales have in common. This leads to reducing the connotative
description of music from 26 or more concepts to possibly 5-10 relatively
independent concepts.

Given six levels of patterns, a factor analysis was generated for
each level and one for the combined levels. By doing a factor analysis
for each level the amount of variance contributed by each level could be

checked. Thus, the major design involved generating 7 factor structures.

Association Analysis

 

Hypotheses were tested based on the associations that subjects
made between Specific patterns and scales. That is, what meanings were
attached to the scales?

The nature of the first hypothesis was that a significantly greater
number of people judged the scale relevant than judged the scale irrelevant.
In terms of the nominal level "scale" assumptions: any response not in
the middle category or step is related (or relevant), the data on each
scale can be placed into two groups (middle step/non-middle step). To
test this hypothesis, a binomial (two category) exact test of statistical
significance was most apprOpriate and a z approximation could be used as
the group n was sufficiently large (79). This data is available from the

frequency distribution for all subjects on each pattern and each scale.

82

In this case, 18 x 26 = H68 binomial tests were computed. Since the +

and - direction categories were collapsed and since there was no interest
nor likelihood of cases of significantly irrelevant results, the test could
be one tailed and a .01 level of significance was selected.

A further hypothesis was tested for subject direction consensus
in meaning, again using 2 approximation to the binomial two category
exact test. This hypothesis is that a significantly greater number of
subjects will judge the pattern in one direction or the other. So a two
tailed test is called for, still only operating with nominal level
assumptions (+1-). Again, 468 tests were computed and since this is
rather exploratory research, it was felt that the level of significance
for the direction relations be lowered to .05.

Lastly, it was of interest to find out which music patterns correlated
most and least in terms of all the scales on each of the connotative
dimensions. In order to do this, factor scores were obtained for each
music pattern in each dimension. ~flThis implied 18 ° 5 = 90 scores. Then
the characteristics of the low and high scoring patterns were compared
and contrasted in terms of variable value differences. Actually, these
results were referred to only when they differed from the results obtained
in the above single adjective associations.

In summary, this data was collected during June, 1966 at Michigan

State University. It was then submitted to a computer for data processing.

CHAPTER VI

RESULTS

There are two sets of results to be reported: What Specific
connotative meanings do the melodic patterns communicate? What is the
dimensional structure of melodic connotation? The results of the second
question will be considered first because this study offers more complete

information about it.

The Dimensions of Melodic Connotation

 

Prerequisite Statistical Test

 

First, a description is necessary of some results which are a pre-
requisite to consideration of the factor analysis results.

The total set of data was obtained from three matched classes, each
of which was presented a different order of the 18 patterns to find out
if pattern contexts would significantly influence the judgments about the
meanings of many patterns. Twenty-six one-way analysis of variance
designs were used, one for each scale as dependent variable. These results
reveal significant differences (.05 level) only on a small group of scales:
floating-dragging, pleasant-unpleasant, beautiful—ugly, rugged-delicate,
gay-pensive and happy-sad. So it seems that context effects occur only in
relation to some kinds of connotation.

The next point is to see which of the 18 pattern - pattern context
combinations had the most influence on these scales. This produces 108

analysis of variance designs. The results show that very few of the

83

8H

pattern-pattern context combinations contribute to the significant
differences in scale use. The overall significance of the first four
scales are produced by patterns 3, 5, 10 and 13. For the last two scales
it is produced by patterns 3, 5, 10 and 15. A further study of
regularities in the ordering of these patterns reveals that they are

1) patterns in at least one of the orders near the first of the series,
or 2) first on a level, or both. This information will be useful when
preparing future studies. However, these results do not necessarily
reduce the value of the factor analysis results because this depends

on correlations and not directly on means.

Given three 26 x 26 correlation matrices, a homogeneity of
correlation test was administered, cell by cell in groups of three. The
prior results with means on certain scales does suggest that heterOgeneous
correlations are most likely with those six scales. The results showed
21 (out of 325) significantly different cells where 90% were contributed
by these six scales. But 16 significant results could be expected by
chance alone (.05 level). Considering this, it appears that it will not

be improper to study the overall factor analysis results.

Factor Analysis

 

In order to determine the dimensional nature of melody, seven factor
analyses were applied to seven 26 x 26 correlation matrices, one for each

of the six levels of melodic structure and one over all combined levels.

Results Over Combined Levels
The results of the factor analysis of the combined levels will be

presented first. In developing the hypothesis about the dimensions, some

85

complexity was expected. The hypothesis in its simplest form was a six
factor structure: evaluation, mood, potency, activity, tautness and
stability. In its more complex form: evaluation, mood, dynamism, activity
and tautness, and stability. How well did the hypothesizing do?

The resulting factor structure turned out to support much of the
hypothesis. The five factor solution was chosen, based on various criteria:
1) Select as many factors as are psychologically meaningful; 2) Select
factors only when they add a significant amount of explained variance
(2-3%); 3) Kiel-Wrigley criteria of factor selection: at least three
variables which have their highest loading on the factor.

Some of the five factors are easier to "label" than others. Since
this research is somewhat exploratory in purpose, it seemed useful not to
try to reduce the number of factors too severely until more work has
been done. This present set of factors may provide useful hypotheses for
future research. For an initial summary of the nature of the factors,
four were clearly identified: evaluation; mood, stability plus tautness;
dynamism. A fifth factor was more difficult to identify. (See Appendix IV)

These five factors accounted for 58% of the response variance on
the overall factor analysis, which is more than reported in Osgood's re-
search. But here there is a more intense focus on one symbol system and
so it should be expected. Evaluation is the first and strongest factor,
accounting for 21% of the variance, which is over one-third of the total
of 58°. The second factor, mood, accounts for 14% of the variance. The
third accounts for 9% of the variance and the fourth and fifth account

for 8% and 6% reSpectively.

86

A separate discussion of the characteristics of each factor will be

apprOpriate.

Table 2. Aesthetic Evaluation

 

I II III IV V h2
18 Graceful - awkward .82 .03 -.02 .18 -.ll .73
12 beautiful - ugly .80 .16 -.02 .16 -.12 .72
19 good - bad .72 .23 .00 .25 -.13 .65
3 pleasant - unpleasant .68 .uo -.06 .1u -.10 .66
7 tender - violent .68 .07 .07 .13 -.36 .62
22 delicate - rugged .72 -.07 -.ou -.31 -.02 .62
21 majestic - plain .58 -.23 -.28 .22 .10 .53
2, floating - dragging .68 .29 -.ll .05 .16 .58
n aSpiring - resolving .59 .30 -.ll .09 .19 .50

 

Aesthetic Evaluation: Graceful-awkward and beautiful-ugly have the highest

 

loadings and the purest pattern and consequently deserve the most con-
sideration when defining the underlying factor. Thus the factor might
best be labelled as aesthetic evaluation.

Actually, all the rest of the scales have high loadings but they
are less pure: good-bad, pleasant-unpleasant, tender-violent, delicate-
rugged, majestic-plain, floating-dragging and aSpiring-resolving. In the

case of several of these scales, this represents a divergence from the

87

hypothesized grouping on the evaluation factor. Three out of four of

the scales originally categorized as activity (tender-violent, floating-
dragging, aSpiring-resolving) resulted in being highly evaluative. Also,
some scales considered originally as being on a separate potency factor
(majectic-plain, rugged-delicate) resulted as highly evaluative. When
potency and activity scales are loaded highly on the same factor, this
implies the more general notion of dynamism. However, a separate dynamism
factor appeared and its scales turned out not to have significant loadings
on evaluation. Based on Osgood's hypothesis about the evaluative
dimension collapsing onto the most significant features of a code, a sub-
set of dynamism seems to be most important.

A possible rationale is available for these seemingly contradictory
resale. Possibly, the subset of hypothesized dynamism scales which loaded
highly on evaluation are less abstract and have a more anthropomorphic
(social) meaning than those which remained to define the dynamism factor.
Thus they are not as independent of evaluation judgments. Especially
when some of the scales used poles such as dragging, plain and violent.
These scales might be characterized as describing relatively active time
oriented (rather than static spatial oriented ) referent behaviors. Further,
they describe subtle, SOphisticated cues that are relatively observable

<x1the surface of referent events. Further research is needed to improve

this conception.

88

Table 3 . Mood-emotion

 

I II III IV V h2
10 gay - pensive .22 .83 -.07 .13 .ou .75
26 happy - sad .21 .80 -.02 .18 .ou .73
5 humorous - serious .00 .80 -.07 -.Ol .03 .65
ll extroverted - intro-
verted .05 .60 -.26 .21 .ll .48
23 friendly - ominous .us .65 -.ll .10 -.1u .68

 

Mood-emotion: Scales which have their highest loadings and are most pure

 

on this factor are humorous-serious, happy-sad, gay-pensive. Others are
extroverted-introverted and friendly-ominous. Active-passive also loads
fairly high on this factor. In terms of the relations between the scales,
are all highly intercorrelated. Thus this factor contains adjectives
describing basic emotions and personality features.

Pleasant-unpleasant also has a substantial loading on this factor,
and is the only one of the strongly evaluative scales for which this occurs.
Also, friendly-ominous has a relatively high loading on evaluation. This
may be due to the fact that these are labels which refer to both internal
emotional-arousal and emotional qualities of music.

In general, the hypothesis was supported that mood is independent

of evaluation in relation to an essentially connotative code such as music.

89

Table u. Stability-tautness

 

I II III IV v h2
6 steady - changeable .09 .03 .69 .16 -.12 .52

l deliberate - impulsive .03 -.14 .60 .l7 -.19 .us
in formal - passionate -.20 -.08 .64 -.ou .15 .H8
8 tight - loose -.36 -.27 .55 .02 .29 .59
13 stiff - elastic -.H7 -.30 .53 -.07 .22 .6u

 

Stability—tautness: Several scales from the hypothesized stability dimension
are the purest and have the highest loadings on this factor. They are
steady-changeable, deliberate-impulsive and formal-passionate. Several
tautness scales tight-loose and stiff-elastic, also have their highest
loadings on this factor.

