SPECTROGRAPHIC ANALYSIS OF THE ACOUSTICAL PROPERTIES OF
SELECTED VOWELS IN CHORAL SOUND
APPROVED:
Graduate Committee:
Major Processor
Minor Professor
Commxttee Member
X
(P /a
Committee Member
^1
Dean of the College of Educiiti<E>n
Dean of the Graduate School
SPECTROGRAPHIC ANALYSIS OF THE ACOUSTICAL PROPERTIES OF
SELECTED VOWELS IN CHORAL SOUND
DISSERTATION
Presented to the Graduate Council of the
North Texas State University
in Partial
Fulfillment of the Requirements
For the Degree of
DOCTOR OF EDUCATION
By
William A. Hunt, B.M., M.M.E,
Denton, Texas
June, 1970
TABLE OF CONTENTS
Page
LIST OF TABLES
iv
LIST OF ILLUSTRATIONS
V
Chapter
I.
II.
III.
INTRODUCTION
Statement of the Problem
Purposes of the Study
Background and Significance of the Study
Limitations
Basic Assumptions
Instruments
Procedures for Collecting Data
Procedures for Analyzing Data
SURVEY OF RELATED RESEARCH
PROCEDURES
1
16
•
34
The Sample
The Evaluation
The Analysis
IV.
ANALYSIS OF THE DATA
44
The Evaluation
The Spectrographic Analysis
V.
SUMMARY, FINDINGS, CONCLUSIONS, AND
RECOMMENDATIONS
93
Summary
Findings
Conclusions
Recommendations
APPENDIX
^
109
BIBLIOGRAPHY .
175
ill
l i s t hf t a b l e s
Page
Table
."Judges' Ratings of Vowel Examples,, Giving
;the Number - and Percentage ±n Each
ICategory .
.
45
;Judges1 Ratings off Vowel Examples,, Giving
:the Percentage .of Accuracy and the
Correlations Between the Two Evaluation
-Sequences .
.
.47
III. /Number and Percentage of Errors in Judges'
/Identification of Vowel Sounds . . . . . .
.
48
Summary of Judges' Ratings of Vowel Examples
by Category and Educational Level . . . . .
49
Judges' Ratings of Vowel Examples., Giving
-the Number and Percentage for each Vowel
and Voice Combination in the Junior High
School Sample
......
49
Number and Percentage of Vowel Sounds
Incorrectly Identified in the Junior
High School Sample
52
Judges' Ratings of Vowel Examples, Giving
the Number and Percentage for Each Vowel
and Voice Combination in the Senior High
School Sample . . . . . . . . . . . . . . . .
53
Number and Percentage of Vowel Sounds
Incorrectly Identified in the Senior
High School Sample . .
54
Judges' Ratings of Vowel Examples, Giving
the Number and Percentage for Each Vowel
and Voice Combination in the College
Sample . .
55
IV.
V.
VI,
VII.
VIII.
IX.
IV
LIST OF ILLUSTRATIONS
Figure
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Page
A Spectrogram of the Vowel (a) Sung on the
Pitches f1 'and g1 .
. .
4
A Chart of Average Frequencies of Vowel
Formants
5
A Spectrogram of a Showing That Voice
Pitch Can Be Independent of Vowel
Color
6
A Spectrogram of a-eh Showing That Vowel
Color Can Be Independent of Voice Pitch . .
6
An Illustration Comparing the Notations of
Musical Intervals With the Calibration
Scale of the Sound Spectrogram
9
An Example of the C Major Scale as Sung by
the Male voices Alone
35
An Example of the C Major Scale as Sung by
the Female Voices Alone
35
An Example of the C Major Scale as Sung by
the Male and Female Voices Combined . . . .
36
An Illustration of the Method of Alternating
the Notes of the Scale Between Two Groups
of the Same Educational Level . . . . . . .
37
An Example of the Natural Harmonic Series
Beginning on the Pitch A, 110 Hz. . . .
.
57
Average Frequencies of Vowel Formants
Illustrated in Musical Notation
58
A Spectrogram of the Vowel eh (e) Sung by
College Female Voices on the Pitch A
(Approximately 440 Hz)
59
A Spectrogram of the Vowel eh (e) Sung by
College Female Voices on the Pitch A
(Approximately 440 Hz)
61
Figure
Page
_14. . A -Spectrogram of ±he Vowel ah ((a.}) Sixng
:by Junior High School -Female Voices
:on .the Pitch F (^jprrtxiraatfiily 349 Hz) .
fRank of 2 . .
..........
119
.15. .-AJSpec-trogram of the Vowel aih .(a) Sung by
-Junior High School Female Voices oil
-the Pitch c (Approximately 523 Ha) .
Rank of 2
. . . . .
120
-16. . A ISpectrogram of the Vowel ah (a) Sung
by Senior High School Female Voices
on the Pitch c (Approximately 523 Hz)
Rank of 2 . ..
121
17. -A-Spectrogram of the Vowel ah (-a) Sung by-Senior High School Female Voices on
the Pitch A (Approximately 440 Hz)
Rank of 4.5 .
122
18.
19.
20.
21.
22.
23.
A Spectrogram of the Vowel ee (i) Sung
by College Female Voices on the
Pitch D (Approximately 294 Hz)
Rank of 4.5 . . . . . . . . .
123
A Spectrogram of the Vowel ee (i) Sung
by Junior High School Female Voices
on the Pitch A (Approximately 440 Hz)
Rank of 7 . . . . . . . . . . .
124
A Spectrogram of the Vowel eh (e) Sung
by Junior High School Female Voices
on the Pitch c (Approximately 523 Hz)
Rank of 7
125
A Spectrogram of the Vowel ee (i) Sung
by Senior High School Female Voices
on the Pitch c (Approximately 523 Hz)
Rank of 7
126
A Spectrogram of the Vowel ah (a) Sung
by Junior High School Female Voices
on the Pitch E (Approximately 330 Hz)
Rank of 11
127
A Spectrogram of the Vowel ah (a) Sung
by Junior High School Female Voices
on the Pitch A (Approximately 440 Hz)
Rank of 11
128
VI
Figure
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
Page
A Spectrogram of the Vowel ah (a) Sung
by Junior High School Female Voices
on the Pitch B (Approximately 494 Hz)
Rank of 11
129
A Spectrogram of the Vowel ah (a) Sung
by College Female Voices on the
Pitch c (Approximately 523 Hz)
Rank of 11
130
A Spectrogram of the Vowel ee (i) Sung
by College Male Voices on the
Pitch c (Approximately 261.5 Hz)
Rank of 11
131
A Spectrogram of the Vowel ee (i) Sung
by Senior High School Female Voices
on the Pitch D (Approximately 294 Hz)
Rank of 14.5
132
A Spectrogram of the Vowel ee (i) Sung
by Senior High School Female Voices
on the Pitch F (Approximately 349 Hz)
Rank of 14.5
133
A Spectrogram of the Vowel ee (i) Sung
by Junior High School Female Voices
on the Pitch F (Approximately 349 Hz)
Rank of 18
134
A Spectrogram of the Vowel ee (i) Sung
by Junior High School Female Voices
on the Pitch G (Approximately 392 Hz)
Rank of 18
135
A Spectrogram of the Vowel ee (i) Sung
by Junior High School Female Voices
on the pitch B (Approximately 494 Hz)
Rank of 18
136
A Spectrogram of the Vowel ee (i) Sung
by Junior High School Female Voices
on the Pitch c (Approximately 523 Hz)
Rank of 18
137
A Spectrogram of the Vowel ah (a) sung
by Senior High School Male Voices on
the Pitch D (Approximately 147 Hz)
Rank of 18
138
Vll
Figure
Page
2 34. .A-Spectrogram of the "Vowel all 5(a) Sung
Jby Senior High School Female Voices
:on the Pitch F (Approximately 349 Hz)
TRank of 23
............ . . . . . .
139
135. ;-A -Spectrogram of the "Vowel ee
Sung. .
by College Female Voices on the Pitch
.A (Approximately -4-4D Hz). Hank of 23 . .
140
136. . A Spectrogram of the Vowel ee (1') Sxrng
:by College Female Voices on the Pitch
x: (Approximately 523 Hz). Rank of 23
141
. .
37. -A Spectrogram of the Vowel ah (a) Sung by
,College Male Voices on the Pitch A
(Approximately 220 .Hz.)-. Rank of 23- . . .
"38.
.39.
40.
41.
42.
43.
44.
142.
A Spectrogram of the Vowel ah (a) Sung by
College Male and Female Voices Combined
on the Pitch Coded c (Approximately
261.5 and 523 Hz),. Rank of 23
143
A-Spectrogram of the Vowel eh (c) Sung by
College Female Voices on the Pitch B
(Approximately 494 Hz). Rank of 26 . . .
144
A Spectrogram of the Vowel ee (i) Sung by
College Female Voices on the Pitch G
(Approximately 392 Hz). Rank of 27. . . .
145
A Spectrogram of the Vowel ee (i) Sung by
Junior High School Male Voices on the
Pitch C (Approximately 131 Hz). Rank
of 191
147
A Spectrogram of the Vowel ah (a) Sung by
Senior High School Male Voices on the
Pitch B (Approximately 247 Hz). Rank
of 191
148
A Spectrogram of the Vowel ah (a) Sung by
Senior High School Male Voices on the
Pitch A (Approximately 220 Hz). Rank
of 191 . . . .
149
A Spectrogram of the Vowel ee (i) Sung by
College Male Voices on the Pitch D
(Approximately 147 Hz). Rank of 191 . . . 150
v m
Figure
45.
46.
47.
48.
49.
50.
51.
52.
Page
A Spectrogram of the Vowel ee (i) Sung by
College Male and Female Voices Combined
on the Pitches Coded D (Approximately
147 and 294 Hz). Rank of 191
151
A Spectrogram of the Vowel eh (e) Sung by
Junior High School Male and Female
Voices Combined on the Pitches Coded C
(Approximately 131 and 261.5 Hz). Rank
of 194
152
A Spectrogram of the Vowel eh (e) Sung by
College Male and Female Voices Combined
on the Pitches Coded as C (Approximately
131 and 261.5 Hz). Rank of 195.5
153
A Spectrogram Of the Vowel ah (a) Sung by
College Male and Female Voices Combined
on the Pitches Coded as E (Approximately
165 and 330 Hz). Rank of 195.5
154
A Spectrogram of the Vowel e*2 (i) Sung by
Junior High School Male Voices on the
Pitch B (Approximately 247 Hz). Rank
of 198.5
155
A Spectrogram of the Vowel ah (a) Sung by
Junior High School Male Voices on the
Pitch C (Approximately 131 Hz). Rank
of 198.5
156
A Spectrogram of the Vowel ah (a) Sung by
Junior High School Male Voices on the
Pitch B (Approximately 247 Hz). Rank
of 198.5
157
A Spectrogram of the Vowel eh (e) Sung by
College Male Voices on the Pitch D
(Approximately 147 Hz). Rank of 198.5
. . 158
53.
A Spectrogram of the Vowel ah (a) Sung by
College Male Voices on the Pitch C
(Approximately 131 Hz). Rank of 201. . . . 159
54.
A Spectrogram of the Vowel eh (e) Sung by
Junior High School Male Voices on the
Pitch G (Approximately 196 Hz). Rank
of 204
IX
160
Figure
Page
555. -A -Spectrogram of the Vowel eh le) Sung by
Junior High School Male Voices on the
.Pitch B (Approximately 247 Hzj) . Rank
:af 204 . .
161
556. . A .Spectrogram .of the Vowel e±i (e) Sung by
Junior High School Male and Female
Voices Combined on the Pitches Coded D
(Approximately 147 and 234 Hz) . Rank
of 204 . •>•>•••»•<•».* • . • » • . . .
162
"57. -A Spectrogram of the Vowel eh. (E) Sung by
Senior High School .Male Voices on the
Pitch D (Approximately 147 Hz). Rank
:of 204
163
558.
:59.
60.
"61.
62.
63.
A Spectrogram of the Vowel eh (e) Sung by
College Male and Female Voices Combined
on the Pitches Coded E (Approximately
165 and 330 Hz).. Rank of 204
164
A Spectrogram of the Vowel ee (i) Sung by
College Female Voices on the Pitch C
(Approximately 261.. 5 Hz). Rank of 207 . .
165
A Spectrogram of the Vowel ee (i) Sung by
Junior High School Male and Female
Voices Combined on the Pitches Coded C
(Approximately 131 and 261.5 Hz). Rank
of 209
166
A Spectrogram of the Vowel ah (a) Sung by
Senior High School Male Voices on the
Pitch E (Approximately 165 Hz). Rank
of 209 . .
167
A Spectrogram of the Vowel ah (a) Sung by
College Male and Female Voices Combinedon the Pitches Coded C (Approximately
131 and 261.5 Hz). Rank of 209
168
A Spectrogram of the Vowel ah (a) Sung by
Junior High School Male and Female
Voices Combined on the Pitches Coded A
(Approximately 220 and 440 Hz). Rank
of 211
169
x
Figure
64.
65.
66.
67.
68.
Page
A Spectrogram of the Vowel ee (i) sung by
Senior High School Male Voices on the
Pitch E (Approximately 165 Hz). Rank
of 213
170
A Spectrogram of the Vowel ah (a) Sung by
College Male Voices on the Pitch E
(Approximately 165 Hz). Rank of 213 . . .
171
A Spectrogram of the Vowel eh ( ) Sung by
College Male and Female Voices Combines
on the Pitches Coded F (Approximately
174.5 and 349 Hz). Rank of 213
172
A Spectrogram of the Vowel ee (i) Sung by
College Male Voices on the Pitch C
(Approximately 131 Hz). Rank of 215.5 . .
173
A Spectrogram of the Vowel ee (i) Sung by
College Male Voices on the Pitch E
(Approximately 165 Hz). Rank of 215.5 . .
174
XI
E H a F H » I.
INTRODUCTION
"Vocal blend is a primary concern for teachers of choral
music :at :all educational IBTTBIS. Much has been theorized
..about /good choral blend and how it may best be obtained.
.These theories have been largely subjective, depending for
effectiveness upon the aural perception of the individual.
Vocal :blend can be described as the sound produced by the
mixing xcf the various characteristics of voices.
An aurally
pleasing sound is said to have good blend, an unpleasant
sound to have poor blend-
Little empirical evidence has
been utilized in support of any of the various theories of
choral blend.
There is a need for the gathering and codifying of
empirical data which can be used as a scientific basis for
the formulation and evaluation of theories and techniques
of achieving good choral blend.
Such information, by
separating the factual from the supposed, can help disperse
the mysticism which has existed among teachers of singing
and lead to more thorough understanding and clearly defined
processes for the teaching of choral singing.
Statement of the Problem
The problem of this study was a spectrographic analysis
of the acoustical properties of selected vowels in choral
sound.
Purposes of the Study
The purposes of this study were (1) to categorize
examples of vowel sounds by means of subjective evaluation,
(2) to ascertain by spectrographic analysis the distinguishing characteristics of the acoustical properties of the
examples in the categories, (3) to determine the similarities
and dissimilarities which exist within and between the
categories, and (4) to analyze the implications of the
findings for the teaching of choral singing.
Background and Significance of the Study
"The problem of blend is the mixing of all types of
voices, colors, and intensities into one common tone quality
that is the best possible for that particular group" (4,
p. 147).
The sound of the voice when used in singing has
duration, intonation, loudness, vibrato, quality of tone o.r
timbre, and vowel sound.
When voices are combined into a
choral group the resulting quality of sound depends upon how
well the characteristics of the voices blend and/or unify.
In choral singing, unity should be sought in duration,
intonation, and loudness.
The vibrato should be used for
expression and should be governed by intensity, vocal timbre,
and musical style.. Vocal timbre should be regarded as an
-individual property and sh-oul-d not be modified for the sake
:df ^achieving choral IhlencL
The vowel sound, which is common -
-to rail voices, is variable and can be used as the.vehicle
ffcrr Tthe mixing of individual timbres into a blended sound.
"There is almost unanimous agreement that one of the most
important factors, if .not -the most important factor in the
achievement of choral blend is unity of vowel" (22, p. 40).
^Research studies using the sound spectrograph to
^analyze individual voices have indicated that vowel sounds
.are measurable concentrations of energy at certain frequency
levels within a tonal spectrum.
The strongest of these con-
centrations, the formants, determine the vowel sound which
is perceived.
Howie and Delattre explain the formant theory in the
^following manner:
Briefly, the theory of formants is as follows: The cavities of the vocal tract possess
natural notes of resonance which "pass" or reinforce certain harmonics of the vocal chord tone.
These emphasized harmonics form concentrations
of acoustic energy at frequency regions on the
spectrum corresponding to the natural notes of
resonance of the cavities. The term formant
refers to the selective resonance in a particular
frequency which characterize the timbre, or
color, of vowels, while the higher formants
contribute mostly to the timbre, or quality,
of the individual voice (6, p. 6).
The pitch of the formants can be calibrated in cyclespex-second, or Hertz, on the sound spectrogram.
The intensity
of the formants can be measured in decibel units by making
an amplitude section on the spectrogram.
This is illus-
trated in a study by Taff:
fps-
. >X5
n
C1'^;
n
ft
2XO
v:oo
I5i
Fig. 1 — A spectrogram of the vowel (a) sung on the
pitches f' and g', with sections made from the low and high
points of the vibrato wave. The baseline and the first two
formants are indicated (19, p. 8).
Each vowel is characterized by two well-defined
formants.
When a specific vowel is spoken or sung, these
two formants should be in a specific relationship.
These
relationships are shown in the following chart made from the
finding of a study by Peterson and Barney (11, p. 182).
The symbols identifying the vowel sounds in the chart
are from the International Phonetic Alphabet.
These symbols,
along with English equivalent sounds, were used for the
identification of vowels in this study.
2 SCO
—• — -i-4~ ~~~rr
£
Z~tl
%OOG>
IS'oo
;..l.
":n:~r:r
; • • ! /Of®
foot
i K
r-M-f-
Ufa
l&iO
t—I—.r-r — !^S*J©-' 4 6 O b m b b b . " ~ , f e ^ o
— 7 3 o _
-••."** ¥?o
STOO X7o
:.i. ,
,.:Srfro—um 3o©
a
/
/
-£':-;:as-1 ^
^
TV
iu
A
iiee.i L/t bet boZ Bteh Jess hocK boat itit Bett
Fig. 2 — A chart of average frequencies of vowel formants
When there is a modification of the vowel, the formant
frequencies should also modify, or deviate, in a predictable
direction.
Examination of numerous sound spectrograms (13)
indicates that this modification is smooth and continuous
and independent of the pitch being sung by the voice.
following information is from a study by Delattre.
Obviously formant frequency is independent
from the fundamental frequency, subjectively
called voice pitch and whose variations make
"intonation" in speaking and "melody" in singing. For example, if the utterance of an "a"
vowel is sustained while the pitch is made to
fall, the spectrum will show that all the
formants of the "a" keep constant frequencies:
The
Fig. 3 — A spectrogram of a showing that voice pitch can
be independent of vowel color: The voice melody frequency
falls by about one octave (about 240 cps to 120 cps), but
the formant frequencies of a (about 700, 1200, and 2400 cps)
remain unchanged throughout.
Conversely, if the pitch is sustained while
vowel color is changed—let us say, from "a"
to "eh"—the first formant falls by some 200
cps and the second rises by some 600 cps while
all the harmonics keep a constant frequency.
zm
isoo
<rY 3
a
a
i
£
1200
<-n
Fig. 4 — A spectrogram of a-eh showing that vowel color
can be independent of voice pitch: The frequencies of
formants one and two change from about 700 cps and 1200 cps
for a to 500 cps and 1800 cps for eh but the"voice melody
frequently remains unchanged at about 110 cps throughout.
Changes in formant frequency are due only to
changes in the shape of the vocal tract cavity
or cavities; changes in pitch frequency to
stretching of the vocal chords. If the two
physiological events are independent, so are
the acoustic results of each event—formant
frequency changes and overtone frequency
changes (3, pp. 4-5).
These studies have been concerned with the recording
and analyzing of the sound of a single voice.
The concern
of this study was to determine in part what happens to vowel
formants when a number of voices are combined as in a choral
group.
This study attempted to analyze and define the dis-
tinguishing characteristics of the acoustical properties of
the vowel formants in choral sound, and to determine the
effect of the combined deviations of the formants upon the
blend of choral sound.
In utilizing the empirical evidence
of the spectrogram and the subjective judgment of a jury of
experts, this study came to bear upon the theory that "the
most important factor in the achievement of choral blend is
unity of vowel" (22, p. 40).
Limitations
This study was limited to the analysis and evaluation
of 216 tape-recorded examples of choral sound.
An equal
number of examples, seventy-two each, were obtained from
junior high school choirs, senior high school choirs, and
college choirs, There is no reason to suppose that the
analysis of blend of choral sound would have differed in
significant ways should a larger number of examples have
been used.
Basic Assumptions
It was assumed that the members of the choral groups
used as examples sang during the recording session in a
manner not different from the manner in which they would
sing in rehearsal or performance.
It was also assumed that
the members of the jury were competent and that they responded
honestly in the evaluation of the examples.
It was further
assumed that the sound spectrograph used in the study was an
adequate instrument for analyzing choral sound.
Instruments
The sound spectrograph is a device developed at the Bell
Telephone Laboratories for use in teaching the deaf to speak
with normal sound (8, 9, 12, 13, 14, 17).
duces a graphic record of vocal sound.
This device pro-
The visual comparison
of spectrograms has been used to teach the deaf to speak
with specific speech patterns.
The machine has been used
frequently to analyze the solo singing voice and has been
accepted as an adequate instrument for use in comparing
visual analysis with aural evaluation of sound (1, 2, 5, 7,
10, .15, 16, 18, 19, 20, 21) .
As the study progressed, a limitation of the instrument
was discovered.
For the'purpose of musical analysis, the
scale of the spectrogram is extremely distorted.
In the
present spectrogram, all musical intervals from the low
threshold of hearing to approximately the C above Middle C
are compressed into the lowest calibration band.
Figure 5.)
(See
In musical notation this distance is approximately
tooo
SToo
£To& &
i
tOO 0
as-#©
1000
area
f ooo
Fig. 5—An illustration comparing the notation of
musical intervals with the calibration scale of the sound
spectrogram.
five octaves.
The second calibration band encompasses
approximately one octave.
The third calibration band en-
compasses an interval of a fifth, the fourth band encompasses
an interval of a fourth, the fifth band a major third, the
sixth band a minor third, etc., continuing in the same
diminishing proportions as the natural harmonic series.
For
10
purpose of analysis of musical sound, it would seem more
appropriate if the scale were calibrated to be more closely
aligned with the acoustical intervals of music.
The instrument the jury used is included in Appendix A.
The instrument instructs the jury to evaluate the examples
on a three-point scale:
I, Good; II, Acceptable; III, Poor.
The jury was also asked to indicate the particular vowel
sound which was perceived.
Procedures for Collecting Data
Choral groups used as sources for examples were selected
with the approval of the dissertation advisory committee
from successful school programs.
Examples were obtained
from junior high schools for examples of voices in the
changing stage, from senior high school for examples of
voices in the formative stage, and from college for examples
of voices in the mature state.
Each of the choral groups
was between forty and fifty members -in size.
To obtain the examples of sound, each of the choral
groups sang, unaccompanied, the C major scale.
The scale
was sung on each of .the following vowel sounds:
ee (i) ,
as in beet; eh (e), as in bet; and ah (a) , as in Bach.
scale was sung in the following combinations:
The
(1) male
voices alone, (2) female voices alone, and (3) male and
female voices together.
