MSU

LIBRARIES
n

 

 

 

RETURNING MATERIALS:
P1ace in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

 

~ MAR 2 7 2006

 

 

 

 

EXPLORING MUSIC PREFERENCE:

AN EMPIRICAL STUDY OF R LEXICON FOR TIHBRE

By

Thomas Gerard Faee

A THESIS

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

MASTER OF ARTS

Department of Telecommunication

1988

ABSTRflCT

EXPLORINS MUSIC PREFERENCE:

AN EHPIRICRL STUDY OF A LEXICON FOR TIMBRE

By

Thomas Gerard Faes

This thesis studies the relationship of two communicative phenomena,
music and words that common people use to describe music preference. fin
experimental design was used to test whether the lexical set of words
that describe music preference vary with timbral variations. Using the
music performance telecommunication standard called MIDI, timbre was
varied while other performance attributes were controlled. Using one-way
analysis of variance} significant differences were found between test
timbre descriptions using the words: Exciting. Fun, Intense, Melodic,
and Relaxing, although only Intense and Melodic showed significant
difference across the combined variance of all 4 conditions. Using a
traditional lexical set of words that describe style of music,
significant differences across timbre conditions were found using the
words Rock and Jazz. The results demonstrate the viability and utility
of using precise instrumentation. experimentally controlled testing, and
listeners own vernacular to guage music preference as a function of music

message components.

C) Copyright 1988

Thomas Gerard Faes

Dedicated to that human sensation
and communication
that accompanies hearing
some new music
which you like upon first listening
and

continue to like for a lifetime

iv

Acknowledgements

The lineage of any human is relevant to the present and for this

reason let me thank my father, Frederick Ferdinand Faes Jr. for his
musical inspiration he has been throughout my life. As a Junior high
school band director during the day and professional musician during the
night, he showed me what a sensation music can be to the human spirit.
Uith the lessons received at a young age, discipline was also learned.
Uhen the recordings of his school’s bands were brought home for me to
practice with, I truly learned about time-shifting with recording
technology. In a pragmatic sense, my father told me in the early years
of playing drums, that the important thing in drumming a beat is not to
worry about how fancy you can get but rather how steady your time is, and
most important 'Juet don't stop.‘ It must be his Swiss desire for time
consistency. In addition, I have shared the stage with my father and
have grown very used to top quality musicianship and saxophone playing
in general. His fascination with history, discovery and exploration has
also had great influence.

On the other side of the genetic coin are the influences of
happiness, adaptability, sensitivity, and an incessant desire to further
analyze, understand and link human behavior to physical explanation.
Helen Sullivan Fees is all this linked together from combining a
French farmer and Irish businessman. Protect thyself from useless
influence, and watch your 'tona of voice' are her favorite axioms.
Adhering to this has allowed me to always find brilliance in the
influence from my immediate group of significant others. In conjunction

with this goes a sense of humor that knows no boundaries. When the chips

were low, mother can always explain the good side of it. Her constant
desire to practice and study health care nurtured not only a mind for
scientific inquiry but also a warrant to perform in the interest of human
welfare. In addition to her practiced skills her motherly qualities
raised an entire family no less and even though as strange and all-
american as any other family, contained a feeling a coherence to a social
cause, usually as individuals, but usually towards the same cause. These
siblings get a mention. Kathy was the reference from ahead in life, what
life could be like, and how the woman's role had changed. From the first
sounds of the flgg1_1hg_flggtlgg album to the recent cassette
correspondence of Pat Hethaney, Kathy has been a stable reference point.
Even when her global position changes, it seems to drew me toward her for
a brief visit and maybe do something as important as buying a synthesizer
that can be used to test timbre. Kathy is a discoverer, and has always
made me welcome from wherever she was discovering. Fred is the light-
hearted one that has regulary reminded me to 'don't worry about a thing,’
it will be alright. Uhile Kathy was listening to the Doors and
contemplating serious issues, Fred was listening to 'Uild Thing' or
dancing up a storm to James Brown. Dr. Fred has also been an academic
influence by helping discuss career goals and conveying fields of
interest that he had opportunity to study. One was anthropology which
still to this author is of great importance to human inquiry. John, is
the adjacent note in life that has been in touch with me for as long as I
can remember. Had him and I not been adjacent, we might understand each
other better but we make up for that by being constantly activity
oriented when together, having good times in the widest variety of

settings including the outdoors, on the bandstand, or Just getting

vi

together for a hockey game. He is the adjacent one on the bandstand that
helps me hold the beat together in a music from the times of swing,
latin, or dixialand. Ue are a rhythm section in music and in life.

Never playing the same riff, but never too different either. Recently I
thank him for his valuable insight on writing and the importance of doing
each draft from start to finish, no stopping. My family was my first
laboratory and all helped tasting.

Robert E. Rockwell was a fire chief and gunsmith by trade and his
leadership skills and manual artistry were unsurpassed. Uhether it be
discussing a welding torch, rifle or deer antler rub, he always
demonstrated grace while handling observation. He believed that there is
always a solution, and you’ll eventually find it. If you don't have
the part, make it and he regularly taught respect and experience in the
outdoors, for he believed the adaptive skill helps one for life. Uhen
you are stuck, and can’t possibly get out, laugh about it for a little
while and then wait and think about how to get out. Eventually you’ll
get out. Be careful and than be careful again, and always value double
checking. Uith his practiced rules from the trades of emergency, danger
and explosives the lessons learned were always of a safe, adventurous,
Journey through mechanics, history, geography, and exploration.

Academic influences have been great, but of course there are too many
to discuss here. There are significant points in life where you take a
great turn in personal growth. In this light, first let me thank the
Catholic nun, Sister James who encouraged this author in first grade to
go ahead and perform the Beatles 1w111_gnfl_§hgu1 during "show and tell.“
Although the mime of the song, the cardboard drums and the plywood guitar

probably didn’t resemble the Beatles, it did allow for a young child to

vii

experiment with music, and see what the reactions of the other students
were to this, at the time, whirlwind of talent called the Beatles.

James Halter Dinkelackar, Ph.O. is an unbelievable occurence of a
friend, teacher, fellow musician, scientist, and futurist that any human
could have the opportunity to meet once in life. If in the course of any
project one becomes overwhelmed with the number of "stones still left
unturned,“ talk to Jamie. He will quickly demonstrate first that half
of the stones still unturned are irrevelent to your analysis and don't
need turning, and the rest can be raised partially to obtain the data
needed without expending energy to entirely turn them over. Jamie has
been my mentor in scientific research, exemplifying that social science
can be as scientific as the physical sciences, as long as you know your
goals, measurement and analyses. Of course the write-up of your results
is also an important step in science and for this reason Jamie regularly
reminded me that these writing skills would only be acquired in graduate,
school, which he'd state is required by any individual desiring
excellence and innovation. In many ways Jamie has been the reflective
force, or maybe more accurately human amplifier, ravarb unit, delay,
stereo chorus, or any other ”idea processing” of many interactions that
molded the ideas in this work. If you need a different view, Jamie
regularly has one to offer while at the same time if you Just want
support and amplification of a clear idea, he will do that also. As
often goes with such a mind of inquiry, a healthy future view has always
been there with a happy outcome for humanity. Space is the place, solar
energy stations in orbit, commercialize space and shut down outdated
dangerous means of generating energy. We have witnessed the U.S. swing

to the right in the United States, the Olympic boycotts, right on thru to

viii

the recent arms limitations agreements and Soviet withdrawl from
Afghanistan. Together as humans, united in force. Hey all other humans
on spaceship Earth have the opportunity to interact with such another
special being.

Jeffrey 'Prog' Marriott is a brilliant computerist, historian and
comrade in global debate and contemplation. Since meeting Jeff and
working side by side in the data processing environment we have
discovered how two people of drastically different orientations and
disciplines can augment and balance each other. He is the epitome of
logic and insight, and always seems to ask that right question which
steers one's mind to the critical issues. His companionship in outdoor
excursions has always been a valued break from the mundane schedule and
workflow of graduate school. Prog knows no limits in his ability,
because he will always acknowledge and.pursue the 'otherwise.‘ He
practices counter attitudinal advocacy like no other, such that if you
don’t talk about something, it won‘t exist and if you discuss a plan
you‘ll soon see it happen. I thank Jeff for his consistency.

Brenda Harie Rugenski is a recently found friend with whom I shared
many long ago experiences from our Detroit pasts. Because of our similar
symbol sets, and her incredible ability to focus all energy on others in
any interaction she’s involved in, quite often she focused that energy on
this author, to believe in himself and let no one interfere with the
absolute requirement of self-determination. In addition, the need to
quickly get to the heart of the matter, and not dispose oneself to
auxiliary distractions was at the forefront of messages from Brenda
Harie. Brenda is a valuable partner in mind and spirit. Our ability to

discuss major decision points in this graduate program and thesis when we

ix

did not formally share the symbol sets of music, computers and research
is Just another example of the wonders of human communication with a
brilliant mind. Brenda’s brilliant mind can talk about anything, always
leading to progress. I am thankful to her for the imperitive scrutiny
with which observations got filtered into decisions, so that this work
closely fit the personal goals of the author. Teaching to be in tune
with others brings you in tune with yourself is the rule of Brenda.
Timing is everything and if miss a chance to go over the rainbow you may
lose it for a lifetime, so don't hesitate.

Hy committee members have helped me finish graduate school and in
many ways that is all that counts. Dr. Thomas Huth is an absolute joy to
interact with, in either specifying legal interpretations, referencing
discovery or running with creativity. His constant enthusiasm for one's
own ideas made the whole project fun to do and sparked the desire for
the next one. The various points at which we met were succinct in time
with a status report of progress and a plan for the future. Progress
was regularly made under his guidance and questioning. Gary Reid was my
overall influence in audio production here at M.S.U. and is very good at
what he does in production and teaching. His ability to entertain while
teaching makes the ideas flow into the mind without much effort, while
his wealth of production ideas always offer Just one more approach to
offer as a solution in the studio session. Whatever you want to do is
'cool' and once you were done it sounds "pretty hip,‘ were the
encouraging words anyone needs to keep pursuing their production dreams,
and Gary was always there to offer such encouragement and insight. Two
other faculty members deserve mention. Dr. Larry Redd has been an

inspiration since the seeds of these ideas were planted ‘0 years ago

especially when he helped me plant the first 'music inquiry crop' back
than. Upon returning to graduate school he has readily made himself
available to me for fine tuning my music industry and research ideas and
I am very grateful to him for that. Dr. Carrie Heater has been a
reference point at many times for experimental design, analysis and
innovation. Carrie has that artistic ability to bring out the best in
you through sound logic and unlimited horizons. Carrie is the driving
force behind maintaining a technology environment for the aspiring
communication scientist, and as a result you can get familiarized with
Just about any computer technology through Carrie.

There are a whole host of other people that I would love to discuss
in great length, but I will acknowledge them by name and two words that
describe prominent charactiristics which I associate with them and have
influenced me as I finish this step in life. These wonderful humans are:

Jeffrey Bagnasco: order and harmony, Hark Blankenship: philosophy
and guitar, Kimberly Downing: drive and perseverance, Hark Fitzpatrick:
interpret and environments, Uilliam Lusk: excellence and rhthym,

Hike Lyon: calm and invention, Jean HcCasey: content and kindness,

Pete Racine: dreams and peace, Frank Rapin: Joy and enterprise,

Joey Reagan: organize and produce, Virginia Trierwieler: alert and
questioning. I am thankful for the virtues they have shown me. Lastly I
am simply thankful for the regular communication with my Brittany
Spaniel, Ponty, whose regularity demonstrates a healthy adaptation, whose
rhthyms are sometimes more sensitive than people, and whose sounds can be
as revealing. He supplies an important reference for determining which
of our human behaviors are human and which are Just the result of being

members of kingdom animalie.
Rock On' and Jazz, tgf

xi

13W

I. Introduction . . . . . . . . .

Music Communication . . . .

Music Code Systems . . . . . . . . .
Lexicon for Feedback . . . . . . . . . .
Music Technology . . . . . . . . . . .
Timbre: Digital Control . . . . .
Timbre Design . . . . . . . . .

Music Programing . . . . . . .

II. Literature Review . . . . . . . . . . .

Introduction . . . . . . . . .

Music Messages . . . . . . . . . . .
Tone . . . . . . . . . . . .

Timbre . . . . . . . . . . . . . . . .
Lexicon of music . . . . . . .

Music Preference . . . . . . . . . . . .
Topic Synthesis . . . . . . . . . . . .
Mass Mediated Music . . . . . . . . . .
Youth Culture . . . . . . . . . . . . .
Measuring Music Preference .

III. Methods . . . . . . . . . . . . . . . . .

Introduction . . . . . . . . . . . . . .
Timbre . . . . . . . . . . . . . .
Preference Lexicon . . . . . . . .
Music Preference . . . . .

Ratio Scales . . . . . . . . . .
Analysis . . . . . . . . . .

The Research Question and Hypothesis
The Test Timbre and Music . .

The Recording Session . . . . .

MIDI Control of Experimental Performance
The Listening Tests . . . . . . . .
Self-Report Instrument . . . . .

IU. Results, Discussion and Conclusions

Results . . . . . . . . . .

Lexicon for Preference . .
Hypothesis Testing . . . . . .

Like and Pleasing . . . . . .

Style Lexicon: Findings . . . . . . .
Femiliarity:_Uhere . . . . . . . . . .
Demographics and Other Findings

Discussion . . . . . . .
Recommendations . . . . . . .
Conclusions . . . . . . . . . . . .

U. References . . . . .
Footnotes . . . . . . . . . . . . . . .
Bibliography . . . . . . . . . . . . . .

xii

‘

001%!“

10

I4

22

22
23
24
25
30
37
41
42
48
49

53

53
53
55
S7
58
80
63
64
SS
65
BS
68

SS
70
74
77
78
79
80
82
91
94

98
98
99

LIST OF TABLES

Table Number Title Page
1 Music Technology Evolution: The Past 100 Years 11
2 Components of Bells, Brass, and Strings S4
3 Sample Demographics BS
4 Sample Music Exposure Profile 70
5 Mean Rating of Lexicon for General Preference 7i
6 Comparison of Rank Order and Ratio Scales 7i
7 Factor Analysis Summary 73
8 Lexical Difference of Timbre by Timbre 74

Combinations
9 Lexical Differences of Timbre by Timbre 75
IO Preference Lexicon Means by Timbre 76
II Groups Means of Music Style By Timbral Group 79

xiii

Figure Number

I

2

LIST OF FIGURES

Title
Model for Music Communication
Expanded Model for Music Communication
Hevner AdJective Checklist

Three Dimensional Plot of Trumpet Timbre

xiv

Page

31

55

Intcggggtign. Timbre is part of musical messages that describes how
different musical sounds are distinct (Herder, 1975). Whether it be a
bass drum beat, a bright trumpet sound or a simple sine wave, the timbre
can be described. This description can be provided linguistically and
by mathematical formulations. Code systems are used to discriminate
different timbre. Producers may use the word brightness, engineers may
talk about decibels and musicians of richness. Listeners describe their
preferred music with a shared vocabulary or lexicon which for the music
researcher is a vital symbol set for communication with producers and‘.
listeners. Timbre is a lexicon of music producers and performers,
however the common music listener does not use such formal lexical terms
for music. Instead the average listener will use a set of words shared
with peers to describe music one 'likes.' In describing music and
accompanying socialized responses, this code system is very valuable.
The lexicon of music preference is a measurable phenomenon. As music
changes so does preference. Timbral changes often accompany changes in
technology, since the creation of new types sounds can be the direct
result of new technology. With the introduction of new digital
production and telecommunication technologies, new timbres are invented
everyday. Timbre is changing. This thesis studies whether timbre is
related to musical preference. The focus is to measure the relationship
between a music production variable: timbre: and, a music listeners’
variable: the lexicon for music preference. If a relationship exists
between timbral structure and listeners’ preference then utility is
provided for those whose goal is to produce, select or perform music.

With such an analytical tool, gatekeepers of music messages could then

formulate production and programming decisions based on a music

parameter: timbre.

flggig_fignagniggtign, Music is an old form of communication. In some

sense it was the first form of communication that could travel over long
distances, or "telecommunication." The sounds of drums and horns have
been used for centuries to summons the members of language communities
(Fishman, 1972), using recognizable signals, distinct from other
communities. From the Greek lyre, to the Japanese koto and the American
Indian drums, different peoples create different sounding timbre. Music
has effects on listeners’ physical and social behavior, resulting in
ritualistic behavior such as dance and singing. In addition to these
physical behaviors, social behavior such as talking and expressive
reactions among listeners, is commonplace. A simple music
communication model may include a message, media transmission,
listeners, and feedback. This process can be depicted as:
Music -> Transmission -> Listeners
. 1
Feedback
Model for Music Communication
Figure 1

Music is defined as the science or art of ordering tones or sounds
in succession, combination, and temporal relationships to produce a
composition having unity and continuity (Webster, 1985). Four
properties are vital components of music: I) pitch, 2) intensity,

3) timbre, and 4) duration (Harder, 1975). Different cultures differ in

their definitions of tone or what combinations of tone are meaningful

and how unity and continuity is obtained, however universally music is
controlled sound over time, in all cultures (Merriam, 1984). Music of
many styles is communicated in many forms and technology. Technology is
available for the.performing musician and the recording engineer alike
to help produce the music. Such tools as studio electronics and
synthesizers aid in the creation of the message, where transducer
technologies such as microphones, recording heads and speakers allow
conversion of analog signals of one form to another, while magnetic
recording and compact disc media are used to record messages. Listeners
hear music broadcast on radio and television plus hear music stored in
the various playback formats such as compact and vinyl discs, or
magnetic tape. At this point any meaning received is described in
feedback. .

A goal of mediated music communication is to match the reproduction
quality of recorded message to the original message. Much music
communication and telecommunication technology has been developed to
retain as much of the original message as possible through the
progression of media, transducer and transmission technologies, i.e
microphones, magnetic tape, radio, television and speakers. This
reproduction quality goal is related to a fundamental communication
goal. That is to transmit messages which convey meaning to listeners.
Feedback is used to identify communicated meaning which can then be used
to formulate messages in future interactions. Music preference is one
type of music listener feedback, and it has been traditionally reported
in the form of 'charts'(l) which are trade magazine rating systems of

which there numerous attempts. There have been intermittent cases of

music research such as street testing and interviews used, but thus far
have not become popular in industry. Quite recently music research has
been adopted by radio program directors. As economic factors play a
more significant role in an industry with increasing supply of music and
consumer music duplication with decreasing purchase demand, the need for
more formalized music feedback becomes apparent. Refining the above
model the production and media technologies now discussed can be

inserted as follows:

Music Storage & Transmission Listeners
Composition \ / vinyl disc \ cable \
Performance I -> I compact disc -> broadcast I -) Listeners
Recording I \ magnetic tape I playback / 9 dance

‘ 9 charts

I 9 research
Feedback

Figure 2 Expended Model for Music Communication

flggig_gggg_§¥gtgag. Western civilization standards of'music science“
were derived by Pythagoras in the identification of the harmonic
overtone series about 550 BC (Finney, 1935). This established
fundamental harmonic relationships of music in Western civilization that
later were further described with counter-point and chordal harmony
systems (McHose, 1947). Code systems standardized the succession and
combination of tone in a unifying and continuous function. These code
systems also enabled a process of communicating with the composer's
idea, by reading music notation. Since the notational advance music of
Bach’s system of harmony, music code systems have remained the same,
while timbre has been modified and the "electronic epoch“ has dictated

an electronic path of development for music reproduction, storage and

transmission and with synthesized creation of music. Broadcast
transmission via the radio spectrum is now a prominent part of the
commercial music industry and music cultures in general. The code
systems have evolved over the centuries, effecting communication with
composer, through many timbre and electronic devices to be passed onto
listeners.