Tautness scales are also highly negatively related to evaluation.
Thus, the elastic and loose poles are positively correlated to pleasing
melodies.

The intercorrelations between the tautness and stability adjectives
are relatively low, so this suggests that they are both correlated and
define components of a third dimension, such as the frequently used style

categories of classic - romantic.

90

Table 5. Dynamism

 

I II III IV v h2
25 strong - weak -.05 .16 .06 .80 -.O2 .68
17 solid - hollow .10 .03 .25 .69 .05 .55
24 active - passive .ll .47 -.16 .54 .19 .59
15 lucid - obscure .24 .20 .Ol .44 .05 .29

 

gynamism: Potency scales actually load the highest and purest on this
factor with scales such as strong-weak and solid-hollow. Active-passive
also loads high and is highly intercorrelated with strong-weak.

Also, lucid-obscure, has a moderate loading on this factor. 08good
used a similar scale, clear-hazy, which was loaded on potency. Actually,
this scale explained variance or communality is very low and may not have
been a meaningful adjective pair for the subjects.

Generally, dynamism implies that strength and excitement seem to go

together in music as was hypothesized.

 

TdfleB. ““““““““

16 sharp ~ blunt .25 .03 .13 .25 .55 .45

 

91

(6&gessive-defensive or social activeness): This factor is defined
principally by rushing-leisurely which is the purest scale and has the
highest loading.

Other scales having their highest loading on this factor are rash-
cautious and sharp-blunt. Rash—cautious also has a moderate loading on
the stability-tautness factor as was hypothesized. However, the inter-
correlation matrix showed that it is highly correlated only to rushing-
leisurely.

In general, the nature of this factor is still in doubt. Leisurely-
rushing may be similar to the fast-slow scale used previously by Osgood
which was loaded highly along with rash-cautious and sharp blunt on a
factor labelled aggression. Mitchell (56) states that much modern music
appears to communicate this emotion. However, by defining it as social
activeness relates it to dynamism although its correlation with the

active - passive scale was not very high.

Results For Each Level

The next question is how much the factor structures at each level
vary from the overall factor structure as presented? Possibly, some clues
may appear when studying these six structures to enhance or alter the
present interpretation of the overall structure. It may be found that
one structure is not sufficient.

On the following table the solution for each level is outlined. If
there is no re-statement of the scales making up the factors, it implies
that it is the same as the overall structure. However, the contents of

the fifth factor are always listed.

92

TABLE“? Factor-Structure for Each Level

 

% explained

variance
1. Single Tone Level: I Evaluation 21
II Mood 15
III Stability-tautness, plus potency 10
IV Rushing-leisurely, violent-tender
rash-cautious 92_
53
2. Two Tone Level: I Evaluation 20
II Mood, plus Stability-tautness 18
III Dynamism 09
IV rushing—leisurely, violent-tender, 08
rash-caut ious , sharp-blunt,
impulsive-deliberate
'5—5'
3. Short Motive Level: I Evaluation 22
II Mood 16
III Stability-tautness 11
IV Potency 07
V rushing-leisurely, violent-tender, 07
rash-cautious ___
63
4. Long Motive Level: I Evaluation 19
II Mood 14
III Stability-tautness, plus solid-hollow 09
IV Dynamism, plus rugged-delicate 07

V dragging-floating, resolving-aspiring,
introverted-extroverted, (goes to IV
or I in four factor solution 07
VI rushing-leisurely, sharp-blunt, rash-
cautious, tight-loose, (these go to

factor IV in five factor solution). 05
51-
5. Phase Level: I Evaluation 24
II Mood 15
III Stability-tautness 08
IV Dynamism 09
V rushing-leisurely, tight-loose, sharp-
blunt, rash-cautious 96_
62
6. Melody Level: I Evaluation 19
II Mood 18
III Stability-tautness, plus cautious-rash 08
IV Dynamism 09

V rushing-leisurely, tight-loose, sharp- 07
blunt, and passionate-formal 61

93

It looks as though there is remarkably little variation in the
(imensional Space across melodic levels. There are four factors which
appear regularly and distinctly. An index of factor structure of
similarity was computed by correlating the first four factors across all
the 6 levels. Given a lower bound of good fit = .72, the actual index
turned out to be .88. Next, the explained variance across all levels
varied from a minimum of 52% to a maximum of 63%. The ranking of factors
in terms of amount of explained variance remained relatively fixed. The
actual scale loading values do not even vary appreciably on the first
Ibur factors and thus have the same scales as highest loaders.

Some more Specific analysis is warranted. First, most of Osgood's
three major dimensions of connotation did not appear in a simple fashion.
Only evaluation appeared on all levels. Activity was related continually
across all levels to three different factors- evaluation, mood and
dynamism. (Of course, Plutchik (67) did posit an intensity or activity
dimension underlying his emotion dimensions.) Potency also turned out to
be related to two factors, evaluation and dynamism, which it dominates.
However, the dynamism factor did appear on every level, except that on the
lowest level it collapses with stability-tautness.

The mood or emotion factor appeared even on the lowest level which
was not expected. The highest scale loadings and communalities were
continually on this factor. Thus, these scales appear to be the easiest
to use. Also, as was predicted, it was continually relatively independent
of the evaluation factor. This is possible since evaluation judgments in

aesthetic Situations can be based more solely on abstract surface patterning

94

and less on "survival" meanings such as emotion states.

It seems that stability-tautness Should be recognized as a basic
dimension of music connotative communication. It appears as an independent
factor on all but the two tone level, where it collapses onto mood. The
prediction was that tautness would appear on the lowest level and it did,
along with stability. Stability scales continually are the higher loaders
on this factor. In terms of explained variance it is always greater than
the dynamism factor. It seems that this factor might be rec0gnized as a
factor that appears most strongly only with the music code. Also, it may
define two components of the classic-romantic style categories. Then too,
it may be basic to the kinetic-syntactic dimension of aesthetic communi-
cation referred to by Meyer (52). Still another interpretation is that
it is a reflection in music of the general need for balance or control as
was found in research with facial expression (64).

The fifth factor cannot immediately be recognized as a basic
dimension of music communication. It appears on all levels but level four.
The content in terms of scales varies somewhat. It is correlated to various
other factors, such as dynamism, tautness (tight-loose on the three higher
levels), and with evaluation (tender-violent on the three lowest levels).

It would be useful in relation to the overall results and eSpecially
in terms of factor V to mention that this research was pre-tested on a
small group of advanced musicians (n = 11) from the Michigan State Music
Department. The experiment went smoothly and the results are rather
interesting. First of all, the first four factors were almost identical

to those reported in the overall result, except that active-passive dr0ps

95

away from factor II (mood) and has a much higher loading on factor IV
which makes this factor more clearly defineable as dynamism. Most
importantly, no factor V appears and sharp-blunt goes to factor III.
Cautious-rash along with leisurely-rushing goes to factor IV.

Of course, this is only a beginning but some powerful regularities
turn up. Further analysis of the relations between music and connotative

structural elements may improve the ability to describe these results.

Melodic Pattern Connotations

 

Attention will now shift to what the melodic patterns communicate
in terms of interjudge agreement about scale relevancy and direction
associations. This will be divided into two parts: 1) overall view of
pattern and scale relevancy and direction consensus. Relevancy, in terms
of the 5 step scale, means that a Significant number of people marked any
step but the neutral, middle step. Direction consensus means that a
significant number of peOple marked a scale in one direction or the
other; 2) the most intense directions of connotation associated with

the patterns and factors are identified and analyzed.

Relevancy and Direction Consensus

 

The following table contains the patterns of relevancy and dh1£Tiou
consensus. The following coding procedure appeared simplest. If the
pattern-scale relationship is non-relevant, then an "X" is used to indicate
this. If the pattern-scale relationship Showgzdirection consensus, then

an "-" is used. Thus, an empty cell implies relevancy and direction

consensus .

96

Table 8 Overall Relevancy and Direction Consensus

 

SOPOfimCI-‘mifim

Scale

 

graceful - awkward
beautiful - ugly
pleasant - unpleasant
good - bad

tender - violent
delicate - rugged

- majestic - plain

floating - dragging
aspiring - resolving

><><><><><>< I"

><><

Melodic Pattern

 

XX

X

l><><>< X

X

10

11

XX

12 13 14 15 16 17 18

 

@003

gay - pensive

happy - sad

humorous - serious
extroverted - introverted
friendly - ominous

><><><><><

><><><><

><><><><><

><><><><

 

smwamnwc3r<dwpwomrmn

mmmndcmd

steady - changeable
deliberate - impulsive
formal - passionate
tight - loose

stiff - elastic

strong - weak
solid - hollow
active - passive
lucid - obscure

XX

XX

 

.ruShing - leisurely

rash — cautious
sharp — blunt

97

There are quite a number of observations that can be made across all
the scales and patterns. First, only low and high relevancy and direction
consensus extreme Shall be studied. (Quantitative data Shall be stated
in the following paragraphs by the number in the brackets.) Based on
these results, additional comments will be made about the generality of
the factor structure across levels and about a new conception of levels to
take account of an important change in meaning.

First, low-relevant patterns are #1 (5), #3 (9) and #7 (14). High
relevant patterns are #17'(26), #12 (25), #13 and #14 (24), #8 and #6 (23)
and #10, #16, #18 and #4 (32). In general, the low levels of melodic
structure are low relevant and the higher levels are high relevant. Most
patterns generally turned out to be rather relevant.