Each of the choral groups sang eight
examples of each vowel in three combinations, giving a total
11
of 216 examples to be analyzed and evaluated.
There were
seventy-two ^examples of each vowel soimaL'
.The ^tape-recorded examples selected -were cut into
fifteen-iinch llengths and attached to sixty—inch timing-tapes.
The examples'were identified, .numbered and cataloged.
To
eliminate -the ±>ias of identification in the sequence of
analysis ;-,arid revaluation, a table of random numbers was
utilized "to /determine the order in -which the examples- were ..
presented. The examples were presented in one continuous
tape.
The .examples were subjected tD evaluation by a jury of
seven experts on the variables of quality of blend and
intelligibility.
Each judge was asked to rate the quality
of the.choral blend and to identify the vowel sound which he
heard.
The jury consisted of seven choral directors selected
with the advice of a faculty committee from the School of
Music at North Texas State University.
The jury was
assembled in the Recital Hall in the School of Music at
North Texas State University to perform the evaluation.
Each member of the jury made his evaluation independently and
without consultation.
According to his own criteria, each
member of the jury rated each example either "Good,"
"Acceptable," or "Poor."
To check the reliability of the
evaluation the entire sample was played and evaluated twice
and the percentage of agreement was calculated.
12
Sound spectrograms for the analysis were made on the
Kay Electric Sound Spectrograph.
The examples were trans-
ferred from the tape play-back equipment to the sound spectrograph by direct wire connection.
Spectrograms and ampli-
tude sections were made of each of the examples.
Procedures for Analyzing Data
The data obtained from the jury of experts were entered
into tables.
The tables indicate the number, and percentage,
of ratings in each category for each judge; ther percentage
of accuracy in identifying the vowel sound intended; and the
percentage with which the judge agreed with himself in the
two evaluations.
The tables also indicate the ratings of
examples from each of the instructional levels and the
percentage of accuracy between the vowel sound intended and
the vowel sound perceived.
The spectrograms were placed into categories according
to the data obtained from the jury of experts.
The spectro-
grams were examined in detail to determine the distinguishing
characteristics of the acoustical properties of each example.
The distinguishing characteristics were described and defined
to determine what similarities and dissimilarities existed
within and between the categories.
CHAPTER KEBLIOGHaMSS
1.
Appleman, Dudley Ralph, "A Study by Means of Planigraphy
-Radiograph and Spectrograph of the Physical
JChanges Which Occur During the Transition from
:the Middle to the Upper Register in Vocal Tones,"
unpublished doctoral dissertation, Indiana University, Bloomington, Indiana,, 1953.
2.
Arment, HolJLace Elbert, "A Study by Means of Spectrographic Analysis of the Brightness and Darkness
Qualities of Vowel Tones in Women's Voices," unpublished doctoral dissertation, Indiana University,
Bloomington, Indiana, 1960,
3.
Delattre, Pierre, "Vowel Color and Voice Quality," The
Bulletin of National Association of Teachers of
Singing, XV (October, 1958)., 4-7.
4.
Helvey, Kenneth W., "A Study of the Methods of Choral
Tone Production of Selected Choral Directors," unpublished master's thesis, University of Southern
-California, Los Angeles, California, 1952.
5.
Hohn, Robert W., "A Study of the Relationship Between
Vowel Modification and Changes in Pitch in the
Male Singing Voice: A Spectrographic Analysis,
Reinforced by the Judgment of Phonetics Experts,"
unpublished doctoral dissertation, Indiana University, Bloomington, Indiana, 1959.
6.
Howie, John and Pierre Delattre, "An Experimental Study
of the Effect of Pitch'on the Intelligibility of
Vowels," The Bulletin of National Association of
Teachers of Singing, XVIlf"1'May, 1962), 6-97"
7.
Johnson, Merion.James, "An Investigation of the Effect
of Training in the Articulation of Vowels by the
Speaking Voice Upon the Articulation of Vowels by
the Singing Voice," unpublished doctoral dissertation, Indiana University, Bloomington, Indiana,
1966.
8.
Koenig, W., H. K. Dunn, and L. H. Lacy, "Sound Spectrograph," The Journal of the Acoustical Society of
America, XVIII (July, 1946), 19-49.
13
14
9.
Kopp, G. A. and H. C. Green, "Basic Phonetic Principles
of Visible Speech," The Journal of the Acoustical
Society of America, XVIII (July, 1946), 74-89.
10.
Perkins, William H., Granville Sawyer, and Peggy
Harrison, "Research on Vocal Efficiency," The
Bulletin of National Association of Teachers of
Singing, XV (December^ 1958), 5-7.
11.
Peterson, Gordon E. and Harold Barney, "Control Methods
Used in a Study of the Vowels," The Journal of the
Acoustical Society of America, XXIV (March, 1952),
175-184.
12.
Potter, Ralph K., "Introduction to Technical Discussion
of Sound Portrayal," The Journal of the Acoustical
Society of America, XVIII (July, 1946), 1-3.
13.
, George A. Kopp, and Harriett Green
Kopp, Visible Speech, New York, Dover Publications,
Inc., 1966.
14.
Riez, R. R. and D. Scott, "Visible Speech Cathode-Ray
Translation," The Journal of the Acoustical
Society of America, XVIII (July, 1946), 50-61.
15.
Scott, Anthony, "A Study of the Components of the
Singing Tone Utilizing the Audio Spectrum Analyzer,"
The Bulletin of National Association of Teachers
of Singing, XXIV (May, 1968), 40-41.
16.
Scott, David Watters, "A Study of the Effect of Changes
in Vocal Intensity upon the Harmonic Structure of
Selected Singing Tones Produced by Female Singers,"
unpublished doctoral dissertation, Indiana University, Bloomington, Indiana, 1960.
17.
Steinberg, J. C. and N. R. French, "Portrayal of Visible
Speech," The Journal of the Acoustical Society of
America, XVIII (July, 1946), 4-19.
'
18.
Sullivan, Ernest G., "An Experimental Study of the Relationships Between Physical Characteristics and
Subjective Evaluation of Male Voice Quality in
Singing," unpublished doctoral dissertation,
Indiana University, Bloomington, Indiana, 1956.'
19.
Taff, Merle E., "An Acoustical Study of Vowel Modification and Register Transition in the Male Singing
Voice," The Bulletin of National Association of
Teachers of Singing, XXII (December, 1965), 8-11,
35.
15
20.
Triplett, William M. , "An Investigation Concerning Vowel
Sounds xm High Pitches," The Bulletin of National
Asso:cia±ion of Teachers of Singing, XXIII
(February, 1966), 6, 8, 5D-
21.
van denlBexg, Janwillem and William Vennard, "Toward an
Objective Vocabulary for Voice Pedagogy," The
Bulletin of National Association of Teachers of
Singing, XV iFebruary, 1353),,, 10-15, 24.
22.
Wyatt, liaxry W. , "Blend in Choral Sound," unpublished
master "KS t.hesis, School of Music, North Texas
State .University, Denton-, Texas, 1967.
CHAPTER II
SURVEY OF RELATED RESEARCH
The sound spectrograph, which was developed initially
for speech analysis, has been increasingly used in the field
of vocal music.
Discoveries about speech sound have had a
great influence upon the study and teaching of singing.
The
sound spectrograph can show in an objective way the acoustical properties of the voice as it is used in singing.
Previously, teachers of singing have had to depend entirely
upon subjective evaluation to determine the quality of sound.
In essence a sound spectrograph is a machine which
graphically represents a complex sound wave in its component
parts.
These component parts can be measured in terms of
duration, pitch, and intensity.
Several steps are involved
in the making of a sound spectrograin.
The first step in the process is to record on the
internal turntable the sound which is to be analyzed.
The
recorded sound is then replayed continuously until the
graphic representation is complete.
The second step involves
the playing back of this sound through a variable filter.
The current from this filter activates a stylus which records
on paper the intensity and duration of frequency.
The paper
is on a drum which reveolves at the same speed as the tape
loop.
16
17
The filter is set first at a low frequency,
# 50
cps, or Mz, aand :is -then progressively raised until all frequencies rpreaent in the complex wave axe represented on the
graph.
This procedure is accomplished automatically on the
current models :oif the machine,. A narrow band filter, 50 Hz,
gives more -detail on the graph, while a wide band filter,
300 Hz, :teiids rto .give more concentration and definition to
the resonance isaxs on the graph..
The turntable has a dura-
tion of 2.-4 ^seconds and the time required for making a sound
spectrograph _is about thiree minutesEach :sourid, vowel or consonant.., lias a distinctive conformation .on the spectrogram.
Vowels are distinguished by
horizontal baxs of varying height and intensity.
Of the
bars, or formants, the second is the most variable in the
case of vowel sounds.
In studies by Potter, Kopp, and
Kopp (13) the second formant is referred to as the hub and
~
~
/•
- k
is considered the most significant part of the vowel change.
The first formant remains at a fairly constant frequency
range while the second formant, the hub, has a fairly wide
frequency range.
The second formant can be as high as 2290
Hz for the ee (i) vowel and as low as 840 Hz for the awh (O)
vowel (11, p. 182).
Consonants are also recognizable on the spectrogram.
Un-voiced consonants appear as "white noise"; voiced consonants have some of the properties of vowels; and other
consonants are combinations of white noise and vowel characteristics.
18
In the Journal of the Acoustical Society of America in
1946, a group of articles appeared:
"Introduction to
Technical Discussions of Sound Portrayal," Ralph K. Potter
(14); "Portrayal of Visible Speech," J. C. Steinberg and
N. R. French (16); "Sound Spectrograph," W. Koenig, H. K.
Dunn, and L. H. Lacy (8); "Visible Speech Cathode-Ray
Translator," R.. R. Riez and L. Scott (10); "Visible Speech
Translators with External Phosphors," Homer Dudley and
Otto O. Gruenz, Jr. (4); and "Basic Phonetic Principles of
Visible Speech," G. A. KOpp and H. C. Green (9).
These
articles were the result of experiments which had been conducted in association with the Bell Telephone Laboratories,
Incorporated.
These experiments, begun in 1941, were instigated for
the purpose of helping persons with hearing and speech problems.
A language of "visible sound" was formulated from
the experiments.
Persons who are unable to hear can learn
to speak with specific speech patterns by comparing their
own "visible sound" with a model.
The information in these articles and other related information went into the making of a book, Visible Speech,
by Potter, Kopp, and Green (13).
The authors present a very
complete analysis of all vocal sound as it appears on the
spectrogram.
A complete alphabet of sound is given in
standard English and in regional dialects.
The book presents
the sounds combined into words and short sentences.
A
19
perusal of vthis book indicates that it could be of great.
value -to ^singing -teachers to give a better understanding of
vocal sarnxd.
A:good 'discussion of methods for conducting an experiment in-the .evaluation .of vowel sounds is "Control Methods
Used in :.a "Study of Vowels" by Gordon E„ Peterson and Harold E.
Barney til).. 3?his article gives a detailed description of
the steps .and pxocedures of a study utilizing spectrographic
analysis :and-subjective judgment..
A total of seventy-six
speakers recorded a randomized list of the following words:
heed, Jiid, head, had, hod, hawed., .hood., who'd, hurd, and
heard. -Each -speaker recorded two listsT resulting in a
total of 1/520 recorded words.
These words were randomized
and a group :of seventy listeners, some of whom were also
speakers, were asked to indicate on paper each word they
heard as the edited tapes were played back.
Comparisons were drawn between what the speaker intended
to say, what the listener thought the speaker said, and what
the spectrograms indicated that the speaker had said.
This
was a carefully constructed experiment which produced valid
results.
A number of articles on the use of the sound spectrograph in the analysis of the voice have also been published
in the Bulletin of the National Association of Teachers of
Singing.
In an article on the subject of vowel color and
voice quality, Pierre Delattre (3) demonstrates with the aid .
20
of the spectrogram that voice pitch and vowel formants are
independent of each other.
When a particular vowel is
maintained while the voice changes pitch, the vowel formants
remain constant.
Conversely, when the voice pitch is held
constant while changing the vowel sound, the formants change
but the overtones of the voice remain unchanged.
Delattre
states that "the color of the vowels—what causes one vowel
to be perceived as linguistically the same or different from
another—is mainly characterized by formant frequency" (3,
p. 5).
Delattre goes on to make a distinction between
formants which distinguish vowel color and those which distinguish voice quality.
It is reported in this article that
the first two formants determine the vowel color and the
"voice quality in singing is mainly characterized by the
two or three formants whose frequencies are just above the
vowel formants" (3, p. 5).
Also presented in this article were X-rays which were
previously published in a study by Raoul Husson.
These
X-rays show different mouth formations used in speaking and
singing the same sounds.
The X-rays indicate that resonance
factors in the vocal.cavity are more efficient in singing
than in speaking.
An article concerning research on vocal efficiency by
Perkins, Sawyer, and Harrison (10) was a report of a study
conducted by speech therapists in an attempt to objectively
define "good quality" in the vocal sound.
21
-In .this study, thirty-two males with low-pitched voices
wex€ sselexrted as subjects,. They were to sing three test
vowels, tee, ;ah, oo) , in -both an efficient and an inefficient
manner. .X-ray photographs and tape-recordings were made
simultaneously of each .sound. The tapes were judged by a
panel :and rated as efficient or inefficient sounds.
The
X-rays were measured to determine physical differences made
between drhe two types of sounds.
The tapes ivere also
analyzed :on the sound spectrogram and line spectrum to
determine -the acoustical differences between the two types of
sound. The line spectrum produces a picture of the sound
at a -given -instant and shows amplitude in a linear graph.
The -spectrogram shows amplitude in dark or light lines.
The
two -types :of graphs indicate that efficient sounds are rich
in harmonics and inefficient sounds are almost devoid of
harmonics.
"Toward an Objective Vocabulary for Voice Pedagogy" by
van den JBerg and Vennard (20) was a study conducted to objectively measure with X-ray and spectrogram various vocal
faults which have been subjectively described by singing
teachers.
.
Common vocal problems in breathing, attack, regis-
tration, resonance, and articulation were examined.
The
faults were divided as Hypofunction (under-doing) and Hyperfunction (over-doing).
Subjective vocabulary connected with
these faults was given:
breathiness, yelling tone, shallow
tone, white tone, throaty tone, etc.
22
Each of the vocal faults was demonstrated for the X-ray
and spectrogram.
The study describes the physical and acous-
tical characteristics of each of the subjectively described
vocal problems.
The authors state that the study was an
attempt to formulate a tentative vocabulary with objective
definitions.
The authors suggest that other studies could
eventually refine the vocabulary to a point of usefulness to
the profession.
In 1962, Howie and Delattre collaborated on a study of
the effect of pitch on the intelligibility of sounds (6).
This study was an attempt to discover why intelligibility
varies in certain ways with changes of pitch.
The study was
based on the measurement of the formants in spectrographic
analysis.
It has been established that the first two formants
determine the vowel color and that the upper formants have to
do with voice color.
Formants for certain vowels have a
relatively fixed frequency resonance (11, p. 183).
It has
been established that the change of formant range changes
the vowel.
It was determined in the Howie and Delattre
study that if the pitch sung by a subject is higher than the
first formant of the vowel, the color of the vowel is
changed.
The fundamental frequency of the pitch which is
sung would, in such a case, become the first formant.
The
second formant of the vowel would thus be closer to the
fundamental, and the vowel color would be altered.
23
The vowel sound which lost intelligibility most quickly
as the voice :pitch was raised was the vowel ie li).
The
vowel which ..retained intelligibility at "the highest pitches
was the vowel ?ah (:a).. As pitch rises higher all vowels come
to resemble ?ah fa),, rand from high C (1,050 Hz.) up all vowels
were identified ias ?a:h (a) , regardless of the vowel that was
intended by .the. singer..
The conclusions x>± the study were that vowels generally
lose intelligibility as pitch rises, and that recognizable
vowels cannot :be produced when the fundamental frequency of
the pitch sung :±s :appreciably higher than the first formant
of the particular vowel.
A follow-xip "of the Howie-Delattre study (6) was an
investigation concerning vowel sounds on high pitches conducted by Triplett (19).
In this study, two female singers recorded five vowels
on
(middle C) , E^,
, C,-, E^,
, and Cg (high C).
Spectrograms and amplitude sections were made of all vowels
sung.on Cg (High C).
The recordings and spectrograms were
taken to Delattre for examination.
Listening observations
were made with the following results:
when the observer
heard the initial attack of the sound he could identify the
intended vowel; when the recording was begun just after the
attack, the observer identified all sounds as ah (a).
spectrograms showed the following changes:
the initial
The
24
attack showed formants resembling the intended vowel but
quickly changed to formants resembling the vowel ah (a).
The conclusions of the study were that for more intelligible sounds on high pitches the singer should make the
attack of the tone be as nearly as possible the intended
vowelf after which the singer may relax into the ah (a)
sound without losing intelligibility.
Merle E. Taff conducted a study of the acoustics of
vowel modification and register transition in the male voice
(18).
In this study five baritones and five tenors sang
ascending octave scales in two chromatically adjacent keys.
This gave sound examples of every semitone of the octave
range.
The tapes were played for observers who were asked
to judge where the register transition took place.
The
tapes were arranged in random order and each one was played
twice in succession.
The tapes were analyzed by spectro-
graph and the decisions of the observers were tabulated and
compared to the findings of the spectrographic analysis.
The results indicate that all singers in the study made
significant changes in the formant frequencies when singing
from the lower to the higher register.
The formant fre-
quency changes varied with the vowel and with the person.
This study used charts and scattergrams to illustrate the
types and locations of the changes in vowels that were made.
The average transition point for tenors was approximately a
minor third higher than for baritones.
Closed vowels (ee
25
and oo) are modified at lower pitches than open vowels (ehf
ah, and oh).
A study by Vennard and Irwin was concerned with speech
and song as compared in sound spectrograms.
In this study
one phrase of words was both spoken and sung in rhythm.
The
recordings of the phrase were transferred to the spectrogram.
The spectrograms were measured in several ways to determine
the relationships of the two examples.
The authors also
quoted considerable material concerned with the use of the
spectrogram and vowel formant study.
The following is their
summary:
To summarize: Song is at least 10 decibels
louder than normal speech in this case. It prolongs both vowels and consonants but especially
vowels far beyond the duration of speech. Its
sound energy is concentrated in fewer and more
powerful formants. . . . The slogan, "Sing as
you speak," or at least "Sing as you should
speak" holds good in most vowels and almost all
consonants (21, p. 23).
A doctoral dissertation by Appleman was entitled "A
Study by Means of Planigraph, Radiograph and Spectrograph
of the Physical Changes Which Occur During the Transition
from the Middle to the Upper Register in Vocal Tones" (1) ..
The purpose of this study was to determine what physical
changes occur when a singer makes a transition from an open
to a closed, or covered, vocal tone.
X-ray and radiograms
were made simultaneously with tape-recordings of six subjects,
The subjects sang predetermined vocal patterns on each of
26
three vowels.
The tape-recordings were analyzed by sound
spec t rqgr-aph.
JT.be coral ocavities, as s.hnwn ±n the Z-xays and radiograms , werce rmeasured and a comparison -was made between the
open arid xJtosed positions.
The findings were that the oral
cavity -is :e3qpanded during the physical adjustment called
"covering."
A dissertation study by Wooldridge (23) was concerned
with the nasal resonance factor in the sustained vowel tone
in the singing voice.
This study was an investigation of
resonation ;.in the nasal cavities and the relative effect on
the oral vowel J.n singing.
A sample of singers was recorded singing the five
primary vowel sounds.
Each vowel was sung once with the
nasal passages open and once with the nasal passages closed.
The recorded examples were analyzed by sound spectrograph to
determine differences.
It was concluded that there was no
significant difference between the pairs of vowels and that
the term "nasal resonance" is without validity in describing
voice quality in the singing voice.
In 1956, a doctoral dissertation study was conducted by
Sullivan (17) on the subject of the relationships between
physical characteristics and subjective evaluation of male
voice quality in singing.
sang three vowels:
and F (34 7 Hz).
In this study eighteen subjects
ah (a), ee (i) , and oo (u), on A# (234 Hz)
Recordings were made and analyzed by sound
27
spectrograph on twenty-two variables of pitch, intensity, and
location of partials and formants.
The recordings were
played for a jury which made subjective evaluations of the
voice qualities.
product-moment.
The data were correlated using the Pearson
The results indicate that the strongest
formant above the second vowel formant had the greatest
effect on voice quality.
In general, as the intensity of
this formant increased-, the tone was ranked higher in quality.
"A Study of the Relationship Between Vowel Modification
and Changes in Pitch in the Male Singing Voice:
A Spectro-
graphs Analysis, Reinforced by the Judgements of Phonetics
Experts" was a dissertation study by Robert W. Hohn (5).
This study was concerned with finding out what changes might
occur in formant patterns at different pitch levels as
singers intend to produce the same vowel.
Excerpts were
taken from available LP recordings and analyzed by sound
spectrograph.
A jury of trained phoneticians were asked to
judge what vowel was sung in each example.
The findings of
the study were that all singers showed significant changes
in formant frequencies on all vowels at the three pitch
levels studied.
Formant frequencies for the lowest pitches
were nearest to the frequencies of the spoken vowel.
The
jury showed greater agreement on low pitches than on high
pitches.
David W. Scott conducted a dissertation study of the
effect of changes in vocal intensity upon the harmonic
28
structure of selected singing tones produced by female
singers. 3?his .study was an effort to determine the effect
of intensity ±n the production of the tone upon the acoustical characteristics of the tone as shown on the sound
spectrogram.
Ten subjects sang three vowel sounds:
ah (a)., f-and oo (u) , at three intensity levels:
ee (i),
+13 db, +5 db,
and —"3 :"db, :on the pitch C (523 Hz).
The recorded sounds
were analyzed by sound spectrograph-
The findings were
that.as ±he tone became less intense, all the partials
except ±he zfLirst decreased in strength, with the higher
partiais ^disappearing first.
The first partial remained
constant at. all intensity levels.
A dissertation study by Arment (2) utilized the sound
spectrograph to investigate the brightness and darkness
qualities .of vowel tones in women's voices.
Five subjects,
sopranos, recorded 240 tones of six vowels.
Each vowel was
sung "as bright as possible," "as dark as possible," and
without distortion.
The recordings were played for a jury
and were judged as "very bright," "very dark," or "neither
bright nor dark."
The findings indicated that brightness,
or lack of brightness, can be shown on the spectrogram and
can be detected by the ear.
Brightness depends upon the
presence of strong high harmonics and symmetrical distribution of narrow formant bands.
Darkness of tone shows a
general lack of high harmonics and relatively broad formant
bands.
29
James R. Whitworth conducted the following study: ."A
Cinefluorographic Investigation of the Supralaryngeal
Adjustments in the Male Voice Accompanying the Singing of
Low Tones and High Tones" (22).
This study involved the
making of X-ray moving ..pictures of ten subjects as they
caused their voices to glide from low to high pitches on
each of three vowel sounds:
ah (a), oh (o), and ee (i).
The X-ray films were analyzed and measured to determine what
changes took place within the oral cavity as the pitch
changes took place.
"An Investigation of the Effect of Training in the
Articulation of Vowels by the Speaking Voice Upon the
Articulation of Vowels by the Singing Voice" was a dissertation study conducted by Johnson (7) . This study was conducted as an experiment, using two groups from a university
choir.
The groups were assigned by a random method and
equated by voice-type, aural acuity, pitch discrimination,
and vowel perception.
The experiment was concerned with the
pedagogical concept "sing as you speak."
The experimental
group was given six weeks of training in the spoken articulation of the cardinal vowels.