The code systems for listeners in assessing the impact of music, can
be physical behavior and/or verbal lexicon. The code system for
feedback is primarily the code systems of the listeners, however in
specialized test settings, the coding scheme can adhere to more formal
measures available from survey techniques, end/or laboratory measuring
devices such as a spectrometer or listening questionnaires. There are
code systems used for each step of the music communication process. In
this thesis, two code systems will be measured to see if there is a
relationship between them: I) a production parameter timbre, and 2) the
listeners' verbal lexicon. These can be viewed as two communicative
phenomena of the message and the feedback. To measure this
relationship, some recent digital coding of music performance has added
further scientific controls.

Currently, digital coding of music performance information for
communication between composer and performer plus between performers
enjoys continuing development. This coding has directly impacted the
timbre of music we hear. Group music performance communication
traditionally is a point to multipoint system, from composer to
orchestra members via notated music sheets while other messages about

performance dynamics and tempo changes are sent visually directly from

the conductor. In the late 1800’s smell combos became common, and an
improvisational group communication process became described as "Jess“,
now referred to as Jazz. Here a communication process between
performers was recognized (Merriam, 1964), including not only code
systems in the music itself but also haptics, facial gestures, and body
language. Recently music communication has been advanced with new
digital coding schemes, with a telecommunication standard called MIDI
(Moog, 1982) which performs digital interperformer communication. This
impacts musical timbre by enabling many previously thought separate
timbre to be combined in time and space, into one timbre. In addition,
a microcomputer based MIDI interface can automate(2) music performance
which aside from reducing the need for human performers, music
performance messages encoded digitally reduce noise, and expend
flexibility through enhanced editing capabilities. Code systems
describing music become more precise in time. This has particularly
effected musical control of timbre as music instruments become
constructed of digital circuits and the codes systems of ggcfigpmgngg and

cgggcging converge into one integrated code system.

Lg;iggn_19:_£ggflhggk, The lexicon of music preference is a valuable

code system for feedback between the listener and music sender. It
identifies the meaning decoded by the listeners. Verbal labels used by
individual listeners to describe the impact of music can be similar with
others who are members of the same population. It is these verbal
labels which are shared with others that makeup the lexicon of music. A

Preference meaning or rating can be derived from inquiries of 'music

they like," and then the lexicon becomes a lexicon for‘music preference.
In the music communication process, the sender of the music is attuned
to and receives listener feedback. Traditionally live audiences
responses could dictate to the composer their next compostion, and the
performer what to play next. Dance is also an means for feedback for
audience, as is singing. Feedback has taken form in the “call-in
request lines“ of broadcast radio music and more recently music
preference research has been operationalized for commercial music and
radio program decision making.

Music listeners hear music and describe it with the available
lexicon. Music preference research usually classifies music by songs,
artists and styles of music. Feedback on these can be obtained, however
listeners are able to identify further domains of what they hear in
music. They will describe the music with a mixture of colors, textures,
flavors and moods (Toole, 1981) plus some reactions and music
parameters. When asked to describe the 'music one likes,“ these
responses shared with others can reveal a whole host of other music
parameters such as happy, loud, beat, exciting and fun. Scaling music
preference represented by these words can yield a higher level of
precision than style, artist or song title by describing the sound
contained in the music, as well as some moods and indicators of
activity.

By systematically varying a music parameter such as timbre, and
measuring the lexicon of music preference, a relationship of message
structural parameters and lexical feedback parameters is studied. By

testing this relationship with systematic variation of timbral music

content, the actual social referential definition of these different
timbre can be studied. In this way using a more precise set of
preference measures, the lexical set of music preference becomes a
scale for comparing different songs, of any artist, style, or timbre.
It would be feasible then to examine if Jazz listeners in Los Angeles
find bell timbres “exciting“ while to the New York Jazz enthusiast
defines 'exciting' with a brass timbre. This exemplifies a case where

timbral preference would vary with geography.

flggig_lgghnglgg1, The progression of audio telecommunication technology
has enabled new forms of music content in both production and
performance. The late nineteenth century was met with the introduction
of an array of audio production technologies. Alexander Grahm Bell
originally labelled his invention the 'harmonic telegraph,‘I which was
meant to act as a device for the hearing impaired, for transmission of
phonetic verbal audio signals. It was only after realizing commercial
viability of telephone that the 'harmonic telegraph“ invention halted.
The audio frequency spectrum was established as an electronic medium for
human communication. At the same time were other ventures into
assessing what listeners actually did hear in music, and in physical
sounds or tones. With microphones and wireless transmission (Fessendon)
music became a broadcasted phenomenon in 1906 (Head 8 Sterling, 1987:
Hybels C Ulloth, 1978). Since then broadcast technology for audio has
produced microphones, magnetic tape, multi-track recording, and
amplification. The evolution of audio technology has not only
established audio frequency spectrum electronically as a means for human

communication, but has also enabled new forms of music and music

performance to evolve. Amplification enabled such mass communication
phenomenon as the outdoor concert festival. The digital coding
capability on magnetic tape eliminates noise. The digital coding of
performance information has enabled music timbres to be created and
transferred in various telecommunication media, around the world. Since
the late nineteenth, telecommunication technologies have almost always
had on an impact on music.

As technology advances, so have the available musical instruments.
Molding and molding of new metals led to balls of brass, while with
inventions of levers we received pianos and saxephones, and as discussed
already telecommunication has had a regular impact on music in the last
century. During the electronic revolution transducers technologies that
have affected timbre include distortion, wah-wah, and reverb along with
and through the added response of the dynamic moving coil and studio
condenser microphones, increasing reproduction quality. Not only
could Voice be carried over longer distances, but could also create
sounds such as 'crooning' (Frith, 1986) that were not quite
characteristic of gngnglifiigg long distance vocal sounds (Russell,1986).
By 1964 voltage controlled oscillators led to our first electronic
synthesizers (Moog, 1982). With the development of musical synthesis,
the idea of making the sound attributes independent of the physical
structure of the performance instrument now became evident. Up until
this time the timbre of music had frequently been associated with the
acoustical properties of instruments. Now multiple timbres can be
produced by one instrument. With the influx of digital integrated

circuits, it is not uncommon to find a sythesizer containing 128 sounds

10

built-in to the machine, with more sounds available via memory

cartridges, digital cassette tape, or floppy diskette.

Iiagcg;_nigjtgl_§gntggl. The formation of a standard for digital musical

instrument telecommunication is available via a protocal called MIDI, a
nmemonic for musical instrument digital interface. This interface
communicates performance information between MIDI compatible equipment,
such as computers and other synthesizers. Now, computer communication
has become commonplace and “countless thousands of music producers have
embraced its capabilities“ in recording studios as control centers for
music performance (Moog, 1996). Digital control of performance allows
a music that can be easily produced with much less cost of music
production. Today it is possible to record a sequence of your
composition on microprocessor based computer system utilizing no
microphqne transmission of signal to tape. Eliminating a microphone
from the recording process has been common for some time, known as the
“direct injection method“ (Runstein, 1986),.but now tape is being
eliminated. The advantage of this is a reduced noise floor and
elimination of transient microphone responses and studio noises. One
new characteristic of the “MIDI recording“ is that the information being
recorded is digital instructions of performance and not an analog of the
music. Therefore, requiring one to travel with instruments and computer
to the studio to playback the performance instructions. Once there
however, they can produce a master that never travelled through the air
and has virtually little noise introduced into the signal. This digital

recording is only one outgrowth of the MIDI revolution (Baird, 1986),

Year

11

Technology

Impacts

 

1876
1906
1948
1960
1965
1970

1980

1982

Harmonic Telegraph

Radio (Fessendon)
Television

Multitrack Tape Recording
Electronic Music Synthesis
Cassette Tape

Music Television
Digital Synthesis
Digital Tape

MIDI

another is timbre.

Audio Spectrum

Music Broadcasts

More Radio Availability
Program Diversity

New Sounds

Profits Decrease
Copyrights and Ownership
Sales Drop

New Medium of Music Promo
Natural Sounds

Less Noise

Music Performance Code
Timbre Communication
Studio Computing

MIDI telecommunicates timbre in various forms. Through the ability

of simultaneous musical instrument control,

extensive mixtures and

combinations of the timbres are enabled. In the MIDI specification, a

communication chip is on-board any MIDI compatible synthesizer machine.

This chip enables the communication of musical information between two

or more machines, simply with the insertion of a MIDI cable between

successive machines.

A microcomputer at the heart of a MIDI system is

becoming a valuable musical performance network, allowing timbres that

are so extensive in numbers, mixing and matching with each other via a

musical telecommunication standard (MIDI), all separate from any

magnetic recording tape.

It is possible to transfer timbre between

synthesizers of the same manufacturer, and this alone has impacted the

timbre we hear.

One can purchase tapes and memory catridges of new

timbre or timbre can be immediately transferred between some instruments

via a MIDI cable demonstrating that the telecommunication of timbre.

The music industry is feeling MIDI’s effect, and the listeners are
hearing its effect on timbre.

From a scientific control standpoint, the level of performance
control with MIDI has specific research implications. The cost of
producing test music can be cheaper to compose, record, and
systematically vary. With MIDI control a researcher can vary a variety
of performance parameters, systematically whether it be timbre, tempo,
melody, or mode. The scientific control of various timbre is obtained
through MIDI’s ability to work simply between two or more synthesizers
making possible simultaneous performance and therefore control. The
live musician performance is less controllable than a machine
‘performence, and therefore there is reason for attention to to be given
to MIDI as a music research instrument. MIDI can provide controlled

performance of different timbre.

Ijahng_ngljgn, One major outgrowth of early electronics was and still
is the ability to appeal to large groups of people using amplification.
At this time a discrete set of musical instruments could be used
electronically. These usually included guitar, bass guitar, organ and
later electronic piano. Later harpsicord and clavinet were often
included as alternative timbres that could be switched on electric
pianos and organs. The number of instruments available was limited,
but the impact of electronic timbres became prevalent with the addition
of each new instrument.

A major area of development in the decade of the 1970's was

extensive sound processing and sound synthesis. Sound processing of

W_-

13

“wah-wah', distortion, flanging, phase-shifting were all brought within
control of musicians, each modifying the wave form of sound. In
addition new in-line studio gear such as delays, exciters, gated reverb,
and limiters became utilized. All of these processed the signal of the
soundwave and always modify the timbre. When sound synthesis became
available, it was quickly used as a primary instrument in new styles(new
wave) and for complimentary filler sounds such as strings in other
styles(disco). By the 1980’s, timbres available in synthesized form
increased enormously. In addition digital coding enabled timbres to be
programmed, stored, retrieved, editted or written to tape storage for
input to other synthesizers of the same manufacturer.

Today the range of available sounds increases with the addition of
each new synthesizer but also due to a transformation of musicians'into
regular telecommunicators. They not only can receive new timbre via
telecommunication media, but create new timbre through “real-time
layer“ (Anderton, 1986) of two timbre into one, control multiple timbre
simultaneously and exchange timbre via electronic messaging systems.

On these messaging systems are timbre libraries. Resembling the
computer software industry, the music sound design industry has adopted
computer based electronic bulletion board systems for methods of
disseéinating and exchange of musical information from timbre to machine
performance to musician teleconferencing. In the transfer of
information there are many new timbres transferred from one side of the
country, to the other, simply via telephone exchanges. Music
performance and timbre is becoming programmable, and musicians are

becoming software engineers in the process. Through the whole available

14

range of telecommunication and sound design technologies, timbre is a
music parameter under great change. All this new timbre has meaning

since people apparently like what they hear in new timbres.

flggig_flcggcgaing. Some music is preferred by listeners more than other
music. Listeners like some music at certain times more than other
times. The identification of audience segments is important for a
programmer (Dinkelackar, 1982) so they can program to their target
market segment. The task of programing music for any potential media
demands the ability to play preferred music of certain market segments
at the right time. Within these constraints, the broadcast radio music
programmer assembles pleylists made of new releases and established
songs that are believed to be preferred by the target market. This
“believed preference“ can be formulated from a variety of sources
including past record sales, chart performance, survey and music testing
research.

The recording industry traditionally uses chart performance as a
measure of music preference (Hesbacher, 1978). Chart performance has
been referred to as an indicator of music preferences of-the populace
for over thirty years. There are flaws in this logic. Being an
indicator of preference made up of three sources of information: radio
station playlist, juke box playlists, and record sales, the record
rating system peculiarly mixes cause and effect in calculating aggregate
music preference (Cable, 1977). Playlists do not represent actual
sound exposure and the sales outlets sampled are often unrepresentative
of all sales outlets (Hesbacher, Downing, Berger, 1978). Record, tape

and compact disc sales are also flawed as indicators of music preference

15

since many people like what they hear on radio but may never buy it to
hear, however when offered an opportunity to record a friends copy, they
gladly do so. This ownership and preference go undetected by sales
estimates. The impact of consumer cassette tape equipment has helped
change the accuracy of sales as indicator of preference. The
recording industry has traditionally used a flawed indicator for music
preference. The music preference consumer research efforts likewise
have inherent limitations.

Commercial music testing is done in two ways: 1) Hook research, a
call-out method of testing commercially released music samples over the
phone, has been popular for the last ten years with radio stations: and
2) Auditorium testing, testing released music in an auditorium setting
with incentive driven subjects there for hours at a time. “Hook“
research claims to study those components(choruses, melodies, riffs)
that catch the attention of the listening public, however this is done
in the context of currently released music. These elements are called
hooks due to their ability to capture and hold the interest of
listeners. The hook is usually repeated several times within a song
(Lull, 1987). “Hook-research“ can determine whether a released song is
“liked“ by a certain audience. Other preference parameters are
measured such as familiarity, and airplay evaluations as in “I hear it
too much“, “I would like to hear it more.“ This has value for the radio
programmer in deciding what particular selection from the recording
industry will be played but it does not give a programmer music content
preference indices that can be applied to all music, but rather to only

the songs played. In addition it does not always supply to the

16

programmer the music preference “taste culture“ indicators of their own
station audience (Fox, Wince, 1975). Since the programmer must make the
link between their station and the audience (Hennion, Meadel, 1986), and
this link can depend on music preference for certain “meaning“ triggered
by a word, a phrase, a chord, a riff, or (even) a combination of sounds
(Harmon, 1972), a warrant for studying whether music preference
indicators in the actual music content exist. Hook research more aptly
measures ratings of newly released music, while “oldies“ released music
research uses another popular method of commercial music research,
auditorium testing.

Auditorium testing is the other popular method of music research
used in any regularity, and in spite of its advantages, it still has
major economic drawbacks. Auditorium testing is very expensive. It
allows more complete portions of the music samples, and for a much
broader frequency band at playback, compared to the telephone used in
hook research. The music programmer using research has an idea of
listeners’ preferences that the programer without research does not.
Market research of this nature better equips the music programer for
program decision-making. There are currently financial constraints to
performing listening tests, due to the number of individual songs that
must be measured. Music production parameters such as tempo, key, and
timbre are unstudied by the commercial arms of the music industry. The
only research available for music program formulation is either very
expensive or logically flawed or both. With some of the new
capabilities provided by refined measurement scaling, experimental

devices in music performance (MIDI), and expanding timbral textures the

17

message can become varied for scientific experimentation, with
exceptional control. The level of analysis of the message can then
increase it precision over traditonal methods.

Music components have not been used for classifying test music,
thereby requiring the testing of each song title. Music preference
ratings and research efforts have used “song title“ as their independent
variable, and have classified songs by artist and by style. These
nominal level classification measures are difficult to code and
therefore are not in much research. When they are used, its risky due
to high variability in coding such a complex phenomenon. Subsequently,
the “song title“ is usually the independent variable measure. Research
efforts using the song title as the independent measure have inherent
cost imbalances when the number of songs selected for a weeks period of
programming could number in the thousands for many music gatekeepers
(radio, recording, production).' The resultant cost of measuring an
adequate sample of music becomes enormous. If the lexicon of music
preference could be found to relate with specific, tested musical
structure parameters then artists selections could be more reliably
classified, without having to test every song to be measured.

Performance of test music for research has always posed a challenge
for the performer at controlling performance sufficiently enough to play
all performance variables identical while varying the experimental
variables. Subsequently the breadth of research available concerning
the systematic variation of musical message parameters within a song,
has not been as prevalent as the measurement of simple tones, complex

tones, different songs, intervals, chords and modes. This is simply

18

because the systematic variation of musical performance is very
difficult, without introducing unsystematic variation due to the
influence of the human performer during each “supposed identical“
performance. MIDI technology now enables control of that systematic
variation. When combined with a lexicon for music preference used to
describe music people “like,“ a new classification scheme becomes
available linking together music production, and listener feedback.
This would therefore enable a musical message to be rated in preference
for musical structures. To do this it could be quite helpful to remove
one significant factor in preference, familiarity.

Using released music for listening tests, has certain drawbacks.
Music preference can change in time and as music becomes more familiar.
To assess the impact of music’s structure on preference one very .
significant factor of preference is familiarity, and if not removed then
past exposure frequency could become a large basis for variance in past
studies of music preference. Current music research techniques are
developed for purposes of obtaining an audience for radio, and not
necessarily the music itself. Although these are both related, radio
sells the audience to advertisers and thereby wants to study how an
audience can be identified, accessed and maintained. Familiarity is
important to audiences. The music listener preference independent of
social setting and occurence (exposure) would be preference for the
structure of the music itself. When measuring listeners’ reactions to
songs they have already heard, a whole range of other variables become
introduced into the music preference equation, including familiarity,

social peer group evaluation, and simple associative memory phenomena.

19

The current commercial research has proven valuable but certain
questions cannot be answered with the amount of information supplied
from this research (Anthony, 1987). An improvement of “music research“
-rather than radio audience music research is to systematically vary the
independent variable while measuring the lexicon of music preference.
Musical messages are created, produced and programmed across limited
channels and times. The music listener hears this music and responds
via feedback mechnisms. One type of feedback is the music preference
ratings such as charts, which in spite of being massive institutions
dedicated to music preference assessment, the actual scientific research
efforts have been limited in breadth, methods, rigor, and scope for
purposes of producing music research. The level of research studying
audio phenomena is not in proportion to video phenomena (Atkin, 1983),
with a general absence of pop music studies in academic research
(Goodwin, 1986). This makes the task of the music programmer especially
difficult, since many decisions must be made with very little research.
Besides the programmer, though is a whole other host of music
gatekeepers that also could benefit from refined music research.
Besides always having a natural imbalance between messages (supply) and
outlets (channels), music in its technological evolution has become more
varied and more easily created and produced. One music production
parameter undergoing substantial change right now is timbre, and it is
the goal of this thesis to pursue the exploration of a relationship
between a message parameter timbre and a lexical feedback known as music
preference. Many music communication parameters such as sender,

message, medium, and listener(receiver) have been researched, primarily

20

within other disciplines such as engineering, music, psychology, and
sociology. Within this decade two volumes of key communication
journals, and volume of a sociology Journal have dedicated entire
volumes to audio and musical phenomena. These are:

e

*MWM.