Low relevant scales are tender-violent (6), lucid-obscure and
beautiful-ugly (7) plus good-bad and rash-cautious (9). High relevant
scales are majestic-plain, strong-weak, deliberate-impulsive and stiff-
elastic (18), pleasant-unpleasant, steady-changeable, tight-loose and
solid-hollow (l7) and graceful-awkward, gay-pensive and active-passive
(16). High relevant scales are Spread among all factors except factor V,
but the low relevancy scales are contained only in factor I, IV and V.

The general summary evaluation scales in factor I turned out to be low
relevancy. In terms of factors, high proportion of low-relevancy factors
are factors V, I and II (25-28%) while factors IV and III have the lowest
proportion of non—relevancies (l) and 10%, reSpectively.

Next, which patterns and scales are low and high consensus? The
low consensus patterns are #3 (45%), #6 (48%), and #1 (50%). The high

consensus patterns are #4 and #16 (95%). In general, the low levels of

98

melodic structure are again low in consensus and higher levels are high
consensus. For instance, level one is 65% and level five is 77%.

Low consensus Scales are solid-hollow (41%), lucid-obscure (43%),
sharp-blunt (50%), steady-changeable (53%), tight-loose (59%), stiff-
elastic (61%), formal-passionate (64%), and rash-cautious (66%). High
consensus scales are floating-dragging,tender-violent, beautiful-ugly
and good-bad (100%), pleasant—unpleasant, graceful-awkward and gay-
pensive (96%), happy-sad and delicate-rugged (92%), humerous-serious
(90%), friendly-ominous (86%) and introverted-extroverted and rushing-
leisurely (84%). Here it appears that factors I and II are high con-
sensus while factors III, IV and V are low consensus. This is interesting
since the scales on the latter three factors presumably refer more to
observably "surface" pattern cues, about which more intersubject
pagreement might have been expected.

Combinations of high and low patterns and scales on both dimensions
can be revealing. First, are there any high relevancy and high consensus
melodic patterns? #4 and #16. These then can be interpreted as the
clearest and broadest in meaning. These patterns have cues that activate
all five dimensions of connotation.

What are the low relevancy and low consensus patterns? #1 and #3.
These then are the vaguest and least connotatively codable patterns.

What patterns have high relevancy but low consensus? #6. This
pattern might be considered the most ambiguous and the one which exposes
the most individual differences. Most people thought that many scales

were relevant to this pattern but did not agree as to the direction of

99

meaning. IS it surprising that this pattern turned out to be a
"medium" two tone length?

What patterns are low relevancy but high consensus? Possibly #7.
This pattern is narrow but clear in meaning, primarily for factor III.
One could call this a "pure" factor III pattern.

With reference to the scales, which are high-relevant and high
consensus? pleasant-unpleasant, graceful-awkward, gay-pensive and possibly
rushing-leisurely. If deliberate-impulsive and strong-weak (which are
high relevant but not quite as high consensus) are also included, the point
can be made that this set of scales included all the highest leaders
with the purest patterns on each of the five factors. 50, in general,
these scales map onto some dimension and across all levels of structure,
and general agreement exists about the direction of meaning. Scales
such as these which are good operational definitions for each dimension
would be used if only a very short form of the music semantic differential
was permitted for testing purposes.

Next, which scales are low relevant but high consensus? Tender-
violent, beautiful-ugly and good-bad. These scales have the narrowest
band of use in relation to the levels of patterns. In the present case,
this portion of the evaluative factor only maps onto extended melodic
sequences, the middle and higher levels of structure.

Which scales are high relevant but low consensus? Solid-hollow,
steady-changeable, tight-loose, and stiff-elastic. Interestingly, these
scales primarily come from factor III. Possibly, individual differences

most influence these scales. The differences might be due to the

100

variation in the learned, general cultural context of meaning for these
scales. Therefore, these scales may suggest different pattern variables
for different people.

Finally, which scales are low relevant and low consensus? Lucid-
obscure and rash-cautious. These are scales which might be thrown out,
especially the former which is not pure and has neither a high factor
loading nor a high Communality. Probably, the choice of labels was bad
in these cases. For instance, lucid-obscure might better have been
clear-hazy or Simple-complex.

Some general conclusions are: 1) When factor I and II scales are
used, there is consensus but they are not used as much (low relevant)
on low levels; 2) Factors III and IV seem to be relevant to all levels
but somewhat low consensus in meaning on all levels; 3) Factor V is
medium relevant and medium consensus.

Based on these conclusions, insight into a new conception of levels
of structure will offer a clearer relationship to the connotative factors.

First; pattern #1 had the lowest consensus and relevancy in terms
of absolute numbers (3). Next fewest was #3. Both related primarily to
factors III and IV and only slightly to factor I. Pattern #7, which
featured a more extended but highly redundant sequence, revealed
essentially the same configuration of meaning.

These patterns feature "central" and Simple dimension variable
values. That is, on the Single tone level, less pattern internal
structure exists. This is also true for very redundant extended sequences

such as pattern #7. However, pattern #2 contained a high pitCh and a

101

long duration and all five factors appeared. Consequently, a more
extreme deviation in variable value is required to have a broad
connotative reSponse attached. This is the method applied in a primitive
langIage with the "yell".

On the other hand, what are the characteristics of the higher level
patterns? Most high relevancy and high consensus patterns were from
the higher levels. Generally, most two tone, motive, phrase and melody
patterns continually revealed a considerable eXpansion of meaning, even
without any extreme variable values. Thus, on higher levels it requires
less extreme deviations to obtain a broad connotative reSponse because
more internal elements and relationships exist. These patterns have
gestalt properties of continuity, contrast, completeness, closure, etc.

Given the original conception of melodic levels (# of tones), all
the factors relate to all the levels. That is, it is possible to
communicate on all factors from any of the six levels. Another con-
ception is needed to make a distinction within the music code that will
take account of an important change in meaning.

To take account of this particular change in meaning, only two
levels will be posited. If the nature of the variable value besides
the number of tones is taken into consideration, then a low level can be
defined which will only relate to factors III and IV. That is, factors III
and IV relate to all levels, but the low level meaning is entirely

confined to these factors.

n,

Thus, factors I and II, whose appearance on the lowest levels in

terms of the original definition, was only based on a few significant

102

associations to melodic patterns, would not be associated with the
lowest level in terms of the new conception. It seems that these two
factors contained scales which had the most "social" meaning (e.g.,
Specific emotion states). Thus, this level distinction takes into account
amount of social meaning. The social meanings are not related to the
low level. Or, a positive correlation exists between social meaning
and internal structure.

It is difficult to say whether this conception of levels will be
easy to operationally define. Further research is required to find the
Inoundaries. If they can be defined clearly, then it would be more

feasible for composers to take account of it.

Specific Nature of Pattern Connotation

 

Each pattern will be presented in music notation with a
description of its Specific adjective meaning organized by factors and
levels. The underlined adjectives are most intensely related to the
pattern and contribute the most to the description of its meaning. The
distribution of subject responses to this underlined subset (96 out of
257 total Significant direction consensus adjectives) is in Appendix V.
The tabulated results for all 257 adjectives is in Appendix VI.

The Specific connotations associated with each pattern are very
interesting. However, for the present we will seek more general findings

which cut across many melodies and adjectives.

103
Table 9. Melodic Pattern Meanings

5° 3 T i L.
1

2 3

 

     

 

I plain I pleasant, beautiful, I pleasant, plain
III stiff graceful, tender III steady
IV strong aspiring, floating, delicate IV strong

II pensive
III steady, formal, tight,

 

 

 

 

 

   

 

 

 

deliberate
IV strong
Two Tone Level V $£$§E£E$X-
4 5
, i_J
| J
1 .1
O
I unpleasant, awkward, I pleasant, graceful, I pleasant, floating,
dragging, plain, rugged, plain, floating graceful, delicate,
resolving II gay, extroverted, aSpiring
II serious, pensive, sad, friendly, happy II friendly, serious
ominous, introverted III deliberate, loose, III formal
III steady, deliberate, stiff, elastic IV active
tight, formal IV strong, active V sharp
IV strong, rugged, V leisurely
passive
V blunt, leisurely and
cautious '

Short Motive Level 8
7

   

I unpleasant, awkward, I pleasant, gpaceful, I pleasant, aSpiring,

 

 

 

 

 

dragging good, aSpiring, rugged, plain
II pensive floating, tender II gey, extroverted, happy,
III stiff, steady, tight, II serious, friendly, friendl
deliberate, formal sad, pensive III elastic
IV strong, solid III deliberate IV strong, solid, active
IV solid V sharp, rash

V leisurely

104

Long Motive Level

10 11 12

 

 

 

 

I awkward, rugged, I gracefpl) beautiful, I pleasant, graceful,
violent good, floating, good, beautiful,
II pensive, ominous, pleasant, aSpiring floating, majestic,
serious II gay, friendly, extroverted delicate, tender,
III stiff, tight, happy aSpiring
deliberate, formal III elastic, loose II sad, pensive, serious,
IV strong, active IV strong, active, lucid introverted
V rushing, rash, sharp V leisurely III steady, deliberate

IV strong, lucid, solid
V leisurely, cautious

 

Phrase Level

13

 

14
I pleasant, floating,

4/34

    

 

 

aSpiring,

graceful,

50°,

beautiful,

majestic,

delicate, I pleasant, ugly,

tender awkward, bad,
II sad, rugged, dragging

pensive, plain

serious II sad, ominous,

intro- EEESive, serious

verted III tight, stiff,

friendly changeablr, formal I pleasant, graceful, good,
III steady, IV floating, plain

deliberate V Sharp II happy, gay, friendly

elastic III steady, deliberate

loose IV strong, solid, active
IV V leisurely

V leisurely
cautious

105

Melody Level

16

 

 

 

 

I pleasant, graceful, good, floating, aspiring, rugged, plain; II gay, happy,
humorous, friendl , extroverted; III deliberate, steady, loose, elastic
IV strong, solid, active, lucid; V leisurely

 

l7

 

 

A

'7 ..., VIEW/:1”: I

   
 

 

 

I pleasant, graceful, beautiful, good, floating, aspiring, delicate, tender
II pensive, sad, introverted, serious; III deliberate; IV V leisurely,
cautious, sharp

18

 

 

 

 

 

 

 

 

I unpleasant, awkward, dragging, bad, resolving, plain, rugged;
II sad, pensive, serious, introverted, ominous; III deliberate, formal, tight,
stiff; IV active; V

106

What are the general relationships between melodic pattern
dimensions (rather than levels) and connotative adjectives? Two
approaches exist: 1) From the point of view of the Sign system, given
certain patterns and variable values, what is the variation in connotative
scales and factors? 2) From the point of view of the significant system,
given certain dimensions and scales, what is the variation in patterns
and variable values? Actually, this kind of analysis can barely be
started here because it requires a much larger set of patterns.