Recordings were made of each
of the subjects, both before and after the period of experiment.
Sound spectrograms were made of the recorded examples
and were analyzed to determine the improvement made by the
subjects.
The result showed that the experimental groups
30
improved in spoken articulation but showed little improvement in singing^articulation.
A comprehensive study of the aspects of choral blend is
a master1 s .thesis ;by Wyatt (24).
This study ^was based upon
a review of lliter.at.ure of the acoustics of music, a review
of writings-relating to the achievement of blend in choral
sound, and .a .report :of the results of ja. questionnaire which
was sent to choral music authorities throughout the nation.
The value ;a"f .this study was in the extensive compilation of the - opinions and experiences of many of the contemporary choral music authorities-.
Wyatt presents in a
clear manner what choral authorities consider desirable and
important in choral blend, and the authorities' recommendations concerning :the achievement of blend in choral sound.
The aspects :of choral blend which were discussed were
rhythmic unity, vowels, quality of tone, vibrato, intonation,
selection and classification of voices, seating and standing
arrangements, balance, and room acoustics.
The information,
which was collected from books, articles, unpublished theses,
and the questionnaire, was primarily subjective in nature.
In view of the studies discussed above, it is apparently
a logical step to determine what application the sound
spectrograph has to the analysis of choral sound.
The aim
of this study is to use the sound spectrograph to analyze
selected vowels in choral sound in order to determine
acoustical differences between "Good" and "Poor" choral
blend.
CHAPTER BIBLIOGRAPHY
1.
Appleman, Dudley Ralph, "A Study by Means of Planigraph
Radiograph and Spectrograph of the Physical
Changes Which Occur During the Transition from the
Middle to the Upper Register in Vocal Tones," unpublished doctoral dissertation, Indiana University,
Bloomington, Indiana, 19 53.
2.
Arment, Hollace Elbert, "A Study by Means o.f Spectrographs Analysis of the Brightness and Darkness
Qualities of Vowel Tones in Women's Voices," unpublished doctoral dissertation, Indiana University, Bloomington, Indiana, 1960.
3.
Delattre, Pierre, "Vowel Color and Voice Quality," The
Bulletin of National Association of Teachers of
Singing, XV (October, 1958), 4-7.
'
4.
Dudley, Homer and Otto 0. Gruenz, Jr., "Visible Speech
Translators with External Phosphors," The Journal
of the Acoustical Society of America, XVIII (July,
1946), 62-73.
5.
Hohn, Robert W., "A Study of the Relationship Between
Vowel Modification and Changes in Pitch in the
Male Singing Voice: A Spectrographic Analysis,
Reinforced by the Judgment of Phonetics Experts,"
unpublished doctoral dissertation, Indiana University, Bloomington, Indiana-, 1959.
6.
Howie, John and Pierre Delattre, "An Experimental Study
of the Effect of Pitch on the Intelligibility of
Vowels," The Bulletin of National Association of
Teachers of "SingiriqI""xviII (May, 1962f, 6r9~
T
7.
Johnson, Merion James, "An Investigation of the Effect
of Training in the Articulation of Vowels by the
Speaking Voice Upon the Articulation of Vowels bv
the Singing Voice," unpublished doctoral dissertation, Indiana University, Bloomington, Indiana,
1966.
8.
Koenig, W., H. K. Dunn, and L. H. Lacy, "Sound Spectrograph," The Journal of the Acoustical Society of
America, XVIII (July, 1946), 19-49":
—
31
32
9.
Kopp, G. A. and H. G. Green, "Basic Phonetic Principles
:cr"f Visible Speech," The Jonrinal of the Acoustical
^Society of America,, XVIII {July, 1946) , 74-89.
10. IPexJcins, William .R.„ SrarooJ-le Sawyer, and Peggy
Harrison, "Research rm Vocal Efficiency," The
:Bulletin of National Association of Teachers of
Singing, XV (December,, 1958j)» 4-7.
11. rPetecson, Gordon E. and "Harold L. Barney, "Control
Methods Used in a Study of the Vowels," The Journal
:of the Acoustical Society of America, XXIV (March,
JL952) , 175-184..
12. .Potter, Ralph K.,•"Introduction to Technical Discussion
:o.f Sound Portrayal," The Journal of the Acoustical
"Society of America., XVIII (July, 1946), 1-3.
13.
, Gfeorge A.. Kopp, and Harriett Green
PP' Visible Speech., New York, Dover Publications,
-Inc., 1966.
Ko
14.
Riez, R. R. and D. Scott, "Visible Speech Cathode-Ray
Translation," The Journal of the Acoustical
Society of America, XVIII (July, 1946) , 50-61.
15.
Srcott, David Watters, "A Study of the Effect of Changes
of Vocal Intensity Upon the Harmonic Structure of
Selected Singing Tones Produced by Female Singers,"
unpublished doctoral dissertation, Indiana University, Bloomington, Indiana, 1960.
16.
Steinberg, J. C. and N. R„ French, "Portrayal of Visible
Speech," The Journal of the Acoustical Society of
America, XVIII (July, 1946), 4-19.
17.
Sullivan, Ernest G., "An Experimental Study of the Relationships Between Physical Characteristics and
Subjective Evaluation of Male Voice Quality in
Singing," unpublished doctoral dissertation,
Indiana University, Bloomington, Indiana, 1956.
18.
Taff, Merle E., "An Acoustical Study of Vowel Modification and Register Transition in the Male Singing
Voice," The Bulletin of National Association of
Teachers of Singing, XXII (December, 1965) , S^l.
35.
33
19.
Triplett, William M., "An
Vowel Sounds on High
National Association
(February, 1966), 6,
Investigation Concerning
Pitches," The Bulletin of
of Teachers of Singing, XXII
8, 50.
20.
van den Berg, Janwillera and William Vennard, "Toward an
Objective Vocabulary for Voice Pedagogy," The
Bulletin of National Association of Teachers of
Singing, XV (February, 1959), 10-15, 24.
21.
Vennard, William and James W. Irwin, "Speech and Song
Compared in Sonagrams," The Bulletin of National
Association of Teachers of Singing, XXIII
(December, 1966), 18-23.
22.
Whitworth, James Ralph, "A Cinefluorographic Investigation of the Supralaryngeal Adjustments in the Male
Voice Accompanying the Singing of Low Tones and
High Tones," unpublished doctoral dissertation,
State University of Iowa, Iowa City, Iowa, 1961. .
23.
Wooldridge, Warren Buchanan, "The Nasal Resonance
Factor in the Sustained Vowel Tone in the Singing
Voice," unpublished doctoral dissertation, Indiana
University, Bloomington, Indiana, 1954.
24.
Wyatt, Larry W., "Blend in Choral Sound," unpublished
master's thesis, School of Music, North Texas
State University, Denton, Texas, 1967.
amPTEit i n
PROCEDURES
The Sample
-The choral groups which were used as sources of examples
were -seiexrted with the approval of the dissertation advisory
committee from successful school programs.
Tape recordings
were.made :of several public school and college choral groups.
Each :df "the groups had a minimum of forty and a maximum of
fifty members performing for the recording.
The tape record-
ings were monaural and at a tape-speed of 7.5 inches-persecond.
.For ±he recording, the choral groups were assembled on
standing choir risers.
The girls stood on the front two
rows and the boys on the back two rows.
The members of the
choral groups were instructed that they were participating
as subjects in an experiment having to do with the blend of
choral sound as related to particular vowel sounds.
They
were asked to sing the vowel sounds as their own director
had instructed them.
They were asked to achieve a good
ensemble effect in attack, duration, intonation, loudness,
and vowel sound.
Each group was conducted by its own
director.
34
35
Each of the choral groups sang without instrumental
accompaniment the notes of the C major scale.
The scale was
repeated using each of the following vowel sounds:
ee (i),
as in beet; eh (e), as in bet; and ah (a), as in Bach.
The
scale, on each of the vowels, was sung by the male voices
alone, the female voices alone, and by the male and female
voices together.
The male voices sang the scale as illus-
trated :
•v—J—j
:/—o,
H
r
J*"""'
£>• J
-P'-r
rP-
Z + -
\
f
Fig. 6—An example of the C major scale as sung by the
male voices alone.
the female voices sang the scale in the same manner:
331
Fig. 7—An example of the C major scale as sung by the
female voices alone.
36
and the male and female voices sang together as follows:
_CL
m
,
—t-VK~
I
O '
| l®'
Fig. 8—An example of the C major scale as sung by the
male and female voices combined.
The pitch for the scale was given from a piano prior to
the singing of each scale.
Each note was sung approximately
three seconds in duration, at a moderate intensity.
A
breath was taken before the attack of each note of the scale.
The choral groups were encouraged to perform at their best.
No attempt was made to structure any of the recording
sessions to obtain either "good" or "poor" sounds for the
sample.
For the purpose of controlling acoustical quality, tape
recordings having extraneous sounds or other quality deterrents were rejected.
Those recordings which were a true
representation of the choral group were retained for the
study.
It was determined that the sample would be comprised
of examples from one junior high school group, two senior
high school groups, and two college groups.
37
From each of the educational levels there were to be
eight examples of each of the three vowels, sung in the
three combinations of voices.
For the purpose of selecting
the correct number of examples from the two senior high
school groups and the two college groups, a method of
distribution was devised.
The method was devised to avoid,
insofar as possible, any structuring or pre-judging of the
examples of choral sound.
The college group which was recorded first was designated CI, and the other college was designated C2.
In the
same manner, the senior high school groups were designated
SI and S2.
The designations "1-head," "1-tail," "2-head,"
and "2-tail" were written on individual slips of paper and
placed in a bowl.
For each scale, one slip was drawn and a
coin was tossed to determine which group would begin on the
first note of the scale.
The succeeding notes of the scale
alternated between the two groups in a systematic manner
as illustrated:
-aat
cr
CI
CX
11
tx
CI
CX
Fig. 9—An illustration of the method of alternating
the notes of the scale between two groups of the same educational level.
3®
This procedure was carried out for each of the scales
in the college and senior high school sample.
Since the
one junior high school sample furnished the correct number
of examples for the study, all the examples from the group
were included in the study.
A listing of the distribution
of all the examples is included in Appendix B.
The sounds which were used as examples were isolated
and extracted from the original tape.
After each example
was identified, the tape was cut just before the attack x>±
the sound and the leading end of the example was attached to
a sixty-inch length of timing-tape.
Each example was coded
and cataloged and rolled onto a storage reel.
This process
was continued until all examples were individually coded,
cataloged, and placed on a storage reel.
To eliminate the bias of identification in the sequence
of evaluation, the examples were assembled in a random
sequence.
To determine the random sequence the catalog
numbers which had been assigned to the examples were listed
in numerical order.
Included with the catalog number was
the coded identification of each example.
numbers (1, pp. 179-180) was utilized.
A table of random
It was decided to
begin with column nine, row five, to utilize three columns
and read downward.
This procedure was continued until.all
the catalog numbers had been assigned a random sequence
number.
A list of the numbers which were assigned is in-
cluded in Appendix B.
The examples were removed from the
39
storage reels, arranged in the random sequence and replaced
on the storage reels.
So that the examples would have a uniform duration
which would accommodate the capacity of the spectrograph, it
was necessary to cut the example tapes in uniform lengths.
Since the time duration of the sound spectrogram is 2.4
seconds, and since space was needed for the insertion of a
calibration scale, it was decided that the optimum duration
for the example would be 2 seconds.
With the tape-speed of
7.5 inches-per-second, a fifteen-inch length gave the correct
duration.
Each example in the random sequence was measured and
cut fifteen inches from the leading end of the example.
The
trailing end of each example was spliced onto the leading
end of the timing-tape leader of the following example.
The
result was one continuous tape, presenting the 216 examples
in the random sequence, each example two seconds in duration,
separated by eight seconds of silence.
The Evaluation
The sample was subjected to evaluation by a jury of
experts, on the variables of quality of blend and intelligibility of vowel sound.
The members of the jury were asked
to rate the quality of the vocal blend and to identify the
vowel sound which he perceived.
40
The jury consisted of seven choral directors selected
-with ~the :advice of a jCoimrii±-tfie of faculty members from the
; School ctif :Mus±c of North .Texas State University.
The members
of :the ; jury as selected incGLuded four faculty members and
three ::dcrctoral candidates..
The jury was assembled in the
Recital Mlall of the School of .Music for the purpose of
evaluating the recorded examplesPrepared instructions explaining the procedures of the
evaluation were furnished each .member of the jury.
The
instructions were read aloud and points of clarification
were made.
The members of the jury were seated separately
and did.not consult on the ratings.
According to his own individual criteria for choral
blend, each member of the jury rated the examples by
circling a symbol representing the vowel sound in the
appropriate column.
vowel symbols:
The judge could circle one of three
"ee," "eh," or "ah" in one of three columns;
"Good," "Acceptable," or "Poor" for each example in the
sequence.
A copy of the prepared instructions and an
example of the rating sheet are included in Appendix A.
The tape, which was thirty-six minutes in duration, was
played straight through; with a rest stop at the half-way
point.
To afford a check on the reliability of the evalua-
tion, the entire sample was played and evaluated twice in
the same manner.
Each example was heard only once during
each evaluation sequence.
41
The Analysis
For the purpose of acoustical analysis, sound spectrograms were made of all examples on a Kay Electric Sound
Spectrograph.
The examples were transferred from the tape
play-back equipment to the internal recording turntable of
the spectrograph by direct wire connection.
A standard
setting for play-back and record loudness levels was maintained insofar as was feasible.
Adjustments were made when
necessary to obtain the best quality spectrograms.
For the purpose of isolating machine noise from the
recorded sound, spectrograms were made of the machine
characteristics alone.
Machine characteristic examples were
made with the play-back and record loudness levels at the
maximum setting, the minimum setting, and at the standard
or average setting that was used with the examples.
A
sample spectrogram was also made of the spectrograph alone
with no input connected.
Two types of spectrographic display were made of each
example
The type 1 display shows time duration horizontally,
pitch frequency vertically, and intensity as shading between
gray and black.
A white area indicates no sound, a gray
area indicates a low intensity sound, and a black area
indicates a high intensity sound.
The type 2 display is a
cross-section showing intensity horizontally and frequency
vertically.
The time spot of the cross-section can be
selected at various intervals along the time axis of , the
42
spectrogram.
To ascertain if there were acoustical differ-
ences -in -the eexamples between -the sound at the moment of
attack and:.a£ter ±he attack., x;ross-ser;ticms were made at the
attack arid ^..approximately 1 to 1..2 seconds after the attack.
Each spectrogram contains both ±he "Type 1 and the Type 2
displays.
A-.'frequency calibration scale,, in bands of 500 Hz, was
included-an^ea-ch spectrogram.
After -each example had been
recorded:on-the internal turntable., the calibration tone
was recorded _d.uxing the '. 4 second interval between the end
of the examples and the repetition of the attack.
The
spectrograph paper was placed on the dxum in such a way that
the calibration scale appears at the beginning of each
spectrogram.
As each spectrogram was completed, it was identified
with the sequence number, catalog number, identification
code and the tabulated evaluation of the judges.
When all
spectrograms had been made they were rearranged in the
sequence of the catalog numbers.
CHAPTER BIBLIOGRAPHY
1.
Mouly, George J., The Science of Educational Research,
New York, American Book Company, 1963.
43
CHAPTER IV
ANALYSIS DP THE DATA
The -pprobJbem of this study was a spectrographic analysis of-the ^acoustical properties of selected vowels in
choral -sound. Examples of the vowels ee (i), eh (e), and
ah (a), were obtained from choral groups from junior highschool, senior high school, and college.
The purposes of
this study were to categorize -the examples of vowel sounds
by means :df ^subjective evaluation., to ascertain by means of
spectrographic analysis the distinguishing characteristics
of the acoustical properties of the examples in the categories , and to determine the similarities and dissimilarities which :exis.t within and between the categories.
The Evaluation
In the process of evaluating the recorded examples, the
taped sequence was played and evaluated twice.
Explicitly,
each member of the jury heard and evaluated each of the 216
examples twice.
Consequently, each judge gave a total of
432 ratings, two for each example.
The ratings were tabulated
to determine how many ratings each judge gave in each category.
Table I contains these tabulations.
It is apparent from-the tabulations that a diversity of
criteria were utilized in the evaluation of the examoles.
44
45
TABLE I
JUDGES' RATINGS OF VOWEL EXAMPLES, GIVING THE NUMBER AND
PERCENTAGE IN EACH CATEGORY
I
"Good"
II
"Acceptable"
Number
Percent
A
B
C
D
E
95
165
146
54
22.0
38.2
33. 8
12.5
41.0
215
49.8
146
162
293
48
33.8
37.5
67. 8
11.1
2.3
28.5
176
177
40.7
41.0
25.5
1,217
177
10
123
F
G
Total
770
Number
Percent
Judge
40. 2
III
"Poor"
Number
Percent
122
121
124
85
28.2
207
246
132
28.0
28.7
19.7
47.9
56.9
30.6
1,037
34.2
A large difference can be noted between Judge E and Judge F
in the ratings placed in the "Good" column.
Judge E rated
41.0 percent of the examples as "Good," which Judge F, in
hearing the examples, rated only 2.3 percent as "Good."
Judges A, B, C, and G were fairly similar in the evaluation.
Of this group, Judge A gave the lowest number of
"Good" ratings, and the highest number of "Acceptable"
ratings.
Judge D was unusual in that he rated a high per-
centage in the "Acceptable" category and a low percentage in
the "Good" and "Poor" categories.
On the other hand,
Judge E was inclined to rate most of the examples as either
"Good" or "Poor" and rated only a small percentage in the
"Acceptable" category.
The tabulations indicate that
46
Judges D and E exhibited opposing inclinations in rating the
examples.
Judge IF .placed the lowest percentage of ratings in the
"Good" rcategory and the highesf percentage in the "Poor"
category. TThis seems to indicate -that Judge F utilized a
more severe ..-set of criteria than "the other judges.
The above tabulations were an indication of how the
judges agreed with each other in "the evaluation of the
examples. The ratings were also computed to determine how
each judge "agreed with himself..
As previously stated, the entire sample was played in
the random sequence and evaluated twice, each judge rating
each example twice.
The ratings were tabulated to determine
if each judge gave the same rating to the same example both
times, or if he gave different ratings to the same example.
Table II gives the percentage of accuracy between the two
sequences.
The figures indicate whether the judge gave the
same rating both times, whether he deviated by one rating,
e.£. , I to II, or II to III; or whether he deviated by two
ratings, £.£., I to III, between the two sequences.
To
determine a reliability coefficient, Pearson's ProductMoment was computed.
in each category:
Score values were assigned to ratings
a score of three for a "Good" rating,
a score of two for an "Acceptable" rating, and a score of
one for a "Poor" rating.
The correlation for the panel of
judges is reported.in Table II.
47
TABLE II
JUDGES' RATINGS OF VOWEL EXAMPLES, GIVING THE PERCENTAGE OF
ACCURACY AND THE CORRELATION BETWEEN THE TWO
EVALUATION SEQUENCES
Agreement
Deviation of
"1 Rating
Deviation of
2 Ratings
Reliability
Coefficient
Pearson's
r
Judge
Number
%
Number
%
Number
%
A
B
C
D
E
F
G
128
137
109
149
143
144
147
59.3
63.4
50.5
69.0
66. 2
66.7
68.1
87
70
98
67
38
72
66
40.3
32.4
45.4
31. 0
17. 6
33.3
30.6
1
9
9
0
35
0
3
0.4
4.2
4.2
0.0
16.2
0.0
1.3
Panel
957
63.3
498
32.9
57
3.8
.56
It is apparent from the tabulations that the judges
were fairly high in accuracy.
It should be noted that
Judge D, who rated the highest percentage of examples in the
"Acceptable" category, and the lowest percentages in the
"Good" and "Poor" categories had the highest percentage of
agreement.
It should also be noted that Judge E, who rated
the lowest percentage of examples in the "Acceptable" category and the highest percentages in the "Good" and "Poor"
categories, had the highest number of split ratings.
How-
ever, Judge E did maintain a high percentage of agreement.
Judge F, who rated only 2.3 percent of the examples as
"Good" and 56.9 percent as "Poor," apparently maintained the
48
same criteria throughout both evaluation sequences.
He
gave the same rrating two out of three times for the two
sequences, = arid zicLrd not deviate by .moxe than one rating.
As indicated in Table I, 25,.5 percent of the ratings
given for :the-sample were "Good.," 4D-2 percent were "Acceptable," and 1 3 4 p e r c e n t of the ratings given -were "Poor."
Table II -indicates that the percentage -of agreement for all
the judges was .'63.3 percent.
All the judges deviated by one
rating an~avexage of 32.9 percent of the time, and deviated
by two ratings-only 3.8 percent of the time.
The reliability
coefficient 'for the panel was .56..
Table "III indicates the accuracy with which the judges
correctly identified the vowel sound..
The figures in this
TABLE III
NUMBER AND.PERCENTAGE OF ERROR IN JUDGES' IDENTIFICATION •
OF VOWEL SOUNDS
Judge
Number
Percent
Judge
Number
Percent
A
B
C
D
-1
1
2
2
0.4
0.4
0.9
0.9
E
F
G
5
2
1
2.3
0.9
0.4
table indicate how many errors were made by each judge in
the identification of the vowel sound.
The percentage
figure given is for the entire sample which was heard twice.
Judges A, B, and G marked a vowel sound incorrectly once
49
each, and Judges C, D, and F marked vowels incorrectly
twice each.
Judge E, who had the largest number of split
ratings marked five vowel sounds incorrectly for the highest
percentage of error of 2.3 percent.
In tabulating the.ratings according to educational
level, the figures in Table IV indicate that the distribution
TABLE IV
SUMMARY OF JUDGES' RATINGS OF VOWEL EXAMPLES BY CATEGORY
AND EDUCATIONAL LEVEL
1
I
"Good"
Level
Number
Junior High
Senior High
College
263
269
238
Totals
770
III
"Poor"
II
"Acceptable"
Percent
34. 2
34. 9
30.9
100.0
Number
379
422
416
1,217
Percent
Number Percent
31.1 •
34.7
34.2
100.0
366
317
354
1,037
35.3
30.6
34.1
100.o'
of "Good," "Acceptable," and "Poor" ratings was fairly even.
Of the ratings given in each category, roughly a third of
the number was from each of the three educational levels.
This would seem to indicate that all three educational
levels were equally represented in the sample.
It further
seems to indicate that the examples, which were presented in
a random sequence, were evaluated with the same criteria and
without bias.
50
To consider the distribution of the sample in detail,
the - ratings -were -tabulated .by instructional level, voice
combination, land vowel sound.. These -tabulations are presented ..in .Tables V through XX.
At-the : junior high school level,, Table V» the figures
indicate -.that :the female voices furnished the highest percentage ;of-examples in the "Good" category.
The ah (a)
vowel was .rated the highest, the ie (1) vowel next highest,
and the^eh. (e) the lowest.
The .male voices were signifi-
cantly llower -in the number of "Good" ratings with only 15
percent rrated in that category,.
For the male voices in this
portion:of .the sample, nearly half of the examples were rated
as "Poor."
For the male and female voice combination, the ratings
inclined toward the low ratings of the male voices.
These
examples were not rated nearly as high as were the female
voices alone.
This seems to indicate that, in this case,
the "Poor" vocal blend of the male voices had more effect
upon the vocal blend of the combined voices than the "Good"
vocal blend of the female voices..
The incorrectly marked vowels were also tabulated by
educational level.
In Table VI, it is reported that the
largest number of incorrect markings were for vowel sounds
sung by the female voices alone.