*MIAJWW

The music related industries are in their dawn of music preference
research. Research agendas in the communication related academic fields
are just beginning to study music. In an increasingly competitive
industry, music research is a necessary requirement for rational
business practice, yet the commercially available research lags behind
television research of exposure, effects, and preference. What is.
available is very expensive and in many ways does not measure the
fundamental music preference referent, the music structure. The same
technologies involved in creating more varied musical messages make.
possible new avenues for music research. Both creation and control is
possible on one music structure parameter: timbre. At the same time
trends in social research have resulted in more powerful lexical
measures. With such research available music production and programing
decisions can refer to scientific relationships concerning the most
universal communication (Lull, 1982), to which listeners' ears do not
close.
This thesis studies two communicative phenomena: 1) Music and

2) Words (lexicon) that describe music. The technology available make
possible experimental controls to test this relationship. With such

controls in music research the ability to abstract findings to other

21

music becomes at least theoretically logical, therby impacting the
researh community. This community would benefit from added precision

and refinement of the music preference research process.

Chapter Two

Literature Review

The music communication process described earlier borrows from
Shannon and Weaver’s Ihg_flg1hgng1jga1_lhggpx_g£_§gmngnjgatlgn (1949).
The sender and message classification have been combined, and then
articulated to represent the production of musical signals and symbols.
Using this perspective, research in this topic area focuses on the
message, media, receivers (listeners), and feedback. The message
component for analysis in this work is the timbre of music, the media
although not an active element in the focus of this work is necessary to
review for an adequate understanding of music preference. The receivers
on listeners have a variety of characteristics that in this review will
relate with the message and media, however the primary element of
listeners for this work is the lexicon of music preference, the feedback
measure of musical timbre being tested.

The western musical message structure has been systematized through
code systems. The social system of music science has for centuries made
many assumptions about the meaning contained in music structure, and
regularly made recommendations to composers on how to get the “desired
effect.“ As a result there is an enormous amount of social learning
that has occurred based on these cultural traditions. With the onset of
electronic audio in the late nineteenth century, a conceptual merger
between music science and physical acoustics occurred. Research
concerning what the listener heard became the focus. “What is sound?“
was the question and the focus became the message. The twentieth
century obsession with analyzing humans as individuals, led to the

22

23

receiver as the point of interest by studying “How does the perception
of an auditory stimulus compare to the structure of the stimulus?“.

This are led to the works concerning frequency response and pitch
(Stevens, 1935: Stevens, Uolkmann, and Newman, 1937) and equal loudness
contours (Fletcher, Steinberg, 1924: Fletcher, Munson, 1933). There is
also scattered throughout music education research, an interest in
musicians, a message sender, in how their reception charateristics
differ from non-musicians. Quite recently the communication and
sociological fields have gained interest in popular music, coinciding
with the introduction of broadcast radio channels, low cost playback
equipment, and youth markets in the popular music recording industry.
Out of this grew a strong recorded music-radio industry (Hesbacher,
1978). The goal here is to review those studies relevant to the stages
of music communication under study in this thesis. These are:
timbre(message), music preference(feedback), and the lexicon of music

preference(code system), and the listeners.

flggig_flg§ggggg. Music messages are made up of combinations of multiple
dimensions. Music is social, requiring rules for conduct, specifying
means of ordering and systematizing its constituents (Cross, 1985).
Musical messages are composed using varying degrees of rules and
conventions for organizing basic constituents or components. Sets of
components are labelled differently, depending on the social systems
using the components. Each social system has different expectations and
goals, and subsequently has need for ways of packeting music into

information about the structure of music. To review research on music

24

messages and the receiver one must analyze the components common across
social systems. Timbre is common across many social systems. The study
of timbre is undertaken by physicists, acousticians, audio engineers,
‘music educators, socialogists and communication researchers.

It was stated earlier that music is “the science or art of ordering
tones or sounds in succession, in combination, and in temporal
relationships to produce a composition having unity or continuity.“
Music than is tone over time, in serial and parallel configurations
(Harder, 1975) having unity and continuity. Defining “unity“, as
“a combination or ordering of parts in a literary or artisitic
product ton that W or W
gfifiggt.“ Defining continuity as “an uninterrupted connection,
succession, or union“, the definition of music is tone over time in
combinations that are identified in succession and harmony as belonging
together, by a certain social system. The review will now turn attention
to the one fundamental component of music, and timbre: tone. The
discussion will then review research of the tonal characteristic of
timbre both as a physical structure and a received structure of musical

Inna. Music is patterning of tones. Helmholtz, in the late nineteenth
century, pioneered a work to identify the relationship between the, at
the time, separate fields of physical acoustics and music science.
Across these two fields Helmholtz identified three distinguishing
characteristics of tone:

1) force,

2) pitch,

3) quality
(Helmholtz,1862)

25

In this case, force is measured by amplitude of a sound-wave, pitch by
the frequency of the sound wave, and quality by multiple dimensions:
spectral content, temporal dynamics, and auxilary noises. Helmholtz
explained spectral content is made up of various individual simple
tones, known as a fundamental and one or more partial tones, known as
upper partial tones. Others refer to these upper partials as overtones
(Lundin, 1953). The entire set of tones including the fundamental tone,
can be referred to as the hapngnjg_ggnjgg. Temporal change in the on-
set of tone is known as the envelope, made up of an attack, internal
dynamics, and decay elements (Runstein, 1986). The attack is the
beginning portion of the sound wave, internal dynamics are temporal
changes in intensity of the individiual partial tones, and decay is how
the sound ceases. For example, trumpets have-a fast attack and are
known to emphasize the upper partials at high levels of intensity, and a
decay quickly (Beauchamp, 1975). The third component is auxilary
noises. With the strumming of a guitar, the bowing of a cello, or the
“slapping“ of'a bass string, noises are introduced at the attack,
decay, and everywhere in between on some instruments. These are
separated out due to uniqueness in combinations of partial tones over
drastic changes in intensity.

These are the three components making up what Helmholtz called the
quality of the sound where Helmholtz directly linked quality to timbre
but chose not to use the word timbre due the generalizability of his
findings to tone theory rather than music. Helmholtz use one of the
earliest scientists to make note of the actual formulation of complex

waveforms, in which quality, exemplified by timbre, was one of three

26

main components of organized tone. Helmholtz in his initial sentence
discussing the definition of “quality“ says: “When we hear notes of the
same force and same pitch sounded successively on a piano-forte, a
violin, clarinet, oboe, trumpet, or by the human voice, the character is
the musical tone of each of these instruments, not withstanding the
identity of force, is so different that by means of it we recognize with
the greatest ease which of these instruments were used.“

(Helmholtz, 1862).

Seashore in 1939 expanded the definition of tonal quality by saying
it was made of two fundamental aspects, namely, (1) Timbre, which is
the simultaneous presence or fusion of numerous simple tones at a given
point in time, and (2) Sonance, the successive presence of changing
timbre, pitch, and intensity. Here we can see that tonal quality is
again referenced to as a major component of timbre. The concept of
timbre was again linked to quality, and has been traditionally present

since the early days of music and aesthetic research.

Ijnhng. Mursell (1937) says “probably the most fundamental and
universal source of interest in listening to music is the tonal content
itself organized in terms of its volume and timbre.“ He states:

that when enjoyment arises out of nothing but the

shifting volumes and qualities of .the tonal content

.there will be no definite awareness of musical shapes or

design. It is a characteristic evoking a low level of

response which on one hand hardly is a basis for

musicality, but on the other hand is the ultimate

foundation of all music (Mursell, 1937).
“The subtleties of timbral discrimination are, to some degree, the very

essence of musical interest“ (Naumann, Wagoner, 1985), yet timbre has

often been the miscellaneous category for describing the receivers’

27

physical attributes of sound, “gathering into one bundle whatever was
left over after pitch, loudness, and duration had been accounted for“
(Dowling and Harwood, 1986). This is partially the result of the
research technology that was available in the past. It could only
systematically vary pitch, loudness, and duration.

Timbre, has also been paralleled to the organization of speech,
containing steady-state vowel sounds, onset-transient-consonant tones,
and noise bursts. In audiology and psyco-acoustics, lingual
organization of vocal timbre is made up of 1) spectral composition of
tone in its constituent simple tones, 2) the onset of a. complex tone
called the attack-decay envelope, and b. partial tones at certain levels
of intensity, and 3) accompanying noises traditionally associated with
the method of generating sound. Here we can see similarities in
findings of musical and vocal timbre. Timbre is identifiable and has
been discriminated into different timbre.

Efforts in assessing what listeners hear, particularly in timbre
have been performed. Multidimensional scaling has done with a set of 16
instrument tones (Gordon 6 Grey, 1978: Grey, 1977) applying multi-
dimensional scaling to a set of 16 instrument tones that included both
transient and steady-state portions, found three prominent dimensions:

1) spectral energy distribution in steady state
2) onset and offset patterns of the spectral components
3) noisiness of attack

Miller and Carterette (1975) varied frequency of the fundamental
tone in the harmonic series, shape of spectrum envelope, and attack-
decay functions, and they found that the strongest dimension was

fundamental frequency, the second dimension separated envelopes into

28

piano-like spectrum envelopes from all others, and the third dimension
separated sounds with strong attacks from those with what was called
“trapezoidal onsets and offsets. Trapezoidal refers to a smooth attack
and decay of the entire waveform. In a similar effort, varying number
of harmonics (upper partial tones), attack envelope, and relative
pattern of onset times of different harmonics, Miller and Carterette’s
results showed that listeners’ first dimension separated clusters having
three, five and seven harmonics, a second dimension separating the five
harmonic tones from the other two types, and a third dimension of
overall attack envelope ordering: piano, brass, and smooth. Their
conclusions were that spectral content and attack functions are the
primary determinants of timbre. Gray in 1978 showed that when
synthesized tones are present in a musical context, certain aspects of
musical tones can be treated as more or less irrelevant in the
determination of timbre, especially subtle fluctuations of parameters
during steady-state. Chowning (1973) and other engineers have found
similar results in the invention process of “real sounding“ electronic
synthesizers, using various methods of sound generation. However at the
same time, certain determinants of timbre are required to emulate “real“
acoustically produced timbre. Results such as these leads one to
believe that to research timbre and its effect on music preference, the
definition of timbre can be limited to those attributes in different
sounds that are a function of spectral composition, and on-setloff-set
(attack/decay) of the sound wave.

To analyze music let the fundamental definition be that music is

composed of tone over time. Further tone has been decomposed to reveal

29

a major component, timbre, that is mainly discriminated by two
dimensions, 1) Spectral composition, and 2) Time-based on-set/off-set or
attack/decay. Instruments are discriminated by listeners according to
the way in which they combine these two dimensions. The importance of
these research efforts show that timbre has been identified structurally
in the context of music listening, contributing to the overall auditory
stream recognized and differentiated by listeners.

When applied to the instruments that form sounds of different timbre,
each instrument can be defined in terms of its component spectral make-
up, and its attack-decay envelope. The classical families of acoustical
instruments of strings, brass, percussion, and woodwinds, are
differentiated by their unique combination of these two timbral
components, but there are some exceptions to adherence to this makeup in
some family members. For this reason the classification of timbre is
best kept with actual auditory data emitted from the instrument, and not
the classical method for categorization by family, and simple
simultaneous-

Musical timbre is recognized in various forms of ordering.’
Traditional forms of ordering the sound waves of tone accompanied their
method of tone creation, however now with synthesized timbre the
creation of tone is electronically based, and perceptual validity of
synthesized timbre has been reliably demonstrated (Risset, 1983). The
research question posed in this work is “can timbre effect independently
a lexicon used for music prefernce.“ We will new address the “unity
and continuity“ part of our definiition of music, namely what do

listeners label and prefer in music, thereby denoting this unity and

30

continuity.

Lngggn_g£_flugig, Music is a social phenomena which comes in various
forms. At the initial level of a musical lexicon, we have a judgement
of preference, with binary attributes, 1) Yes: the sound is music, or
2) No: the sound is NOT music. To aid cultures and sub-cultures in
communicating about music they often acknowledge a conceptual level of
“musical style“ or “musical genre.“ Labels for western civilization
styles have numbered many over the years, such as Classical,
RhythmGBlues, Funk, Soul, Rock, RockGRoll, Pop, Jazz, Folk, Bluegrass,
Latin, New Wave, Punk Rock, and New Music (Baumann, 1964: Fox & Wince,
1975: Fink, Robinson 6 Dowden, 1985). These are used to label different
styles. Within each style of music, a certain set of musical timbres
are created and combined. With these combinations, music of each style
is sought out by avid listeners, and avoided by others. Style of music
is a nominal level categorization of musical messages, that has been
utilized quite widely by listeners, researchers and industry alike. The
criteria for being a listener of one style over another is that one
“likes“ this style of music. The style label carries meaning to humans
but the meaning can be limited. This is demonstrated when an avid
classical attends a classical concert but walks away from the show
slightly discontented by what they heard. The classification of style
is limited in its identification of preferred music gggtggi.

The lexicon of music preference is actually the listeners’
calibration of information or meaning they decode from the music. To an

avid “pop“ listener, improvisation may indicate jazz, while to others,

use of strings may mean “classical.“

31

To further communicate about

musical meaning, a lexicon with further descriptors than “style“ is

required. Early efforts at using verbal descriptors for musical

emphatic
exalting
majestic
martial
ponderous
robust
vigorous

agitated
dramatic
exciting
exhilirated
impetuous
passionate
restless
sensational
soaring
triumphant

awe-inspiring
dignified
lofty

sacred
serious

sober

solemn
spiritual

k ' t
bright
cheerful
98V
happy
joyous
merry
delicate
fanciful
graceful
humorous
light
playful
quaint
sprightly
whimsical
calm
leisurely
quiet
satisfying
serene
soothing
tranquil
dreamy
longing
plaintive
pleading
sentimental
tender
yearning
yielding
dark
depressing
doleful
frustrated
gloomy
heavy
meloncholy
mournful
pathetic
sad

‘tragic

32

analysis began with Hevner (Hevner, 1935), who attempted to ascertain a
relationship between moods and music by using an adjective list
constructed of 67 words, arranged in eight clusters. Revised a number
of times to increase the consistency between adjectives within clusters,
results using this list have problems due the lack of using descriptors
familiar to the listener respondent (Meyer, 1956). However as a
paramont study in the area Hevners efforts have been drawn upon again
and again. Figure 3 lists the Hevner Adjective Checklist.

In the spirit of what Aristotle called ellicited mood (Tame, 1984),
the emotional or mood meaning of music has been of great academic
interest. The actual claim of this meaning or informaton being actually
emotional, or charaterizing mood qualities specific to music itself is
debated. Meyer (1956) states that much of mood derived from music is
the result of associative mechanisms linking the sound with past
sentiments. Rather than eliciting natural emotional reactions, music
depicts conventionalized modes of behavior called “designative emotional
behavior.“ In addition Meyer claimed that the communication of moods
and sentiments tends to become standardized, such that certain “musical
devices become formulas which indicate a culturally codified mood or
sentiment“ (Meyer, 1956) This culturally codified mood, is encoded in
tones with.grammer of “acceptable usage“. This can change in time and
does, leading to scenarios where “what was not allowed some time ago may
now be viewed as acceptable“ (Farnsworth, 1969).

The semantic concepts of “connotation“ and “denotetion' organize
different types of meaning found in music. Connotative meaning can be

equated with the idea of associative meaning, while denotative meaning

" 1: .ml

33

of music is that created within the structure of the music itself
(Nielzen 6 Cesarac, 1983). Meyer’s used the term “embodied“ to refer to
what we have presented here as denotative meaning.

Mean ratings of large groups can (also) serve as a qualitative
description of music (Nielzen and Cesarac, 1983). Osgood’s semantic
differential has been used (Gray and Wheeler, 1967: Crozier, 1974), for
measuring meaning. In a study done by Nielzen B Cesarac, three main
factors were found when scaling mood:

Factor 1: Tension-Relaxation,
Factor 2: Gaiety-Gloom
Factor 3: Attraction-Repulsion
This study was meant to find dimensions in the emotional experience of
music, i.e. the designative-connotative meaning in music. In a similar
effort to assess the connotative meaning (McMullen, 1982), more than
connotative is acknowledged when he mentions descriptions of the music
being heard. He states that:
When many individuals, if not most, describe their
response(s) to music, words such as relaxing, sad,
or majestic are a common part of their vocabulary.
These words often are the only means some individuals
possess to communicate to others either how they
feel personally or to describe the music being heard.
McMullen points out how the denotative meaning can be described by the
sensuous or physical attributes of the stimulus, and the formal which
refers to the relationships between the physical objects or the
perceptions of these relationships. In the development of a list of
verbal descriptors for measuring the connotative words used for
describing music, it is noted that “connotative words do not have the

same meaning for all humanity“, and “we do not know for certain if the

word happy for one person refers to the same internal experience for

34

another individual.“ Therefore, verbal descriptors of denotative and
connotative meaning are proposed and used but it has been proposed that
these may vary across individuals and populations, and the connotative
labels of music effect can be classified by two main dimensions:
activation and evaluation. This is similar to the results of the
Osgood, Suci and Tannenbaum research of 1957. Their three primary
dimensions were: evaluation, potency and activity.

Maher and Berlyne (1981) used rating scales classified in three
main groups: an evaluative Judgement group, collative or information
properties of the stimulus, and reports of internal states. In this
study a list of paired descriptors were found to be related to the
difference in tones(intervals) used in musical stimuli. Hair (1991)
found the usage of the words: loud, soft, fast, slow, high and low, in a
design comparing childrens' with adults verbal identification of musical
meaning. Nierman (1985) found that students ratings of music were
sometimes too low on several items simply because students were not
familiar with the musical vernacular descriptors for such concepts. In
this case, it was stated that some denotative meaning of music is not
simply described by the average individual, while “on the other hand,
items involving components of nggnggg and tjmggg were very easily
perceived by most students regardless of the vernacular used.“ The
limitation of listeners’ labelling processes of musical meaning may
actually be limited, yet be informative about what listeners are not
limited in identifying.

Meaning in music both connotative and denotative has been mesaured

using an appropriate lexicon for each area. Because of the inherent

35

difference between verbal language and music, particulary in
aggregate’s lack of musical vocabulary, the boundaries between
connotation and denotation become blurred. The connotative meaning
includes associative moods, while denotative meaning concerns the
musical ordering of tone. A limited number of words describing these
different domains of meaning, common across respondents in the
aggregate, are quite often the only means one possesses at describing
the meaning, or information decoded from music. These descriptions do
vary by population and age where some denotative descriptions are easier
for some musical parameters than others. Musical timbral structure is
an identifiable denotative dimension, that may be related to its verbal
denotative and some connotative lexical counterparts.

The measurement of preference components may be a function of
physical components, so that verbal connotative and denotative items are
the result of physical structure or connotative designative moods
associated. To address this one must investigate how the physical
makeup of the message is related to the verbal lexicon. Music style can
correspond with timbral elements, however as styles evolve subsequent
message-lexicon relations change. It has been shown there are methods
for analyzing timbral elements, and musical meaning via a lexicon for
music. Music preference is a special case of meaning. It supposedly
predicts exposure patterns to music, as well as other social affiliation
and status codes. In the same light music preference is the same
construct at work when one ordering of sound is identified as noise
while another is music. In this case preference is descriptive of the

music structure, at the fundamental distinction between music and noise.