The Sign system approach will be taken first. What scales and
factors are strongly related to certain variable values which appear in
a set of patterns? In terms of the most general relationship, what
variable values seem to dominate the overall meaning of any pattern
context in which they are present? That is, certain adjectives would
appear whenever a certain variable value is used.

Actually, it is unlikely that any completely context independent
(unconditional) variable value associations to adjectives will be found
when comparing all possible combinations of values across the entire
music population. Thus, inferences from the present results must be
withheld until there is further research. The following table contains

some of those which were found in the sample.

107

Table 10 Pattern-Scale Associations

 

Timbre
piano: pensive, awkward, dragging and ominous.
violin: floating, tender, and majestic.
clarinet: floating, plain and friendly.

Pitch
conjunct patterns (4, 7, 9, 16): plain, steady and deliberate.
up direction patterns (6, 8, ll, 12, 17): aspiring.
down direction patterns (4, 18): resolving.
narrow pitch range patterns (4, 7, 9, ll, 15, 16): strong and solid.
wide range patterns (6, 8, 10, l3, l4, 17, 18): (not with potency
scales).
minor patterns (12, 18): sad, serious, and introverted.
major mode patterns: both happy (16) and sad (13).
atonal patterns (10, 14, 18): awkward, bad, rugged, dragging,
pensive, stiff, tight, formal and sharp.
tonal patterns (ll, 12, 13, l6, l7): graceful, beautiful, good,
floating, elastic, loose, steady and leisurely.

Duration
staccato patterns (5, 6, 9, 10, ll, 15): strong and active.
legato patterns (2, 4, 12, 13, 17, 18): strong.
simple repeated duration length rhythm patterns: deliberate and strong.

 

Some observations might be made of these results. First, the pitch
dimension variables feature the most unconditional associations. They
cut across all dimensions of meaning. Only the tonality variable is
related broadly to the evaluation dimension, as well as to all other
(iimensions. So this is an important variable on all dimensions which is
not surprising considering the importance of the concept of tonality in
western music.

Secondly, only a few duration variables appear. Those are associated

primarily to dynamism scales.

108

Thirdly, timbre is primarily associated with evaluative and mood
dimensions. This is likely to occur with the more "static" music
variables.

Lastly, it seems that if one side or pole of a scale is associated
with one extreme variable value, the other adjective pole is not
necessarily associated to the variable value at the opposite extreme.

Not we will turn to the second Significant approach. Given four
connotative factors defined in terms of subsets of adjective scales, what
is the variation in pattern and variable values? That is, what pattern
and variable values most frequently appear with each factor?

The description of each factor will only be based on those scales
which load most highly on it (except when factor scores are used).
Patterns will be divided in terms of associations with either "positive"
or "negative" direction on the factor. The pattern in first position is
the most intensely associated with the adjectives. The variable value

associations will be taken from these patterns.

Factor I: Aesthetic Evaluation

graceful - l3, 2, 8, l2 awkward - 1, 18
beautiful - l3, 2, ll, 12 ugly - --
good - 13, 2, 11, 16 'bad - 18
pleasant - l6, 5, ll, 15 unpleasant - 7
tender - 13, 2, 8 violent - --
delicate - 13, 2, l2 rugged - l8
majestic - -- plain - 1, 3, 18
floating - 11, 6, 16 dragging - 18, 7

aspiring - 9 resolving - --

109

Positive evaluation associates most intensely with familiar,
romantic, tonal music. Negative evaluation is most strongly connected
VVith both modern, atonal music and very redundant patterns. In general,
this can be conceptualized as two extreme endpoints of a pattern in-
formation continuum where the lowest evaluation is at each end and the
highest evaluation is in the middle information region. That is, a

curvilinear relation between evaluation and pattern information is evident.

Factor II: Mood

gay - 16, 9 pensive - 14, 18, 7, 4
happy - 16, 9 sad - 18, 14, 4
humorous - l6 serious - l2, 4, 8, 10,
l3, 17, 18
extroverted — l6, 9 introverted - --
friendly — l6, 5, 9, ll ominous - 18

Positive mood is most strongly associated with relatively fast
speed of tones, clarinet and staccato, while negative mood is most strongly

associated with slower Speed and legato.

Factor III: Stability-Tautness

steady - 7
deliberate - 7, 4

formal - 7, 18, 4

tight - 7

stiff - 7

110

Only the positive pole has any intense adjective associations.
Thus some less intense direction consensus associations with the negative
pole (shown in Appendix V) will be used for contrast. Positive stability
strongly relates to very redundant patterns with few pitches, repeated
Single duration length, and tonal. Positive tautness also includes
atonal patterns. Negative tautness is associated with patterns that
feature properties much like those associated with positive stability.
So the cues that map into both sides of tautness scales seem to map into
one side of the stability scales. Since the tonal-atonal variable and
the melodic contour variable (redundant to complex) seemingly do not

influence positive tautness, it may be affected more by duration cues.

Factor IV: Dynamism

strong - 16, l, 3, 5, 9, 10, 15
solid - --
active - 16, 9, ll

lucid - ll, 16

Positive dynamism is strongly related to features such as staccato,
relatively fast Speed, submetric duration lengths, narrow-medium pitch
range. Furthermore, it can be tonal or atonal. This suggests that
dynamism relates rather equally to both pitch and duration dimensions.
There is practically no intense use of negative dynamism. Factor scores
give some evidence that patterns #7 and #18 feature variable values that
are negative dynamism. These variable values are generally the opposite

of the above values associated with positive dynamism.

111

Factor V

leisurely - 13, 12, 8, 5, 2
cautious - 13, 12, 4
blunt - --

This pole of factor five is strongly allied with Slow, romantic,
tonal patterns. For purposes of contrast, a look at the less intense
associations with rushing and rash (Appendix V) find atonal and complex
contours with wide pitch range. In general, the associations are with
both pitch and duration variables.

An interesting result has appeared with the last three factors:
Only one scale direction has been used extensively. A discussion of this
is contained in Appendix VII.

Some observations about factor - pattern meanings are in order. First,
no low level pattern is the most intensely associated pattern with any
of the factors. Patterns such as 13, 16, 7 and 18 are important.

All factors are associated with pitch variables. It appears also
that duration variables are important with all factors. Possibly, even
though both pitch and duration relate to all factors, they are not always
equally important? For intance, it appears above that mood may have
stronger relation duration than pitch. More research is needed to answer

this question.

Summary

Generally, both the Sign and Significant oriented analyses will lead

to one overall description of music code - connotative meaning relationships.

112

Relationships of the widest possible nature have been described in the
present study. In the future, conditional relationships will need to be
studied.

Overall, it seems that all the underlying dimensions of the music
Sign system (pitch, duration, timbre, loudness) probably are associated
with all the underlying dimensions of the significant system (evaluation,
mood, stability-tautness and dynamism). A more exact description of
the relationships will reveal much more complexity. This future research
requires the use of much larger samples of patterns on all levels, within
and across all components, such as harmony, texture, etc.

In conclusion, the results of this study offer more information
about the inner structure of music connotation but less information about
music pattern relationships with connotation. This is due to the nature
of the music pattern sampling which focused on only covering the breadth
of the connotative Space and therefore did not use all components nor
variable values in many pattern contexts. Future research of this latter
nature depends for its success on the completion of research of the

former nature.

CHAPTER VII

DISCUSSION

Introduction

 

This chapter contains two sections: an integrated summary of findings
and future research needs. Each section will be divided in terms of
relations within the connotative component and relations between the
melodic and connotative components. Further, the section on future re-

search needs includes two parts: theory and methodology.

Summary of Findings

 

Dimensions of Music Connotation

The broadest generalization is that music appears to have a good
deal of meaning. Further, when decoding music on the presently selected
number of scales, there seems to be a considerable amount of correlational
redundancy. Only a few general groupings appeared. Four and possibly five
cimensions have been found to describe music connotative structure. The
first four are: aesthetic evaluation, mood, stability-tautness and
dynamism. Each of these dimensions contained many high loading scales,
so defining the dimensions was not difficult. Also, each had many
relatively pure scales on it. Thus the independent character of each
could be identified.

An interesting finding is that evaluation turned out to be the
strongest dimension and graceful-awkward turned out to be the highest
loading scale on the evaluative dimension. Mood was the second most

important factor and stability-tautness and dynamism were third and

113

114

rather equal.

The evaluation dimension contains scales which were originally
crnsidered to be activity, potency and tautness scales. However, these
escales appear to be less abstract and have more of a "social" component,
thus being more evaluative in character. It is suggested further that
these scales are related to more temporally active and potent social events.