None of the vowel sounds
which were sung by the male and female voices together were
51
TABLE V
JUDGES' RATINGS OF VOWEL EXAMPLES, GIVING THE NUMBER AND
PERCENTAGE FOR EACH VOWEL AND VOICE COMBINATION IN
THE JUNIOR HIGH SCHOOL SAMPLE
"Good"
Vowel
Number
"Acceptable"
Percent
Number
"Poor"
Percent Number
Percent
Female Voices
ee
eh
ah
Sub-Totals
62
52
69
55.4
46.4
61.6
32
33
38
28.5
29.5
33.9
18
27
5
16.1
24.1
4.5
183
54.5
103
30.7
50
14.9
44.6
39. 3
42.9
55
53
55
49.1
47.3
49.1
163
48.5
Male Voices
ee
eh
ah
7
15
9
6.3
13.4
8.0
50
44
48
Sub-Totals
31
9.2
142
42.3
Male and Female Voices
ee
eh
ah
19
13
17
17.0
11.6
15.2
49
48
37
43.8
42.9
33.0
44
51
58
39.3
45.5
51.8
Sub-Totals
49
14.6
134
39.9
153
45.5
263
26.1
379
37.6
366
36.3
Totals
marked incorrectly, and only one vowel sound was marked incorrectly for the male voices alone.
At the senior high school instructional level the
female voices again produced the highest percentage of
52
TABLE VI
NUMBER. AND :EERCENTAGE OF VOWEL JSDU.NDS XNCOREECTLY IDENTIFIED
:IN THE JUNIOR .HIGH SCHOOL SAMPLE
Vowel
Number
Percent
"Female Voiceas
^ee
.eh
4
2
;
3.6
1.8
Male Voices
ah
1
examples-in ;the "Good" category..
0.9
(See Table VII.)
In this
case the .ee (.i) vowel was rated the highest, the ah (a)
vowel next, and the eh (e) vowel was again rated the lowest.
In contrast to the junior high school sample, the
examples sung by the male voices alone,, and the female voices
alone, were rated higher than were the examples sung by the
male and female voices together.
In each voice combination
in the high school portion of the sample, the. ee (i) vowel
and the ah (a) vowel were rated fairly evenly.
The eh (e)
vowel in each case was rated lower than the others.
The incidents of vowel sounds being marked incorrectly
in the senior high school portion of the sample were
isolated and fairly evenly distributed between the different
voice combinations.
As in the junior high school sample,
53
TABLE VII
JUD
p®p'
° F V ° W E L EXA MPLES, GIVING THE NUMBER AND
PERCENTAGE FOR EACH VOWEL AND VOICE COMBINATION IN
THE SENIOR HIGH SCHOOL SAMPLE
"Good"
Vowel
"Acceptable"
Number
Number
"Poor"
Number
Female Voices
00
0h
ah
49
39
44
Sub-Totals
132
43.8
34. 8
39. 3
39.3
51
40
53
45.5
35.7
47.3
33
15
10.7
29.5
13,4
144
42.8
60
17.9
12
Male Voices
27.7
17.9
28.5
Sub-Totals
44.6
41.1
41.1
44.6
24.7
Male and Female Voices
19. 6
33.0
46.4
36. 6
19.6
Sub-Totals
Totals
16.1
26.7
45.2
31.5
more vowel sounds were marked incorrectly for the female
voices than for the other voice combinations (see Table VIII)
The examples from the college level choral groups were
rated in much the same proportions as those from the senior
54
TABLE VIII
NUMBER- AND PERCENTAGE OF VOWEL SOUNDS INCORRECTLY IDENTIFIED
IN THE SENIOR HIGH SCHOOL SAMPLE
Percent
Number
"Vowel
JFemale Voices
:eh
ah
2
\
1
|
1.8
0.9
. Male Voices
:ah
1
0.9
Male and Female Voices
1
fie
high school choral groups.
0.9
These ratings are presented in
Table "IX.
As ±n the senior high school sample, the female voice
combination received the highest number of "Good" ratings
in vocal blend.
The male voices were rated the next highest.
The male and female voice combination was again rated the
lowest.
It was apparently significant that the eh (e) vowel was
consistently rated the lowest in vocal blend.
The one excep-
tion was in the junior high school portion of the sample
where the eh (e) vowel was rated slightly higher than the
other vowels sung by the male voices alone.
In the evaluation of the sample, each judge was asked
to rate, according to his own criteria, the quality of the
55
TABLE IX
JUDGES' RATINGS OF VOWEL EXAMPLES, GIVING THE NUMBER AND
PERCENTAGES FOR EACH VOWEL AND VOICE COMBINATION IN
THE COLLEGE SAMPLE
"Good"
Vowel
Number
"Acceptable"
Percent
Number
"Poor"
Percent Number
Percent
Female Voices
ee
eh
ah
Sub-Totals
48
29
40
42.9
25.9
35.7
40
55
55
35.7
49.1
49.1
24
28
17
*21.4
25.0
15. 2
117
34.8
150
44.6
69
20.5
Male Voices
ee
eh
ah
25
18
30
22.3
16.1
26. 8
36
45
41
32.1
40.2
36.6
51
49
41
45.5
43. 8
36.6
Sub-Totals
73
21.7
122
36.3
141
42.0
44
57
43
39.3
50.9
38.4
Male and Female Voices
ee
eh
ah
17
11
20
15.2
9.8
17. 9
Sub-Totals
48
14.3
144
42.8
144
42.8
238
23.6
416
41.3
354
35.1
Totals
51
44
49
45.5
39.3
43. 8
vocal blend of each of the sounds presented.
The jury per-
formed this task with an overall percentage of accuracy of
63.3 percent.
The examples from each of the educational
levels, when presented in random sequence, received approximately the same kinds of ratings.
This would seem to indicate
56
that the adjudication was valid and that the sample was
well^distributed among the three educational levels.
TThe ^examples were arranged ±n rank order in the following, manner.
.The examples with the greatest number of "Good"
ratings rarid the least number of "Poor" ratings were ranked
the highest.
The examples with tire smallest number of
"Good"-ratings and the greatest number of "Poor" ratings
were-ranked the lowest.
The examples were ranked progres-
sively ifrom the greatest number of "Good" ratings downward
to .the rgrea~test number of "Poor" ratings.
The ratings and
the .ranking of each example are listed in Appendix B.
lit was determined, by the number of "Good" ratings,
that .the :first twenty-seven samples in the rank order were
rated high enough to be considered as the "Good" examples
for ;the spectrographic analysis-
A similar number of
examples, twenty-eight, in the lowest rank order were considered as the "Poor" examples for the spectrographic
analysis.
All of the remaining samples were considered in
the "Acceptable" category.
The Spectrographic Analysis
In the spectrographic analysis of the individual
examples, terminology referring to the natural harmonic
series, vowel formant frequencies, and the spectrographic
display will be utilized.
In order to clarify the meaning
and application of the terminology, the following explanation is offered.
57
The natural harmonic series, or overtone series, consists of a fundamental frequency and the subsequent overtones.
The overtones will be multiples of the fundamental
frequency in the order of 1, 2, 3, 4, 5, 6, 7, 8, etc.
With a fundamental frequency of 110 Hz, the natural harmonic
series would be 110 Hz, 220 Hz, 330 Hz, 440 Hz, 550 Hz,
660 Hz, 770 Hz, 880 Hz, etc.
These frequencies converted
into musical notation would be
gfo
U o TIP
o
—4^|*ZEE
-
ilto °
WW
¥
no
Fig. 10—An example of the natural harmonic series
beginning on the pitch A, 110 Hz.
The natural harmonic series will continue in these same
proportions to as high a frequency as the tone generator will
produce.
Of the tones illustrated above, the seventh tone is
out-of-tune with modern musical instruments.
As the harmonic
series continues into higher multiples, several overtones
will be out-of-tune.
The natural harmonic series is present
in all musical sound, excluding electronically produced
58
"pure" tone.
The intensity of each of the harmonics will
vary with different instruments and voices.
Vowel formants have been discussed in previous chapters.
In Chapter I, average frequency ratings.were reported for
the first and second formants of several vowels.
The average
frequencies for each of the vowels when converted into
musical notation would be as follows:
!>•
it
&>
•d-
JSSm
—
.
-2SL
J2I
ae
i>e«t
bit
i>ct
Ut
cU
bOSS
W.
A
t>oot U t
1
Bert
11- Av
4. •
7 erage frequencies of vowel formants illustrated xn musical notation.
It is understood that the average frequencies illustrated above are from a specific sample, and that the notation used is approximate to the average frequency.
Varia-
tions of these frequencies would be expected within any
specific sample.
Two types of spectrographs display were utilized in
the analysis of the examples.
The Type 1 display shows
frequency vertically, time duration horizontally, and intensity as shading between gray and black.
The Type 2 dis-
play is a cross-section showing frequency vertically, in an
inverted scale, and intensity horizontally.
59
In the Type 1 display the frequency calibration scale
is in units of 500 Hz.
(See Figure 12.)
The lowest band on
Type 2 bupU.*f
O'BAICImhL
fto
8t*>
£©«>£>
i
^
^
5
" °
tfAo
VI
%%0 $
n
W&t>•a
V
IWO §
*
5*
nto v|
—1380
©OO
I 3foo
'
^ * eoo^—|
Q 3.S&0 ^
S I OOO
<t
- ? o-g&elmt.
iuedm&'O
Tyfrc 1 btapUy
-T/iwc
158 - 121 - C1G - eh - A - 4
8
2 - 78
Fig. 1 2 — A spectrogram of the vowel eh (e) sung by
college female voices on the pitch A (approximately 440 Hz).
the spectrogram is the baseline.
The baseline is machine
noise and is not a part of the recorded sound.
In this
example, the fundamental voice pitch is approximately 440 Hz
60
The first bank above the baseline is the fundamental pitch.
Each of the frequency bands is a multiple of the fundamental
frequency.
The intensity of each frequency band is indicated by
the darkness of the line on the spectrogram.
A black area
indicates a high intensity sound, a gray area indicates a
low intensity sound, and a white area indicates that there
was no sound.
Figure 13 indicates that the first and second
frequency bands, the dark black color, were high in intensity.
The third frequency band is of a gray color, indicating a
lower intensity sound.
The white areas between frequency
bands indicates that no sound was present at those frequencies.
Time duration is indicated on the horizontal
axis from left to right.
On the spectrogram, five-eighths
of an inch is equal to one-tenth of a second, and six and
one-half inches is equal to one second.
In Figure 13 the wavy lines in the higher frequencies
indicates that individuals within the choral group were
singing with noticeable vibrato.
Because the frequency
scale used on the spectrogram compresses the musical intervals in the low frequencies and spreads the musical intervals in the high frequencies, the vibrato is visible on the
spectrogram only in the high frequencies.
The Type 2 display is a cross-section of the sound,
which in this case corresponds exactly in time with the
Type 1 display.
In the Type 2 display, the frequency scale
61
| Type 2 btspOitf
Xnfe,rt**T
^latgp^
3atfo *
3f i>o
4^»o
3®fd 3
Z&fo
IKK) |r
.,,...
....
(7b0
^
1$
S4K0m«-O
7ypg 3 bopUy
158 - 121 - C1G - eh - A - 4 8 2 - 78
„ ni F i g l 1 3 A spectrogram of the vowel eh fr)
k„
o ege ;emale voices on the pitch A (approximately 440 Hz).
is inverted, with the baseline at the top and the high frequencies at the bottom.
(See Figure 13.)
The intensity of
each frequency band is indicated by the length of the line
from left to right.
m
Figure 13 the first and second
frequency bands are high in intensity and the shorter lines
62
are lower in intensity.
The white areas between the fre-
quency bands indicates that no sound was present at those
frequencies.
The sound spectrograms of the twenty-seven "Good"
examples were examined to determine the most logical sequence
of presentation for this discussion.
It was determined that
the examples would be grouped first by voice combination,
then by pitch, vowel sound, and educational level, respectively.
Listed with each example as it is discussed is (1) the
catalog number, (2) the random sequence number, (3) the code
for the educational level and voice combination, (4) the
vowel sound (with incorrect vowels in parentheses), (5) the
code for the pitch, (6) the ratings of the judges (with
ratings for the incorrect vowels in parentheses), (7) the
rank of the example, and (8) the page number of the example
in the Appendix.
Spectrograms of the examples categorized
as "Good" are included in Appendix C in rank order.
Female voices on the pitch c.—The first group of
examples from the "Good" category to be considered is of
female voices alone.
The examples of the pitch coded as c,
approximately 523 Hz, present each vowel sound, and represent each of the educational levesl.
All of the examples in this group are characterized by
narrow, well defined frequency bands.
These frequency, or
resonance, bands are steady dark lines with white area
63
between.
There is no diffusion of the resonance bands which
would be indicated by gray area around and between the bands.
In each case in this group, the resonance bands are evenly
spaced and clearly defined.
The resonance bands begin at approximately 523 Hz which
is the fundamental pitch intended by the singers.
The first
resonance band above the base line on the spectrogram is the
fundamental pitch.
There are as many as seven or eight
bands above the base line in some, but not all, of the
examples.
Each of the frequency bands are equidistant from
the adjacent band or bands.
The:distance between the
resonance bands is equal to the frequency of the fundamental
pitch,, in. this case 523 Hz-
For this first group of examples,
the fundamental pitch is the first resonance band at approximately 523 Hz.
1„Q46
The second resonance band is at approximately
Hz:, the third band is: at". 1., 5 6 9 Hz, the fourth at
2 , 0 9 2
Hz,, the fifth at 2,615 Hz, the: sixth at 3,138 Hz, and the
highest visible resonance band: lies at approximately 3,66 2 Hz.
This sequence of frequency bands: follows exactly the natural
ELarmanic series for the given.fundamental frequency.
This
general discussion applies to: the: following examples in this
group and will not be restated:"with, each example.
While the frequency, spacing^ and clarity of the
resanan.ee bands are uniform forrthe examples in this group,
the intensity of the bands varies from one example to the
next.. To best illustrate this-variation, each of the
examples are discussed individually.
64
The explanation of the identification code below is as
follows.
The example below is catalog number 8, it was 197th
in the random sequence, and was recorded by the female voices
of the first junior high school group.
The vowel sound
ee (i), which was identified as eh '(e) by one of the judges,
was sung on the note C above middle C.
The ratings from the
two evaluation sequences were 9 "Good," 3 "Acceptable," and
2 "Poor."
One of the "Poor" ratings was given to the vowel
sound which was incorrectly identified.
ranked 18th in the sample.
is included on page
This example
The spectrogram of this example
in Appendix G.
8 - 197 - JIG - ee(eh) - c - 9 3 1(1) - 18 (p. 137).
In this example the resonance band of the fundamental
is the most intense.
The second band is approximately one-
half the intensity, and the third is no more than one-fourth
the intensity of the first band.
lightly visible.
A fourth resonance band is
None of the higher resonance bands appear
on this spectrogram.
The first resonance band, 523 Hz, is
the first formant for the vowel, and the fourth resonance
band at approximately 2,092 Hz is the second formant.
The average frequencies for'the distinguishing formants
for the ie (i) vowel fall at approximately 270 Hz and
2,290 Hz, a difference of 2,020 Hz.
In the case of this
example, the difference between the highest and lowest
resonance bands was approximately 1,560 Hz.
that this variation would make the vowel more
It is possible
difficult to
65
distinguish.
Note that one of. the judges incorrectly
identified the vowel.
80 - 84 - S2G - ee - c - 10 3 1 - 7 (p. 126).
For this example there are five resonance bands visible
above the baseline of the spectrogram.
The fundamental band
serves as Formant 1 and the fourth band serves as Formant 2
for the vowel.
These two resonance bands are more intense
than the other bands..
152 - 118 - C2G - ee(eh) - c - 7 (1) 6 0 - 23 (p. 141).
In this example there are six resonance bands clearly
visible.
The fundamental.is the strongest in intensity.
The succeeding bands reduce proportionately in intensity.
Vibrato wave, indicated by the curving line, is visible in
the higher frequencies.
The: vibrato wave is characterized
by an increase and decrease of, intensity as the vibrato
pitch crosses the resonance: band.
The vibrato wave is
visible on the spectrogram only as the frequency of the wave
matches the frequency of:," the:.resonance band.
Here, as in
ex:ample 8 above, the vowel! souad was incorrectly identified
by ane of the judges.
16 - 145 - JIG - eh. — c:: — 10 3 1 - 7
(p. 125).
In. this example, the; first-and second resonance bands
are tlxe more intense.. The; third band shows a steady
intensity.. The fourth, and: sixth .bands are less intense
than the other bands.
There is not a band visible in the
axea where the fifth band, should., be.
For the vowel, the
66
fundamental, 523 Hz, is the first formant, and the third
resonance band, 1,569 Hz, is the second formant.
24 - 124 - JIG - ah - c - 12 2 0 - 2 (p. 120).
This example is characterized by the prominence of the
first two resonance bands.
defined.
The third band is fairly well
The higher bands are visible only in short traces.
The fundamental and the first overtone are equal in intensity
and serve as the first and second formants for the vowel.
96 - 53 - S2G - ah - c - 12 2 0 - 2 (p. 121).
This example is also characterized by the prominence of
the first two resonance bands.
The third and sixth bands
are barely visible on the spectrogram.
resonance bands visible.
There are no other
The first two resonance bands,
which are equal in intensity, serve as the distinguishing
formants for the vowel.
The frequencies are 523 Hz for
Formant 1 and 1,046 Hz for Formant 2.
168 - 119 - C2G - ah - c - 9 5 0 - 11 (p. 130).
This example has very strong first and second resonance
bands.
The second resonance band is slightly more intense
than the first.
The third and sixth bands are clearly
defined and relatively strong.
appear only in traces.
The other frequency bands
There is vibrato wave in the higher
frequencies which appears as intensity reinforcement for
the resonance band.
The first two resonance bands are the
distinguishing formants for the ah (a) vowel.
67
Female voices on the pitch B.—The next group of
examples to be considered is of female voices alone, on the
pitch B.
As in the above group, the resonance bands are
well defined and clearly spaced with no gray areas between
the bands.
The resonance bands are all multiples of the
fundamental frequency, 494 Hz.
7 - 95 - JIG - ee (eh) - B - 6(3) 3 2 - 18 (p. 136).
The only prominent resonance band is the fundamental
tone.. The second band is readily visible, but it is of low
intensity.
traces..
The third and fourth frequency bands appear in
The higher frequency bands are not visible on the
spectrogram..
Only one, the lower., of the two distinguish-
ing formants for the vowel ee (i.) is present.
the vowel sound was difficult to_ identify.
Apparently
Three of the
judges, indicated that the vowel-heard was eh (e).
153 - 21 - C2G - eh - B — 8 5 1 - 26 (p. 144).
This is a complex spectrogram, with a total of eight
resonance bands visible.
The: frequency of the highest band
is approximately 4,000 Hz.., Resonance bands one, two, three,
ficrax:,, and" six are the stronger: bands.
The fundamental,
4S4; Hz,, is the first formant. forrthe vowel.
resonance, band is Formant. 2-..
The fourth
The: fourth band is wider than
ttfee other: bands and has a. slight;division.
Apparently part
©f the band is strengthened:" by - the. pitch harmonic and part
is strengthened by the vowel formant resonance.
The fre-
quency of the fourth band is. approximately 1,900 Hz.
68
23 - 104 - JIG - ah - B - 9 5 0 - 11 (p. 129).
This example of the ah (a) vowel has two prominent
resonance bands.
The fundamental tone at approximately
494 Hz and the second band at approximately 988 Hz are the
strongest.
These two bands serve as the distinguishing
formants for the vowel.
A third band, at 1,482 Hz is also
fairly well defined.
Female voices on the pitch A.—The following group of
examples of female voices singing the pitch A, possesses
the same general characteristics attributed to the groups
discussed above.
The resonance bands are clearly defined
and equally, spaced.
The fundamental frequency for this
group of examples is 440 Hz.
6 - 66 - JIG - ee - A - 10 3 1 - 7 (p. 124).
The fundamental tone is the only prominent resonance
band on this spectrogram.
The second band is approximately
one-third the intensity of the fundamental.
appear only as traces.
Formant 1.
The other bands
The fundamental tone serves as
There is no frequency band which can serve as
Formant 2 for the vowel ee (i).
150 - 26 - C2G - ee - A - 8 6 0 - 23 (p. 140).
In this example there are seven visible frequency bands,
The strongest bands are the fundamental, 440 Hz, which is
Formant 1; the second band is the first overtone at 880 Hz;
and the fifth band at 2,200 Hz is Formant 2.
69
22 - 182 - JIG - ah - A - 9 5 0 - 11 (p. 128).
This example of the ah (a) vowel has three prominent
resonance bands.
Bands one and two are equally strong; band
three is approximately two-thirds as intense.
bands are visible, but are not consistent.
Other higher
It is not clear
which bands are the distinguishing formants for the vowel.
Bands two and three appear to be closest to the average
frequencies for the distinguishing formants for the ah (a)
vowel.
94 - 156 - S2G - ah - A - 10 4 0 - 4.5 (p. 122).
This is a fairly complex spectrogram, with ten resonance
bands visible.
4,000 Hz.
The highest of the resonance bands is above
The first, two, bands are equal in intensity, the
third is approximately two-thirds as strong.
five, and six are barely visible.
Bands four,
Bands seven, eight, and
nine are approximately one-fourth the intensity of the
fundamental.
The second: band at 880 Hz, apd the third band
at 1,,760 Hz serve as. F.ormant 1 and Formant S, respectively.
Female voices on the; pitch G . — T w o examples of female
voices of the pitch G:/, 3921HZ, w„ere categorized as "Good."
Berth, examples displaydiearly defined, evenly spaced
Eestanance bands.
5 - 33 - JIG - ees — G : - 9 3 2 - 18 (p. 135).
This example from the junior high school sample is
characterized by a strong fundamental tone and two weak
70
overtones.
The fundamental serves as Formant 1 for the
vowel, but there is no frequency band to serve as Formant 2.
149 - 192 - C1G - ee - G - 8 3 3 - 27 (p. 145).
This spectrogram is characterized by several strong,
high resonance bands.
There are eleven resonance bands in
all, the highest is above 4,000 Hz in frequency.
A con-
siderable amount of vibrato wave is visible in the higher
frequencies.
The vibrato wave appears to broaden and
diffuse the resonance bands.
The fundamental tone stands
out as the strongest resonance band.
Above the fundamental,
the second, fifth, sixth, seventh, and eighth bands are
equal in intensity at about half the strength of the fundamental.
The third and fourth bands are visible in traces.
The fundamental is Formant 1, and the sixth resonance band,
2,352 Hz, serves as Formant 2.
Female voices on the pitch F.—For female voices alone,
four examples of the pitch F were categorized as "Good."
The general acoustical characteristics are the same as those
mentioned for the groups discussed above.
4 - 58 - JIG - ee - F - 9 3 2 - 18 (p. 134).
This example displays a strong fundamental frequency
band at approximately 349 Hz.
The second resonance band,
at approximately 698 Hz, is only about one-third the intensity
of the fundamental.
intermittent.
The higher resonance bands are weak and
In order to have a clearly formulated ee (i)
71
vowel, there should be a fairly strong resonance band at
2,000 to 2,200 Hz.
In this example there is a weak, but
traceable, resonance band at approximately 2,094 Hz.
76 - 39 - S2G - e e - F - 9 4 1 -
14.5 (p. 133).
This spectrogram displays twelve resonance bands.
ever, only the fundamental is steady and strong.
resonance bands are visible as light gray traces.