36

Music preference can be a binary measure on one dimension, or ratio
level on numerous verbal dimensions. Aggregate listeners can hear the
difference. It is here where a lexicon for music preference becomes
valuable as an identifier of music structure. At the same time
addition, preference has certain social redemption, status and group
membership value, such that a lexicon for preference could hold certain
social relations

Presently in a world with virtually unlimited timbre when just 20
years ago the distinct types of timbre in music could be counted on
one’s fingers, one must wonder if the use of new timbre incringe on the
boundaries between styles such that the meaning in music is more finely
discriminated, beyond style. Another view is that people could always
hear timbre but could not describe it. If it did not very much then
perhapsthere was no need to further describe it. Regardless, timbre is
an active component of music structure and perception. How the common
listeners describe music may hold the tools necessary to identify timbre,
and timbral differences. In many ways these are often the only lexical
terms available to the average listener. The focus of this work
addresses whether timbre exists in the reception of music as witnessed
in listeners’ feedback using a lexical measure of music preference.
This is of great interest to the music communication researcher,
producer, and programmer by studying the relationship between a highly
variable music parameter (timbre) and a lexical set of music preference

items.

37

flggjg_ficgjgcgngg. Music preference is the set of behaviors and
perceptions involved in identifying music one likes, and listens. Music
which people like is indeed music to which they will listen. One study
(Fink, Robinson, and Dowden, 1985) found that attitudinal structures for
music they liked and behavioral structures for the music they would
attend at live performance, yield the same multi-dimensional structure of
the types of music for the aggregate audience. Two primary dimensions
were identified as, 1) Formality-Complexity, and 2) Ecological-Geographic
were found. These findings, based on a large sample size (N-17,254), in
a United States national survey, identify one very fundamental finding of
music preference attitudes and exposure behavior. The two are congruent.

Models for the music preference communication process have found that
music preference is related to three main factors: 1) message, 2) social
system within which the music is active, and 3) the receiver. Wapnick
(1976) describes three categories of variables related to formation of
attitudes of preferences for music: 1) musical, 2) situational, and
3) subject. LeBlanc (1980) describes the sources of variation in musical
taste to be the music, environment and the listener. It is noteworthy
that for purposes of this work that both these models identify the
' physical properties of the music as a single and unique component in the
determination of music preference.

In a refinement of these models for purposes of communication
research, we will collapse two of the main components of music preference
from these two models: the individual (subject,listener) with the group
(situational, environment). This yields a simpler model of two

components: message and listeners. It has been noted both in social

38

research (Soloman, 1985) and music preference (Taylor, 1985: Burton,
1985), that research across studies often lacks continuity, and within
studies lacks comprehensive designs. As Taylor states:

the design of experiments often leads to a body of

disjointed research literature that is difficult for the

music practitioner to use.
In the mass media fields music practictioners include composers,
musicians, producers, and programmers. In a related field,
anthropology, cultural inquiry often begins with a wholistic approach to
analysis, then followed by quantitative analysis. This is implemented
for reasons of limiting bias in initial assessment of not only the parts
within a system but also the system itself. Music, a very complex social
phenomenon, can be viewed as'a “social organism“ (Ourkeim, 1834), and
this wholistic cultural analysis technique bears merit. A frequently
utilized dictum in anthropology is “Occam's Razor“, which the states the
“simplest of competing theories be preferred to the complex,“ or that
“explanations of unknown phenomenon be sought in terms of known.
quantities“(Webster, 1985). LeBlanc has attempted to state a complex
model for music preference and then operationalize a systematic attempt
to sort out and examine these variables that impinge on the listener’s
domain. This has led to focus on generic styles of music preferences
(1979), effects of style (1981), tempo (1981, 1983), performing medium
(1981, 1983), vocal vibrato and vocalist’s sex (1986). General
preference was for popular music styles, faster tempos, instrumental
mediums, low amounts of vocal virbeto, and male vocal timbres. LeBlanc

has done much to contribute to a comprehensive approach for the research

of musical preference, and these studies found that instrumental mediums

39

and male vocalists both are examples of timbre as an active part of music
preference formulation.

Other efforts to discern items of relevance to music preference have
been done, but to summarize them is beyond the scope of this work. The
thrust of this work is to focus on a relationship of timbral message
attributes and a listeners’ lexicon for music preference. It is with
this focus that we continue the final steps of our literature review.
This section will summarize those attempts at studying how musical
preference varies with musical message structure, particulary timbre, and
how transmission and listeners effect this process. Although tranmission
is not formally part of the research focus here, it is a relevant
variable in the spirit of exploration and in the control of directly
pertinent research variables: music and listener feedback.

As mentioned previously, LeBlanc found a significant difference in
preference for different performing mediums. In his definition, medium
was Operationalized as vocal or instrumental. Viewing the voice as a
unique timbre of its own, the performing medium varied was actually a
dual category measure of timbre: 1) instrumental and 2) instrumental plus
vocal. The results demonstrated a higher level of preference for
instrumental timbres. LeBlanc (1979) showed in a study of fifth
graders that their favorite music genres (style) of music were, in order
of popularity: pop, rock, ragtime, Dixieland, march, and country-western.
Christenson, DeBendittis and Lindlof (1985) summarize in anlgggglg
Ugg_g£_1hg_flggjg_flggig (1983) that children use the words “sound“ and
“beat“ in determining their preference. Chistenson and Lindlof report

that six to thirteen year olds were more likely to cite “the sound“ or

40

“the beat“, than the lyrics as reasons for what they liked about music.

Boyle, et al. (Boyle, Hosterman, Ramsey, 1981), studying factors
influencing pop music preferences of young people, found that between
respondents of grades 5, 7, 9, 11, and college, that “grade 11 subjects
rated instruments as significantly more important factor than did
subjects in grades 5, 7, and 9.“ Males rated instruments and peer
influence as more important than females. Factors related to popular
music that separated age groups such as peer influence and dancebility
diminished in effect from the lower three grades to the upper two. The
factor of hearing it on the radio was significantly more important to
grades 7 and 9 than to the upper grade levels (11 and college). In
summary, Boyle et al. conclude that “the study confirmed that both
sociocultural and structural factors were viewed by young people as
important in influencing their preferences for tunes, although structural
reasons were generally viewed as more important than sociocultural.“
Boyle, at al discuss the importance of using a “low theory“ made up of
verbalizations which are the only symbols available to respondents, in
describing their music listening preference. He states that much
attention in studying this domain is needed.

Baumann (1964) found in his epoch work, that younger teenagers prefer
popular selections as compared to other styles. As age increased there
was increased preference for other styles, especially classical items,
although the classification of music styles beside classical needed
clearer definition. Baumann found significant differences in
preferences of high and low socio-economic groups for some kinds of

music, where across all popular music recordings high 959 listeners

41

liked the category more, while seven of the twenty musical selections,
were rated as “liked most“ by more low SES listeners. The level of
formal training in studying certain instruments varied by $65, and was.
attributed to the opportunities provided by the higher incomes. The most
popular instrument studied within each SES group was piano, while the
next most popular instrument in low $66 was the guitar and for high 966,
it was the trumpet. Geography played a role as a factor in gggglgc music
style preference where eastern (U.S.) teenagers favored it significantly
more, and western teen-egers had a greater liking for classical. This
was previous to the major shift of much of the popular music recording
industry to southern California, which may have eventually increased

western preference for popular styles.

Iggjg_§xnthgg;g. Popular music preference research identifies two main
sources of variation, the message and the listening social group,
although one other major source of variation resides in industry. Though
some researchers will separate the individual from the group in modelling
musical preference, for sake of simplicity in modelling socially derived
preference, the listener and the social situation or environment will be
combined. This refinement is warranted in the mass communication
research environment, for two reasons. First, there are notes about the
already scattered nature of social and music research and 2)the social
nature of music preference. This refinement may allow for consolidation
of former scattered variables into a more cohesive whole. Prominent
message variables are: instrumentation, style, tempo, vibrato, radio
play, and familiarity. These factors do engage with certain sub-

groupings of all listeners described by: age, formality, geography, sex,

42

socioeconomic status, and peer approval. This is the set of factors that
can be relevant to our two stage breakdown of preference.

Transmission factors, however, are not the focus for this work. This
thesis simply measure the relationship of the music message with the
feedback mechanism most natural to the music listener, the lexicon of
music preference. The message parameter under investigation is timbre
and it will be varied systematically, and measured with the lexicon of
music preference to discover if this lexical set varies systematically.
The investigation of the variables involved though will continue with
review of other stages in the music communication process prior to music

preference lexical feedback: media transmission and listeners reception.

flagg_flggjgtgg_flugjg, Some of the earliest forms of mass communication
over great distances were music (Carrington, 1949). The beats of drums
notified early man of activity for purposes of defense, work, or
celebration rituals (Merriam, 1964). The introduction of telephone made
,the audio spectrum an electronically transmittable phenomenon. Radio
made possible transmission of music, as a type of programming using both
recorded and live performances. With the introduction of commercial
television in 1948 (Hesbacher, 1978: Charron,Sandler, 1983: Rothenbuhler,
1987), much broadcast and advertising emphasis transferred to the
television, making radio an alternative, rather than primary medium. A
working arrangement developed between the recording industry and radio.
Additionaly, radio had less access control exerted over it by others in
the household, particularly the heads of household (Lull, 1985). This

occurred as a result of the independence of access listeners: anywhere

43

and anytime. As a result uncontrolled adolescent exposure became
prominent and popular.

Radio established a relationship with recording producers and a
listening audience. The records played on the radio reached an
adolescent audience that became a major audience segment by 1957, when
the percentage of popular music oriented Top-40 radio formats were on the
upswing (Denisoff, 1973). The producers of recorded music “had replaced
the music publishers as the dominant institutions of the music industry.
Control of sheet music was replaced by control of radio and recorded
copies, a change of access control from one medium to another (Laing,
1986). In order to get attention on the national musical soundscape,
music has to succeed in penetrating the repertoire of existing record
companies and of radio broadcasting, or to create its own apparatus of
production and distribution (Regev, 1986). The performance of music on
radio became primarily a recorded message, mostly due to the reduction in
programming cost that could be realized(Hesbacher, 1978). Christensen et
al. (1985) found that children listen to radio primarily for music. The
major portion of the total radio listening audience became adolescents.

As the recording industry members have varied in numbers, increases
in industry concentration have led to decreasing diversity of programming
types (Peterson, Berger, 1975: Rothenbuler and Dimmick, 1982). However,
Rothebuhler and Dimmick stated that “song type may succeed market
concentration as the primary predictor of change in popular sound
recordings“ (Anderson, Hesbacker, Etzkorn, Oenisoff, 1980). The
classification scheme used for “song type“ has three indicators: 1)vocal

style, 2)instrumentation, and 3)song structure. The recording industry

44

admits relatively few pieces of music to its channels, than mass produces
and aggressively markets those few (Cheffee,1985). Music marketing is
composed of promotion and radio play, commonly referred to as “plugging“
(Cable, 1977). In the marketing process of the recording industry, as
in many industries, little effort is dedicated to assessing consumer
(listener) preference, frequently adopting an explanation for musi -
preference as “People don’t know what they like 7 They like what they

know“ (Denisoff, 1973). Artists' music is selected, produced and

 

promoted. The primary promotion medium is radio. Radio promotion is
repeated exposures of the new releases, and has even led to a format
entirely dependent for programming from newly recorded popular music, Top
40, which is now often identified as Contemporary Hits Radio (CHR).
Eventually some music becomes preferred after intense promotion, or in
time as the music reaches listeners via other radio formats. The number
of music compositions “tested“ by its listening audience is limited, and
exposed in a repetitive manner. As has been stated the extent of
diversity of music programming can vary with both song type and industry
concentration where song type is classified using indicators of vocal
style, instrumentation, and rhythmic structure, and where industry
concentration is measured with accepted concentration ratios.

Repetition of music has been shown to be effective in the development
of “liking“ (Hargreaves, 1984: Seashore, 1938: Lundin, 1967: Meyer, 1956)
and familiarity(Russell, 1986). Popularity and “pleasantness-
unpleasantness“ of repeated music trends have been discussed by Lundin.
These trends exhibit an inverted-U function. “Popular music tends to

reach the maximum of pleasantness at an early repetition, whereas

45

classical selections reach their effective height with later
performances“ (Lundin, 1967). Hargreaves studied how this repetition
was related to message ggnglngty and Lgmjljgcjty, which are inter-
related. As familierity increases, though objective complexity stays the
same, subjective complexity decreases. It is hypothesized that there is
some optimum level of subjective complexity which is reached over time,
as familiarity increases, and from which preference evolves. Uerveer,
Barry, and Bousfield (1933) in an early effort distinguished between
continuous repetition and repetition at intervals, and found both to have
a positive effect on ratings of “pleasantness“ on a 20-point scale.
Hargreaves (1984) recently has found that repetition at intervals is more
significant than continuous repetition, with an interaction effect
between these two types of repetition. The results were interpreted as
further support for the inverted-U theory. Repetition thus plays a
factor in music preference.

The working relationships between radio and records enable a mechanism
for repitition of popular music selections, via broadcast radio. This
affects preference for that music, due to the interaction of familiarity
and liking. The ability to access their audience through available mass
media technology: radio, makes possible a peculiar metamorphosis of
recorded music and broadcast radio. This is such that it integrally'
intertwines these two industries to the point that recording industry
concentration ratios are related to radio programming diversity (Peterson
et al. 1975, Rothenbuhler and Dimmick, 1982). The findings indicate
though that song type may still serve as a better predictor of

programming diversity, and one component of song type is instrumentation.

46

Radio serves as a medium upon which other new technologies get
superimposed making a web of technology for highly repeated, limited
number of musical selections. With a virtually guaranteed audience in
the early days of the relationship between radio and records, technology
was devoted to the reproduction technology.

Reproduction technology has its impact on timbre. The microphone,
enabled new methods of vocal delivery that was associated with certain
emotional qualites that could not be obtained with musical instruments or
live acoustic performances (Frith, 1986: Russel, 1986). With the
introduction of recorded music, listeners were allowed to listen to their
own copies as much or as little as they wanted. This technology effects
the exposure rates of radio broadcast music. Radio besides repeatedly
exposing listeners to-a limited set of songs, enables a regular medium
for newly recorded music. Magnetic tape enables new types of music
messages by making the recording studio a production tool for multi-track
tape reproduction. This enables the musician to simultaneously perform
various tapes tracks for a comprehensive unified result. The use of tape
also enables types of tape editing, splicing, and musical sounds in
general. Recorded music has exhibited its influence on timbre. By
reversing sections of tape the envelope and spectral contents of sound
can be reversed in time. This effect was high during the psychodelic age
of rockbroll in the late sixties, when the Beatles, Jimi Hendrix, and
Spirit all used such effects to influence the timbres in their music. In
addition each year new effects have been introduced, starting a
progression from wah-wah, distortion, and sustain to stereo chorus,

flanger, and now a whole range of digital controls such as delay and

47

reverb. Some of these effects are a static conversion, while others
allow for analog control with a foot pedal that work as accelerator
pedals in an automobile. Each of these “effects“ modifies timbre.
Telecommunication technology is integrally tied to the development of
music, particularly timbre. It affects message content through
instrumentation, recording, and microphoning technologies, while the
message exposure is modified through playback recording, and broadcast
technology. Exposure is now influenced by instruments with a variety of
“preset“ rhythms, and “one key“ playing capability. More people are able
to play music now. Mediated music has an inherent dynamic to modify
itself into continuously innovative musical forms, and these are quite
often tied to the potentials enabled from rising technologies. As has
been pointed out this is not even peculiar to electronic music. The rise
of the small combo after the invention of the saxophone, led to a
completely new style of music, jazz, whose fundamental goal is to modify
itself. These new technologies introduce new timbres into music, from
which old styles are made into new styles, and old styles remain marked
by their timbral elements. In conjunction with this though are listeners
that react and respond to mediated music for a variety of reasons
including behavioral access and effects,' structural content, and peer
approval. One group of listeners that is especially attentive to newly
released recorded music and subsequently radio programing are the youth
of the world. Music carries meaning for youth in a variety of ways.
These meanings will continue to change as they progress through life.
The meanings are carried in simple repeated messages that literally get

hammered, into the inner workings of social youth (Lull, 1987). This

48

fundamental human relationship of music and age is our next and last area

of focus.

1ggth_§gltgng. Radios, compared to television, are more available to
people, due to lower cost, and simpler technology. Popular music is
referred to as the “voice of the youth“ (Chaffee, 1985), which is
uncontrolled by parents especially since music can be taken outside the
realm of the household, made portable in cars and in one's hand, as in
the “transistor radio“ craze of the 1960’s or the “ghetto blaster“ craze
of the 1980’s (Christenson, Debendittis,1985). When cassette tapes
became popular, the control listeners had over their music programming
became even greater by making their own choice, their own copies at their
own time (Frith,1986). In addition, popular music tends to have less
rigid social norms of lyric content, and of visual displays during
performances. Out of these norms and actions, sub-cultures become
associated with each (Lull,1987). Here is a socializing force. Peer
approval is another active social mechanism (Hanser,1974: Alpert,1982) in
making musical judgments and Judgements of popular music preference.
Preference for music perpetuates music communication, and preference is
a social phenomenon. Preference is a topic commmunicated between
listeners sometimes with language, and sometimes without. An assignment
of symbols, though, demonstrates an extent of probability of “shared
meaning“. That which is “real“ is something not only which has been
symbolized, but also shared and agreed upon with others. Only through
sharing our observations and agreeing on them does an observation become
“real“ (Einstien,1952). To ascertain relationships requires meaning to

transmit from respondent to researcher, using a shared symbol set. Using

49

this the social scientist more closely measures social phenomenon. Once
given the shared symbol set, and an identifiable phenomenon, the
scientist is ready to measure conditions, whether they be controlled or

naturalistic. To do this requires measurement instrumentation.

flgg;g:ing_flggig_flcg£gngngg. In identifying music preference, “most

experimental studies present a music stimulus and then call upon them to
respond to that stimulus in some way.“ (LeBlanc, 1984) Some
operationalize the preference as “liking“ (LeBlanc, 1982), while others
discuss it as a behavioral act of “choosing, esteeming, or giving
advantage to one thing over another.“ (Kuhn,1981) Others have
operationalized preference as the amount of “pleasingness“ (Seashore,
1938: Russell, 1986) however Russell found that “pleasingness', does not
particularly predict the music one prefers. Other measures are behavior
observation techniques that derive the preference from frequency of
occurence such as overt selection via experimentation equipment. There
is much use of measures of how much do you like a certain musical
stimulus, or ratings on a continuum of like-dislike. In both cases they
have used measurement scales with ranges of 1-5, 1-7, up to 1-20.