In contrast to the results obtained when working with spatial non-
verbal codes, activity neither appeared independently nor as the strongest
dimension. As hypothesized, a dynamism dimension did appear. This is
testimony to the interdependency of music properties. It also suggests
that activity is more significant in spatial oriented codes and potency
is more Significant in temporally oriented codes, because potency
dominates the dynamism dimension.

In terms of notions about melodic levels used in this research,
rjemarkably little variation appeared in the dimensional structure across
levels. Even though variation exists in the direction consensus and
intensity of scale use, the correlational use shows that the four main
dimensions continually are present. However, it was revealed that the
appearance of evaluation and mood was not based on many Significant
associations to melodic patterns on the lowest level.

This is the first empirical demonstration that people out up the
world somewhat in accordance with the different aesthetic theories. It
appears that the referential position is independent of the absolute-
expression position. The mood dimension corresponds to the former
position and stability-tautness and dynamism correSpond to the latter position.

Further, the absolute-expression position has the most generality across

115

levels. The referential position appears not to be relevant to the lowest
level.

Lastly, the scales based on the early categorical research in music
expression by Hevner and others turned up primarily on the mood and
evaluation dimensions. A suggestion can be made about how these two
independent dimensions underly the circular arrangement of adjective
categories. The mood dimension would be placed vertically through the
middle (6:00 - 12:00) and the evaluation dimension would be placed

horizontally through the middle (9:00 - 3:00).

Melodic Pattern - Connotative Relations

 

Only a few general findings can be given here. The broadest finding
is that all patterns had some meaning.

More Specifically, to make an oversimplified but possibly useful
generalization, low level patterns are low-relevant and low consensus
while high level patterns are high relevant and high consensus.

Two patterns on the Single tone level related only to dimensions,
III and IV, while all other patterns (except one) on all levels related
to all dimensions.

Consequently, another conceptualization of levels has been offered
in association with connotation. Both the internal structure (more than
just number of tones) and the dimension of meaning must be taken into
account. All melodic structure is meaningful on dimensions III and IV.
Dimension I and II only associate significantly with patterns which have
a certain amount of internal structures. Thus the phoneme - morpheme

level distinction is relevant only to the referential aesthetic position.

116

Given the notion of a neutral least meaningful region in terms of
variable values, a certain amount of vertical or horizontal extremeness
is needed in order to activate the entire connotative Space. For example,
a single tone implies little horizontal complexity, so proportionally
more vertical extremeness is required if there is an intention to
communicate on all dimensions. Of course situations may appear where
the intention is to communicate on only a few dimensions.

Findings about the relationships of patterns and melodic variables
to connotative dimensions can only be glimpsed at. In general, all basic
melodic dimensions are associated with all dimensions of meaning.

However, the results suggest that pitch (eSpecially tonality) and duration
are most importantly associated with most dimensions of connotation.
Possibly duration is most strongly associated with the mood dimension

and the tautness aspect of dimension III. From the opposite point of view,
evaluation and dynamism appear to have the broadest set of relationships

to melodic pattern dimensions.

Future Research Needs

 

Dimensions of Music Connotation

 

Theory

The most obvious need is to measure the meaning of more components
of music and manipulate more variables, separately and in combination, in
order to see if the present set of dimensions is generalizable or if some
alternations are needed.

It appears that the original, relatively small set of 26 scales will

need some modification and possibly enlargement. This implies increasing

117

the boundaries of conceptualization of music connotation. Of course, as
the scales change, the dimensional nature may alter somewhat and the
relative importance of dimensions may change in terms of amount of ex-
plained variance.

The goal is to explain as much response variance as possible in
relation to that aSpect of connotative reSponding which iscnmmon to most
all music. This implies that an explanation of 100% of the reSponse variance
will never be attained. The difficult question is to Specify the
probable cutoff point. For instance, 75%? AS more scales and more
music are used some idea about this may develop. Of course, dimensional
structures only in relation to Specific music style subsets using only
more specifically relevant scales could account for more variance.

Some changes in scales can be suggested based on the present
results: 1) The need for more adjectives related to higher level, more
(amplex melodic shape properties such as angular-rounded, ornate (or
fancy) - plain, vigorous-placid, etc. 2) The need for novelty dimension
adjectives such as interesting—boring and expected - surprising. 3) The
need for more timbre-texture-orchestration adjectives such as lush-
austere, warm-cold, hard-soft, etc. 4) The need for some uni-polar
labeling such as steady-unsteady, active-inactive. 5) A change in some
of the ambiguous polar labels such as obscure, blunt, resolving, hollow.
6) The deletion of some Scales such as lucid-obscure. 7) The addition
of scales to help clarify the nature of the dimensions, eSpecially

stability-tautness and evaluation.

118

The present use of verbal language in music appreciation courses
might now be discussed. Roughly, music appreciation verbal material
language can be divided into four categories:

1) technical, structural terms such as interval, rondo, etc.

2) historical, sociological terms, about social context of composer.

3) emotion or mood terms, such as delicate, tender, etc.

4) abstract form terms, such as elastic, floating, etc.

In music appreciation courses, it seems that the first two
categories are stressed the most. The third category is given a little
attention but the fourth category is slighted.

It is suggested that the last two categories have the most con-
summatory value. They make up the dimensions of music connotation. It
is assumed that these adjectives are used by most people in relation
to many objects and events and the pattern correlates may be relatively
familiar to all and similar among all even though obtained from various
different Specific referents. Thus, the use of these adjectives in
conjunction with structural terms may facilitate music pattern concept
attainment which will be maximally rewarding in relation to listener-
consumer use of music.

A practical future goal is to develop an instrument for measuring
music connotation. It could have a variety of uses: 1) measure the
meaning of a single piece of music; 2) measure the meaning of a sample
of pieces from one composer to assess the central tendency in connotation
from this composer. 3) measure Similar aSpects of a style or period

of music; 4) measure the connotative sensitivity of a student, proceeding

119

on to a study of individual differences in relation to connotative
reSponses to music; 5) measure the amount of composer-listener
connotative communication, by comparing the responses of composers

on the instrument to the responses made by listeners on the instrument.
This would offer composers a broader range of feedback about the meaning
of their composition, in a relatively effortless manner. Actually,
after the measuring instrument has been checked for reliability
(homogeneity and stability), then only a few, maybe as few as eight or
1en, of the main defining scales on the dimensions will be necessary in
any measurement Situation.

The main application of this instrument should be in the area of
public school music education. Music appreciation courses need to accept
a broader notion of the meaning of music. The teaching of music as a
feature of the general education of all students in primary and secondary
schools is important because it is the single, most frequently used art
form. So whatever aesthetic sensitivity children develop depends
primarily upon their contact with music.

This immediately suggests developmental research. In what sequence

do dimensions of connotation appear? Which appear first with children?

Or, what are the main dimensions of meaning in children's music? There
might be a most frequent order of appearance among the four dimensions.
Does evaluation appear first with the implication that tastes and
preferences precede all other meaning? Or do the dimensions with the most
generality appear such as Stability-tautness and dynamism?

This study may have some relevance to audio-visual instruction

research. This study has~found four independent variables with which to

120

determine the effectiveness of methods of instruction using music such
as music and film.

Work might begin on typologies of adjectives, using the high
loading and pure adjective scales on each of the four dimensions as
referent axes. Then a complex adjective having impure loadings and
possibly having Specific denotational relations, such as "heroic", could
be defined in terms of its amount of dynamism, evaluation, stability-
tautneSS and mood. Given the measurement of many of the frequently used
but complex adjectives in association with music patterns, a basic
question would be how many clusters appear and what is the nature of
each cluster? Also, in relation to the general categories of emotion,
how many are frequently relevant in music communication?

This research has limited itself to a concern with only one stage
or component of receiver decoding-connotation. In the future, as the
rmnsurement of pattern perception improves, it will be interesting to
study the interdependencies among these components. In the present study
it has been assumed that most people are operating similarly with respect
to pattern perception. This is less likely to be correct as the patterns
become increasingly complex because differences in taste, attention,
familiarity with music styles among other variables begin to exert a
stronger influence. But if connotative discriminations are not as
precise as pattern structural discriminations, then some variation among
individuals' perceptual behavior will not affect connotative behavior.

Lastly, when future research uses a much broader range of subjects,

a closer study can be made of the relation of the dimensional structure

121

based on grouped data across all subjects to the dimensions generated by
each subject. A factor analysis can be run for each individual. Then
factor structure comparisons can be made to get directly at individual
differences in factor structures. To the extent that there are
differences, personality variables and other sociological variables

such as occupation, role, etc., might be possible intervening influences.
In the present research it was likely that all the groups were quite
similar, but there were not independent measures on, for instance,
personality variables to check this possibility. Several of the
dimensions might have been generated due to several major differences in

personality among the subjects.

Methodology

One possibly important influence on the nature of the dimensional
results is the particular design: factor analysis, based on a product-
moment correlation matrix, A check on this influence can be made by
replicating the research using another technique for analyzing the data.
For instance, a multi-dimensional scaling approach could be used, such
as the method of triads (I). In this design the subjects generate their
own attributes (or scales) along which the patterns are compared, and
direct comparisons are made between patterns so that judgments are
relative, rather than absolute. Another technique is the Guttman non-
metric factor analysis of paired comparison data (32). Still another is
McQuitty's hierarchical classification procedure (50).

The information obtained about the possibility of some patterns

being more context-dependent than others on certain connotative scales

122

must be taken into account when preparing future experiments. One way
is to present a group with a brief subsetcf patterns that reveal the
structural range and boundaries in, for example, a pre-test instruction.
Another way is to place the context dependent patterns near the last

of the series.