How-
The upper
The
strongest of the higher resonance bands is less than onefourth as strong as the fundamental frequency band.
This
resonance band is at approximately 2,094 Hz and serves a
Formant 2.
20 - 158 - JIG - ah - F - IX 2, 0. - 2 (p. 119) .
This example displays three- equally strong resonance
bands at frequencies of 349 Hz, 69.8: Hz, and 1,047 Hz.
These
resonance bands are the fundamental: pitch and the first two
overtones in. the natural harmonic: series.
The first over-
tone, 698 Hz.,, is Formant 1, and." the; second overtone, 1,047 Hz,
is Formant Z far the vowel ah (a;).. The higher resonance
bands diminish, quickly in intensity.. There is an isolated
resonance band" at approximately 3.", 0311 Hz.
This band does
not appear: an; the spectrogram until! about one-third of the
time duration; has elapsed.
52 - HOD - S2G - ah - F. — 8:
6:
0: — 23 (p. 139) .
&s in; the example immediately/above, there are three
equally strong resonance bands at 349 Hz, 698 Hz, and 1,047
Hz.
These, resonance bands serve; as; the fundamental pitch,
72
Formant 1, and Formant 2, respectively.
A fourth resonance
band/ 1,396 Hz, is also fairly strong.
This band serves as
a harmonic reinforcement for the fundamental tone.
The
ninth band is also steady and fairly strong at approximately
3,041 Hz.
A strong band at this approximate frequency is
characteristic of a resonant vocal sound.
Female voices on the pitch E.—Only one example for
female voices alone'on the pitch E was categorized as "Good."
The approximate frequency of the pitch E is 330 Hz.
19 - 41 - JIG - ah - E - 9 5 0 - 11 (p. 127).
In this example there are two equally strong resonance
bands at 330 Hz and 660 Hz.
There is another band which is
only slightly less in intensity at approximately 990 Hz.
Two weaker resonance bands are visible, one at 1,320 Hz, and
one at approximately 2,970 Hz.
The second resonance band
serves as Formant 1, the third band serves as Formant 2 for
the vowel sound ah (a).
Female voices on the pitch D.—Two examples which were
sung on the pitch D, approximately 294 Hz, were categorized
as "Good."
Both are examples of the "ee (i) vowel.
74 - 18 - S2G - e e - D - 9 4 1 - 14.5 (p. 132).
This spectrogram displays a strong fundamental tone and
little else in the way of resonance.
There are intermittent
resonance bands at approximately 588 Hz and 882 Hz.
bands seem to alternate in occurrence.
These
Only twice, and for
73
a very short duration, do these two bands occur at the same
time.
The fundamental tone serves as Formant 1 for the
vowel.
There is no visible Formant 2 for the vowel ee (i).
146 - 5 - C2G - e e - D - 1 0 4 0 - 4.5 (p. 123).
This spectrogram displays several well defined resonance
bands in a pattern which seems to be indicative of the ee (i)
vowel.
The fundamental tone is a strong resonance band at
approximately 294 Hz.
The second resonance band at 588 Hz
is a little more than one-half the strength of the fundamental.
The third band is approximately one-fourth the
strength of the fundamental.
The fourth, fifth, and sixth
bands are only visible in very light traces.
The seventh,
eighth, and ninth resonance bands are approximately onefourth the strength of the fundamental.
As many as thirteen
resonance bands are discernable, the-highest band is at
approximately 3,822 Hz.
The fundamental tone, at 294 Hz, is
Formant L„ and the eighth resonance: band, at 2,352 Hz, is
Eormant 2 far the ee (i) vowel.. '
Male and" female voices on the-pitch £.—Only one example
of male and. female voices combined: was: categorized as "Good."
B i s example has the well defined, evenly spaced resonance
bands which, were characteristic: ofthe: examples of female
wuices alone.. The resonance bandss are -more closely spaced
ctae to- the: lower" fundamental pitch-sung by the male voices.
There are two natural harmonic series present simultaneously.
74
One series is in multiples of the fundamental frequency,
approximately 261.5 Hz, sung by the male voices.
The other
series is in multiples of the fundamental frequency, approximately 523 Hz, sung by the female voices.
Beginning with
the second resonance band in the harmonic series for the
male voices, every second resonance band will coincide with
a band in the harmonic series for the female voices.
216 - 130 - C1M - ah - c - 8 6 0 - 23 (p. 143).
Above the baseline of the spectrogram, the first
resonance band, at approximately 261.5 Hz, is the fundamental
frequency sung by the male voices.
The second resonance
band, at approximately 523 Hz, is (1) the fundamental frequency sung by the female voices, (2) the first harmonic
overtone of the frequency sung by the male voices, and
(3) the first formant for the vowel sound sung by the female
voices.
The second band is also the strongest in intensity.
The third resonance band at approximately 784.5 Hz is the
second harmonic overtone for the•frequency sung by the male
voices.
This band is almost twice as strong as the funda-
mental sung by the male voices.
Since this band is not re-
inforced by a pitch overtone of the frequency sung by the
female voices, it possibly is reinforced by a vowel formant
frequency.
The third resonance band is possibly Formant 1
for the vowel sound sung by the male voices.
The fourth resonance band at approximately 1,046 Hz is
very slightly less in intensity than the strongest band.
75
This band is (1) the first harmonic overtone for the frequency sung by the female voices, (2) the third harmonic
overtone for the frequency sung by the male voices, and
(3) Formant 2 for the vowel sound sung by both the male and
female voices.
In all, there are sixteen resonance bands visible on
the spectrogram.
Hz.
The highest visible band is above 3,500
The frequency ratings of all the bands are multiples
of the fundamental frequency sung by the male voices.
The
even numbered bands, 2, 4, 6, etc., have frequency ratings
which are multiples of the fundamental frequency which was
sung by the female voices.
The stronger resonance bands
are those which were, reinforced by both the natural harmonic
series of both the male and female voices.
Male voices on the pitches c, A, and D.—Three examples
sung by male voices alone were categorized as "Good."
The
acoustical characteristics of these examples are similar to
those of the examples of J the female voices alone.
Due to
the difference in. frequency ratings of the fundamental tones,
the spacing of the^ resonance bands is much closer with the
male voice examples:, than:-, it was with the female voice
examples.
For proportionately resonant sounds, twice as
many resonance bancfe can: be visible in the same frequency
. range.
76
176 - 97 - C2B - ee - c -•9 5 0 - 11 (p. 131).
The first seven resonance bands are fairly well defined.
The strongest bands are the first, second, and seventh.
The
resonance bands above the seventh, approximately 1,931 Hz,
become diffuse and not well defined on the Type 1 display.
However, the cross-section Type 2 display indicates strong
resonance peaks at regularly spaced intervals.
The resonance
bands are clearly visible up to approximately 3,500 Hz.
This indicates that the sound was fairly complex and rich in
overtones.
The fundamental frequency, 261.5 Hz, serves as
Formant 1, and the seventh resonance band at approximately
1,841 Hz serves as Formant 2 for the vowel ee (i).
190 - 100 - C2B - ah - A - 8 6 0 - 23 (p. 142).
In this example there are five strong resonance bands
in the lower frequency range.
The first band is the funda-
mental frequency at approximately 220 Hz.
The second band
is the first harmonic overtone at 440 Hz.
The strongest of
the bands is the third band at approximately 660 Hz.
This
resonance band is the second harmonic overtone and is
Formant 1 for the vowel ah (a).
The fourth resonance band
is a harmonic overtone at 880 Hz~. The fifth band, at
approximately 1,100 Hz, is Formant 2 for the vowel as well
as a harmonic overtone.
A strong resonance band also occurs
at approximately 2,640 Hz.
This resonance band, which is
approximately one-half the intensity of the fundamental, is
characteristic of a resonant vocal sound.
77
114 - 150 - S2B - ah - D - 9 3 2 - 18 (p. 138).
-This example is characterized by closely grouped, strong
resonance bands in the lower frequency range.
mental frequency is approximately 147 Hz.
The funda-
The succeeding
frequency bands are multiples of the fundamental frequency.
The strongest of the bands are the fundamental and the first
harmonic overtone.
The fourth resonance band, at approxi-
mately 58 8 Hz, serves as Formant 1 for the vowel.
The
seventh band, at approximately 1,029 Hz, serves as Formant 2.
The area on the spectrogram between 1,029 Hz and 2,500 Hz
is devoid of resonances.
There are. traces of resonance in
the area around 2,500 Hz.
AIL the examples discussed above were categorized as
"Good" by a majority of ratings by the jury of experts.
There were twenty-seven examples categorized as "Good."
ranks of the examples were from 2. to:27.
The
The examples to
be discussed below were categorized as "Poor" by a majority
of ratings of the jury of experts... A total of twenty-eigm
examples with, ranks from 191 to 225i 5 are considered in thisx
category.
Spectrograms of the examples categorized as
"Poor" are included in Appendix
arranged in rank order.
The examples categorized as; "Good" were characterized
by clearly defined resonance- bands; which were evenly spaced
and separated; by white area.. The; resonance bands for the
examples categorized as "Poor" in vocal blend are not as
clearly defined.. Generally, in. this: category the spectrograms display little clearly defined: white area within the
78
frequency range of the resonance bands.
The resonance bands
of the natural harmonic series stand out as the strongest
frequency bands, but the areas between them are filled in
with lower intensity frequency bands.
The result is that
much of the information on the spectrogram is attributable
not to natural harmonics or vowel formants, but to other
undefined sources.
Several of the spectrograms indicate that sound was
produced at every definable frequency level from the fundamental tone to as high as 3,000 Hz.
In most cases, the
Type 1 display presents a confusing mass of black and gray.
White area occurs only in areas which are devoid of resonance
bands.
The Type 2 display indicates, in most cases, that
there are strong peaks of resonance at the frequency levels
of the overtones of the natural harmonic series, but that
these resonance bands are obscured at the base by resonances
of lower intensity.
The result is a continuous band of
resonances from the lowest to the high frequency.
Male and female voices combined.—The examples to be
considered in the "Poor" category are grouped by voice combination, pitch, vowel sound, and educational level.
The
first group to be discussed is of male and female voices
combined.
70 - 27 - JIM - ah - A - 1 0 13 - 211 (p. 169).
In this example there are six prominent resonance bands.
The lower bands are strong, the higher bands decrease quickly
79
in intensity.
The resonance bands are at frequencies which
are multiples of the fundamental frequency of 220 Hz which
was sung by the male voices.
These resonance bands are well
defined at the peaks, but the bases are connected by lower
intensity frequencies.
The area between the resonance peaks
is gray with little area which could be described as white.
Traces of resonance bands are visible up to 3,000 Hz.
How-
ever, after the sixth band, at approximately 1,320 Hz, the
intensity level is very low.
Formant 1 and Formant 2
apparently coincide with the second and fourth resonance
bands.
142 - 9 - SIM - ah - A - 0. 5: 9 - 191 (p. 149).
This example is also characterized by six prominent
resonance bands.
These bands are fairly well separated by
white area, but the Type 2 display indicates that the bands
are of irregular configuration,. The irregular shape
apparently indicates a lack of:" alignment of pitch frequencies
and vowel formant frequencies.
In this example there is a
sharp drop in. intensity in the: resonances above 1,320 Hz.
There are traces of resonance, bands as high as 4,000 Hz.
In this example the third resonance band at 660 Hz, and the
fourth resonance band at 88X): Hz: ar.e^ the strongest in intensity.
It is nat clear which bands: serve.: as: formants for the vowel.
204 - 171 - C2M - eh - FT — 0: 11 13 - 213 (p. 172).
This example is characteristic of the elements
attributed to the "Poor" category.. The frequency bands are
80
almost continuous from the fundamental pitch of approximately 174 Hz to above 3,500 Hz.
The resonances decrease in
intensity as the rate of the frequency increases.
One small
band of white area occurs at approximately 1,900 Hz.
This
is the only white area within the range of the frequency
bands.
The third resonance band at approximately 524 Hz is
the strongest in intensity.
It is not apparent which of the
resonance bands are the vowel formants.
203 - 4 - ClM - eh - E - 0 3 11 - 204 (p. 164).
This spectrogram displays continuous frequency bands
from approximately 165 Hz to above 4,000 Hz.
There are no
white areas apparent within the frequency range.
The third
band at approximately 49 5 Hz and the fourth band at approximately 660 Hz are the strongest of the resonance peaks.
The
Type 2 display indicates that the resonance bands are irregular in shape and intensity.
The intensity of the
resonance peaks decreases and the rate of the frequency increases.
211 - 92 - C2M - ah - E - 1 3 10 - 195.5 (p. 154).
This spectrogram indicates continuous frequency bands
from approximately 165 Hz to about 1,320 Hz.
white area from 1,320 Hz to about 2,310 Hz.
There is a
A lightly shaded
gray area extends from approximately 2,310 Hz to about 3,500
Hz.
There are resonance peaks at 49 5 Hz, 660 Hz, and 825 Hz
which are equally strong in intensity.
These resonance peaks
81
are not separated by white area on the spectrogram.
The
vowel formants are not discernable.
58 - 112 - JIM - eh - D - 0 3 11 - 204 (p. 162).
In this spectrogram the first four resonance peaks are
equal in intensity.
The bases of the resonance bands are
heavily connected by bands of lower intensity frequencies.
Above the fourth band there is a sharp drop in intensity at
all frequencies.
There is a fairly strong band at approxi-
mately 1,470 Hz which is apparently the second formant for
the vowel.
The location of Formant 1 is not apparent.
194 - 50 - C1M - ee - D - 0 5 9 - 191 (p. 151).
This spectrogram displays four strong resonance bands
which are connected by lower intensity frequency bands.
In
the crass-section Type 2 display, the frequency bands which
lie between the resonance peaks appear to be approximately
©ne-hialfr the strength of the strongest peak.
From the base
©f the cross-section to the, half-way point in intensity, the
spectrogram shows continuous: frequency bands.
In the area
between 588 Hz and 1,500 Hz:, almost no resonances are indicated-
From 1,500 Hz to ai point: above 4,000 Hz the spectro-
gram shows a lightly shaded"gray/area without definitive
resonance bands.
The lack of: resonances between the fre-
quencies af~ 58:8 Hz and 1,5.00: Hz: is. indicative of the ee (i)
vowel.. However, it is not- apparent which of the resonance
bands serve
as the distinguishing formants for the vowel.
82
49 - 60 - JIM - ee - C - 0 2 12 - 209 (p. 166).
This example presents a cluster of resonances in the
.lower frequency range, from approximately 131 Hz to 524 Hz.
Above that point very little is visible on the spectrogram.
There are light gray areas at about 1,800 Hz and about 2,200
Hz.
The resonance band at approximately 2,20 0 Hz should
serve as Formant 2 for.the vowel sound.
57 - 80 - JIM - eh - C - 2 2 10 - 194 (p. 152).
The strongest resonances are in the lower frequency
range.
The strength of the frequency bands decreases sharply
after 500 Hz.
The area above 500 Hz is lightly shaded in
gray on the spectrogram, with few definable resonance bands.
A fairly strong frequency band at approximately 1,300 Hz is
apparently indicative of the vowel sound.
201 - 36 - ClM - eh - C - 1 3 10 - 195.5 (p. 153).
In this spectrogram the beginning of the frequencies
presents a cluster of resonances ranging from approximately
131 Hz to 800 Hz.
As the time duration continues, additional
resonances appear in the higher frequency range.
After
approximately one second has elapsed, the spectrogram shows
a continuous band of frequencies' ranging from the fundamental
to above 3,500 Hz.
The least intense of the lower frequencies
appears at approximately 1,000 Hz.
The strongest of the
higher frequencies appears at approximately 1,824 Hz.
This
resonance band should serve as Formant 2 for the vowel eh (e).
83
209 - 183 - C2M - ah - C - 0 2 12 - 209 (p. 168).
• This example presents a mass of resonance bands with no
separation from approximately 131 Hz to 7 86 Hz.
From approxi-
mately 786 Hz to about 2,000 Hz there is no evidence of
frequency bands.
Above 2,000 Hz there is a light gray area
indicating low intensity frequencies in a range up to about
3,500 Hz.
The vowel formants are concealed in the mass of
resonances in the lower frequency range.
Male voices alone.—The next group of examples to be
considered in the "Poor" category is of male voices alone.
These examples display in general the same: acoustical confusion that was apparent in the above examples of male and
female voices together.
31 - 154 - JIB — e e - B - 0 4 1 0 — 1SJB'.5.: (p. 155).
In this example the resonance bands are: fairly well
defined and separated more or less clearly.
However, only
the fundamental and the first overtone are:, steady in intensity.
The higher overtones are strong, but intermittent.
There is considerable gray area between, the.- first two
resonance bands.
This: indicates that, there-were lower in-
tensity frequency bands which would interfere, with the
stability of tlxe pitch... The fundamental! tone serves as
Formant l r and. a. fairly steady band air. approximately 1,976
Hz could serve as: Formant 2 for the vowels.
84
39 - 106 - JIB - eh - B - 0 3 11 - 204 (p. 161).
. This example is very much like the one described just
above.
The first six frequency bands are fairly well
defined in pitch, but are unstable in intensity.
The
strongest of the bands are the fundamental tone at approximately 247 Hz, and the first harmonic overtone at approximately 494 Hz.
Isolated well above the lower resonance
bands is a fairly strong band at approximately 2,470 Hz.
The second resonance band at 494 Hz serves as Formant 1, and
the sixth band at approximately 1,482 Hz serves as Formant
2 for the vowel.
47 - 176 - JIB - ah - B - 0 4 10 - 198.5 (p. 157).
This spectrogram displays five strong resonance bands
in the lower frequencies.
The peaks of the resonance bands
are strong, but there are lower intensity frequency bands
between the stronger bands.
These five bands range in fre-
quency from approximately 247 Hz to 1,235 Hz.
The area
above 1,235 Hz is devoid of frequency bands up to approximately 2,717 Hz.
At this point, and above, five more
resonance bands are faintly visible.
Apparently, the second
resonance band at 494 Hz and the~fourth band at 988 Hz serve
as the distinguishing formants for the vowel.
119 - 162 - SIB - a h - B - 0 5 9 - 191 (p. 148).
This example is much like the one immediately above.
The resonance bands appear to be stronger in intensity.
The
frequencies of the resonance bands are the same, the same
85
area is devoid of frequency bands.
The intensity of the
resonance bands decreases and the frequency increases.
It
is not clear which of the resonance bands serves as the
distinguishing formants for the vowel.
37 - 56 - JIB - eh - G - 0 3 11 - 204 (p. 160).
This spectrogram presents a cluster of resonances with
no separation in the lower frequencies.
The resonance peaks
of the first three bands are strong and steady.
three bands appear weaker and more unstable.
The next
The seventh
band is fairly strong and apparently serves as Formant 2 for
the vowel.
The third resonance band at; approximately 588
Hz apparently serves as Formant 1 for the vowel eh (e).
99 - 52 - SIB - e e - E - 0 1 1 3 — 21.3 (p. 170).
In this spectrogram there is a cluster of frequency
bands near the baseline.
Three resonance bands are dis-
cernable, at frequencies of 165 Hz, 330. Hz, and 495 Hz.
Above this point,, the next resonance: bands of significant
strength occur at approximately 1,900" Hz:.. The band at 1,900
Hz and the ones above, appear only in.traces.
The second
resonance band at 330 Hz and the thirteenth band at approximately 2,145 Hz appear to be the distinguishing formant
bands for the vowel171 - 138 - C1B - ee - E - 0: 0: 114,— 215.5 (p. 174).
This spectrogram displays a continuous band of sound
from the fundamental, approximately 165 Hz, through the
fifth resonance band at approximately 825- Hz.
Another
86
continuous band begins at approximately 1,485 Hz and runs up
to approximately 3,600 Hz.
The intensities of the resonances
in the higher frequencies are very strong compared to other
examples in the sample.
There is a strong resonance band at
approximately 1,815 Hz, which is apparently the second
formant for the vowel.
The second resonance band at approxi-
mately 330 Hz would serve as the first formant for the vowel.
115 - 125 - SIB - ah - E - 0 2 12 - 209 (p. 167).
A mass of seven closely connected resonance bands form
the lower part of this spectrogram.
These bands range in
frequency from approximately 165 Hz to 1,155 Hz.
A higher
group of resonance bands of lower intensity range from
approximately 2,500 Hz to about 3,800 Hz.
The area between
the two groups of resonances is mostly white, with traces of
gray.
It is not apparent which of the resonance bands are
the distinguishing formants for the vowel.
187 - 216 - C1B - ah(?) - E - 0 1 11(2) - 213 (p. 171).
This spectrogram in most respects is similar to the one
immediately above.
In addition to the characteristics
already described, there is in this spectrogram evidence of
extraneous noises.
It is apparent from listening to this
example that the tape recording itself became damaged in
the process of handling in the experiment.
Note that the
example was the last example in the random sequence.
Note
also that one of the judges failed to give any identification
of the vowel sound in either evaluation sequence.
87
170 - 114 - C2B - e e - D - r 0 5 9 - 191 (p. 150).
A cluster of resonances, ranging from 147 Hz to 588 Hz,
forms the lower part of this spectrogram.
A second smaller
cluster lies between 1,764 Hz and 2,058 Hz.
A third cluster
of resonances begins at approximately 2,352 Hz and continues
to above 3,500 Hz.
The second formant for the vowel would
be in the second cluster of resonances at approximately
1,911 Hz.
The first formant would be in the first cluster
of resonances at a frequency of approximately 294 Hz.
106 - 205 - SIB - eh - D - 0 3 11 - 204 (p. 163).
The resonance formants in this spectrogram are diffuse
and difficult to distinguish.
The lower frequencies are
grouped together in resonance bands extending from approximately 147 Hz to 58 8 Hz.
Two other areas of strong resonance
occur at approximately 1,617 Hz and/ 2,940 Hz.
The first
formant of the vowel would be in the: lowest group of
resonances at a frequency of approximately 441 Hz.
The
second vowel formant would be in the. resonance area at a
frequency of approximately 1,617 Hz:..
178 - 208 - ClB - eh - D - 0: 4, ID"- 198.5 (p. 158).
This spectrogram exhibits ah unbroken string of frequency bands from approximately 147" Hz: to well above 3,500
Hz.
TftiLs indicates that all the_ frequencies in the spectrum
between. 147 Hz: and 3,500 Hz were-, audible.
This configura-
tion would closely fit thecfefinition of "noise," as opposed
to musical sound.. There is a resonance peak at approximately
88
441 Hz which is stronger than the surrounding resonance peaks.
This frequency apparently serves as Formant 1 for the vowel.
Another strong resonance peak at approximately 1,617 Hz is
apparently the second formant for the vowel.
25 - 184 - JIB - ee - C - 0 5 9 - 191 (p. 147).
Four strong resonance bands are closely clustered
between 131 Hz and 524' Hz on this spectrogram.
Another
group of low intensity frequency bands begins at approximately 1,562 Hz and continues to above 3,000 Hz.
The lower
resonance bands are high in intensity, but are not separated
by white area.
The higher resonance bands are separated by
white area, but are of low intensity.
The third resonance
peak, at approximately 39 3 Hz, is apparently the first
formant of the vowel.
Formant 2 is apparently at a fre-
quency of approximately 1,824 Hz.
169 - 155 - ClB - e e - C - 0 0 1 4 - 215.5 (p. 173).
This spectrogram exhibits a strong black band of
resonances from the fundamental frequency of approximately
131 Hz to 524 Hz.
Another cluster of lower intensity
resonance bands begins at approximately 1,690 Hz and extends to well above 3,500 Hz.