In another vein of music preference research, certain factors
believed to be relevant are rated using an ordinal scale of “like least“,
“like“, and “like most“ (Baumann, 1964). From this type of scale again,
the researcher is able to identify relative ordering weights of these
factors. As has been the problem with such scaling data of “believed to
be relevant“ dimensions, the ratings may be made on factors that are

actually irrelevant. To avoid this, the communication fields influence

50

social research to develop a “shared symbol set“ with respondents, prior
to measurement instrument design. In this way the dimensions yielded,
contain a relevancy at the on-set of testing in the research process.
This derives from the fact that the factors being measured are shared
between researcher and respondent. After assessing this symbol set, it
than is a logical step that each member of the evoked set of descriptors
can be scaled to yield measures. In addition more finely calibrated
scales are a natural outgrowth of a dimension already measured as
relevant, and questions of “Is this relevant?“ can be further advanced to
“How much of this is relevant?“. The question now is “Is a scale of 1-
to-7 adequate for respondents to discriminate their responses on these
relevant dimensions7“.

Scaling relevant verbal dimensions has been done successively using
nominal categories, ordinal, interval and ratio levels. The semantic-
differential and Likert scales are popular in social science. Many
concepts relating to domains as diverse as message exposure, uses and
gratifications, attitudes and opinions, and product purchase patterns
have been scaled. A continual shortcoming of ordinal and interval level
data is its inability to yield measures which discriminates sub-groups by
“How much.“ At a basic level, this problem stems from the exclusion of
a real zero point, in spite of the simple logical deduction that if a
scaled item is “not applicable“ then it does not exist as a variable
relevant to the inquiry. The definition of zero is “the arithmetic
symbol of 0 denoting the “absence of all magnitude or quantity“

(Webster,1985). This is the case for a verbal concept which is not

applicablel“ One of the things in measurement is the assignment of

51

symbols to some phenomenon. The exclusion of a zero point symbol from a
measurement scale automatically includes all factors in any solution.
By including a zero point in a scale, than the lowest rating on such a
scale enables that item to drop out of the equation and thereby allow a
simpler, more accurate solution with less variables. Research is
scientific, and parsimony, or simplicity is to be opted over complexity
in formulating scientific explanation (Popper, 1959). The application
of the symbol zero is further warranted in the arena of music preference
research, since the message itself is a demonstrated, physically
quantifiable phenomenon.

In the works of 9.5. Stevens (1935) the development of 0 to 10 scale
proved quite useful in early studies of psycoacoustics. Auditory
phenomena are quantifiable as a ratio-level measure. Though measures
used by Stevens were primarily with regard to pitch-detection on “how
much is heard,“ the application of this measurement scale to an evoked
set of music preference descriptors is logical since these are listeners’
own music descriptors and these are being used to define music, a
physically quantifiable phenomenon. As music preference is found to be
described by message structure, mood, and social behavior, all of these
phenomena have been measured using ratio-level scales, yielding useful
data of “how much“ each concept applies. From such ratio-level measures
other questions can be addressed through analytical techniques of how the
set of concepts exist as a set. It has been shown that timbre is
related to music structure and music preference, and preference is a
function of music structure, media transmission and social group. The

logic of this work is to measure how music preference varies with timbre

52

using a lexicon of music preference.

Timbre is multi-dimensional, and the words used to describe music
preference can cover multiple lexical domains. Though respondents do not
have the capability to describe formally the relations between the
different physical componentsof varying timbres, they do have a
generally agreed upon ability to discriminate timbre. They know when the
timbre has changed, and have come to define certain styles of music by
certain timbre. People prefer some timbre over others. Timbre allows
for a certain quality of music. Timbre physically is a multidimensional
multi-variate phenomenon, and certain magnitudes can be ascribed to
different timbres in each dimension. Timbre is multidimensional, and
likewise the aggregate description of musical preference is multi-
variate. For a comprehensive solution to their relationship, it is
logical to measure how this set of lexical music preference concepts
varies with different timbre. If a difference is found, then further
analyses could be warranted. A difference in the relationship between
music preference descriptors and music timbral elements, across different
timbral variations, can begin to provide insight into this traditionally
salient music component and quite recently a telecommunicated musical

phenomenon.

Chapter Three

Methods

Lntnggugtjgn. The measurement of a music preference lexicon as a
function of timbre requires the specification of our independent
variable, musical timbre, and the dependent variable which is a lexicon
for music preference. The independent variable physically describes the
music message using a production component, timbre. The dependent
variable in this study is the lexicon of musical preference, a lingual
component. To examine a tradition, a style lexicon will be used, which
is still often one's first description of music preference. This can be
a valuable indicator of a listener's popular music sub-culture
preference. In addition at the heart of this study is the use a lexicon
of music preference, which is found in raw form to be unique from style
for most listeners in the aggregate, identifying a different conceptual
understanding of music. By applying ratio level measures to this lexicon
of music preference, and using a physically describable timbre in a
controlled music setting it is possible to explore if there is a

relationship of preference with timbre.

Ijnhcg. Timbre will be differentiated by selecting experimental
conditions that differ in both of the two primary dimensions (Gordon and
Grey, 1978: Moorer, 1975: Charbonneau, 1981):

9 Spectral Composition
0 Attack Functions

The spectral composition is described by its harmonic composition of up!
many harmonics and which ones it has from the theoretical harmonic
series. The attack function has been found to order into three relevent

53

54

types: fast piano-like attacks, brass like attacks (which is a bit
slower), and smooth or trapezoidal attack-decay. In exploring whether
there is a difference between timbres, the two main dimensions spectral
composition and attack-decay will be combined in such a way as to present
different timbre. There are three categories of traditional methods for
generating sound: 1) percussive, 2) string, and 3) wind (Hall, 1980). An
example of each has been selected: 1) balls, 2) strings, 3) brass. Each
of these differ from each other in both spectral composition, and attack
decay functions, as Table 2 indicates. Figure 4 shows the reader a
three-dimensional plot (trumpet) of volume by frequency by time, to
supply a graphic representation of both spectral content and attack

functions, and how they interrelate over time.

Table 2 - Components of Bells, Brass, and Strings

 

 

Timbre Spectrum Attack
Bells 3 harmonics Very Fast
Brass 7-11 harmonics Fast
Strings 5-8 harmonics Smooth

 

Dowling & Harwood (1986)

Although some multi-dimensional research of timbre perception points
to piano as a distinct type of timbre in spectral envelope, we will
translate this finding into using bells instead which still have the same
type of attach/decay however the spectral content is a bit simpler to
synthesize. The validity of this timbral difference is retained since
balls are a unique sounding timbre with a quick attack, and exponential
decay. Bells are also a familiar timbre that has been especially popular
with the onset of synthesized and layered (telecommunicated) timbre.

Thus the substitution of balls for piano simply changes our harmonic

 

55

with the onset of synthesized and layered (telecommunicated) timbre.

Thus the substitution of balls for piano simply changes our harmonic
composition while keeping a relatively “piano-like“ attack, and also
presenting a highly relevant timbre in the context of popular music. Our
operationalization of all timbres is synthesized timbre, on a digital

synthesizer using a frequency modulation technique for synthesis.

B:gjg:gngg_Lg§jggn. A lexicon for preference is that set of words used

by the aggregate for describing their preference of certain phenomena or
lingual domain, be it adjective, noun, or verb. Our domain is music
preference and as we have already seen the lexicon used most often by

listeners, researchers and industry is that set of words used to describe

   
  
    

Alpliluda
(Linear)

’//.r};:—: .

   

7,

  

J/Z)

        
 
  

u, I :2,” .
////

w-sww-o-pw-svs—s-o-o»

 
 
 
 

Each slice = 33 I sec

“Enqucncy (Linear)

 

(Figgs, 1981)

56'

genre or style. In addition music preference tends to have various sub-
styles or sub-cultures active within it at any one time. This set of
words is important in initially describing one’s music preference. In
addition a different set of words can be obtained by asking listeners
“What words do you use to gaggnjhg the music you like?“, which yields a
set of words that further describes music communication. Words such as
beat, loud, fast, relaxing and exciting, are characteristic of this
domain of describing the music they “like.“ These descriptors often
indicate a level of “meaning“, in both denotative-embodied and
connotative-designative ways. To assess the actual preference for any
music selection it is important to determine if indeed “description of
the meaning of preference“ for a particular song is the same as “liking“
that song. Therefore it seems important in this exploratory effort to
measure both. In addition both “liking“ and “pleasant“ have been used to
describe preference and both these will be used.

The style and preference lexicon have been collected from the
population under study by asking them “What kind of music do you like?“,
and “What words do you use to describe the music you like?“,
respectively. A quick sort was done and the resultant lists represent
lexicons of style and preference respectively. By selecting those
descriptors that occurred at least twice, the following lists identify
the following music styles and lexicon of preference:

Style blues, classical, country, folk, funk, jazz,
new age, new wave, pop, punk, soft-rock, rock

Meaning beat, dance, exciting, fast, fun, happy,
intense, loud, melodic, relaxing, slow, upbeat

Although “pleasing“ was only identified by one of our intercepted

57

respondents, it is included here since it has been a popular measure in

the past for measuring preference.

flggjg_ficgtgcgngg. Music preference has been frequently measured with
simple descriptive statistics of style preference, and how these interact
with listener variables. The listeners’ variables of interest to music
preference are: 1) Age, 2) Socio-Economic Status, 3) Peer group, 4) Media
Usage, 5) Familiarity, and 6) Dancing frequency. Although findings of
sexual differences, have not conclusively been found across studies, some
studies have indicated differences by sex at certain stages, such as the
change of 11 years to 13 years old. We will include sex this for
purposes of validation. In addition as another social indicator the type
of housing (on-campus, off-campus) is included as measure of potential
differences, a variable that has special significance in this year
proceeding the arrival of an FM radio station on campus.

Exposure of the music message is a strong determinant of preference,
when exposure is repeated. As was discussed in Chapter Two, repetition
and familiarity are integrally interelated over time. As repitition
continues, familiarity goes up and in certain cases, preference goes up,
although in certain cases when a message is “too complex“ familiarity
will go up but preference will not. In this light, the test music to be
used is “simple-uncomplex“ music, made of an 8 beat hook. Popular music
frequently takes the form of music accompanied by lyrics. Lyrics were
not included in the test music because this experiment is designed to
measure music timbre and not limited to popular “song.“ The test music

was made of harmonically “simple-uncomplex“ structured music portion of a

58

typical “pop song,“ without the sung part of the music.

The model for the operationalization closely conforms with that found
in Anderson, Hesbacher, Etzkorn, and Oenisoff (1980). The “pop song“ is
a simply defined combination of musical parts, particularly having a
beat, chordal progression, and a melody. Quite often vocal timbres are
used for the melody. The combinations of these variables contribute to
“song type“ of which there are two recognized type of songs in any style
of popular music: ballads and upbeat. “Tempo basically distinguishes
ballad and upbeat songs: other critical determinants of song type are:
song structure, vocal style, interpretation, rhythm, instrumentation and
arrangement.“ (Anderson, Hesbacker, Etzkorn, Oenisoff, 1980). A song
type category schema referred to by the recording industry includes
'instrumentation, song structure, and vocal style. Since previous
research (LeBlanc, 1981, 1983) shows a preference for instrumental music
over combined instrumental/vocal music, and it can be stated with
confidently that by testing instrumental music, the musical form of
preference is being explored is simpler, vocal timbres were not included
in this study. A popular song was produced that matches the style “pop
upbeat“, containing “rhythm and blues“ components active in the “hook“
bass line, accented by the corresponding rhythmic and melodic chordal
“hook“ phrase. This “pop upbeat“ song structure, is composed of a simple
ABAB pattern, with only two predominant harmonic chords used in each of
the A and 8 patterns, thereby keeping harmonic variation to a minimum

(Krumhansl, 1985).

flgtjg_figglgg, As has been mentioned earlier, if there is a difference

between the preference lexicon of popular music preference for different

59

timbre, this difference may not be identified with traditional levels of
measurement. The exclusion of a real zero point often includes more
variables in a solution then may be the case if variables have been
quantified as “not existent“ or zero (0). In conjunction the independent
variable of timbre is a physically describable music attribute, that
although cannot be measured unidimensionally, does very in magnitude in
each of its two prominent perceptual dimensions.

Since the primary timbral dimensions, 1) spectrum frequencies, and
2) attack-onset have been quantified and measured using ratio-scales,
their use is proposed for use here. To measure a phenomenon, such as
the relationship between sound and words, the words also can be treated
as ratio-level phenomenon. As has been previously discussed, assignment
of number range of 0 to 10 has been done to sound and words using a
direct magnitude estimation scale. The need for ratio-level measures
lies in the crux of our research problem, of whether people discriminate
timbre using a set of words. Such precise word usage may only be found
with ratio precision. In exploring such a relationship, the measurement
tools must be able to discriminate such a relationship, if any exists.
The level of precision afforded with such precise levels can always be
modified in the future if our precison “overshoots' our phenomenon. In
this exploratory effort, measurement will include an absolute zero-point,
and a scale up to 10 continuing in the tradition of listening tests
(S.S.Stevens, 1935, Toole, 1982). The multivariate nature of timbre is
acknowledged, and used.

A lexicon for music preference carries meaning of music, and meaning

at times “does not exist.“ In a music-word relationship this can be due

60

to a mismatch between the set of sounds presented and the set of words
used describe the sounds. If meaning does exist the combinations of
words of this set used to describe any type of timbre in music, can be
more accurately analyzed with available numerical analysis techniques.
The lack of ratio measures being used somewhat derives out of the
available analysis technology available. Twenty years ago the level of a
computational technology was not what is available to them now in this
“micro-millenium“ (Evans, 1979). Ratio-level scales were used for
measuring the lexicon of popular music preference. If this varies with
timbre, the instrumentation for music preference has been numerically set
up to “match the concept(music preference) to a good numerical scale.“
(Miller, Nicholsen, 1976) and reflect the physical nature of the timbral
stimulus. Therefore we can represent numerically the degree of
preference and obtain further meaning of this preference by measuring not
only “What's relevant?“ but for those concepts that are relevant, we can

determine “How much.“

finglxgig. Given measures of listeners' ratings of a preference lexicon,
the nature of the relationship with timbre in the test music was analyzed
with three analytical techniques. The measure of difference between
independent conditions was done using an ANOUA, for each descriptor
across all conditions. To further refine our understanding of any
differences, contrasts were calculated for each descriptor, across all
paired conditions. To measure the set “as a set“ othe; analysis
techniques were used.

As a validity check the lexicon of music preference ratings of general

61

musical preference were checked against the ratings of the timbre
conditions, to check for colinearity. The dimensional nature of the
lexicon for preference was analyzed using a factor analysis, which can
identify how the set of words, in particular combinations, explain
variance in the ratings measured. Factor analysis analyzes “the degree
to which clusters of intercorrelated variables may represent fewer
underlying, more basic, hypothetical variables.“ (Williams, 1979). This
acts in this study as a check of dimensional validity with general
findings in semantics, and can also identify inter-relationships between
members of the music preference lexicon.

As a comparison with traditional methods, a rank order of the lexicon
for general music preference was measured along with the ratio level
measures. The two will be charted to check if the ordinal relationships
among lexical members is retained in the rdsults using ratio measures.
Another comparison with traditional measures was the scaling of style for
each timbre condition, to see how this set of descriptors can
discriminate timbre, compared to the lexicon for music preference. For
each style lexicon, an analysis of variance was calculated.

The numerical nature of the independent variable, timbre, could have
been translated into ratio-level measures which could then enable
regression analysis. This is outside the scope of this work, but if
findings warrant futher investigation, the analysis could yield those
coefficients of pertinent descriptors used in discriminating timbre, thus
formulating a equation for music preference as a function of timbre.
Since the nature of our independent variable is bi-dimensional the

regression equations could be fit to each dimension of timbre. Since‘

62

the nature of this study is not one of prediction, this is the reason for
not including regression analysis in this thesis. The analysis of other
“listener“ variables will be cross tabulated with our primary dependent
preference variables to check if there are any relationships, in an
effort to validate our findings against findings from other scientific
efforts. These cross tabulations will calculate chi-square ratios of
observed over expected occurences.

There were four (4) analysis techniques used determine if any
differences in the music lexicon for preference, as individual variables
and also a set of descriptors, across timbral conditions. T-tests,
correlation afforded us univariate analytical tools, while the set of
words were analyzed as a set using the multi-variate tecchnique of factor
analysis. Besides analyzing the data for determination of support or
unsupport of the research hypothesis, other analyses were run to allow
for comparison of findings with traditional techniques. Lastly some

traditional demographic crossbreaks were performed.

All statistical tests were calculated with a level of significance of
p - .05. This was selected for two reasons. First, this level of
significance although representing a “wrong guess“ 5% of the time, is
currently used as convention in the social sciences. Second, since the
nature of this thesis is exploratory, the purpose is to discover
relations, and therefore in early discovery certain relations that may
not hold to linear solutions, must be retained for further replication at

higher levels of significance in the future.

63

Ihg_flggggcgn_flgggtign. This thesis investigates whether any
relationship is found between musical timbre and a lexicon of music

preference.
The central research question under examination is:

Do people discriminate between music with different
timbre using a lexicon of music preference?

To refine this question we state:

Is there is a relationship between the preference
lexicon and timbre of music ?

Ihg_flggggcgh_flxgg1hggig. Defining the research hypothesis with an
exploratory focus, simply tests whether there is a difference in usage of

the music preference lexicon across timbral conditions.

Hypothesis I: There is a difference in the lexicon that describes
different timbre.

p <> p <> p <> p

Null Hypothesis: There is no difference in the lexicon that describes
different timbre.

P ‘ P ' P ' P

This hyphothesis was tested first by seeing if there is any
significant difference across all conditions, and then further analysis
tested the nature of any differences found. This is the extent of our

exploratory research question and hypothesis.

64

Iggt_1ingcg_gng_flugjg. The timbres used in the research tested three

types of different timbre. To vary timbre at least 2 different timbre
conditions are needed, however to be sure that timbre is what is being
varied then 3 timbre conditions are selected. Due to the exploratory
nature of this effort, for the sake of parsimony other conditions are not
desired. This research uses three timbres: bells, brass, strings. The
timbres used were factory pre-sets from three instruments commercially
available from the same manufacturer: a Yamaha DX-27, DX-100, and FB-01
FM digital synthesizer. The sounds were labelled as “Brass 2“ on DX-100,
“Lo Strings 2“ on the FB-01, and “Glocken' on the DX-27. A pre-test of
these three test timbre was performed to validate the differences in
these three timbre. The pre-test respondents were asked “How different?“
the timbres were from each other. Differences were a scaled on 0 to 10
scale. The bells and brass, and balls and strings were always different,
but the strings and brass closely resembled each other. To correct for
this in the actual test sessions the equalization in the high-end (10-
15Khz) was boosted during the mixdown of the brass melody track, a
technique used by Wapnick (1980) for varying timbral spectrum. In this
way the brightness characteristics of the brass (Beauchamp, 1982) was
accentuated. The synthesis technique used is a frequency modulation
technique using 4 sets of oscillators, known as operators. The
synthesizers are all members of the Yamaha 4-operator based synthesizer
family. The sound synthesizers used to generate the different test
conditions are identical.

The test song is made up of a 16 measure, 4/4, two-chord, key of G

major heavy bass line funk verse segment, and an 18 measure, two chord,

65

key of G minor, latin-feel part (pop music's chorus). This music is
designed to be upbeat enough to sufficiently arouse the test listeners
attention. This is accomplished by using simple harmonic structures

built over two chords, fairly upbeat (less than 100 beats per minute)

tempos, prominent bass pattern, and solid drum beat.