Next, there may be a need to alter the form of the scale as it is
presented in the measuring instrument. First, the + and - cues to scale
idirection may have added an evaluation bias. Secondly reSponses in the
center step or category are ambiguous because they can mean several
different things: irrelevancy; don't know; equal amount of both poles,
etc. The problem is to pull these necessary aSpects of reSponding
apart without increasing appreciably the subject task time, fatigue
and possible error. I

Empirical studies of the most common adjective bi-polar relations
used in relation to music are needed. Some adjectives probably have
multiple relations to several other adjective contrasts. It may be
important to select a certain pairing in order to avoid decoding
difficulties, for example, when one pole is very clear but in association
with an unexpected contrasting adjective, confusion results. Thus, a
poor selection on one pole ruins the data on both poles.

An important task is to check the reliability of the instrument.
Reliability contains two components - homogeneity and stability. Alternate
forms can check homogeneity while a test-retest procedure can check
stability. The simplest immediate way to begin is to repeat the pre-

sentation of several patterns within the context of an experiment.

123

There are many other methodological problems, such as the many
assumptions underlying the scale: equal interval steps; polar endpoints
are Opposite and equidistant; etc. Much work is presently being done

to test these assumptions.

Melodic Pattern - Connotation Relations

 

Theory

In the present research, overall music structure has been sliced
first in terms of components and then sliced across music styles as
best as was possible from within the component. It is hoped that the
melodic component will turn out to be sufficient in obtaining some
notion of all the dimensions of music connotation. By using only one
component it was possible to delimit the number of dimensions, levels
and variables more quickly. Then variable values could be selected based
on a clear notion of what would be needed in order to have a repre-
sentative sample. Due to this approach the ability was improved to
compare and contrast the resulting connotative meanings of patterns and
levels, since all the patterns were defined in terms of the same set of
variables.

The most obvious future research need is to manipulate more
variables Simultaneously. Other connotatively significant components
and dimensions such as overall loudness, performer variables, harmony,
texture and orchestration, must be measured both in isolation and in
combination to find Out about their connotative significance. Unlike
non-verbal communication research in an area such as facial expressions

where there is an observable referent which can guide the selection of

124

significant variables, almost every conceivable aspect of music structure
may have some significance in terms of connotation.

New music structure variable can be created. There are many ways
to operationally define variables: non-metric categories, quantitative
measures of the ordinal or interval metric variety, log formulas, ratios,
internal structure probabilistic relations, complex concepts with multiple
variable indicators, etc.

For instance, Dallin (19) states that there are 4-5 different
categories of melodic contours. Each category can be viewed as using
many different variable values in a set of configurations which have some
basic meaning Similarity among them and some meaning difference from any
of the other categories. Without elaborating on the details at this
point, this definition Should be subjected to empirical testing in terms
of hypothesized differences in connotative meaning across categories.
After presenting all possible combinations of values on appropriate variables
to a group of subjects, who shall respond on several scales, it will be
interesting to find out if 4-5 hypothesized clusters turn up in the
reSponse data. Possibly the data on pattern meanings will lead to
collapsing some of the categories or enlarging them and even ranking them.

Within the above example appeared the basic approach to determine
the conditionality of pattern variable value relations to connotative
meanings. That is, as variable value combinations vary, how do connotative
meanings vary? If the meaning of one value, from each of several variables,
is measured in isolation, what is the resultant meaning when all the

different values are combined? The goal here is to predict meaning change.

125

For instance, value combinations on several variables may produce a
summative increase in meaning on a scale. For example, if an increase
in rhythmic activity accompanies a rise in countour to the peak of a
melody, the effect is made stronger.

In general, mathematical models of intra (besides inter) - language
sequential and simultaneous combinatory rules involved in part—whole
organization within the connotative component can be develOped. Possibly
only certain classes of mathematical functions will fit the data and
predict behavior change. There are two kinds of predictive situations:
1) quantitative combinations: additive or multiplicative within a Single
scale; 2) non-quantitative combinations: emergence of new adjective scale
relevancy. It may be more difficult to predict meaning change on some
dimension scales than others. Meyer (51, ch. 8) thinks it will be
easier to predict meaning change on scales from dimensions III and IV
which appear to be more closely related to more observable surface
features of music patterns.

When much of the above research has been accomplished, then large
scale studies of music styles can be attempted. Modern jazz, folk music,
romanticism, etc. can be measured and compared and contrasted in terms
of scales and dimensions of meaning. The connotative distances between
styles can be obtained.

The unique stylistic features of various well-known composers can
be identified in terms of connotative meaning differences. For instance,
Paisley (65) suggests that a distinction should be made in terms of
major and minor encoding habits of composers. Possibly the minor, more

subtle encoding habits may have a different connotative meaning than the

126

major encoding habits, thus turning out to be the interpretation that is
used when referring to the most pleasing aSpect of a particular composer.
Trends in style change within composer, School, etc. can be
studied in terms of trends in meaning change.
Also, styles can be compared in terms of the nature and amount of

uncertainty and ambiguity of meaning for various receiver pOpulationS.

Methodology

Facet Design (27) is applicable to the present interest in
manipulating many variables in all possible combinations. Given sets of
‘values in each variable (facet), all possible combinations of values
will be generated (variables). With relatively exhaustive manipulation,
multi-dimensional interaction effects can be studied. Correlations be-
tween generated variables may appear as a "Simplex" or possibly a
"circumplex".

Also, analysis of variance can be used in conjunction with a facet
design to determine the relative influence of different combinations of
variable values on variation in meaning.

Finally, multiple regression can be used to establish quantitative
predictive equations, where several pattern variables which are highly
correlated to variation on some connotative scale are employed in linear

combination to predict changes in connotative variation.

BIBLIOGRAPHY

l. Abelson, R. P., and Sermat, V. "Multidimensional scaling of facial
expression," Journal of Experimental Psychology, 1962, 63, 546-554.

 

2. Aldrich, V. C., PhiloSOphy of Art. Englewood Cliffs, New Jersey:
Prentice-Hall, Inc., 1963.

 

3. Attneave, Fred, "Some Informational aSpectS of visual perception",
Psychological Review, 1954, 61, 183-193.

 

4. , "Perception and Related Area", in (ed.) Koch, S.,
PsycholOgy: A Study of a Science, Vol. 4. New York: McGraw-Hill
Book Company, Inc., 1962.

 

 

5. Barlow, H., and Morgenstern, S., A Dictionary of Musical Themes.
New York: Crown Publishing Company, 1948.

 

6. Barzun, Jacques, Music in American Life. New York: Doubleday, Inc.,
1956.

 

7. Beardsley, M. C., Aesthetics. New York: Harcourt, Brace and Co., 1958.

 

8. Berlo, D. K., The Process of Communication. New York: Holt, Rinehart
and Winston, 1960.

 

9. Birdwhistell, Ray, "Kinesics and Communication", in Carpenter, E.,
(ed.), Explorations in Communication. Boston: Beacon Press, 1960.

 

lO. Boring, E. G., Physical Dimensions of Consciousness. New York:
Dover Publishing Company, 1963.

 

11. Brown, R. W., Leiter, R. A. and Hildum, D.C., "Metaphores from Music
Criticism", Journal of Abnormal and Social Psychology, 1957, 54,
347-352.

 

l2. Bruner, J. S., "The Course of Cognitive Growth", American Psychologist,
1964, 19, 1-15.

 

13. Campbell, Ivy G., "Basal Emotional Patterns expressible in music",
American Journal of Psychology, 1942, 55, l-l7.

 

14. Carroll, J. 8., Language and Thought. Englewood Cliffs, New Jersey:
Prentice-Hall, Inc., 1964.

 

~\

\

lSSWCofer, C. N., "Experimental Studies of the role of verbal process in
concept formation and problem solving", Annals of the New York
Academy of Sciences, 91, Article 1, 94-107.

 

 

127

16.

17.
18.
19.

20.

23.

24.

25.
26.
27.
28.
29.

30.

31.

32.

128

Cooke, D., The Language of Music. London: Oxford University Press, 1959.

 

Cooper, G. and Meyer, L.B., The Rhythmic Structure of Music. Chicago:
University of Chicago Press, 1960.

 

COpland, A., What to Listen for in Music. New York: McGraw Hill Book
Co., 1938.

 

Dallin, L., Techniques of 20th Century Composition. Dubuque, Iowa:
W. C. Brown Co., 1964.

 

Deihl, C. L., "The Two Worlds of the Artist”, FORMAT, 1966, 2 (#5).

Farnsworth, P. R., "A Study of the Hevner Adjective Check List",
Journal of Aesthetic and Art Criticism, 1954, 13, 97-103.

 

Farnsworth, P. R., The Social Psychology of Music. New York:

 

Dryden Press, 1958.

Fearing, F., "Toward a Psychological Theory of Human Communication",
Journal of Personality, 1953, 22, 71-88.

 

Fishman, J. A., "A Systematization of the Whorfian Hypothesis", in
Smith, A. G., Communication and Culture. New York: Holt, Rinehart
and Winston, Inc., 1966, pp. 505-516.

 

Fishwick, M., "What modern art says about America", Saturday Review,
April 9, 1966.

 

Flesch, Rudolf, The Art of Readable Writing. New York: Harper Brothers,
1949.

Foa, Uriel G., "New DevelOpmentS in Facet Design and Analysis",
Psychological Review, 1965, 72, 262-274.

 

Forte, Allen, Tonal Harmony in Concept and Practice. New York:
Holt, Rinehart and Winston, 1962.

 

Fuchs, Wilhelm, "Mathematical Analysis of formal structure of music",
IRE Transactions of Information Theory, 1962, IT-8 #5, 5225 - 5228.

 

Gibson, J. J., "Perception as a function of stimulation", in (ed.)
Koch, S., PsycholOgy: A Study of a Science, Vol. 1, New York:
McGraw-Hill Book Company, Inc., 1959.

 

Grout, D. J., A History of Western Music. New York: W. W. Norton and
Co., 1960.