The area devoid of frequency
bands between 524 Hz and 1,69 0 Hz is characteristic of the
ee (i) vowel.
Although this example was not aurally pleasing
to the panel of experts, the vowel sound apparently was
distinguishable.
89
41 - 131 - JIB - a h - C - 0 4 1 -.198.5 (p. 156).
• A mass of frequency bands extends from the fundamental,
at approximately 131 Hz, to approximately 1,049 Hz.
The
area from this point to approximately 2,600 Hz is devoid of
resonance.
Beginning at approximately 2,600 Hz, there are
traces of frequency bands up to approximately 3,000 Hz.
The
strongest of the resonance peaks in the lower frequencies
comes at approximately 524 Hz.
at approximately 1,04 8 Hz.
Another resonance peak comes
These two resonance peaks
apparently serve as the distinguishing formants for the vowel
ah (a).
185 - 146 - C1B - ah - C - 1 2 11. - 201 (p. 159) .
In this spectrogram there is also a: mass of frequency
bands in the lower range.
In the Type. 1." display, none of
the frequency bands are distinguishable" as resonance peaks.
The Type 2 display indicates that certain.frequency levels
are more intense than others.
The range: of the mass is from
approximately 131 Hz to approximately 11,431 Hz.
From this
point to approximately 2,34 8 Hz the: spectrogram is devoid of
frequency bands.. Light gray traces: begin.: at 2,348 Hz and
extend to. above 2, 00.0 Hz.
It is not- possible to determine
which of the resonance peaks in the lower: frequency mass
serve as the distinguishing formants: fori the vowel sound.
Female voices alone.—The last example to be discussed
in the "Poor" category is one of female voices alone.
This
90
is the only example of female voices alone which was rated
as "-Poor" by the jury of experts.
145 - 16 - C1G - ee - C - 1 1 12 - 207 (p. 165).
This spectrogram exhibits two strong resonance bands.
The first resonance band is the fundamental frequency of
approximately 261.5 Hz.
The second strong resonance band is
at approximately 523 Hz.
A resonance band at approximately
784.5 Hz appears intermittently.
The area between 784.5 Hz
and 2,092 Hz is devoid of resonance bands.
An area of weak
tracings of resonance bands begins at approximately 2,092 Hz
and continues to above 4,000 Hz.
The fundamental frequency
of 261.5 Hz serves as the first formant for the vowel ee (i).
A relatively strong resonance band at approximately 2,353 Hz
apparently serves as Formant 2 for the vowel.
Although this
example was categorized as "Poor" by the panel of experts,
the spectrographic display is not significantly different
from examples in the "Good" and "Acceptable" categories.
The distinguishing characteristics of the acoustical
properties of the examples categorized as "Good" are consistent within the category.
The resonance bands of fre-
quencies are clearly and cleanly" aligned.
No traces of lower
intensity frequency bands occur between the resonance bands.
The resonance bands follow in each case the pattern of the
natural harmonic series.
The frequencies of the resonance
bands were sequential multiples of the fundamental frequency
in the order 1, 2, 3, 4, 5, etc.
The vowel formant
91
frequencies in each case coincided with the frequency of a
resonance band in the natural harmonic series.
The distinguishing characteristics of the acoustical
properties of the examples in the "Poor" category were also
consistent within the category.
The resonance bands were
obscured and diffused by lower intensity frequency bands
which were not a part of the natural harmonic series.
In
practically all instances conflicting frequency bands were .
present.
These conflicting frequency bands were usually as
much as one-half the intensity of the resonance bands which
were aligned with the natural harmonic series.
Despite the
fact that prominent resonance peaks were visible, it was
often not possible to determine which ones coincided with
the distinguishing formants~ of the vowel.
The category which was termed "Acceptable" could have
been more appropriately termed."neither Good nor Poor."
The acoustical characteristics, of these examples are in
some instances similar to those of the examples categorized
as "Good."
In other instances; the characteristics of these
examples are similar to those: examples categorized as "Poor."
There are apparently no distinguishing acoustical characteristics for the examples categorized as "Acceptable."
It is
apparent that the definitive; acoustical differences occur
between th.e examples categoxizied, as "Good" and the examples
categorized as "Poor."
92
Common to all examples was the change which took place
in the higher frequencies following the attack of the tone.
In each example there were fewer acoustical factors present
at the instant of the attack than there were in the time
duration which followed.
The fundamental frequency and the
first harmonic overtone were present at the attack.
After
the attack of the tone, the other acoustical factors—voice
pitch overtones, vowel formants, resonance formants, etc.-occur in a sequence from low frequency to high frequency.
CHAPTER V
SUMMARY, FINDINGS, CONCLUSIONS,
AND RECOMMENDATIONS
Summary
The purpose of this study was to analyze the acoustical
properties of choral sound.
Several studies, which are
mentioned in previous chapters, have dealt with the acoustical analysis of the solo singing voice.
The subjects of
these studies were concerned with vowel formants theories,
vowel modification, "brightness" and "darkness" qualities of
tone, voice quality, etc.
Heretofore, no study has been
concerned with the spectrographic analysis of choral singing.
Because of its utilization in-previous studies involving the acoustical analysis of sound:, the sound spectrograph
was selected for use in this study as the analyzing instrument.
This study sought to determine: the potential use of
the sound spectrograph in the analysis^ of choral sound.
In
research of the solo voice, the sound: spectrograph, as an
analyzing instrument, has led to gxeater.-understanding of
the singing voice.. The information-, gained, from this research
has been of considerable benefit to: those who teach singing.
The information gained from the spectrographic analysis of
93
94
choral sound should be of further benefit to those who
teach singing, both solo and choral.
The blending of voices in a choral group is of particular
concern to both the choral director and the voice teacher:
to the choral director because vocal blend is a necessity to
good choral sound, to the voice teacher because of his concern with the vocal development of the individual student.
Information which can shed light on the problem of blending
voices should be of benefit to the teaching profession.
It is thought by many choral authorities that the vowel
sound is one of the most important factors in the achievement of good choral blend.
vowel sounds:
This study concentrated on three
ee (i), eh (e) , and ah (a).
These vowels are
common to all modern Western languages, and are three of the
five primary vowels most often used in the teaching of singing.
In order to explore the problem of blending of voices
at different developmental stages, three educational settings
were utilized.
At the junior high school level the voices
are in various stages of change in range and vocal timbre.
At this level few, if any, of the choral students are receiving instruction in solo singing.
At the senior high
school level, the voices are beginning to develop toward
maturity.
In high school many of the choral students begin
to receive instruction in solo singing.
At the college
level most of the voices are physically mature and most of
95
the choral students are receiving private instruction in
solo singing.
These three educational levels were given
equal attention in this•study.
To simplify the tonal aspects of the analysis, no
harmonies were included in the experiment.
The examples
which were gathered for the sample were structured to be
either of unison or octave sounds.
The evaluation, portion of the experiment was performed
by a panel of choral authorities.
Although it is apparent
that a diversity of criteria were utilized in the evaluation,
the application of the criteria was consistent.
Different
trends were apparent in the ratings of the examples.
One
member of the panel rated very few of the examples as either
"Good" or "Poor."
"Acceptable."
He. instead rated most of the-examples as
On the other hand, another member, of the
panel tended to rate all of the examples as either "Good" or
"Poor," and rated only a few as "Acceptable:."
A".third
member of the panel rated very few of the. examples as "Good."
This judge rated over one-half of the examples: as, "Poor," and
only 2.3 percent as "Good."
However, inr the: two. evaluation
sessions, the members of the panel gave the., same: rating to
the same example approximately two out of three times.
The results of tfr£t evaluation indicate: that the ratings
were equally distributed" among the three instructional levels
in the sample.
In each of the categories—"Good,"
"Acceptable/" and "Poor"—each of the educational, levels
96
received approximately one-third of the total number of
ratings in the category.
At all three educational levels the examples sung by
the female voices were rated the highest.
These examples
were consistently rated higher than those sung by the male
voices alone, or those sung by the male and female voices
together.
At all educational levels the eh (e) vowel was
rated lower than the other vowels.
In the factor of the identification of the intended
vowel sound, the accuracy of the judges was quite high.
It
should be pointed out however, that the experiment was not
designed to be particularly sensitive in this factor.
Since
the three vowel sounds used in the study are relatively dissimilar in sound, there should have been little difficulty
in the correct identification of the intended vowel.
It is
worthy of mention therefore, that some of the vowel sounds
were not correctly identified.
The highest percentage of
vowel examples which were not correctly identified were sung
by the female voices.
Findings
In the spectrographic analysis, certain aspects appear
to have significance.
The most prominent aspect seems to be
one of tonal clarity.
The major findings of this study were
1.
Of the examples of choral sound categorized as
"Good," the analysis indicated that all the sound which was
97
present was concentrated into frequency bands which were
exactly aligned with the natural harmonic series.
All of
the known acoustical factors--vowel formants, resonance
formants, the fundamental voice pitch and its overtones—
were all tuned to the same tonal pattern.
The spectrograms
show this tonal pattern to be clear and distinct, without
deviation in frequency.
The resonance bands appear in
multiples of the fundamental frequency in the order of 1, 2,
3, 4, 5, 6, etc.
The intensity of the frequency bands
gradually decreases as the frequency increases.
2.
Of the examples of choral sound categorized as
"Poor," the analysis indicated that there were sounds
present which were not aligned with the natural: harmonic
series.
The analysis indicated that the components of the
natural harmonic series were discernible as peaks of
resonance which were of greater intensity than the other
frequency bands.
However, the lower intensity frequency
bands combined with the higher intensity frequency bands
gives an impression. Q£ acoustical confusion.. The frequency
bands of the natural harmonic series werei present with
considerable intensity,, but were not clear:; and/ distinct, or
free of frequency deviation.
The cross-section:, or Type 2,
display indicated that the higher f requenciess maintained a
high level of intensity.. This intensity • level: is significantly higher than that which was indicated for the examples
categorized as "Good..,"
98
3.
Of the examples of choral sound categorized as
"Acceptable," the analysis indicated that characteristics of
both the "Good" and the "Poor" categories were evident in
variable degrees.
4.
Common to all examples was the basic acoustical
alignment of the natural overtone series.
In the examples of
good vocal blend, the alignment was clear and precise, with
all acoustical factors tuned to the same frequencies.
In
the examples of poor vocal blend, the alignment was unclear
and obscured.
Acoustical factors which were not tuned to
the fundamental frequency were present in considerable intensity.
5.
Of the twenty-seven examples of choral sound which
were considered in the "Good" category for the spectrographs analysis
(a) By voice combination,. the largest number were of
female voices alone.
Of the twenty-seven examples, twenty-
three were of female voices alone, three were of male voices
alone, and only one example was of male and female voices
combined.
(b) By educational level, the junior high school level
had a slightly larger number.
Of the twenty-seven examples,
eleven were of the junior high school level, nine were of
the college level, and seven were of the senior high school
level.
(c) By vowel sound, the smallest number of examples
were of the eh (e.) vowel.
Of the twenty-seven examples,
99
thirteen were of the ee (i) vowel, twelve were of the ah (a)
vowel, and only two were of the eh (e) vowel.
6.
In the tabulation of the total number of "Good"
ratings given by the judges
(a) By voice combination, the largest number of "Good"
ratings were for female voices a] one.
Of the total of 770
"Good" ratings which were given, 432 were for examples of
female voices alone, 187 were for male voices alone, and 151
were for male and female voices combined.
(b) By educational level, the ratings were evenly
distributed.
Of the total of 770 "Good" ratings given,
263 were for the junior high school level, 269 were for the
senior high school level, and 238 were for the: college level.
(c) By vowel sound, the smallest number, of "Good"
ratings were for the eh (e) vowel.
Of the 770. "Good"
ratings which were given., 283 were for the- ah (a) vowel,
280 were for the ee (i) vowel, and 207 were: f:or::the eh (e)
vowel.
7.
Of the twenty-eight examples of choral." sound which
were considered, in the "Poor" category for the: spectrographic
analysis
Ca) By voice combination, the largest; number, were of
male voices alone.
Q£ the- twenty-eight examples.-, sixteen
were of male voices alone,, eleven were of" male-: and female
voices combined, and only one example was of female voices
alone.
100
(b) By educational level,- the largest number of examples
were of college level.
Of the twenty-eight examples,
thirteen were of the college level, ten were of the junior
high school level, and five were of the senior high school
level.
(c) By vowel sound, the number was evenly distributed.
Of the twenty-eight examples, ten were of the ah (a) vowel,
nine were of the ee (i) vowel, and nine were of the eh (e)
vowel.
8.
In the tabulation of the total number of "Poor"
ratings given by the judges
(a) By voice combination, the smallest number of "Poor"
ratings were for female voices alone.
Of the total of
1,037 "Poor" ratings which were given, 431 were for male
voices alone, 427 were for male and female voices combined,
and only 179 were for female voice alone.
(b) By educational leve, the ratings were fairly
evenly distributed.
Of the total of 1,037 "Poor" ratings,
366 were for the junior high school level, 354 were for the
college level, and 317 were for the senior high school
level.
(c) By vowel sound, the largest number of "Poor"
ratings were for the eh (e) vowel.
Of the 1,037 "Poor"
ratings which were given, 396 were for the eh (e) vowel,
325 were for the ah (a) vowel, and 316 were for the ee (i)
vowe 1.
101
9.
In the adjudication of the sample, it was apparent
from the tabulation of the ratings that a diversity of
criteria were utilized in the evaluation of the examples.
However, in the two evaluation sessions, the members of the
panel gave the same rating to the same example 63.3 percent
of the time.
The reliability coefficient for the panel,
computed by the Pearson's Product-Moment method, was r = .56.
According to Borg, this correlation is statistically significant beyond the 1 percent level (1, p. 28 3) .
Conclusions
The conclusions derived from the findings of. this study
indicate the following:
1.
The concept of unity of vowel sound is essential in
the achievement of good vocal blend in choral sound.
2.
The problem of vowel unity in choral sound is a
problem of intonation of vowel formant frequencies.
3.
In order to achieve good vocal blend in.choral
sound,"all of the acoustical factors in the chorallsound must
be aligned with a common, natural harmonic sexi.es:..
It has been borne onxt by acoustical research' that the
sound of a vowel consists: a£ two. distinct: frequency
resonances, referred to as the distinguishing: formants.
The
vowel sound is distinguished by the relationship;: of: the frequency levels of the two. fbrmants.
102
In Chapter I, average frequency ratings were reported
for the distinguishing formants of several vowel sounds.
Research indicates that .when the frequencies of the distinguishing formants occur at 270 Hz and 2,290 Hz, the
vowel sound would be aurally perceived as ee (i).
If the
lower formant. increases in frequency to 390 Hz and the upper
formant decreases in frequency to 1,990 Hz, the vowel sound
would be aurally perceived as ih (I).
Further, if the lower
formant were to move up to 530 Hz and the upper formants
were to move down to 1,840 Hz, the vowel sound would be perceived as eh (e).
From the evidence it would appear that the influential
factors involved are (1) the rate of frequency of each vowel
formant, and (2) the distance, in terms of Hz, between the
two formants.
If the frequency rate of one formant increases
as the other decreases, or vice versa, the distance between
the formants changes.
In this case aural perception would
indicate that the vowel sound had changed from one identifiable vowel to another.
As an example, from eiT (i) , 270 Hz
and 2,290 Hz, to eh (e), 530 Hz and 1,840 Hz.
If the fre-
quency rate of both formants increased, or decreased, at the
same rate, the distance between the formants would remain the
same.
In this case, the aural perception would indicate
that the vowel sound remained the same vowel, but changed ih
what is subjectively called "color."
For example, an ee (i)
vowel of 330 Hz and 2,350 Hz would be "brighter" than
103
average.
An ee (i) vowel of 210 Hz and 2,230 Hz would be
"darker" than average.
However, because the relative
positions of the formants are the same, both sounds would
most likely be perceived as ee (i).
Acoustical research also indicates that voice pitch
frequencies and vowel formant frequencies are independent.
In Chapter I it is pointed out that changes in formant frequencies are due to changes in the resonating cavities of the
vocal tract, and that voice pitch changes are due. to changes
in the positions of the vocal chords.
Voice pitch fre-
quencies can be changed without affecting vowel formant frequencies.
Vowel formant frequencies can be changed without
affecting voice pitch frequencies.
In a choral, group, each
individual produces a voice pitch with the resultant overtones, and a vowel sound with the distinguishing formants.
The problem of vocal blend is one of achieving: the. best
combination of these acoustical factors.
In order to achieve good vocal blend within-, a: choral
group, all the acoustical factors in the choral! sound must
be aligned with a common harmonic series.
rrr oxder:. to pro-
duce a common harmonic series for voice pitch,, each individual in the choral group must sing a pitch.which.is
correctly in tune with, the desired frequency.. Theiovertones
produced with the voice pitch must be of like;, frequencies,
and should be similar in number and in intensity.. The factors
of the voice pitch and its overtones should be; effectively
104
controlled by striving for unity of duration, pitch intonation, and loudness.
In order to effectively align the distinguishing
formants of the vowel, it is necessary that each individual
in the choral group sing a vowel sound which will result in
distinguishing formants of like frequencies.
An ee (i)
vowel which is "brighter" than average and an ee (i) vowel
which is "darker" than average will most likely not result
in distinguishing formants of like frequencies.
Only by
producing all vowel formants of the same frequencies can
the exact alignment of the formants take place.
Only if the
individuals in a choral group sing a like vowel sound can
all of the vowel formants be of the same frequency.
Unity
of vowel, then, would be achieved within a choral group by
all the individuals singing a vowel sound which would result
in distinguishing formants of .the same frequencies.
The problem of the alignment of the distinguishing
formants of the vowel in choral sound is further complicated
if the formants of the vowel must be aligned with the natural
harmonic series of the voice pitch.
The spectrographic
analysis in this study indicated that in the examples of
good vocal blend, the vowel formants were aligned with the
natural harmonic series of the fundamental pitch.
This would
seem to indicate that in those examples the individuals in
the choral group sang a vowel sound which resulted in vowel
formant frequencies which coincided with voice pitch
105
frequencies in a common harmonic series.
The spectrographic
analysis indicated that in the examples of poor vocal blend
many acoustical factors were not aligned with the natural
harmonic series of the fundamental pitch.
This would seem
to indicate that in those examples the individuals in the
choral group sang vowel sounds which resulted in vowel
formant frequencies which did not coincide with voice pitch
frequencies.
In order to achieve good vocal blend, the individuals
within a choral group should sing vowel sounds which will
result in correct intonation of the distinguishing vowel
formants.
In order to achieve correct intonation of the
distinguishing vowel formants and voice pitch frequencies,
the vowel sounds should be modified in "color" to achieve
alignment with the natural harmonic series of the: voice
pitch.
Recommendations
Based upon the findings and conclusions of' this study,
it is recommended that teachers of choral singing; be equally
concerned with intonation of voice pitch factors; and vowel
formant factors.
Acoustical research indicates; that vowel
formants are audible frequencies which can be; adjusted in
pitch by modification of the vowel sound.
As: audible fre-
quencies, the vowel formants can be tuned to a desired frequency in much the same way as the voice pitch.
106
Vowel unity should be presented as a basic concept in
the teaching of choral singing.
The concept of vowel unity
is not one of matching a model vowel, or of matching a model
tone sound.
The concept of vowel unity is one of intonation.
It is recommended that further research into aspects of
choral singing utilize the potential of the sound spectrograph.
Based upon the> findings and conclusions of this
study, the following questions are proposed:
1.
Are certain frequencies of voice pitch more
adaptable to certain vowels than others in the achievement
of choral blend?
2.
If the concept of vowel unity is one of intonation,
how can it be best taught as a basic concept in choral
singing?
3.
Are acoustical factors of overtones and difference
tones a problem of vocal blend when male and female voices
are combined in a choral group?
4.
What effect does the singing of harmony, as opposed
to unison or octaves, have upon the intonation of vowel
formants and the blend of choral sound?
5.
What are the acoustical characteristics of bright-
ness and darkness qualities in choral sound?
6.
What is the effect of brightness and darkness
qualities upon the intelligibility of vowels in choral
sound?
107
7.
Can good vocal blend be achieved more readily with
female voices than with male voices?
ing acoustical factors?
What are the contribut-
What are the contributing subjective
factors?
8.
Can good vocal blend be achieved more readily on
voice pitches in the mid-range than on more extreme high or
low pitches?
What are the contributing acoustical factors?
What are the contributing subjective factors?
9.
Can good vocal blend be achieved more readily on
certain vowels than on others?
acoustical factors?
What are the contributing
CHAPTER BIBLIOGRAPHY
1.
Borg, Walter R., Educational Research, An Introduction,
New York, David McKay Company, Inc., 19 63.
ldfl
APPENDIX A
EXPLANATION:
This study is being conducted to determine, by means of
spectrographic analysis, the distinguishing characteristics
of the acoustical properties of choral sound in relationship to subjective evaluation of vocal blend. This jury has
been assembled to make, individually, a subjective judgment
as to the quality of the vocal blend of each example presented.
The examples are pre-recorded from choirs of five different
school settings, including junior high school, senior high
school, and college. The vowel sounds, ee, eh, ah, were
sung in unison on notes of the middle range C major scale;
in combinations of female voices alone, male voices alone,
and male and female voices together. The examples are
completely randomized as to pitch, vowel sound, voice combination, and school Level.
INSTRUCTIONS:
Each example will be two seconds in duration, followed by an
eight-second pause.. Please evaluate, according to your own
criteria, the quality of the vocal blend of each sound presented. Indicate your rating on the adjudication, sheet by
circling the appropriate vowel symbol in the- appropriate
column.
The entire sample is thirty-six minutes in duration. To
establish reliability the sample will be evaluated twice.
A short rest period will be taken at the mid-point of the
sample and between the two hearings. If at: any point more
time is needed for scoring, the tape can be; stopped
momentarily.
110
Judge
BLEND:
Hearing
I - Good;
II - Acceptable;
I
II
III
ee
ee
eh
'
III - Poor
I
II
ee
ee
ee
ee
eh
eh
eh
eh
eh
ah
ah
ah
ah
ah
ah
I
II
III
I
II
III
ee
ee
ee
ee
ee
ee
eh
eh
eh
eh
eh
eh
ah
ah
ah
ah
ah
ah
I
II
III
I
II
III
fee
ee
ee
ee
ee
ee
eh
eh
eh
eh
eh
eh
ah
ah
ah
ah
ah
ah
I
II
III
I
II
III
ee
ee
ee
ee
ee
ee
eh
eh
eh
eh
eh
eh
ah
ah
ah
ah
ah
ah
I
II
III
I
II
III
ee
ee
ee
ee
ee
ee
eh
eh
eh
eh
eh
eh
ah
ah
ah
ah
ah
ah
I
II
III
I
II
III
ee
ee
ee
ee
ee
ee
eh
eh
eh
eh
eh
eh
ah
ah
ah
ah
ah
ah
-
' III
APPENDIX B
THE SAMPLE
The catalog numbers were assigned by a systematic
method in the following priorities:
female voices, male
voices, and male and female voices together; junior high
school, senior high school, and college.
number was assigned by a random method.
The sequence
The code symbols
for the school groups are J for junior high school, S for
senior high school, C for college; 1 for the first group
recorded, 2 for the second group recorded; G for female
voices, B for male voices, and M for male and female voices
together.
The symbols ee, eh, and ah, represent the intended
vowel.
When a vowel other than the. intended vowel was
marked by a judge,
the vowel symbol' and: the rating which
was marked are shown in parentheses.
The; pitch names repre-
sent the notes of an. ascending C majpx- scale.