Ihg_nggchng_figgngn. This song was performed using a pre-written and

taped rhythm track using a Yamaha DX-27 and RX-17 Digital Rhythm
Programmer. This rhythm track uses a drum, bass, and chording (using the
same timbre of the bass line), to generate the rhythm, tempo, and general
harmonic structure of the song, including a “hook“ in the rhythm of the
bass with chording. Using this track as input to one track on an Otari
MX-5050, 4 track tape recorder, the other three tracks were filled with
simultaneous performance of our three MIDI-compatible synthesizers. The,
outputs from the synthesizers were routed through a patch bay as input
modules on a Soundcraft 12-Chennel Mixing Board, which then re-routed the
signals to the recording machine. All machines were calibrated using 100

hz, 1000 hz, and 10,000 Hz reference tones.

MW Synthesizer: were confioured by

chaining MIDI control information from the master keyboard DX-27 to the
DX-100 and on thru to the FB-01. With this MIDI control of music
information, simulataneous control of each synthesizer enabled identical
performance dynamics in all three experimental conditions. The
experimental conditions had a simple five-note melody with one note
occuring on the down-beat of each measure, except the occurence of two

notes in the last four beats of section A, or verse portion of the song.

66

The melody was played so it was always triggered throughout the first
sixteen beats, or section A. The melody in section B of the song was
played using six notes, in alternating succession of sets of three notes,
against 8 beats, and two chords: 8 flat minor 9 and a G minor. In this
way the song was constructed to maintain as little variance as possible
in other musical structures, yet attain enough variance to match a
popular song.

From the 4-track version of the test song, identical mixdowns of the
rhythm track with each timbre condition wer made to an Otari 2-track
stereo machine. The specifications for the 2-track mix was a 60/40 on
setting of the two panaromic potentiometers (pan-pots). Another way to
look at it is that the rhythm track and melody track had their pan-pots
set at 60 and 120 degrees respectivley within a 180 degree reference
meter for the two. This attempts to eliminate any particular interaction

between timbre and stereo separation.

Ligtgnjng_1ggtg. The population under study are members of the Michigan
State University undergraduate student population. Particularly, the
students selected for testing were members of introductory undergraduate
communication and telecommunication classes. They signed up for one of
four groups of listeners, one control condition and three experimental
conditions: bells, brass, and strings. The listeners in each group heard
a 45 second cut of two complete verses and choruses, with a fade out at
the and of the second chorus. This way a simple-uncomplex “pop hook“
music test is exposed to the listeners, conducting a test of a portion of
a repeated piece of music, resembling the design of a familiar type of

music, “pop song.“

67

The listeners were asked to self-report their response to the test
music. Before any listening tests, subjects were were asked to respond
to two questions. What words do you use to describe music ? What words
do you use to describe the music you like ? These are asked in an effort
to check the calibration of the instrument used for the next set of
questions. One practice question was asked, to familiarize the
respondents with the measurement scales that were used. The practice
question asked the respondents to rate the lexicon of preference for the
song “Jingle Bells.“ It was thought that this culturally familiar song
is familiar enough to most people to serve as a good practice song, even
from memory.

During the test session the listeners were asked to record their
rating of the music on the ratio-scalings of the preference lexicon and
style, at any time during the listening session. The respondents were
able to fill out scales in any order they found convenient. This allowed
for immediate recording of the detection of any of the attributes. At
the end of each listening session, respondents were asked to report their
media use, music listening and purchase patterns, along with other
demographics including age, income level, education level, ethnicity and
sex. College relevant variables such as fraternity/ sorority membership
college class (year in school), housing type were also included to gauge
any music preference indicators against the socially relevant items. The
listening enviroment was the same for each condition, a classroom capable
of holding about 35 people. There was a total of 120 people sampled with

30 in each group.

68

figl£_flgggcj_1n31:ungn1. The self-report instrument is designed to

measure relationship of timbre and music preference lexicon. In addition
it is meant to assess certain listener attributes by known factors in the
music preference phenomenon. Subsequently the instrument first asks a
set of music descriptors as a validity check of how well does our earlier
identified lexicon fit this population. Then the ratings of the lexicon
are assessed using a scale from 0 to 10. A measure of “how much“
preference is taken by measuring how much one “likes“ and how much one
finds this song “pleasing.“

A rank order of preference for musical style and preference
descriptors is done as a validity check against our ratio-measures to see
if the same relations exist. Media use has been checked in general, with
special attention to radio, recorded, and other music listening exposure,
Radio use is measured in hours, recorded music in number of purchases in
a year, with percentage of all purchases made for each type of recorded
music. The other music listening exposure questions are measures of
yes/no to ownership of certain audio components, and attendance at live
music events. In addition a measure of familiarity is taken by asking
various questions of “if you have heard music like this, where?“. Lastly
the measures of age as years, socio-economic status as “Do you have a
job?“, “Is it part-time or full-time?“, and “How much is your income?“
are asked. Music behavioral yes/no measures used of dance, singing, and

playing (Do you play any musical instruments 7).

Chapter 4

Results, Discussion and Conclusions

ngultg. The set of respondents from the undergraduate classes were made
up of 47.52 females, and 52.5% males, average age was 21.2 years and
with an average income approximately halfway between the income
categories of 2 and 3, which interpolates to an average income of 84500.
The sample was predominantly Caucasion, where 88% of the sample
identified themselves as white. Those employed made up 65% of the sample.
The year in school was represented by 22.7% freshman, 26.1% sophmore,
39.5% junior, and 10.9% senior. The percentage of the sample living

on-campus was 601, and off-campus 39.2%.

 

 

Table 3
Sample Demographics (N-120)
I I
Demographic Frequencies
Age Mean: 20.2 years
Employment: 65% Part-time: 64.2% Full-time: .8
Income 81000-83000: 43.81 83000-66000: 35.0
Gender Females: 47.5% Males: 52.5
Hometown Rural: 26.12 Urban: 12.6
Suburban: 61.3
Housing Student House: 15.1% Greek: 3.4
Residence Hall: 60.5 Family House: 2.5
Apartment: 16.8 Other: 1.7
Race American Indian: 0% Asian: 2.5
Black: 8.3 Caucasion: 88.3
Hispanic: 2.5
Year in School Freshman: 22.72 Sophmore: 26.1
Junior: 39.5 Senior: 10 9

Graduate: .8

 

69

70

The basic use of music through various broadcast media is represented
by an average radio listening of 3.4 hours, with a mode of 2 hours.
Music purchasing exposure is represented by an average number of 14.7
separate purchases in a year, ranging from 0 stated by one person, to 65

albums in a year by another.

 

 

Table 4
Sample Music Exposure Profile (N-120)
Demographic Frequencies
Radio Listening: 3.4 hours per day
Music Purchases: Albums: 19 45’s: 6
(avg. I bought) Compact Disc: 12 Cassette: 56
Live Concert: 6 Other: 1

Number of Purchases: 14.7 purchases a year
Hi-Fidelity Media Ownership:

Cassette: 98.3% Compact Disc: 20.8 Turntable: 82.5
Radio: 98.3 Stereo TV: 27.1

 

Lg;iggn_g£_fi:g£gcgngg. The lexicon of music preference, originally
derived from the sample population, was scaled both for the test music,
as a preference description rating, and also the music liked in general,
which will be referred to later as general preference rating. These
measures serve to test the research hypothesis and validate the findings,
and structure timbre-lexicon findings in relation to the general use of
this lexicon for general music preference. General preference ratings
are shown in table 5, and show that the lexical descriptor with the
highest mean is “beat“ at 8.6, with the lowest mean rating being “slow“

at 5.3. The standard deviations for the descriptors ranged from 1.4

71

(beat) to 2.8 (slow). The measures range primarily in the upper half of
the scale, although “slow“, “loud“, “relax“, “intense“, and “melody“ had
at least 201 of its responses falling within the lower half of the 0 to

10 scale. Such precision would not have been available from a 1 to 5

point scale.

 

 

Table 5
Mean Rating of Lexicon for General Preference
Lexical Term Mean
Beat 8.6
Exciting 8.2
Upbeat 7.7
Fun 7.7
Dance 7.3
Fast 7.0
Happy 6.7
Relaxing 6.5
Intense 6.2
Melody 6.4
Loud 6.0
Slow 5.3

 

In using the ratio level scale, a parallel use of rank order scales
was done as a check of the ratio scales reliability. Table 6 shows how

Table 6
Comparison of Rank Order and Ratio Scales

 

Lexical Term Rank Order Ratio Scale

 

Beat 3.9 (1) 8.6 (10)
Exciting 4.6 (3) 8.2 (10)
Upbeat 4.8 (4) 7.7 (10)
Fun 5.1 (4) 7.7 (10)
Dance 5.8 (1) 7.3 (10)
Happy 6.8 (7) 6.7 (5)
Relaxing 7.0 (10) 6.5 (5)
Fast 7.1 (8) 7.0 (8) *
Intense 8.0 (11) 6.2 (S)
Melody 8.1 (12) 6.4 (8)
Loud 8.3 (12) 6.0 (5)
Slow 8.4 (10) 5.3 (S)

 

! (only case of scale ordering difference)

 

72

the two scaling techniques compare. In all cases except one do the
ordering of preference descriptors match, and the difference occurs in
the descriptor “fast,“ where the numerical difference is approximately
1/20th the scales magnitude (.5/10) which is still more precise than
difference between scale points on a rank order scale of 1 to 12 where
the difference between scale points here is 1/12. What this means is
the ratio scale has reflected the traditional rank order measure. In
addition the divergence from the rank order findings continues to hold
value identifying relations between the lexical set and timbre that may
not adhere to linear modelling but perhaps other modelling such as
curvilinear. Though sophisticated non-linear modelling techniques were
outside the scope of this thesis, it is well documented in the literature
of how perception is often found to be in a logorithmic or exponential
relationship with physical referents. With a ratio level measure, such
curvilinear relations can be studied in the future. In addition the
assignment of numbers to concepts by social groups, can often yield
numerical relations that are exaggerated due to certain individual's use
of extreme scores. With the ratio level data, a logorithmic
transformation is made possible. All the information from the rank order
technique has been retained while increasing the precision of measurement
discriminations between lexical measures.

The lexical measures of musical preference were checked on their
intgcglgtggnggg with a Pearson Product moment correlation. The rule
selected for noting correlation coefficients is greater than 0.40
strength at the p - .05 level of significance. The most pronounced

relationships existed between slow and relaxing (0.65), beat and exciting

 

73

(0.48), exciting and fun (0.47), fun and happy (0.46), fun and
dance (0.43), upbeat and dance (0.43), and happy and dance (0.43).

The general musical preference raw data was numerically transformed
using a natural logarithm for the goal of reducing impact of extreme
scores on the mean and subsequent group mean differences. Using the
transformed data, a factor analysis was run on the lexicon of general
music preference. The factor analysis was run with a criterion minimum
eigenvalue of 1.0 and used a varimax method of rotation. The results
yielded a four factor solution, accounting for a total of 66% of the
variance, with each the four factors accounting for the following
percentage of variance: (1) 27.7, (2) 17.2, (3) 12.5, (4) 8.5.

Table 7 summarizes these factorial groupings, including listing the
preference descriptors that loaded at least 0.50 on the rotated factor
loadings. Factor one has high loadings in dance, exciting, fun, fast,
happy, and upbeat. Factor two includes slow, relaxing, and melodic.

Factor three includes loud, intense, and melodic, and factor 4 includes

 

 

Table 7
Factor Analysis Summary
Factor I of Variance Lexical
Explained Loading
1 27.7 dance, fun,

fast, exciting,
upbeat, happy

2 17.2 slow, relaxing,
melodic

3 12.5 intense, melodic,
loud

4 8.5 beat, loud‘

74

beat and loud. The underlying dimensional structure yields factors

which nioht be described by MW. new.
Wis. and W.

flxggthggj§_lggting. The lexical measures of the test music timbral
variants were analyzed individually with a one way analysis of variance
across all paired and non-paired combinations of timbral groups. The
most prominent findings concern the brass timbre with control and other
timbre. Brass timbre is described as less “Intense“ than the bells, and
the strings. Brass is less “Relaxing“ than the strings and the
bells/strings combined means. Brass is less “Exciting“ than the control,
and brass is less “Melodic“ than the strings, and finally brass is less
“Fun“ than the control. The control was also more fun than the combined
means of all timbral groups. The finding for the “Melodic“ descriptor
between brass and strings was not isolated to these timbres. Strings
were found to be more melodic from the combined means of the other two
timbre (brass/bells), plus from each of the timbres brass, and balls.
These results are summarized in Tables 8 and 9. The use of the greater

than (“>“) and the less than (“<“) signs correspond with the

 

 

Table 8
Lexical Differences of Timbre by Timbre Combinations
Control Brass Bells Strings
Control ) Fun ( All 3 timbre combined )
Brass < Intense (combined)

< Relaxing (combined)
Bells

Strings > Melodic (combined)

 

 

75

level of the timbre on the vertical axis of the tables. An example is
that the control timbre is greater in fun than the brass.

Table 8 shows that the control varied in “fun“ from all 3 the timbral
experimental conditions, brass varied in “intensity“ and “relaxing“ from
bells and strings combinations, and strings varied in “melodic“ from the
brass and balls combination. Table 9 shows similar results except the
finding of brass being less exciting than the control condition, and

brass was less “fun“ than the control. Table 9 also

 

 

Table 9
Lexical Differences of Timbre by Timbre
Control Brass Bells Strings
Control
Brass < Exciting < Intense < Intense
< Fun ( Relaxing < Melodic
Bells < Melodic
Strings

 

shows that the only finding of difference between bells and another
timbre, is with strings, in its lessor level of “melodic.“
Null Hypothesis: There is no difference in the lexicon
that describes different timbre.

The results from the above analyses demonstrate support for rejection
of the null hypothesis. The measures of some members of this lexical
set of music preference descriptors, ygnjgthetnggn_tinhpgs. The
descriptors that varied are “intense“, “melodic“, “exciting“, “relaxing“,
and “fun.“ Descriptors that remained relatively stable across timbral

variations were “loud“, “fast“, “happy“, and “dance.“ These serve as

 

76

indicators of control of the apparatus (music) in loudness (loud) and
tempo (fast). Likewise the danceability was apparently unaffected by
timbre as was the level of happy. The stable measures were also
indicative of the scales capability for consistent measurement, compared
to varied measures found. Table 10 lists lexical mean ratings by timbre.

Table 10
Preference Lexicon Means By Timbre

Preference Control Brass Bells Strings

 

Beat 8.9 8.3 8.5 8.5
Slow .9 1.1 1.1 1.7
Upbeat 7.8 8.0 7.3 8.2
Relax 3.2 2.6 3.6 4.3 *
Excite 7.1 5.8 6.4 6.8 e
Loud 6.3 6.3 6.9 5.9
Fun 7.4 6.1 6.9 6.4 *
Intense 4.8 2.9 4.9 5.0 0
Fast 7.8 7.1 7.3 7.0
Happy 6.6 6.7 7.2 6.0
Dance 7.5 7.2 6.4 7.3
Melody 3.4 3.4 3.4 5.0 *

 

0 represents significance at p - .05

It is worthwhile to point out that the findings of differences in
peoples’ descriptions of timbre variations are verified in other
measures. First, the finding of “intense“ and “melodic“, the variates
whose significant differences were found across the combined variance of
all Lianne gngggs, also are described together by factor 3 found in the
factor analysis. Second, the relationship between use of descriptors for
general preference and test music identification, was checked for
colinearity in these two measures of the preference lexicon. There were
not any findings that were indicative of such a trend. The preference
ratings assigned to the test music were different from the general ‘

preference numerical assignments such that the differences across timbre

 

77

were not also reflected in general music preference ratings. For
example, the brass group did not necessarily dislike “less intense“
music, but they did like slower music. Across timbres, brass was found
to be less intense while intensity did not characterize the general
preference of the brass group of listeners, in niether a positive or
negative trend. In the same light, the strings group liked more upbeat,
fest, dance music and this may account for the strings being described as
more relaxing than the brass. The brass condition was found to be less
exciting than the control, while the brass group liked slower music, and
the control group liked less loud: intense: relaxing: melodic and more
dance music. The strings were described as more melodic, and the strings
group liked more upbeat, dance and fast music than the others. This
shows that the application of this word set to the test music, was
different than its application to describing ones’ general musical
preference. There was no apparent signs of a linear relationship between
the general preference lexicon and use of this lexicon for timbre

preference. The instrument was sensitive to the independent measure.

ijg_gng_filggging. The fundamental descriptors for musical preference in
levels of “like“ and “pleasing,“ were scaled across timbral conditions.
The findings are that people significantly liked the control more than
all the timbral conditions combined, and more than the brass and balls
separately, but not more than the strings. The strings are “liked“ more
than the brass. The “pleasing“ ratings of the control condition were
significantly higher than the brass, but not more than the balls or the

strings. The strings were more pleasing than the combined means of the

78

brass and balls. So here we see that two traditional measures, often
used interchangably, yield different groups of variance among timbre
groups. Apparently the “like“ concept identified the bells group as less
likable, but not less pleasing, and strings were more pleasing than the
combination brass/bells, while strings were “liked“ more than only the
brass. The ratings descriptors of “like“ and “pleasing“ varied
between some timbral conditions, however they varied differently from
each other.

The relationship between the “like“ and “pleasing“ descriptors and
the lexicon for music preference was analyzed by calculating Pearson
Product Moment correlations between them. Again using the rule of any
correlation greater than 0.40 at the significance level of p-.05, as
notable, the following relationships are worth noting. “Like“ is
correlated with “exciting“, “upbeat“, “fun“, and “relaxing“. “Pleasing“
is correlated with “relaxing“, “exciting“, and “upbeat.“ The only
difference between “like“ and “pleasing“ here is that fun does not show
up as correlated with “pleasing“, and the strongest relationship with
“like“ is “exciting“ while with “pleasing“ it is “relaxing,“ although

this difference is not significant.

51y13_ngjggn;_ELngLngg. The measures of the set of music styles

presented to the listeners after hearing the test music were analyzed
using a one way analysis of variance for each of the styles across all
the timbral groups. The strings are described significantly more as
“rock“ than brass, bells, and the brass/bells combined means. The brass
was significantly less “rock“ than the combined means of “strings/bells.“

The other style variation across timbre is that brass is less “jazz“ than

79

control, bells, strings and all combinations of the timbral conditions.

 

 

Table 11

Group Means of Music Style By Timbral Group
Style Control Brass Bells Strings
Rock 2.7 1.8 2.5 4.0 *
Blues 1.9 1.3 1.5 2.1
Jazz 6.3 3.1 5.8 6.4 *
Soft-Rock 3.4 2.9 2.7 3.6
New Wave 3.7 3.9 3.4 3.8
Folk .3 .1 .2 .5
Pop 5.2 4.9 4.7 5.1
New Age 3.7 4.2 4.6 4.5
Country .1 .3 .2 .6
Punk 1.5 1.6 1.4 2.?
Funk 5.3 5.9 4.4 5.0
Classical 1.0 .6 1.2 1.0

' significant at p - .05
To check how the style lexicon varied with the lexicon for musical
preference, a Pearson Product Moment Correlation was run between the two
lexical sets. Three significant (0.05) correlations were found, with a
strength of at least 0.30. One is pop with dance (0.46), jazz with

exciting (0.39), and folk and slow (0.30).