 

Guttman, Louis, unpublished report of an analysis of facial expression
data (See reference 34).

129

33. Harmon, H. H., Modern Factor Analysis. Chicago: University of Chicago
Press, 1960.

 

34. Harrison, R., Pictic Analysis: Toward a Vocabulary and Syntax for the
Pictorial Code with Research on Facial Communication. Unpublished
Dissertation. Michigan State University, College of Communication
Arts, Department of Communication, 1964.

 

 

35. Haydon, G., Introduction to Musicology, New York: Prentice-Hall, Inc.,
1946.

 

36. Henkin, R. I., ”A Factorial Study of the Components of Music",
Journal of Psych., 1955, 39, 161-181.

 

37. Hevner, K., "Experimental Studies of the Elements of Expression in
Music", American Journal of Psych., 1936, 48, 246-268.

 

Hochberg, J., Perception. Englewood Cliffs, New Jersey: Prentice-
Hall, Inc., 1964.

38.
39:)Jung, C: G. and C. Karenyi, Essays on a Science of Mythology.
New York: Pantheon, Bolinger Series XXII, 1949.

40. Kandinsky, W., Concerning the Spiritual in Art. New York: Wittenborm,
1963.

 

 

 

41. Kaplan, M., Foundations and Frontiers of Music Education. New York:
Holt, Rinehart and Winston, Inc., 1966.

 

42. Kelly, G. A., A Theory of Personality: The Psychology of Personal
Constructs. New York: W. W. Norton and Company, 1963.

 

 

43. Kennan, K. W., The Technique of Orchestration. New York: Prentice-
Hall, Inc., 1953.

 

44. Klare, G. R., The Measurement of Readability, Ames: The Iowa State
University Press, 1963.

 

45. Kohler, W., Gestalt Psychology. New York: Liveright Publishing
Corporation, 1947.

 

46. Langer, S. K., Philosophy in a New Key, New York: New American Library
of World Literature, Inc., 1951.

 

47:1Langer, S. K., Philosophical Sketches. Baltimore: Johns Hopkins PreSs,
~~»/’ 1962.

 

48. Leonhard, Charles and House, R. W., Foundations and Principles of
Music Education. New York: McGraw-Hill Book Company, Inc., 1959.

 

49.

50.

53.
54.
55.
56.
57.
58.
59.

60.

61.
62.
63.

64.

130

Marquis, G. W., Twentieth Century Music Idioms. Englewood Cliffs,
New Jersey: Prentice-Hall, Inc., 1964.

 

McQuitty, L. L., "Single and Multiple Hierarchical Classification by
reciprocal pairs and rank order types", Educational and Psychological
Measurement, 1966, 26, 253-265.

 

 

 

‘ Meyer, L. B., Emotion and Meaning in Music. Chicago: University of

Chicago, 1956.

Meyer, L. B., "On Rehearing Music", Journal of the American
Musicological Society, 1961, 14, 257-267.

 

Miller, G. A., Language and Communication. New York: McGraw-Hill
Book Co., 1951.

 

Miller, G. A., Pribram, K., and Galanter, E., Plans and the Structure
of Behavior. New York: Holt, Rinehart and Winston, 1960.

 

 

Miller, G. A., "Decision Units in the Perception of Speech", IRE
Transactions on Information Theory, 1962, IT-8, 81-83.

 

Mitchell, Donald, The Langugge of Modern Music, London: Faber 8
Faber, 1963.

 

Morris, Charles, Signs, Language and Behavior. New York: Prentice-
Hall, 1946.

 

Myhill, John, "Musical Theory and Practice", Journal of Aesthetic and
Art Criticism, 1955, 14, 191-200.

 

 

Nettl, B., Theory and Method in Ethnomusicology, Glencoe: The Free
Press, 1964.

 

Newman, W., Understanding Music. New York: Harper and Bros., 1952.

 

Ortmann, Otto, "On the melodic Relativity of tones", Psychological
Monographs, 1926, 35, 1-47.

 

 

Osborne, H., "The Quality of feeling in art", British Journal of
Aesthetics", 1963, 3, 38-53.

 

 

Osgood, C. E., Tannenbaum, P. H., Suci, G. J., The Measurement of
Meaning. Urbana: University of Illinois Press, 1957.

 

OSgood, C. E., "Dimensionality of the Semantic Space for Communication
Via Facial Expressions", Unpublished report. Institute for Communi-
cations Research, University of Illinois.

65.

67.
68.
‘,69.
70.
71.
72.
73.
74.

75.

78.

79.

‘1
I
6 l
6”
‘\--_.’l

J

131

Paisley, W. J., "Identifying the unknown communicator in painting,
literature and music: The significance of minor encoding habits",
Journal of Communication, 1964, 14, 219-237.

 

Palisco, C. V., Harrison, E. L. and Hood, M., Musicology, Englewood
Cliffs, New Jersey: Prentice-Hall, 1963.

 

Plutchik, R., The Emotions: Facts, Theories, and a New Model. New York:
Random House, 1962.

 

Pool, Ithiel de Sola, (ed.), Trends in Content Analysis. Urbana:
University of Illinois Press, 1959.

 

Rader, M., A Modern Book on Aesthetics. New York: Holt, Rinehart
and Winston, Inc., 1960.

 

Rapaport, Anatole, "Music and Abstraction", ETC: A Review of General
Semantics, 1946-47, 4, 182-195.

 

Rasmussen, S. E., Erperiencing Architecture. Cambridge: The MIT Press,
1964.

 

Reed, H. 0., and Harder, P., Basic Contrapuntal Techniques. New York:
Mills Music, Inc., 1964.

 

Reimer, Bennett, "A new curriculum of secondary general music",
Council for Research in Music Education, 1965, Bulletin #4, 11-20.

 

Reti, Rudolph, The Thematic Process in Music. New York: MacMillan
Company, 1951.

 

Rogge, G. 0., "Music as communication with Special reference to its
role as content". Doctoral Dissertation: Univ. of California at
Los Angeles, 1952.

Schlosberg, H., "Three dimensions of emotion", Psychological Review,
1954, 61, 81-88.

 

Seeger, Charles, "Music as communication", In Abstracts American

 

AnthrOpological Association, 60th Annual Meeting, Philadelphia,

 

Sessions, Roger, The Musical Experience of the Composer, Performer,
Listener. New York: Atheneum Press, 1962.

 

Solomon, L. N., A Factorial study of complex auditory stimuli (sonar
sounds). Unpublished doctor's dissertation, University of Illinois.

80.

81.

82.

83.

84.

85.

86.

87.

88.

132

Tannenbaum, P. H., Elliot, T. T., "Factor Structure of Semantic
Differential responses to visual forms and prediction of factor
scores from structural characteristics of the Stimulus Shapes",
American Journal of Psychology, 1963 . 76. 589-601.

Toffler, Alvin, The Culture Consumers. Baltimore, Maryland: Penguin
Books, Inc., 1965.

 

Tomkins, S. S. and McCarter, R., "What and Where are the primary
affects? Some evidence for a theory" Perceptual Motor Skills,
1964, 18, 119-158.

 

Tucker, W. T., Experiments in Aesthetic Communication. Unpublished
doctoral dissertation, University of Illinois, 1955.

 

Van De Geer, J. P., et. al., "The connotation of musical consonance",
Acta Psychologica, 1962, 20, 308-319.

 

Vygotsky, T., Thought and Language. Cambridge: M. I. T. Press, 1962.

 

Wertheimer, M., "Laws of Organization", in W. D. Ellis, Source Book of
Gestalt Psychology, New York: The Humanities Press, 1950, p. 71-88.

 

 

Whyte, L. E., "Atomism, Structure and Form", in (ed.) Kepes, Gyorgy,
Structure in Art and in Science. New York: George Braziller, 1965.

 

Woodworth, R. S., Experimental Psychology, New York: Holt, Rinehart
and Winston, 1938.

 

133

APPENDIX I

Three Group Pattern Order

A B
1 18
2 17
.9. 19.
u 15
5 14
.9 l3;
7 l2
8 11
.9. 19.
10 9
11 8
12 _z_
13 6
14 5
15. _u
16 3
17 2
18 l

10

12

134

APPENDIX II

MUSIC COMMUNICATION STUDY
Department of Communication
Michigan State University

NAME:

 

STUDENT NUMBER: AGE: SEX: MALE FEMALE

 

INSTRUCTIONS:

 

On each of the following pages is a series of adjective pairs. For
each page a melodic pattern will be presented. Between each adjective
pair are five Spaces, for instance:

“I.

1) friendly :__5__3__3__5__} ominous
You will be asked to make one judgment for each adjective pair. If,
to you, the melodic sequence seems very "friendly," you would place
a check in the first Space, under the double plus (++).

If the melodic pattern seems-somewhat ”friendly," but not ex-
tremely so, you would place a check in the second Space, under the single
plus (+).

If the melodic pattern seems somewhat "ominous" you would place
a check in the fourth Space, under the single minus (—).

If the melodic pattern seems very "ominous" you would place a
check in the fifth Space, under the double minus (--).

Finally, "friendly ominous" may not be an adjective pair which is
relevant to this particular melodic pattern. Or, it may be hard to
decide whether it is "friendly" or "ominous." In this case, you would
place a check in the third or central Space, under the zero (0). This
means "undecided" or "irrelevant."

For the melodic sequences in this questionnaire, there are no
right or wrong answers. The best response is what you feel is appropriate.
We are interested in what these melodic patterns mean to you.