The figures
in the rating columns are a summation off the ratings of all
the judges in. both, evaluation sessionsr.. The last column
gives rank of each, example in the sample:.. The ranks were
determined by th.e number of "Good," "Acceptable," and "Poor"
ratings of the judges.
Ill
112
Ratings of Judges
Catalocf Sequence! Schoo
Number Number
Group
1
2
3
4
Pitch
Name
II
III
7
6
4
3
2
2
5
2
67,.5
41
50
18
18
7
18
18
Rank
93
200
7
58
JIG
JIG
JIG
JIG
' 5
6
7
8
33
66
95
197.
JIG
JIG
JIG
JIG
[ ee
I ee
Iee(eh)
lee(eh)
G
A
B
c
9
10
6(3)
9
3
3
3
3
2
1
2
1(1)
9
10
11
12
166
47
86
127
JIG
JIG
JIG
JIG
eh
eh (ee)
I eh(ah)
eh
C
D
E
F
5
6 (1)
4
3
3
9
4
6
137. 5
4
35. 5
0(1)
89
3
35. 5
13
14
15
16
79
78
142
145
JIG
JIG
JIG
JIG
I eh
! eh
jeh
eh
G
A
B
c
6
6
7
10
6
3
2
3
2
5
5
1
41
48
38
7
17
18
19
20
75
108
41 .
158
JIG
JIG
JIG
JIG
ah
ah
ah
ah
C
D
E
F
5
6
9
12
7
5
5
2
2
3
0
0
67.5
44.5
JIG
JIG
JIG
JIG
ah
[ ah
ah
! ah
G
A
B
c
7
9
9
12
7
5
5
2
0
0
0
0
11
11
ee
ee
ee
ee
C
D
E
F
5
6
5
9
11
2
21
22
23
24
77
182
104
124
25
26
27
28
184
105
12
202
JIB
JIB
JIB
JIB
ee
i ee
I ee
ee
C
D
E
F
0
0
2
1
5
6
6
7
9
8
6
5
191
179
148. 5
127
29
30
31
32
37
22
154
59
JIB
JIB
JIB
JIB
ee
ee
ee
ee
G
A
B
C
1
0
0
2
6
6
4
10
7
8
10
2
159.5
179
198.5
96
33
34
35
36
11
68
128
212
JIB
JIB
JIB
JIB
eh
eh
eh
eh
C
D
E
F
3
6
1
2
7
7
8
5
4
1
5
7
121
63.5
114.5
155.5
29.5
2
113
Catalog Sequence School Vowel
Number Number
Group Sound
Ratings of <
Judges
Pitch
Name
I
II
III
Rank
37
38
39
40
56
102
106
40
JIB
JIB
JIB
JIB
•eh
eh
eh
eh
G
A
B
c
0
1
0
2
3
4
3
7
11
9
11
5
204
185.5
204
127
41
42
43
44
131
167
159
85
JIB
JIB
JIB
JIB
ah
ah
ah (eh)
C
D
E
F
0
1
2
2(1)
4
9
8
4
10
4
4
7
198.5
102. 5
109
152.5
45
46
47
48
123
157
176
88
JIB
JIB
JIB
JIB
ah
ah
ah
ah
G
A
B
C
1
0
0
1
6
7
4
6
6
7
10
7
148. 5
163.5
198.5
159.5
49
50
51
52
60
31
191
170
JIM
JIM
JIM
JIM
ee
ee
ee
ee
C
D
E
F
0
1
2
2
2
9
8
8
12
4
4
4
209
102. 5
109
109
53
54
55
56
136
147
175
178
JIM
JIM
JIM
JIM
ee
ee
ee
ee
G
A
B
C
5"
3:
2:
4,
3
7
6
6
6
4
6
4
137. 5
121
148.5
60
57
58
59
60
80
112
164
3
JIM
JIM
JIM
JIM
eh
eh
eh
eh
C
D
E
F
2.
o:
2..
3:
2
3
10
9
10
11
2
2
194
204
96
92.5
61
62
63
64
196
46
211
181
JIM
JIM
JIM
JIM
eh
eh
eh
eh
G
A
B
c
2:
0.
3:
i:
4
6
8
6
8
8
3
7
168. 5
179
81
159.5
65
66
67
68
61
213
179
140
JIM
JIM
JIM
JIM
ah
ah
ah
ah
-C
D
E.
F
3:
2:
3:
4,
3
4
7
6
8
8
4
4
166
168. 5
121
60
69
70
71
72
82
27
115
28.
J"lM
JIM
JIM
JIM
ah
ah
ah
ah
G.
A.
B
c:
I:
n
I:
2:
9
0
4
4
4
13
9
8
102. 5
211 .
185. 5
168. 5
ah
114
Ratings of Judges
Catalog Sequence School Vowel
Number
Number Group Sound
Pitch
Name
I
II
III
Rank
73
74
75
76
72
18
99
39
S1G
S2G
SlG
S2G
ee
ee
ee
ee
C
D
E
F
5
9
1
9
7
4
12
4
2
1
0
1
67.5
14.5
98
14.5
77
78
79
168
163
44
84
SlG
S2G
SlG
SlG
ee
ee
ee
ee
G
A
B
C
3
6
5
10
7
6
8
3
4
2
2
1
121
41
75.5
7
204
24
103
107
S2G
SlG
S2G
SlG
eh (ee)
eh
eh
eh
C
D
E
F
5
3
7
5
5
5
5
6
3(1) 51.5
6
144
2
33
3
55
S2G
SlG
S2G
SlG
eh
eh
eh
eh
G
A
B
88
193
187
169
129
c
6
5
7
1
2
6
2(1)
8
6
3
4
5
49
55
35.5
114.5
89
90
91
92
90
148
30
10
SlG
S2G
SlG
S2G
ah
ah
ah
ah
C
D
E
F
0
6
4
8
7
5
9
6
7
3
1
0
163.5
44.5
89
23
93
94
95
96
87
156
96
53
SlG
S2G
SlG
S2G
ah(eh)
ah
ah
ah
G
A
B
C
4
10
0
12
1
0
3
0
89
4.5
99
2
97 •
98
99
100
43
152
52
71
SIB
S2B
SIB
S2B
ee
ee
ee
ee
C
D
E
F
2
5
0
6
8
7
1
8
4
2
13
0
109
67.5
212
73.5
101
102
103
104
109
126
122
14
SIB
S2B
SIB
S2B
ee
ee
ee
ee
-G
A
B
c
1
7
4
6
5
7
6
8
8
0
4
0
173
29. 5
60
73.5
105
106
107
108
83
295
23
153
S2B
SIB
S2B
SIB
eh
eh
eh
eh
C
D
E
F
4
0
4
3
4
3
7
8
6
11
3
3
140
204 •
71. 5
81
80
81
82
83
84
85
86
87
8 (1)
4
11
2
115
Catalog Sequence School Vowel
Number Number
Group Sound
109
110
111
112
174
38
Ratings of ,
Judges
Pitch
Name
I
II
III
Rank
25
S2B
SIB
S2B
SIB
eh
eh
eh
eh
G
A
B
c
3
1
5
0
5
7
6
6
6
6
3
8
144
130.5
55
179
113
114
115
116
210
150
125
69
SIB
S2B
SIB
S2B
ah
ah
ah
ah
C
D
E
F
3
9
0
6
4
7
3
2
2
12
3(1) 4
152.5
18
209
46.5
117
118
119
120
6
190
91
SIB
S2B
SIB
S2B
ah
ah
ah
ah
G
A
B
c
1
7
0
6
4
4
5
4
9
3
9
4
185.5
35.5
191
46.5
121
122
123
124
73
134
116
42
S2M
SIM
S2M
SIM
ee
ee
ee
ee
C
D
E
F
2
1
2
1
5.
8
7
9
7
5
5
4
155. 5
114.5
127
102.5
125
126
127
128
120
172
141
1
S2M
SIM
S2M
SIM
ee
ee (eh)
ee
ee
G
A
B
c
4
5
3.(1) 4
5
5.
3:
10
5
6
4
1
136
140
51.5
94
129
130
131
132
81
132
203.
198
SIM
S2M
SIM
S2M
eh
eh
eh
eh
C
D
E
F
1_
5
T
6
9
8
5
8
4
173
127
179
102.5
133
134
135
136
51
2
20 €
29
SIM
S2M
SIM.
S2M
eh
eh
eh
eh
'G
A
B
c
li
2.
11
2:
6
51
4:
8.
7
7
9
4
159. 5
155.5
185.5
109
137
138
139
140
160
34
55
70
S2M
SIM
S.2M:
SIM
ah
ah
ah
ah
, C
D
E
F
55
o:
31
li
T
6.
5:
5:
2
8
6
8
67.5
179
144
173
141
142
143
144
13
9
19
94
S2M'
aiM.
S2M:
SIM
ah
ah
ah
ah
G
A
B
51
o;
44
C
4T
8
51
T
6'
1
9
3
4
75. 5
191
71.5
60
62
162
2:
o;
li
116
Catalog Sequence School Vowel
Number
Number Group Sound
Ratings of Judges
Pitch
Name
I
145
146
147
148
16
5
161
61
C1G
C2G
C1G
C2G
ee
ee
ee
ee
C
D
E
F
149
150
151
152
192
26
165
118
ClG
C2G
ClG
C2G
ee
ee
ee
ee (eh)
G
A
B
c
8
8
3
7(1)
153
154
155
156
133
135
180
209
C2G
ClG
C2G
ClG
eh
eh
eh
eh
C
D
E
F
157
158
159
160
143
121
21
57
C2G
ClG
C2G
ClG
eh
eh
eh
eh
161
162
163
164
139
48
151
98
ClG
C2G
ClG
C2G
165
166
167
168
17
35
113
119
169
170
171
172
II
1
10
4
6
III
Rank
1
4
8
6
12
0
2
2
2.07
4.5
78
41
3
6
6
6
3
0
5
0
27
23
133.5
23
2
1
5
3
10
4
9
6
2
9
0
5
96
185.5
84.5
133.5
G
A
B
c
3
4
8
3
6
8
5
7
5
2
1
4
133.5
78
26
121
ah
ah
ah
ah
C
D
E
F
2
5
1
6
6
9
5
7
6
0
8
1
148.5
84.5
173
63.5
ClG
C2G
ClG
C2G
ah
ah
ah
ah
G
A
B
C
5
7
5
9
9
7
7
5
0
0
2
0
84.5
29.5
67.5
11
155
114
138
215
ClB
C2B
ClB
C2B
ee
ee
ee
ee
C
D
E
F
0
0
0
4
0
5
0
6
14
9
14
4
215.5
191
215.5
60
173
174
175
176
32
177
76
97
ClB
C2B
ClB
C2B
ee
ee
ee
ee
-G
A
B
c
3
5
4
9
5
6
9
5
6
3
1
0
177
178
179
180
194
208
188
101
C2B
ClB
C2B
ClB
(sh
C
D
E
F
2
0
1
0
5
4
8
7
7
10
5
7
«2h
.sh
«ah
144
55
89
11
155. 5
198.5
114.5
163.5
117
Catalog Sequence School Vowel
Number
Number Group Sound
181
182
183
184
110
185
64
74
C2B
ClB
C2B
ClB
185
186
187
188
146
149
216
207
189
190
191
192
eh(ee)
eh
eh
eh
Ratings of Judges
Pitch
Name
I
II
III
7(1) 168.5
1
32
5
127
6
140
Rank
G
A
B
C
2
7
2
4
4
6
7
4
ClB
ah
C2B
ah
ClB . ah (?)
C2B
ah
C
D
E
F
1
3
0
4
2
8
1
3
14.4
100
199
8
ClB
C2B
ClB
C2B
ah
ah
ah
ah
G
A
B
c
3
8
4
7
5
6
9
7
6
0
1
0
144
23
89
29.5
193
194
195
196
189
50
214
54
C2M
ClM
C2M
ClM
ee
ee
ee
ee
C
D
E
F
. 0
0
3
1
7
5
7
5
7
9
4
8
163.5
191
121
173
197
198
199
200
20
65
15
201
C2M
ClM
C2M
ClM
ee
ee
ee
ee
G.
A
B
c:
i;
l
6
5:
7
8
6
6
6
5
2
3
130.5
114.5
41
55
201
202
203
204
3:6:
83
4:
17,1
ClM
C2M
ClM
C2M
eh
eh
eh
eh
c;
D
E:
FT
i:.
I:
o:
o:
3
8
3
1
10
5
11
13
195. 5
114.5
204
213
205
206
207
2(78
ia&
67
4:5;
4.9
ClM
C2M
CIM
C2M
eh
eh
eh
eh
G"
A.
K
c:
0
i:
51
3;
9
4
9
7
5
9
0
4
106
185.5
84.5
121
209
210
211
212
183
13:7
92
172
C2M
ClM
C2M
ClM
ah
ah
ah
ah
~ c:
D:
E:
F~-
o:
o:
l.
3:
2
6
3
6
12
8
10
5
207
179
195.5
133.5
213
214
215
216
111
11.7
195
130
C2M
ClM
C2M
ClM
ah
ah
ah
ah
G;
A.
B:
C
I;
3:
4,
8
9
9
8
6
4
2
2
0
102.5
92.5
78
23
11
201
3
81
11(2) 213
7
151
APPENDIX C
This appendix contains sound spectrograms of the
examples categorized as "Good" by the panel of experts.
The spectrograms are arranged in rank order according to
the ratings of the panel.
Of the 216 examples in the sample,
27 were considered in the "Good" category, with ranks from
2 to 27.
Each spectrogram exhibits two types of spectrographic
display.
The Type 1 display shows pitch frequency vertically,
time duration horizontally, and intensity as shading between
gray and black.
The Type 2 display is a cross-section show-
ing frequency vertically and intensity horizontally.
The
frequency scale of the cross-section is inverted on the
spectrogram.
The cross-sections were made approximately at
the attack and 1.2 seconds later.
The cross-section is
shown on the spectrogram at the exact time spot which coincides with the Type 1 display.
Each spectrogram is identified by vowel sound, educational level, voice combination^ pitch frequency, and rank.
118
119
w
CD
o
•H
0
>
1 ' # /
im
a
CD
rH
rd
g
I!
•fi : 1 J. yft•
I :j:W<i
. 0)
m
m
o
o
x:
o
CO
w:
xj
tr>
•H
XI
;|i; |:ij!,!!;
CN
c
i
«x{-
c$
I
u.
?
0
<N
CN
1 I
I
£
3
•
W CN
^ m
fd o
!
X!
fd
1
>i
XI
(xj
4
'Mt
Ms
;
M
O
•H
£
3
xi
fd
:i
!- ;
Vjvs|
-
0
rH
00
to
Go
I/?
o
CNJ
I
!ilf
fi
#
u
ri^s
a
fd
&
rH
Q)
£
•
O —
> N
ffi
0)
o\
X2
4J ^
Co
m
o >i
H
g (1)
fd - P
M fd
tn g
O "H
U X
•P o
0 u
0) PU
O. Oi
w fd
<
1
r
w
fa
^
H
x:
u
-p
• -H
tn a ,
•H
fa d)
XJ
•P
d
o
120
Si!
H
CD
rH
fd
e
a)
IH
H
0
o
XX
u
w
XX
tr»
-H
A
CM
1
O
CM
CM
! 1
I
o
1
xx
fd
l
o
rH
h)
I
CM
r-|
|
CM
M
0
•H
£
•m •
CM
>i
X* ^
0
tn
fX 44
0 a
w fd
fd
•
rCj N
fd su
rH
0
£
0
>
CO
CM
in
>i
•—i
<D CD
43 +J
•P fd
£
m -h
0 X
^ 0
g u
fd p4
U Ok
tT> fd
0 ^
U
•P O
o
<D XX
04 O
w -p
•H
< a
i
i a>
m xj
rH -P
• CJ
tn o
•H
P4 W
<D
O
•H
O
>
121
CD
rH
fd
s
(D
IM
k I
m
'
.
rH
o
o
'
o
w
3
•H
&
U
0
CM
1
O
CM
CM
rH
*H
a
CD
to
•
CM
>i
^ U-i
0
tJ)
a rM
3 a
w fd
&
1
fd
o
*****
1
J2 N
Xi
fd
I
0
CM
CO
|
en
in
I
KO
a\
•
fd «
rH CO
cp cm
£ to
0
> >1
rH
a) cd
XI +J
- P (0
e
<+-l -H
0 X
~ o
6 M
m a<
&-J JH|
t p fd
o —
u
•M
o
a)
a
W
a
x:
o
-}J
•H
<i &
1 CD
VO XJ
rH - P
I
• CI
tr> 0
*H
w
CD
O
•H
0
>
122
T ~ " T
CD
r-i
(d
6
Q)
I
iiii!
S
I
•
H
O
0
r!l
•a
o
w
fell
xi
tn
• H
XJ
U
0
* H
a
.
(L) L O
w
.
in
•
illf
i f
J'
o
' i.
• •' v r
k
;ii
f
'
0
tn
a *•
s
a
t/) r d
I
&
t
xi
rd
r
:
o
CM
W
I
VD
in
i—i
I
'
f
til,
lilkii;
^
o
1—f
<
'l!
I
si'
•
s
>i
rQ
ri'
^r«
<j\
<d
n
--"' •
Xl N
cd w
«H O
0) ^
£^
0
K*
>>1
H
a> <D
Xl -P
- P rd
S
m - H
0
X
_
0
g
M
fd O i
M
0 ,
fcn r d
w
O
u
•P <
a
a> x ;
a * a
W -P
• H
Oi
i
\ <D
xi
H
-P
•
t?>
* H
P4
a
O
'
03
a)
o
• H
o
>
123
cd
- p
a
0
0}
CD
O
" H
o
>
ft 7.55
CD
H
S'i'fitl'i 1
•
it
'
I
rd
e
CD
M-|
0
&
a)
m
r H
•
H
0
f
o
'1
1
> 1
o
t n
a
f
F
.f
o
i
*
)l
»—1
3
to
i
•HI
Q
' Vlf r r i
i n
i
CD
•
[ CD
^
r H
m
CD
CD
CD
$ !'
0
£
1
0
M
CD
&
> £
O
CM
rd
U
JG
4J
1
m
IW
l
0
N
-
W
e
v o
cd
M
r H
cr>
tJ>CN
0
5
!
I
U
;
. 1
!'
-M
> i
r H
a
t'
cd
CD
+ J
a .
rd
w
s
,t.
<
X
1
0
t
M
0 0
C h
r H
Pli
rd
*
^
t n
• H
&
Q
Si
0
• P
• H
&
124
0)
rH
1
g
Q)
rH
0
0
i
&
o
w
.a
tn
•H
u
0
•H
£
rI
i—I
-m
•
r-
>i
JQ M-4
0
tn
£^
CO
3 a
(/) ftf
o
«H
<-~N
I
•r-f
^
i
<D
Q)
t
0
H
.
I Q) ' N
Id) w
i—1 o
0) ^
£ ^
0
>
>1
rH
CP 0)
-P
-M td
e
Xi
I
VD
I
KD
m -h
0 X
0
6 M
«d a *
M a,
t n rd
o —
u
•M
U
a) 42
04 O
CO -P
•H
|
a
1 0)
ov,C
r—1 - p
•
a
&> 0.
*H
'A};
P4 CO
a>
o
•H
n
125
(D
rH
td
S
a)
IW
rH
0
0
A
o
CO
4XT'
3
•H
id
U
0
I. ".¥|f-
r1
rH
CO
o
rH
1
o
1
XJ
<D
1
O
«—I
h>
1
in
i—i
i
«aH
>i
XI ^
0
tn
£
3 a
to fd
to
«w
g
*
*
XJ N-«x
cd a
i—1 CO
0) CM
£ LO
0
> >1
rH
0) <D
43 +>
-P fd
e
m -H
0 X
w 0
e u
fd Qa
KD u ou
rH tr>v fd
0 ~"
M
4oJ 0
CD
04 0
W -P
•H
<• Cu
m
1 0)
o 43
CNJ -p
• CJ
tr> O
•H
fa W
0)
U
•H
O
>
126
<D
H
fd
s
(1)
Ml
V-7.lilff'l
H
0
0
roCj
co
tr>
•H
m
U
0
•aH
r- 0)
w
r1
rH
0
00 tr>
a^
3 a
O
H
CO cd
1
O
<D N
1 I
ICU K
<D
CD rH 00
(1) CM
|
£ LO
0
O > >i
H
»—1
CO (1) CD
•a 4-»
d
i 4J f
6
m -H
00 0 X
_0
1 g u
fd a.
o M
a
fcn rd
00 o
—
•
'*
m
Mm
tm
ffl& 1
•
-P a
o
<u
Oi o
CO +3
•H
< Qa
1
1 <D
r~i >Cj
CNJ -P
•g
t
•HP o
en
Q)
O
•H
O
>
127
<D
—
i 1
fd
' i
E
(D
iW
n
i
\
rH
0
0
XI
O
0)
£5
J Ills
£ ''
Hr
tT>
XI
•H
1—1
H
\
o
;
i
LO
tn
Ch
1
W
0
•rH
a
0 •
•m rH
rH
>H
X* ^
0
a
3 a
a) fd
&
«"—N
«d
^
•
s
1
XI
rd
I
0
i—I
h>
|
H
1
G\
i—1
XI N
cd w
rH O
CD CO
£ co
0
> >i
rH
XI +3
-P fd
B
4H -H
CD CD
0 X
_ 0
e m
fd Oi
5H Ot
tP fd
0~
u
-P w
a
CP XI
a« o
W -P
-H
a*
4
h
:• j
-
• a
i
» <u
CM XI
CM -P
• a
&
•H
0
CO
<u
o
-H
0
>
128
'p y
I B,
1
Hi
m
CD
i—1
0
g5
mCD
i—!
0
0
o
w
I
j
'*M'
•H
u
0
•H
rH a
i—1 0
•nH
rH
f
>
1
o
0
m
tn
Cft 2 £
w &rd
i
< ^rds
l
,£ N
,a cd tn
fd
«—1 o
t <D <n}<
£^
o o
H
> >1
rH
h>
CD CD
I 43 -P
-P g(d
CM
00 U-4 -H
I—1 0 X
_0
I g u
fd a,
CM U 04
CM o
tr>—fd
•
.i'.\
!
!
1
i
H.
Hi?
s
1
m m
I
u
4J
O
<D ^
Qu U
W -P
•H
<l 04
mr ,<aD
CM -P
•a
t7> o
•Hi
W
CD
O
•r-f
0
>
129
2
{
!
.i
h,
.
?
4
%I
'•-J;.' 3
•i
?; i
| 1
> ;>
Vi"-:>y-.
•
.
^
<3)
H
fd
e
m0)
1—1
0
0
r.
&
X
o
,
t/3
ra
-H
&
u
0
! 1
5
ii
•rrl
rH CI
i—1 0 •
•nH
1
i—1
>i
£t
O
O
to tji
a^
<J\
3 £
in fd
&
1
m
fd
i
rC N
XI
fd w
<d
rH ^
i CD Ch
£^
o
0
H
> >1
rH
<D 0)
^ -p
-P fd
B
o
IH -H
rH 0 X
_ 0
1 6 M
fd a,
00 u pu
CsJ tn rd
o—
•p cn
D
0) ^
Oi o
CO -p
•rl
<1 Oi
i a)
^ ,a
CN| 4J
•a
tJ> 0
•H
w
a>
o
-H
o
>
130
a
0
w
C
oD
•H
0
>
a)
rdH
g
CD
IM
CD
CD
rH rH
i—1 i—1
0
1 o
o
>1
m
o\ a
3
1 w
a ^ H
fd rH
-l
-a -aMo
fd rd
M
f rH £
a) fd
0 &&
0
o4 0
>
t CD ^
& N
o\ -p a
rH m ro
i—1
0 CM
t gm
00 fd >i
V0 M rH
rH tr> (D
O +3
-MP f
gd
O
<D*HX
Oi 0
W M
On
< th
I fd
m1
CN o
• rCj
tP O
*H -P
Ot
CD
JCJ
-P
•*W»
131
CD
+3
a
o
XI
I
I
;
Yfiv6
fi h i
* f.