Egajljgcity;_flhgng. One question asked about whether listeners had heard
any music like this before, and if so “where?“. The responses were ‘
classified according to the following nominal categorization: 1) Records,
Tapas, CDs, 2) Radio Music, 3) Radio Ads, 4) Television Programs,

5) Television Ads, 6) Television News, 7) Concerts and 8) Bars. The
findings although not statistically testable due to sampling size, do
offer insight to the applicability of timbral preference findings. For
this reason the following findings are reported if at least 502 of the

test condition group respondents selected a particular response. The

80

control and strings groups found the music they heard to resemble
Records, Tapes and CDs. Although just under 50%, the control, bells, and
strings were thought to resemble music heard on the radio, as Radio
Music. None of the conditions were thought to resemble radio or
television advertisement music. One interesting finding is that the
brass condition was overwhelmingly thought to resemble television program
music, in comparison to other timbre conditions. None of the test groups

indicated any similarity in test music with television news, concert or

bar music. The findings reported here simply exemplify the applicability

of timbral preference findings to particular media channels.

Dgnggcgghjggg_fijngingg. The lexicon of music preference was cross-

tabulatad with the demographics: age, sex, housing(on/off campus),
income. Off campus respondents rated loud and beat higher in the music
they like than did on-campus respondents. Of the off-campus respondents,
50% rated beat as a 10, compared to 26% of the on-campus respondents.
Although not significant sixty percent (60%) of the females describe that
they like (in general) with a 7-10 rating for happy, compared to 441 of
the males. The rating of 7-10 for the descriptor dance was reported by
79% of the females compared to 53% percent of the males. Although not
significant, there appeared to be higher income preference for more

melody and less excite.

flugjg_flghgxigc. The music behaviors of dance, sing, and play (”If you
play music,. . .“) were cross-tabulated by the lexicon for general
musical preference. One finding along the verification domain is that

77% of the people who dance more than once a week, rate dance highly as a

81

component of their general musical preference. This compares to 12% of
those people who dance once a month. Forty percent (402) of people who
dance once a day, rate upbeat as a 10, compared to 16% of those people
who dance once a week. Apparently dancing alot corresponds with liking
upbeat music. Dancing more than once a week, corresponds with liking
music that is fun, however as shown earlier, fun and upbeat are in a
strong positive relationship with each other. Likewise, fast, and happy
music is liked by dancers. The dancing music behavior appears to
correspond with the first dimension of the preference lexicon factor
analysis. One finding related to musicians is that musicians like less
slow music, than do non-musicians. There are findings between this
lexicon of music preference and music behavior, particularly the behavior

of dance, but perhaps even musicianship is related to some words.

MW- Expeaure to varieties of sporting

media coverage was reported by listeners. These yes/no reports were
crosstablulated with the lexicon for music preference, and the findings
are two in number. First, professional baseball(Oetroit Tigers) radio
listeners and college hockey (MSU) television viewers like less slow
music, than those who do not watch such athletic coverage. Second,
college hockey television viewers like loud music more than those who do

not watch college hockey events.

Bgd;g_£xggggcg_gnfl_figyjng, Radio listeners who listen to radio 2 to 4

hours a day rate music they like with dance as a 7-to-10 rating. Of this
group, those who rate dance as a 10 level of preference, makeup 39.2%,

42.1%, and 35.3% for the two, three, and four hour listeners

82

respectively. The rounding point appears to be around 3 hours for the
point where one finds 50% of the listeners liking music with a dance
rating of 7-10. People who listen above or below the 3-4 hour range
seem to rate dance lower in their general musical preference.

People who buy 0 to 10 musical purchases rate “dance“ the highest,
with a decreasing amount of “dance“ preferred by members of higher
purchasing groups. The members of the 0 to 10 music purchases group have
the highest percentage of 7-9 ratings at 77%, where the percentage in the
group buying 11 to 20 music purchases is 66%, and of the group who buys

21 and more purchases, 43% rated dance in the 7-10 range.

niggnggign. This sample of students used the lexicon for music
preference in different amounts to identify the timbral conditions,
particularly the descriptors of exciting, fun, intense, relaxing, and
melodic. Across all timbral conditions' variance (including or excluding
the control) the lexical set that described timbre was made of intense
and melodic. It was previously noted that these two descriptors align
themselves on factor 3 from the factor analysis, which is another
verification of the lexical measurements. Compared to the three factors
identified by Nielzen and Cesarac (1981), 1) Tension-Relaxation,

2) Gaiety-Gloom, 3) Attraction-Repulsion, the factors of the music
preference space may match if factor 1 (dance, exciting, fun, fast,
happy) is viewed as a gaiety-gloom factor, and factor 2 (slow, relaxing,
and melodic) and 3 (intense, melodic, and loud) combined as an intense-
relaxing factor, with factor 4 (beat, loud) as a attraction-repulsion

factor. There is some similarity here but the divergence is holding

 

83

information and error and therefore further investigation is warranted to
identify each. In comparing these factors to those found by Maher and
Berlyne (1981), it might be that the factor 1 found in this study may be
an evaluative judgement group, factor 2 and 3 as an internal states
group, and factor 4 a collative group describing the designative
information. Again there are similarities and divergencies with previous 6 :

studies. The lexicon of music preference did very in part when applied

 

 

to different test timbre, and therefore the research hypothesis is
supported by the findings. At the same time there were some other i
patterns of preference descriptor ratings that are worth discussing.

The findings of the brass being less intense: relaxing: than the
glocken and the strings, and less exciting: and fun than the control,
plus it was the brass that was liked least by the sample, it would
conclude that people did not like the brass version of the song as much
as other timbres due to its reduced intensity (not volume), relaxing,
excitement, and fun. At the same time there may have been another factor
at work, particularly vibrato. The brass timbre used as timbral group 1
from the frequency-modulation synthesizer had been mixed down and
finalized when it was realized that a noticable level of vibrato was
present in the synthesizer timbre. As LeBlanc (1986) found in a study
particularly focused on vibrato, preference was for “minimal levels“ of
vibrato. The amount of vibrato in the brass timbre must be examined to
determine if indeed the amount present was enough to define it as greater
than minimal. It was noticable to this researcher.

The finding of strings as significantly more melodic, is an

interesting finding. The components of melody that were present were so

84

minimal in terms of variation and duration, that in combination with the
slow attack and virtually zero decay from the way it was performed (with
keys depressed always) it may have sounded as if there was no melody.
Compared to a timbre such as the bells which had a ore-programmed decay
rate, one may think that this bell timbre had more of a melodic type
duration, and therefore was more melodic. This was not the case. The
spectral content of the strings being a combination of odd numbered
harmonics, and a slow attack time, were found to be significantly more
melodic than all other timbral conditions, but not more than the control
condition. It should be noted here that Wessel (1979) found some
interesting relationships between adjacent notes (melody) and timbre
space. Though his results showed an effect on adjacent notes when
varying “timbral distance,“ which means timbral variation, there could
have been a related phenomenon at work in the finding of strings being
more melodic. In the findings here though the timbral distance would be
zero between adjacent notes of the single tape track, however all notes
taken adjacently could present some variations in timbral distance
between adjacent notes. This is an aspect of timbre research for the
future, particularly in reference to the lexicon for timbre.

The music preference measures of “like“ and “pleasing“ were found to
vary with timbre. The amount of “like“ was higher for the control,
compared to the brass and the bells, but not the strings. The amount of
“pleasing“ was significantly higher for the control than only the brass.
Although the order of timbral conditions in terms of highest to lowest
was the same for each measure, there were more significant differences

found using the word “like,“ than using the word “pleasing.“ This may be

85

due to the fact that like is an active verb while pleasing is a sensual
state. Perhaps the pleasing state is more sensitive to other components
in the music than timbre, however the liking of the music was apparently
sensitive to the timbral variation. The higher preference for the
control and strings version could be consistent with the findings of
Hargreaves (1984) that the lower the complexity of music, the higher the
preference by the general listener. The control is structurally a less
complex version of the song since it lacks an entire track of produced
music. The strings could be thought of as less complex due to its smooth
ascending attack, which limits temporal variation as discussed previously
in reference to melody. These two versions of the song, if viewed as
less complex could very well explain for their increased “like“ and
“pleasing“ ratings. It can be stated though that listeners liked some
timbre conditions more than others, and also found at least one timbre
“less pleasing“ than other timbres.

The style lexicon revealed differences across timbre, however less of
its members were measured as different across timbre as members of the
preference lexicon. In comparison, the preference lexicon from which 5
descriptors were used differently across timbre compared to two style
descriptors used to identify timbral difference. This also means that
less style descriptors varied, particularly only “jazz“ and “rock“ varied
with timbre. The findings of strings having more “rock“ than the other
two timbre is interesting in that strings do not at first listen sound
like a rockin’ sound, however it should be noted that the spectral
content of the strings closely align with the timbre that so often is

credited with making that “rock sound“ especially early in rock history,

86

the electric guitar. While the strings were high in “rock“, at the
other extreme was the brass, with the bells “rock“ ratings falling in
between, yet different only from the strings. The style lexicon varied
with timbre in 2 of the 12 descriptors used, and besides rock, the
variance of “jazz“ ratings across timbral conditions was significant.

Brass was significantly less jazz than all other conditions, where
the overall mean rating of jazz for this song was 5.4. This could
indeed be consistent with what LeBlanc (1986) states as jazz's use of
“little vibrato“, which if heard by the listeners may have instantly
labelled this music as “non-Jazz.“ Presuming this to be true, than a
timbre with any more than “little vibrato“,*mey be negatively associated
with jazz. The measure of jazz did vary across timbral conditions, and
and vibrato may have been work. It is still the case that whatever
difference was present, be it due to its reviewed timbre structure
(brass), or some other characteristic parameter of the timbre (vibrato),
significant differences were found between brass and every other
condition: control, bells and strings. Therefore style acted as a
usable lexicon for assessing timbral differences, however it did not
identify as many differences in timbral conditions as did the lexicon for
preference.

The lexicon of music preference used in this research demonstrated
its salience to the music preference research domain. It was used to not
only rate general musical preferences, but was also used to rate test
music, and identify differences in the test music. The differences,
corresponding with varying timbre, were significant in 5 of twelve words

from this lexical set: exciting, fun, intense, melodic, and relaxing.

 

87

The strongest differences occurred in two descriptors that occur in the
same factor identified in factor analysis, thereby fortifying these
results and allowing for a synthesis of these results with previous
related findings. The ratio level measurement scale of 0 to 10, appeared
to work well by presenting a scale of the dependent variables, matching
their concept to a good numerical scale. In all but one case in
assessing general music preference did the lexical measures span 0 to 10,
and in all but 4 cases when assessing test music preference.

In terms of standard demographic breakdowns, the finding that females
describe the music they like as more dance than males is significant, and
in many ways makes since from the perspective of socialization theory
where females may traditionally practice such behaviors at an earlier
age. An interesting check on this would be to find out if males state
that they actually dance any less than females. According to this
sample, this comparison was not possible due to sampling limitations.
Related to the dance concept is the reports of high “dance“ ratings for
general music preference, and as stated earlier this served as a good
validity check for the dance concept and lexical measures. Other
validity checks were that people who dance like more beat, upbeat, fun,
happy, feet music. This correspondence with factor 1 from factor
analysis serves as further evidence for labelling this factor an
activation dimension, rather than an evaluative dimension. With this
interpretation there is a convergence of factor 1 and factor 4 into one
combined factor.

The strongest findings from identifying where respondents heard music

like their test music was brass sounded like television programs, then

 

88

control and strings like records, tapes and compact disc music. Though
these findings are not statistically significant, they demonstrate how
music research of an experimental nature could be classified according to
the competing media outlets for music distribution and exposure.

The final point for discussion is the however slight findings of
levels of radio listening and music purchases by the music preference
lexicon. In this case there were no significant findings however most
importantly the utility of the measures in relation to the preference
lexicon measures that correspond with radio listening and music purchases
was demonstrated. The descriptor dance again was found to correspond
with certain levels of radio listening and music purchase behavior,
indicating some sort of correspondence between this overt musical
behavior and overt music exposure behaviors. Besides music exposure
behavior, the media programming breakdowns yielded some interesting
examples of how such a lexical breakdown could make possible music
programming decisions that would acccompany certain program-types. The
findings demonstrate for example that professional baseball radio
listeners and not TV viewers, like less slow music, as do college hockey
television viewers. Perhaps this could indicate that the high paced
music often used under the intros and extros of television baseball
coverage, is not particularly relevant to the television baseball fan,
while it is to the radio baseball fan. In this case tempo may become a
factor for sporting events music. The other finding between programming
type and the lexicon of music preference is that college hockey
television viewers like more loud music than those who do not watch

televised college hockey. This may only substantiate the fact of that

89

old familiar timbre at hockey games, the organ (although many arenas now
use synthesizers) is played blaringly loud because the crowd likes it.
Perhaps loud music accompanies aggressive behavior. An application of
this to televised hockey production would be using louder music beds for
intros and extros for hockey coverage which would appeal to the listening
and viewing audience, in addition to acting according to that old axiom
of advertising: to gain attention. There may be a whole avenue for
research available for hockey music, particulary with regard to timbre.
An article in Sports Illustrated (Kirshenbaum, 1984) deserves mention
here. The article briefly alluded to the impact of an organist’s
selection of “Eye of the Tiger“ (ROCKY III theme song) on the continued
flaring tempers and eventual ejection of two players from a one game.
Likewise another organist's selection of “You Always Hurt the One You
Love“ was credited in another game when the music started after a fight
broke out. The fight quickly petered out after the onset of the music.
These examples demonstrate that others think about the impact of this
music. A question in the context of this thesis concerns whether the
timbre used in the two different games, varied at all from each other.
This area is wide open for development and investigation.

The lexicon of music preference used to measure music preference and
test timbral conditions worked and revealed differences. The differences
found between conditions in reference to exciting, fun, intense, melodic,
and relaxing demonstrated support for rejection of the null hypothesis
which stated there is no difference in the way people use the lexicon
across different timbres. The use of style as a lexical indicator of

timbral difference was less applicable, however it did have applications

Van

90

at identifying timbre variation. It must be pointed out that when there
was a relationship between a style descriptor and timbre, it was a strong
relationship, while the other style ratings were found to be clearly non-
applicable, almost as if style acted as a step function in identifying
timbrel differences. The lexicon of music preference used here ‘
demonstrated a useful technique for assessing musical preference
different from style, and in some ways more precise, since more
descriptors were applicable to more of the conditions, while at the same
time more descriptors varied across the test and control conditions. At
the same time this is an exploratory study, and all findings are expected
to be subjected to extra scrutiny and replication. The use of
traditional preference measures of “like“ and “pleasing,“ showed measures
that they were tapping a similar phenomenon although the differences
between the two measures particularly in the timbres they varied with
warrant further investigation. Examples are whether indeed strings are
more pleasing than other timbre, or that control is liked more and is
more exciting. At the same time the addition of timbrel elements also
affects the complexity of the music, which could become a focus for
future timbre research. Lastly the use of the lexicon of music
preference as a scaling variable in identifying certain demographic, and
media programming exposure patterns although not the focus of this work,
proved to exemplify the potential utility for identifying music
preferences for certain demographic and media program target markets.
Along with the demographic findings is the measure of familiarity.
Although the test must was not expected to be familiar to the listeners,

it did have some level of familiarity in regard to certain classes of

91

music media and outlets. This approach to classifying test music could
help place music of particular timbre in media and programming niches.

The style of the song was originally conceptualized as an upbeat, pop
and funk style. This was verified by the ratings of the upbeat lexical
measure falling in the upper quarter of the measurement scale, and the
style descriptors funk, pop, and Jazz all were rated greater than 5 on
the 0 to 10 scale.

The control apparatus of MIDI for telecommunication of performance
data to all of the timbrel conditions during test music production,
proved quite useful. The control of the performance was brought down to
a digital level of note-on/ note-off, across all timbres to the same
second within 32 nanoseconds, which is the recognized delay time of MIDI
signals between synthesizer nodes. The delay is probably unoticable to
the human ear between the first two or three synthesizers in a chain,
however after that the results may begin to be noticed as “delayed“,
lagging behind or dragging down the tempo. Besides this one drawback of

MIDI, it was adequate for the research task at hand, and worked gegjly.

Baggaggnflgtigng. The use of the lexicon for identifying music preference
and particularly timbrel preference, in this exploratory study, proved to
be quite useful, and productive, although there are some recommendations
for future research that can be categorized as:

1) analytical

2) production control

3) musical components

4) sampling scope

The analytical results of differences across timbre were useful

although analyses of the data were limited due to the nominal level

92

categorization of timbre. In the future it would be quite useful to have
the independent variable scaled in some ratio measures, such as
frequency, intensity, and duration so that more sophisticated analytical
techniques could be used to not only identify differences across timbrel
groups but also identify the trends underlying these differences.
Mathematical functions such as Bessel functions, Fast Fourier Transforms
(FFT), and other techniques using line-segment approximation functions
have been not only used to explain timbre but also to create it in the
domain of electrical engineering, particularly timbre synthesis. With
actual ratio measures equated in mathematical functions the results could
begin to assemble predictive models for music timbrel preference. One
last analytical recommendation stays with convention which is a call for
replication of these findings.

The production of the test music could be controlled in terms of such
factors as vibrato that may have been introduced in timbrel conditions.
One way to identify such extraneous factors before actual test sessions
could be done by using a focus group to review the test conditions before
going into the field for testing. Through the use of such a focus group,
factors as vibrato, loudness, and equalization could be examined. One
note of caution about equalization though, particularly with regard to
timbre, is that the timbre is effected by equalization, and therefore
must be utilized with this in mind particularly for timbre research.

This may be less of a factor where other non-pitch related factors are at
work such as tempo, or beat. In addition to the use of a focus group for
review of the test music prior to testing, another production

recommendation is the use of a complete MIDI studio with computer control

93

of all synthesizers in a star typology network. This configuration first
reduces the maximum performance delay to synthesizers of 32 nanoseconds
and second the computer control enables a wider array of control
parameters not only in performance of the music but also for systematic
variation in timbrel design, melodic components, rhythm components, and
general compositional variations of notes, tempo, loudness (volume) and
sustain (which could be viewed as a timbre decay envelope variant). In
addition, the whole arena of dynamic variation such as timbre changes can
be examined with a extreme amount of control. Using intermediate
processing equipment that responds to MIDI control makes other “mixdown“
parameters such as stereo pan settings feasible as a research topic,
particulary using this digital control. The computer based MIDI studio
can serve than as not only an adequate apparatus for control, but also an
apparatus for innovative controls. With multiple computer-based MIDI
work stations the MIDI research laboratory can create a new age of music
research, from behavioral measures, compared to self-report as was used
in this study.