135

APPENDIX III

Melodic Pattern #

++ + o - -- ++ + 0 - --
deliberate :__}__3__}__3__} impulsive formal :__3__3__}__3__} passionate
floatin :__}__5__5__}__5 dragging Lucid :__5__}“_,__,___obscure
pleasant :_____3__3__5__3 unpleasant sharp :__}__5__u__5__}blunt
aspiring :__}__5__3__5__} resolving solid :__3__}__}__3__}hollow
humorous :__5__}__3__}__5 serious graceful :__}__5__}__}__}awkward
steady :___: _ _: _: _: changeable good :__: _:_: _: _:bad
tender :__3____________ violent cautious :__}__5__3__3__3rash
tight :__}__}__}__}__5 loose plain :__}__3__}__}__}majestic
leisurely :__}__3__5__}__3 rushing rugged :__}__}__}__}__}delicate
gay :__}__}__3__}__} pensive friendly :__}__}__}__}__}ominous
introverted :_____}__}__}__; extroverted active :__3__}_____}__}passive
beautiful : : : : : : ugly strong : : : : : :weak

stiff :__}__}__}__}___ elastic happy

h
DJ
0.

136

APPENDIX IV

Evalua- Stability-
tion Mood Tautness Dynamism
I II III IV V h2
18 graceful - awkward .82 .03 -.O2 .18 -.11 .73
12 beautiful - Ugly .80 .16 -.02 .16 -.12 .72
19 good - bad .72 .23 .00 .25 -.13 .65
3 pleasant - unpleasant .68 .40 -.06 .14 -.10 .66
7 tender - violent .68 .07 .07 .13 -.36 .62
22 delicate - rugged .72 -.07 -.04 -.31 -.02 .62
21 majestic - plain .58 -.23 -.28 .22 .10 .53
2 floating - dragging .68 .29 -.11 .05 .16 .58
4 aspiring - resolving .59 .30 -.ll .09 .19 .50
10 gay - pensive .22 .83 -.O7 .13 .04 .75
26 happy — sad .21 .80 -.02 .18 .04 .73
5 humorous - serious .00 .80 -.07 -.Ol .03 .65
ll extroverted - introverted .05 .60 -.26 .21 .11 .48
23 friendly - ominous .46 .65 -.ll .10 -.14 .68
6 steady - changeable .09 .03 .69 .16 -.12 .52
l deliberate - impulsive .03 -.14 .60 .17 -.19 .45
14 formal - passionate -.20 -.08 .64 -.O4 .15 .48
8' tight - loose -.36 -.27 .55 .02 .29 .59
13 stiff - elastic -.47 -.30 .53 -.O7 .22 .64
25 strong - weak -.05 .16 .06 .80 -.02 .68
17 solid - hollow .10 .03 .25 .69 .05 .55
24 active - passive .11 .47 -.16 .54 .19 .59
15 lucid - obscure .24 .20 .01 .44 .05 .29
9 rushing - leisurely -.28 .04 -.01 -.08 .69 .56
20 rash - cautious -.17 .21 -.38 .09 .50 .48
16 Sharp - blunt _,35 .03 .13 .25 .55 .us
Explained Var. .21 3 TE jfi ‘53— Tag.

137

APPENDIX V
Distributions for Intense Adjective Associations

Single Tone Level

1 2 3
++ + - -- ++ + - -- ++ + - --
plain 27 22 4 1 majestic delicate 17 34 2 0 rugged plain 16 27 5 3 majesti
strong 25 14 9 2 weak graceful 24 35 4 4 awk-
ward strong13 26 9 1 weak
stiff 15 22 9 3 elastic tender 14 31 8 O violent
leisurely ll 32 5 3 rushing

beautiful 13 29 7 3 ugly

Two Tone Level
4 5
++ + - -- ++ + - --

deliberate 31 33 3 3 impulsive pleasant 26 33 7 1 unpleasant
humorous 0 2 42 20 serious strong 13 37 7 0 weak

gay 0 4 40 18 pensive friendly 20 30 100 ominous
formal 16 28 5 2 passion- leisurelyll 32 9 l rushing

ate
cautious 10 35 4 3 rash
happy 0 2 33 15 sad
rugged 13 29 8 O

6
++ + - --

floating 11 30 8 0 dragging
aSpiring 16 30 9 0 resolving

Short Motive Level

7 8
++ + - -- ++ + - --
steady 51 23 1 l changeable leisurely ll 45 3 0 rushing
graceful 0 2 21 27 awkward humorous 1 3 35 18 serious
stiff 43 26 3 l elastic rugged 1 6 35 14 deliberate
tight 37 27 4 O loose graceful 21 39 9 l awkward

gay 0 4 26 19 pensive tender 11 42 8 2 violent

138

Short Motive Level (continued)

9
++ + - --
active 23 36 3 1 passive
strong 14 38 4 2 weak
happy 21 27 5 0 sad
gay 14 38 5 l pensive
introverted l 5 32 12 extroverted
aSpiring 16 25 7 2 resolving
friendly 10 32 7 3 ominous
Long Motive Level
10
++ + - -- ++
strong 19 34 6 2 weak good 15
humorous 2 7 26 18 serious beautiful 12
graceful l2
12 active 19
friendly 14
++ + .. .. lUCid 8
floating 18
humorous 0 0 43 29 serious pleasant 25
leisurely18 41 2 0 rushing
cautious 5 40 5 1 rash
graceful 24 41 6 0 awkward
' tender 17 36 4 4 violent
rugged 4 4 37 12 delicate
beautifulll 33 4 3 ugly
Phrase Level
13
++ + - --
++
humorous 0 0 42 22 serious gay 1
beautiful 25 33 4 l ugly humorous 2
leisurely 19 45 6 0 rushing happy 3
tender 27 41 4 0 violent
rugged 2 3 37 18 delicate
good 24 30 5 0 bad ++
_ graceful 33 35 6 1 awkward strong 15
cautious 11 30 8 rash pleasantl4

11

31
33
43
34
39
35
42
39

uaovu>a>a>aaaau1
NJFJCDNJFJFJFJC>

31
26
31

35 8
43 11

bad

ugly
awkward
passive
ominous
obscure
dragging
unpleasant

25
24
13

pensive
serious
sad

1 weak

0 unpleasant

Melody Level

++
pleasant 37
friendly 36
happy 33
strong 20
good 14
gay 23
humorous l9
introver-
ted l
floating 18
active 25
lucid 12
++
graceful 1
humorous 0
gay 0
floating 1
friendly 1
rugged 19
good 1
' plain 19
formal 18

happy 2

16

+ .—
34 2
30 2
32 2
39 2
40 2
39 5
33 5
5 35
30 5
37 7
35 7

18

+ -
2 3O
4 29
5 37
6 22
7 28
28 6
7 28
24 5
25 5
6 35

I—‘OMOOOO

Ops—am

26
28
21
29
24

15

18

138a

17
++ + -

unpleasant humorous 0 3 40 19
ominous
sad

weak

bad
pensive

serious

extroverted
dragging
passive
obscure

awkward
serious
pensive
dragging
ominous
delicate
bad
majestic
passionate
sad

serious

139

Appendix VI

Melodic Pattern Connotative Meanings

1 2 3

123456789101112131415161718

u

5

6

1 2 3 4

6

123456789101112131415161718

 

 

 

 

graceful X XX X X X X X X X awkward X X X X X
beautiful X X X X X X ugly X

good X X X X X X X bad X X
pleasant XX XX XX X X X X X X X unpleasant X X X
tender X X X X violent X

delicate X X X X X X rugged X X X . X X X
majestic X X X plain X XXX X X X X X
floating X XX X X X X X X X dragging X X X X
aspiring X X XX X X X X X resolvingy_ X X
gay X X X X X pensive X X XX X X X X X X
happy x x x x x sad x x x x x x x
humorous X serious X X X X X X X X X
extroverted X X X X introverted X X X X X X
friendly XX XX X X X X ominous X X X X
steady XXX X X X X X changeable X
deliberate X XX XX X X X X X X impulsive

formal X X XX X X passionate

tight X X X X X loose X X X X
Stiff X X X X X elastic X X X X X X
strong XXXXX X X X X X X X weak

solid XXX X X X X hollow

active XX X X X X X passive X

lucid X X X obscure

rushing X leisurely X XX X X X X X X X
rash X X cautious X X X X
sharp X X X X X blunt X

140
APPENDIX VII
An interesting problem has appeared and Should be discussed briefly.
The following table Shows the overall usage of both adjective poses. The

factors are ranked in terms of equality of usage.

 

+ .-

II 23 36

v 8 15

I 68 33

III 33 10

IV 30 __g.
162' 9

Pattern meanings varied quite a bit for dimensions II, V and I but

not for dimensions III and IV.

 

Actually, within III: + -
Stability 21 l
Tautness l2 9

so for the stability component of dimension III and for dimension IV,
only one direction was used in terms of consensus.

In terms of scales, strong-weak, solid-hollow, formal-passionate,
lucid-obscure, and deliberate-impulsive only resulted in intersubject
agreement in relation to the first pole with the present set of patterns.
For these scales the second pole was always used less. Other scales in
which the second pole was usually ambiguous are leisurely-rushing, sharp-
blunt, steady-changeable, serious-humorous, plain-majestic, aSpiring-
resolving, beautiful-ugly, tender-violent and active-passive.

There are several reasons for treating these poles in this manner:
1) the sample of melodic patterns was not broad enough to map across the
connotative space. That is, the sample of adjectives was too serious,

’ graceful, plain and possibly not sufficiently majestic, passionate,

141

APPENDIX VII (continued)

humorous, violent, weak and passive. Of course, it could be that melody
alone is infrequently associated with these connotations. Maybe these
poles are most frequently associated with other components of music or
when all components appear together.

2) bad choice of labels on the less used polar positions. Possibly
obscure, blunt, hollow, rushing and resolving are vague in meaning.

3) music connotative organization on some of the scales, eSpecially in
factors III and IV, are uni-polar. Some scales might be changed, such

as steady-unsteady, and active-inactive. It could well be that all these

reasons are applicable.

"7'1? 11111111111411“