The recommendations for musical components is actually to expand the
range of music components found in relation with the music preference
lexicon. Identifying, as was done in this study, a subset of lexical
variates that correspond with a music component (timbre) is a model for
other designs. Such designs could investigate those lexical subsets which
vary with other musical components such as pitch (tone or octave), tempo
(time), note and rest length (duration), and mode (scaling temperment).
To expand the applicability of the music preference lexicon into the

research of these other music components is recommended for the purpose

94

of continuing a quest for thorough and comprehensive research designs.
The scope of this study was limited to college students and is
therefore only generalizable to this limited college population. More
sophisticated sampling and larger samples would allow for not only
results generalizable to a larger population, but also allow for sub-
group sizes that would allow statistical analyses of such sub-groups and
appropriate criterion variable. An example breakdown would be the

preference lexicon by radio listening by age.

anglugjgng. In a world with increasing diversity of all forms of
communication, the musical forms of communication we encounter are also
more varying in time. Timbre is one component of music that has been
long recognized for its impact on music. Due to limitations in the
ability to control timbre, the level of rigorous scientific timbrel
control available to the music or media researcher has been limited
except in the area of sound synthesis. Here timbre synthesis and timbre
perception research have been at the forefront of sound synthesis. Now
with the introduction of music telecommunication technology timbre can be
varied while performance remains a constant. Such is the scenario within
which this thesis began. Observing an ever increasing amount of timbre
in a music environment where industry relies on little research and what
it does have available is fairly crude, a technique for scaling
listeners’ verbal descriptions of the music they like had been proposed.
By scaling the actual words (vernacular) that a population uses to
describe the music they like, it has been presented that it could serve
as a ratio-level measure that indicates not only certain denotative

domains but also connotative domains of music meaning and preference. It

 

95

acts as a measure of general music preference plus identifies certain
variations of test music messages.

Through the identification of how these words are rated and rated
across varying timbrel elements in music the warrant for this thesis
research is that results may be obtained that describe the actual timbrel
makeup of songs that are preferred, the nature of the preference and how
much they are preferred. Differing from some present research by
attempting to measure impacts of actual music content, at least as the
average listener would describe it, the scope of such research can expand
into the arena of predictive music research. This results from varying
the music message components and measuring how people use the lexicon of
music preference to describe the variations. This is accomplished by
using a controlled music apparatus, listeners’ own words as feedback, and
precise measurement instrumentation. The significant findings of lexical
timbrel differences warrant further investigation of this research topic.
This can not only identify a lexicon for timbre but also a lexicon for
other music components or general music preference. Used in conjunction
with other music behaviors, exposure or social indicators, the lexicon of
timbre preference begins to provide profiles of different music “taste
cultures.“

The findings of this experimental research demonstrate the usefulness
of new technology while also showing that in the course of increased
control there are other attributes to which the need for more control
becomes evident. In the case of this research, vibrato was suspected as
playing a role in the brass timbre findings. It was only through the

added controls of timbre performance that another variable control was

96

recognized as needed. In this inquiry the variable vibrato takes on
different meaning in music synthesis since it does not actually reside in
the arena of performance as it once had. For example, vibrato in
acoustic wind instruments is an expressive breathing technique using
pulsing degrees of volume and pitch modulation while now electronically
it is a hardware setting for low frequency oscillation. So with the
added control yet another distinction between synthesized performance-
production and live performance-production is discovered. New controls
bring a finer focus to the music preference lab which reveals a need for
further controls and replication.

MIDI control helped research the timbre phenomenon while at the same
time uncovering findings that warrant replication and further refinement.
Significant findings in more than one lexical comparison across timbre,
indicate there is reason to believe this research design has merit and
will work in the future. The stage is set for not only researchers but
for the music and media industry to see and hear that more precise
uggjggl_gzgjgngngg measures than currently in use are possible and
available. These measures not only classify groupings of people, but
provide levels of control and measurement that can provide for predictive
modelling of musical and listener message design. The musical message is
the music structure itself (brass, bells, strings) while the listener
message design is the utilization of the lexicon of music preference for
not only laboratory research but also for eventually verbally
communicating with listeners. Messages could be designed to notify them
of music availability using the symbols they use to describe certain

musical components such as timbre which could now be purposely included

97

in “tailor made“ music compositions.

With a research tool that identifies not only music that people like
but also what ggct of the music they like, hgg they like it and how mush
they like it, the assignment of music preference indices to certain songs
and classes of songs becomes feasible. These classes may become known as
new styles, sub-styles or just continued to be identified by the style
and preference lexicons in use now. Indicators of music behavior (dance)
when combined with such lexical preference measures can enable music
gatekeepers in settings of such music behavior to select the right “tune“
for the preferred behavior be it singing, dancing or exercise. At the
same time lexical indicators of music exposure through radio or music
purchases can act as a guide for the commercial music sector. Radio can
better program to their market segments, and recording companies can
select new music for producing on the basis of the “music.“ For the
music communication researcher the relationship of the words
ujth_1hgn§glygg can act as a general indicator of music preference,
describing how people think and talk about music communication. This
furthers understanding of and better reduces uncertainty in the music
preference research communication process while also expanding the
theoretical framework for studying music preference. For music cultures
and communities it means that the identification of music “they like“ can
begin to correspond with particular music components such as timbre that

have been identified simply by using their own words.

(1) Charts

(2) Automate

Eccimin

Charts are the tabulated listings of released music
ratings. There are such listings in many popular
music industry publications including Billboard,
Cashbox, Downbeat, and Rolling Stone.

Use in reference to automated communications whereby
through a combination of techniques and facilities by
which intelligence (decisions-making) is conveyed
from one point to another without human effort. In

these sense of this thesis MIDI, or Musical
Instrument Digital Interface, in conjunction with
microcomputer control, can automate some music
performance.

98

W

Alpert, Judith
1982 “The Effect of Disc Jockey, Pear, and Music Teacher
Approval of Music on Music Selection and Preference.“
Journal of Research in Music Education 30,3:173-186

Anderson, Bruce: Hesbacher, Peter: Etzkorn, K. Peter and
Oenisoff, R. Serge
1980 “Hit Record Trends, 1940-1977.“ Journal of
Communication Spring 1980

Anderton, Craig
1986 “The MIDI Recording Studio,“ Musician, special ed. ,
1, 31-36

Anthony, Dave
1987 “Research Helps, But Never Underestimate Instinct,“
Billboard, 99, 50:

Baumann, Victor H.
1964 “Teen-Age Music Preferences.“ Journal of Research in
Music Education 75-84

Beauchamp, James W.
1975 “Analysis and Synthesis of Cornet Tones Using
Nonlinear Interharmonic Relationships.“ Journal of
the Audio Engineering Society 23,10: 778-794

Beauchamp, James W.
1982 “Synthesis by Spectral Amplitude and “Brightness“
Matching of Analyzed Musical Instrument Tones.“
Journal of the Audio Engineering Society, 30,6: 396-405

Beal, A. Lynne
1985 “The Skill of Recognizing Musical Structures.“ Memory
and Cognition 13, 5: 405-412

Blalock, Hubert M.
1979 Sggjgl_fitgtjgtjgs, McGraw-Hill Book Company,

New York

Bogart, Leo

1956 Ing_flgg_gfi_lglgyjsjgn, Frederick Ungar Publishing Co.,
New York

Boyle, J. David: Hosterman, Glenn L. and Ramsey, Darhyl S.
1981 “Factors Influencing Pop Music Preferences of Young
People.“ Journal of Research in Music Education
29,1: 47-55

99

100

Cable, Michael
1977 Way—W. U-H- Allen.

London

Carrington, John F.

1949 hiking nouns 9.1 Bibles. Carey Kinoeoete. London
found in lbLflnibmeLcouLduuL Merriam 1964

Chaffee, Steven H.

' 1985 “Popular Music and Communication Research - An
Editorial Epilogue.“ Communication Research 12,3:
413-424

Charbonneau, Gerard R.
1981 “Timbre and the Perceptual Effects of Three Types
of Data Reduction,“ Computer Music Journal, 5,2: 10-19

Chowning, John M.
1973 “The Synthesis of Complex Audio Spectra by Means of
Frequency Modulation.“ Journal of Audio Engineering
Society 21,1: 526-534

Christenson, Peter 6.: DeBenedittis, Peter and Lindlof, Thomas R.
1985 “Childrens Use of Audio Media.“ Communication Research
12,3: 327-343

Oenisoff, R. Serge .
1973 “The Evolution of Pop Music Broadcasting: 1920-1972,“ Popular
Music 6 Society, 2, 3: 202-225

Dinkelackar, James Walter
1982 “Distinguishing Radio Audience Segments by Format Preferences
in the Multichannel Programming Environment: A
Multidimensional Approach,“ Dissertation for Degree of
Ph.D., Michigan State University

Dowling, W. Jay and Harwood, Dane L.
1986 “Timbre, Consonance, and Dissonance,“ chapter in

nggj§_flggnjtign, Orlando, Academic Press, Inc

Durkheim, Emile
1953 Sggjglggy_gnfi_flhjlgggghy, trans. D.F. Pocock, reprinted

with permission of The Free Press, and George Allen 6 Unwin
Ltd., found in a book entitled Emllg_Qchflgim, Thomas Y.
Crowell Company, New York

Einstien, Albert .

1952 - W.
Crown Publishers, New York
Evans, Christopher

1979 Ihg_fljgcg_fliLngigm, Washington Square Press, New York

101

Farnsworth, Paul R.,
1969 The Social Psychology of Music, The Iowa State
University Press, first edition 1958

Figgs, Linda Drake ~
1978 “Qualitative Differences in Trumpet Tones as Perceived
by Listeners and by Acoustical Analysis.“
Unpublished doctoral dissertation, University of Kansas

Fink, Edward L.: Robinson, John P. and Dowden, Sue
1985 “The Structure of Music Preference and Attendance“
Communication Research 12,3: 301-318

Finney, Theodore M.
1935 fl_fligtgcx_gfi_flggjg, United States Armed Forces
Institute, Harcourt, Brace and Company, Madison
Fishman, Joshua A.

1972 Ihg_§ggjglggy_gj_ngggggg, Newberry House Publishers,

Rowley, Massuchusetts

Fox, William S.: Wince, Michael H.
1975 “Musical Taste Cultures and Taste Publics,“
Youth 8 Society, 7,2: 199-224

Frith, Simon
1986 “Art versus technology: the strange case of popular
music.“ Media, Culture and Society 8: 263-279

1987 “The Industrialization of Popular Music.“ an article in

Win. 53-77. ed- James Lull.
Sage Publications, Newbury Park, California, 1987

Gordon, J and Grey, J.M.
1978 “Perceptual Effects of Spectral Modifications on
Orchestral Instrument Tones,“ Computer Music Journal,
3,2: 45-52

Grey, J. M.
1977 “Multidimensional Perceptual Scaling of Musical Timbre.“
Journal of the Acoustical Society of America,
61: 1270-1277

Hair, Harriet I.
1981 “Verbal Identification of Music Concepts.“ Journal of
Research in Music Education 29,1: 11-21

Hall, Donald E.
1980 WWW. Wadsworth Publishino

Company, Belmont, California

102

Hanser, Suzanne B.
1982 “The Effect of Peer Approval and Disapproval on,
Improvement of Pitch Matching and Group Behavior.“
Journal of Research in Music Education 30,4: 221-228
Hargreaves, David J.
1984 “The Effects of Repitition on Liking for Music.“
Journal of Research in Music Education 32,1: 35-47

Hargreaves David J.: MesserSchmidt, Paul and Rubert, Charlotte
1980 “Musical Preference and Evaluation,“ Psychology of Music,
8,1: 13-17

Head, Sydney W., Sterling, Christopher H.

1987 BBQflDQflSIIuE_IN_AMEBlQfli_fl_5ucxsx_9£_Elacicnnic_ncdia.
fifth ed., Houghton Mifflin Company, Boston, Massachussetts

Helmholtz, Hermann L. F.
1862 On the Sensations of Tone as a Physiological Basis
for the Theory of Music, New York, Dover Publications, Inc.,
translated and rendered conformal to the Fourth German Edition
of 1877

Hennion, Antoine and Meadel, Cecile
1986 “Programming music: radio as mediator“ Media, Culture
and Society 8: 281-303

Hesbacher, Peter
1978 “Sound Exposure in-Radio: The Misleading Nature
of the Station Playlist“, Popular Music and Society,
6, 2: 105-117

Hesbacher, Peter: Downing, Robert: Berger, David 6.
1978 “Sound Recording Popularity Charts: A Useful Tool
for Music Research II. Some Recommendations for.Change“
Popular Music and Society, 6,2: 118-131

Howell, Peter: Cross, Ian and West, Robert,

1985 Musical_§icuciuce_and_Qccniiicn. Academic Press.

London

Hybels, Saundra and Ulloth, Dana
1978 Bccadcaa1inc__fln_lniccduciicn_ic_Badic_and_lcicxisiQn.

D. Van Nostranc Company, New York

Kirshenbaum, Jerry
1984 “Mood Music,“ Sports Illustrated, 60: 10

Krumhansl, Carol L.
1985 “Perceiving Tonal Structure in Music,“ American Scientist,
73: 371-378

103

Laing, Dave
1986 “The music industry and the “cultural imperialism’
thesis.“ Media, Culture, and Society 8: 331-341

LeBlanc, Albert
1984 “Selecting a Response Mode in Music Preference
Research.“ reprint from Contributions to Music
Eduction, 11:1-14

LeBlanc, Albert and Sherrill, Carolyn
1986 “Effect of Vocal Vibrato and Performer’s Sex on

Children’s Music Preference.“ Journal of Research in
Music Education 34,4 222-237

LeBlanc, Albert and Cote, Richard
1983 “Effects of Tempo and Performing Medium on Childrens's
Music Preference.“ Journal of Research in Music
Education 31, 1: 57-66

Loy, Gareth
1985 “Musicians Make a Standard: The MIDI Phenomenon,“
Computer Music Journal, 9, 4: 8-26

Lull, James
1985 “On the Communicative Properties of Music."
Communication Research 12,3: 363-372

Lundin, Robert W.,
1953 An Objective Psychology of Music, The Ronald Press
Company, New York

Maher, Timothy F. and Berlyne, D. E.
1982 “Verbal and Exploratory Responses to Melodic Musical
Intervals,“ Psychology of Music, 10, 1: 11-25

McHose, Allen Irvine

1947 IMWWWW.
F.S. Crofts 6 Company, New York

McMullen, Patrick T.
1982 “Connotative Responses to Musical Stimuli: A
Theoretical Explanation.“ Bulletin of the Council
for Research in Music Education Summer:45-55

Merriam, Alan P.
1964 The Anthropology of Music, Northwestern University
Press

Meyer, Leonard 8.,
1956 Emotion and Meaning in Music, The University of
Chicago Press, Chicago 6 London

104

Miller, J.R. and Carterette, E.C.
1975 “Perceptual Space for Musical Structures.“ Journal of
Acoustical Society of America 58: 711-720

Miller, Gerald and Nicholsen, Henry E.,
1976 ngmunjgatign_1gggigy, Addison-Wesley Publishing Co.

Moog, Bob
1986 “MIDI: Musical Instrument Digital Interface“
Journal of the Audio Engineering Society, 34,5: 394-404

Moorer, James A.
1975 “The Synthesis of Complex Audio Spectra by Means of
Discrete Summation Formulae.“ Center for Computer
Research in Music and Acoustics, Department of Music,
Report No. STAN-M-S,

Moorer, James A. and Grey, John
1978 “Lexicon of Analyzed Tones.“ Computer Music Journal
2, 2: 23-31

Mursell, James L.,
1937 The Psychology of Music, W.W.Norton 6 Company, 1937,
New_York

Nielzen, Soren and Cesarec, Zvonimir
1981 “On the Perception of Emotional Meaning in Music“
Psychology of Music, 9, 2: 17-29

Nierman, Glenn E.
1985 “The Development of a Test of Verbal Descriptors in
Music.“ Bulletin for Research in Music Education,
bulletin 84: 21-33

Petersen, Richard A.: Berger, David G.
1975 “Cycles in Symbol Production: The Case of Popular Music“
American Sociological Review, 40: 158-173

Platt, John R. and Racine, Ronald J.
1985 “Effect of frequency, timbre, experience, and feedback
on musical tuning skills.“ Perception and
Psychophysics 38,6: 543-553

Popper, Karl .
1959 The Logic gf Scientific Disggvgcy, Basic Books Inc.,

New York

Rigg, Melvin Gillison
1939 “What Features of a Musical Phrase Have Emotional
Suggestiveness7“, A Bulletin of the Oklahoma Agricultural
Mechanical College, Social Science Research Council of the
Oklahoma A 6 M College, 36, 13

105

Risset, Jean-Claude and Wessel, David L.
1982 “Exploration of Timbre by Analysis and Synthesis.“ The
Psychology of Music, ed Diana Deutsch, Academic Press, 26-54

Rothenbuhler, Eric W. and Dimmick, John W.
1982 “Popular Music: Concentration and Diversity in the
Industry, 1974- 1980. “ Journal of Communication, 32, 1

Rothenbuhler, Eric W.
1987 “Commercial Radio and Popular Music“, in

W. ed- James Lull,
Sage Publications, Newbury Park, California, 1987

Runstein, Robert E. and Huber, David Miles,
1986 Modern Recording Techniques, Howard W. Sams 6 Co.

Russel, P. A. .
1986 “Experimental Aesthetics of Popular Music Recordings:
Pleasingness, Familiarity and Chart Performance.“
Psychology of Music, Society for Research in
Psychology of Music and Music Education. 14:33-43

1987 “Effect of Repitition on the Familiarity and Likability of
Popular Music Recordings,“ Psychology of Music,
15, 187-197

Seashore, Carl E.,
1938 The Psychology of Music, McGraw-Hill Book Company,
Inc., New York

Simon, Paul
1987 Acceptance Speech for BEST ALBUM of 1986 award,
Academy of Recording Artists

Shannon, Claude E., and Weaver, Warren

1949 A Mgthgggtiggl [hggry f9: Qgggggiggtjgg, University of
Illinois, Urbana, IL

Shepard, John: Virden, Phil: Vulliamy, Graham: Wishart, Trevor
1977 MW mm.

Latimer New Dimension Limited, London

Shepard, John
1986 “Music Consumption and cultural self-identities: some

theoretical and methological refelctions“ Media,
Culture and Society 8: 305-330

Tame, David
1984 lb: Segre: nggc gf flggig, Turstone Press, London

106

Taylor, Jack A.
1986 editorial, Forum, Journal of Research in Music
Education,“ 34, 4: 221

Toole, Floyd E.
.1982 “Listening Tests - Turning Opinion into Fact.“
Journal of the Audio Engineering Society 30,6:
482-493

Verveer, E.M., Barry, H., Bousfield, W.A.
1933 “Changes in affectivity with repetition.“ American Journal
of Psychology, 45, 130-134

Waldrop, M. Mitchell
1985 “Why do we like music?“, Science, 3, 36

Wapnick, Joel and Freeman, Peter
1980 “Effects of Dark-Bright Timbral Variation on the
Perception of Flatness and Sharpness“ Journal of
Research in Music Education, 28, 3: 176-184

Wapnick, Joel
1980 “Pitch, Tempo and Timbral Preferences in Recorded
Piano Music.“ Journal of Research in Music Eduction
28, 1: 43-53

Wessel, David L. .
1979 “Timbre Space as a Musical Control Structure,“ Computer Music
Journal, 3,2: 45-52

Winter, James P.
1985 “American Music and Canadian Youth“ Communication
Research 12,3: 345-352

Woram, John M.

1982 magmwunmuam
ELAR Publishing Company, Inc., Plainview, New York

Wooderson, Dawn C. and Small, Ann R.
1981 “Instrument Association Skills: Children in First and
Second Grades“ Journal of Research in Music
Education 29, 1: 39-46

PLEASE NOTE:

The audio cassette is not included
in this material. It is available
for consultation at this author's
graduate school library.

UNIVERSITY MICROFILMS INTERNATIONAL