>
%
;'i
;
x.*,
CO
CD
O
0
>
1
>
:•
1
!
.
*
'
K'i i
|
i< i
J
1
!
Q)
H
rd
E
1
|
a)
tr>
1
<u
1
H
1 H
r—{
i ,—|
0
o
|
|
! 1
>1
i
JQ
| m
tn
1
a
| <J\ 3(0
I°
I
o
*H •
i—1
i—1
I
\Q)
CD l<D m
0)
0
•ii;
1
m
CNJ
u.
i
r(71
1
rH
•—i
a) ^
& a
o rd
> P4
CD
XI •
-p —
N
trs
0
to
B •
td1—i
U vo
tncN
0
U >i
-P i—l
•f ',
u a)
CD 43
Pi <d
to g
•H
X
[ 0
1 u
vo pu
CM &
•
ii H.i i! S|
rd
•H O
(x,
X!
O
4J
•H
p4
132
CD
rH
n3
S
CD
M-4
i—I
O
0
X!
O
w
* M
k
tn
•H
I
M
0
•
in• •CH to
Q) •
rH W ^
rH
1 £* ^
0
H
tP
C M
0 a
<J\ w rd
&
1
•H
P
I CD N
1 I
d) w
CD
<D rH ^
<D
1 £ CM
0
0 > >i
rH
CM
CA <D CD
JCJ -M
1 •P fd
6
00 m -h
H
0 o
X
e
m
1
04
a»
r- M
£r» fd
o
^
u
•P Q
o
<D &
04 O
CO -P
<|
C*OHi
1C
r,Dc;
CM -P
>—"
#
<r~s.
m 1
•it."
•
£
& 0
•H w
a>
o
•r)
£
133
CD
rH
rd
s
CD
m
4; |
\
\'l
\
vli
\
i—i
o
o
&
o
in
&
% -H
ty\
•H "
LO
•
i—1
1
i—I
<y\
0
•H •
£ to
CD «
W ^
rH
>1
rQ
0
tr>
£M
0 a
w fd
&
\
*H
t
CD
CD
(
0
(N
0)
1
cr>
00
\
VD
r-
S
la) n
\Q) ffl
i—i <ys
CD
<: ro
0
> >i
rH
CD CD
JCS -P
-P (tf
6
m »h
0 X
o
e M
co a.
M £X<
tr> rd
0 w
u
•P &4
0
cd .a
a* o
W -P
-H
«<
1
I CD
CO ,d
CM -P
• G
tx> 0
•H
Pn W
CD
o
•H
0
>
134
...m,
>!•
-y
0
rH
grd
0)
m
rH
0
0
o
w
rC
fcr>
•H
0
•H
a
2 •
•noo
rH
J>i
rQ M~l
tn 0
a ^
3 d
CO&ftf
00
rH
I(N
CO
CF>
1 • r-|
*w* #
|0) N
1 10
Q) rH CT*
CD a) ^
& 00
0
f
> >1
O
rH
H
Q) <
D
b ^ -P
-M g(tf
I
-H
CO 0 X
LO
0
E ^
1 fd a.
M 0,
otr>~rd
*—«•*
i
ill
I
!ls'.«! I
$
tSM'M
-p
a
o
Of o
CO -p
•H
< 04
1
1 <])
o\ xi
(N +J
•
G
tn O
•H
Cm W
(1)
O
*H
0
>
135
I7
ii
•X,
& ]
'.'I'!
R
V'
St
I.
%
•u
•U
S
<D
H
td
g
0)
iw
r-i
0
o
rod
w
.c
•H
M
0
•H
£
0 .
•moo
1—1
>1
A O
tr>
a rM
2 a
w &cd
00
1—I
1
CM
CO
as
i N«H
0 —' •
|<D N
I 10 K
<D i—1 CN
CD £
CDoc
or>
O
1 > ><
O
i—1
1—I &Q)+>CD
h)
.p g
fd
1 m -H
co 0 M
CO g 0
n
\
fd Oi
U Qj
in tn rd
0
-UP O
a
0) rtf
0* o
U) -P
< &
y.
I-
I
r a)
o ,d
CO -P
•a
tn o
•H
pM CO
(1)
U
•H
O
>
136
CD
rH
fd
6
a)
m
!
rH
f
o
o
jc;
u
co •
00
rH
i
CN
00
00
V£>
1
CQ
.
-H
,a
0
•H
a
3
•
•r^oo
i
i I
t>i
rQ
0
tr»
a ^
3 £
CO (tf
&
%
1
•H
^
&
0)
n
—'
•
1(1) N
Id) w
Q)
Q)
I—1 ^sf
0) cr>
\
o
O
rH
h)
1
LO
o\
\
r^
&^
£>"1
rH
a) <D
X! -P
4J
g
m -H
0 X
O
S M
«J Or
as
U On
t n cd
0 —
u
•P cq
o
<D XI
a O
CO *P
•H
1
i a)
rH
00 -P
• a
tJ* o
*H
CM W
0)
o
•H
0
137
CD
H
nj
5
0)
m
1
?
r—J
o
o
o
w
CO
, a
tr>
•H
rH
6
u
o
a
3
•
• m o o
r—|
CO
>1
r Q l-N
CT>
o
t p
a
m
a)
a
rd
&
0
O
CD
CD
CD
CD
to
1 0
K
rH
00
CD CM
& LO
o
> >1
0
i—I
r—i
<D
, £
a;
-P
-p
nj
E
H
r o
X
B
«d
8
a*
CO
(d
o
u
—
-P
u
0
0) X I
&
o
CO - p
•H
<
04
1
i
CD
<N X !
00 -p
•
t n
-H
a
O
CO
CD
o
•H
O
>
138
to
<D
0
•H
0
II:
Sill
ll
& IS s 1 f
>
fMU
i! i? $
<D
rH
gfd
I
rH
o0
x;
0
w
1
|
'fit,'
ill
•,'!|if
?ft"wI 1 « *
Kill"
cx>
rH
1
CM
CO
as
1
a
i
XJ
fd
l
m
CM
CO
1
o
in
rH
•H
XI
M
0
•H
a
CD
CO
>i
,Q
fcn .
a oo
3 rH
CO
U-i
—0
fd
rC £
fd fd
rH
a)
& •
o—
> N
|X|
<L)r-p ^
H
O >1
rH
g q)
fd -P
U fd
tp g
O -H
M X
-P O
0 U
<D Qa
Oa PU
w fd
II;!
Hill
MM
1Q
I
*0 x:
ro o
-P
• *H
tX* Oi
"H
d)
£1
-p
a
o
139
a)
rd
e
CD
. M-4
rH
0
O
a
U)
x:
&
•H
H
111
U
0
CO
CM
(
o
vo
00
r
fx*
I
rd
fd
\
o
CO
CM
i
o
H
1
CNJ
CTi
<ir jil
•H
a
cu
.
W CO
J>i
CN
X*
tji
0
£
3 £
co fd
&
rd
-"""V
fd trj
XJ N
r H CTi
(1) ^
£ ro
0
> >i
H
<D (1)
XI -P
-P fd
e
m -h
0 X
^
g M0
fd cu
M 04
tj> rd
0 ^
u
-P Pm
o
C D £l
Oi U
W -p
•H
&
I &<D
CO - p
•a
tn o
•H
Pn w
Q)
O
•H
O
>
140
11
i
a
0
w
CD
0
, -H
O
>
JI i
h I
IS 5S
!;
$ 'i
j
II
0)
I—1 .
d
g
0)
si
i |
u
a)
tr»
0)
rH
CO i—1
CM 0
u
1 >1
o &
VD *7>
00 3
W
1
^ 00
•H CM
<
1 Id)IW0
Q) Id)
M
<D
i 1 £
d
)
I & & fd
O 0
>
C
•
uM
a) —
aN
I ,
-p tc
vo
o
CM
o
^
|
&
o
fd >i
m U
&\ a)
i—I o
+->
u 2
-P 6
O -H
<U ^
0* o
to M
Pu
<1 Cm
m
oo
•
S
^
&
tP U
•rf -M
-H
O*
0)
•
•cps
141
I;!f
P"
£
0
il !
l1
CO
CD
O
•H
0
I
>
00
CM
1
o
KD
1f
*W
rIi
u
i
&
CD
CD
Q)
i
h
1
o
CM
U
CI)
i—1
<d
S
CD
m
CD
tn
(D
H
1—1
0
o
>1
&
u»
£
3
CO
•
^ co
•H
CM
—-
. ^
CD 0
[CD
34
r~i rJ
CD rd
£&
0
>
•
I
CO
1—1
rH
1
CM
if)
i—1
CD —
XI N
4-> ffi
<4-4 CO
0 CM
in
e
<d > i
M «H
cn a)
0 4->
M fd
•P 6
D -H
<D X
a. 0
CO M
< ! Oi
i fd
J ,w>
KD
ro O
• XJ
ti> a
•H -P
PM - H
04
CD
4~>
142
a)
-P
a
0
Xi
(/)
Q)
D
-H
0
i m i
... H > 5 • .
III
i
>
•1 tiiiii,
CD
rH
fd
B
a)
tr>
CO
CM
1
o
V0
oo
Q)
r—1
0
o
>1
XI
tn
3
t/}
1
^
1
Xi
rd
I
PQ
CM
u
1
O
O
rH
1
O
Cft
rH
.5: |
t•;' •
*
«
1
, t&?!h
,
aj oo
cm
M-i
fd 0
H ^
cu £
^ fd
0 &
>
Q)
•
X! —
-P N
ffi
M-l
0 o
CM
E CM
aj
M >i
fcj* rH
0 0)
M -P
-P fd
o g
(I) -H
Oi X
W 0
M
«< Oi
i &
1 fd
rro ^
.^
tnMH
•H 0"
1
1
* -*!)«•
U
-P
-H
CU
143
w
a
o)
•H
0
>
. I
<D
H
fd
e •
CD oo
IW CM
r^f U_j
a0
fd
a) a
rH fd
f
gd &
i: I
CO
CM
1
O
V£>
00
1
Q
i
1
rC|
fd
I
S
I—I
O
0
ro
H
1
vo
H
CM
CD •
tji—
a) N
H
i—i
0 oo
U CN
LT)
rQ a'O
tj> fd
a
3 in
0} •
rH
fd CM
J>i
.3 H
fd.pQ)
rH fd
<D e
& -H
O X
> o
U
0) o»
^ 04
•p fd
M-4
O O
6 *0
fd 0)
tn o
O O
M
-p x:
o o
<D -P
Pl<*H
W p4
*-«•* KD
0)
1
t 4
J
oo
ro
o
11
&r0
•H (1)
Pm £
•H
o
o
144
(L>
rC
•P
a
o
10
(D
O
nihil
•r-f
>o
0)
I—I
fd
e
(I)
CD
tr>
CP
vo
CM
O
O
>h
&
f
in
oo
i
w
m
CO
xI
<D
o
0
U
tn
a
CM
u
?!
Xl VO
d) CM
rH
£(U
O
>
IH
O
M
£
rd
Q) oi
XI
-M
•
t'
CM
m —
o N
6W
LO
H
0
U >i
-P H
a (D
0) - P
Qa (d
w g
y
I
;S
*
1
rd k t
U o\
,i ''
at
«< X
1 o
r m
ov
co Pu
fd
t7> .
-h m
Cm
jC
O
-P
•H
PU
145
"
0)
-CJ
4J
1;
V ' I <
'
> 'i
I *
d
0
;
i
CO
CD
o
0
>
0)
rH
<d
g
0)
III.
CD
tn
a)
«—i
H
0
o
rCN
(
00
>1
,Q
ro
00
£
a
w
1
0
t
<»—
•s
•r~{
N
—
CD
CD
\<D
I CD CM
1
rH
0) o
0
•—1
O
1
CM
0\
i—I
i
Ci
r~f
&
O r*
> G
rd
CD &
JZ
•M
•
M-1
O N
s
fd CN
U (T\
tr»oo
0
U >i
-p t—i
O CD
CD -P
Oi cd
W e
•H
< X
r o
I u
o a
^ 04
fd
«
tr> .
•H 0
P>4
rd
o
•M
•H
APPENDIX D
This appendix contains sound spectrograms of the
examples categorized as "Poor" by the panel of experts.
The spectrograms are arranged in rank order according to the
ratings of the panel.
Of the 216 examples in the sample,
28 were considered in the "Poor" category, with ranks from
191 to 215.5.
Each spectrogram exhibits two types of spectrographic
display.
The Type 1 display shows pitch frequency vertically,
time duration horizontally, and intensity as shading between
gray and black.
The Type 2 display is a cross-section show-
ing frequency vertically and intensity horizontally.
The
frequency scale of the cross-section is inverted on the
spectrogram.
The cross-sections were made approximately at
the attack and 1.2 seconds later.
The cross-section is
shown on the spectrogram at the exact time spot which coincides with the Type 1 display.
Each spectrogram is identified by vowel sound, educational level, voice combination," pitch frequency, and rank.
146
147
<1)
H
rd
O
O
O
W
• * , * ;
&
•f
iu
W-
•Hi
a
XI
U
o
:'' i
•H
a •
3 H
•mcrv
H
CF>
I;
LO
,Q
O
tn
C rM
0 a
w as
&
*H
Q) N
Q) tn
(D CO
£ H
O
> >i
CD a)
rC -p
•P rd
I
e
tH •H
o
6 8
rd Cu
M p4
tn rd
O —
M
-P U
D
0 rC
1
Q* O
W 4J
•H
< 04
1
I <D
rH ^
4-)
• a
o
•H
Pq W
0)
O
•H
O
>
148
111
<D
rH
fd
e
i—i
0
0
&
o
to
1 Ii
lI
<ti1
1 #litIII I
i i n %!
till ill!
11 {I f J
;
I i*i
l
l? -;SFI . <
«
111
r • "
P
•»
it
»!
• 3-
I—1
<y\
i—1
I
ov
m
o
i
CQ
1
i S £i
f1
11, VI
Jl r
li*
1111
fe I
9
? I
,
•H
0
a •
Q) rH
in o\
rH
>1
XX
0
aM
0 a
w fd
P4
rd
»w #
£t
rC N
fd fd W
I rH
CD ^
£ CM
rH 0
cn
> >1
rH
\
CD <D
rCj «P
oa 4-> fd
KD
e
rH m -h
o X
_ 0
1 g
H
!
j G\
fd Qa
I ^ M cu
i—1 tnn fd
0 U
-p m
o
<D ^
a< 0
to -P
•H
Pu
i
1 <D
CM Si
KP -p
•a
tn 0
•H
1*4 W
a>
o
*H
0
>
149
0
rH
rd
e
rH
o
o
O
U)
<
il
J,
t a 1 1 1 : fa
If,, 51, h;' . "'I.
,f, 3,
'
it : f I
I 1 # I
XJ
•H
u
o
•H
a
0*
i—1
•
CD rH
W CTk
rH
1
cr\
LO
0
&
mh
o
tr>
a ^
3 a
w &m
1
rd
rCj N
fd
fd
£
i—l
w
I—i o
CD CM
£ CN
o
> >i
CD CD
-P
- P fd
o\
CN
e
IM - H
O X
g £
fd cu
m a
tr» fd
o —
u
4-> |C
0
CD
I
01 o
to -p
•H
<C th
I
I 0)
CO &
^
4J
•
t n O.
•H
W
CD
1 !
Ill
a
•H
o
>
150
^ '! L'i .
' iu?
flsUi
' H*lt
ill
3V
u
# .bl "i
J# MI 11'
Jfcfe! fe|;f
11 1
0)
XJ
-P
a
0
w
0
o
*H
.0
>
CD
rH
<d
e
a)
I—1 CD
o\
I—1
1—1 rH
0
1 o
as
>1
rQ
in
tr>
o
a
3
l W
Q
*~-s
•H
I
rH
Q)
CD CTi
CD 1 CD rH
1 rH M~l
CD 0
PQ £
CM 0 &
U
> a
fd
1 <D &
XI
-p
•
H
rH
0 N
(
K
B
O
n5 rr- U ^
rH tr> rH
0
U >i
4J rH
O <D
CD 4->
Pu cd
w 6
•rf
C X
1 0
1 u
^ Oi
^ Of
<d
^
* -'^"q
1
ilm
v I i f I-1
•HPS)
•lifflj
i;fH|
illi
liyt
l^lll
Irllf
m i
si I
-jiKi?
'ipni
iJyll
•' .3 $'•" 11
'II® I
• 4' $ i /1
isii
•;!&l|
r$-y% I
m i
M
Wi
*
i'
i i!*$ S; J
&\
•
•
•H Q
P«4
XJ
u
-p
-H
CU
151
(':? f;'!' I'.l 11
jj
11 t•H
li• 11
$1 I:<I 6I IS
? I>. •',
r~H
CT>
0) iH
—
i t
rti mh
e o
0)
M~t M
a
^ fd
a &
rd
\t"\
\
\t
I
I,
If
*;
1*'
'
1*l
IH
CD
1—1
cn
»—(
i1
SM
v* ;L ('$!:•
i !i f rH
m
o
1
a
¥ I'
i
CD
CD
1
S
—
i 1
O
1
o
ID
1
Ch
rH
I ill
! 11 n
jj$J J-1 'i
Q) "sf
0) <N
rH
H T5
o a
a fd
>irrQ ^
1—1
tr>
a >1
0 rH
w o
-p
»
•—
*N
* fd
•H g
w.H
. X
CD 0
10) M
P4
H p.<
a) fd
£—
0
> Q
0) u
^ CD
-P 15
0
u~i a
0
6 u
fd -P
M *H
tn 04
O
M CD
-P £1
O -P
CD
a. a
CO 0
^ Tf
i cd
r a
to -H
lr U*-' f ^
;-i$ 41
^
i •*!
•H
U
ill?
i L:1
•
i—! ^
fd n
e w
f'Hi
)* ? d
i I
$
• o
tn u
W
CD
O
•rH
o
>
152
""TIT
•t •
>
•i
M
*• •
•
^
«
>
ft;;"'',
* 4
Htl$
Cft
H
I
O
i—(
CN
CM
V i:
i
u
\
&
I
h S M
CD
1
s
I—{
h)
1
o
CO
i *
:i *
:
, , ^
r^
in
i
• i^UWI
.?' < V <• i •; •>i i
; ?H|:«
j r* *-.«
, .;
,C' ' I
5 • j'f
i: t;4?
a
XS fd
a&
fd
CD •
i—1
fd n
gW
i—i in
0 .
O rH
& VD
O <N
W
XI £
fd
*PH
rCjmH
M i—1
0
•H >i
CH
3 0
4->
fd
>i g
A -H
X
tn o
a M
3 O*
W 04
fd
•-N.
U)
— U
'd
CD CD
r£j
H 0
(D U
&
o to
> CD
&
o) a
,3 -P
-p -H
&
M-l
0 CD
g^
fd
M fi
tT> 0
0
U T*
-p a>
o £
CD -H
P4^
O O
rij
!
rw
VO 0)
^ o
•H
•O
tJ> >
•r-i •
P-i QJ ^
ir
—I a>
gd r—j
(1) m
m o
1 5 3
7 - f i rprv^mm
m
- v / $ ' 4 >v»* *f
•
! ,
h
l !
M
,
>\ <f
$•<;
•%
?
t ! , !
;
0
; i
f
,
J
t f
£
CD r d
r H P h
fd
e
a)
•
m
—
N
^
a s
a
rd m
»•'}' •*
M
f,
h i
6! ? H i!' h , 4 4 t,
v n i
i,
•;<
i i
'* •
^
1 1
mam
Vf~'
•>' i M i l ?
0)
r H
rd
r H
V£>
< n
e
i n
!
K
f
11^'1
f
f
•
i n
(Ti
•—I
1
o
i—i
i
v&mr
r
x3
CD a
t n rd
CD
r H r H
r H C O
0
r H
O
> 1
> i » H
r Q
a)
- p
C O
rd
G
3
w
r H
•f • , V | J f
'.fe ^ f < ; j
*
I
^ylj'
. £ I •'?; H ' ^ ' F
w
U
~
1
eC|
a)
1
: r j ! f | f t |
K
I I I I t I £ f t
* 'IV1!
(
t
;
i
l
if'iv'Jf
l
i
! f l i ' I
i f
i :
«f ? H f M ' ? ;
' l l ' V t r . tUf!
*' J ' *i I j V - i l - '
H
i i i i
§34,
P i
• ; p j
$
;i ." V\\> '). f t , . ? V,
r*
3 1 $
P
f i f n
i
-
I
f
i
i l l ;
i
v o
0 0
. a —
CD
O
r H
CD 0 )
§
rd
0
>
1 5
CD
0) 1 5
^
• p
0
a
M-i
0
W
CD
6
rd
M
fcn
0
M
- P
O
CD
a i
CO
u
- P
* H
a ,
* C
r
i
r -
1 5
cd
a
- H
r Q
•
t P
• H
& 4
o
o
1
ii - i ' . ' ' ' ! 4 f ^
h. , f \ v » $ , ' 4 . |
'•
S
H
a
E
- H
X
0
M
O i
O .
fd
-
w
I
P
M / f l M
r H
O
C N
CD
&
- P
a
0
CO
•
CD i n
O
•
• h m
O
CTk
J> r H
154
o
<D M
rH
£
(tr ctf
60 &
m
•d
j
\
,
^
ilillf
in«
in
C
HTi
f
o
1—1
CO
1—I
f
W
i
,dfd
1
S
C
ON|
r
CM
<
y»
1
rH
—
iMI
C
dN
fd w
CD CO
rH
O
gfd ro
<D a
D* fd
CDI l-O
—
irH
KD
0 r
H
O
>i
>irH
jq 4J
a)
t£r> erd
3co-HX
—0H
fd p404
—
& _fd
ftf
HW
0) if)
& fd
0> *a
a) a)
ts
A 0
M
- a
mw
a)
^o Xi
e o
fd
+j
H
Cn -H
cu
0M CD
M
- ,a
U -M
C
D
QA C
W 0
< Tf
D
11 C
G
00 * i 1
^ $
C•nOO
•H
cn •
uCD •m
•H in
O
>H
155
;
f,T
X
i
1i
(j:
<D
rH
fd
g
'
;
< '
t
i—i
0
0
rC
O
to
V
'* L • lis!
; *,
•H
;
•
I M1
II m
M
-
*
* I
?
*
i,L
."'Mi
f f f, |
j i1
g
'i
: "1
$
| Kf1;{ $ '• 'i
l
i ; i
* t ! iI' f!
11 I
X
Sl«
f
I
m
•
M
00 0 *
O
oH
\ •aH L•
r
0 00
1 •nc?i
rH
o
>1
1—j
0
Cn
a rM
o
3 a
to nj
&
1
*H ,
CQ —'
I
I Q) N
0) Id) HI
<d
rH ri a) ^
& CN
0
rH £> »>i
h>
rH
a) a)
£
S
-P
I
•P g
rd
in m -h
rH 0 K
^ u0
*
e
rH
CO
<tf
OA
a<
0"^ <tf
0—
-P PQ
o
a) x:
Oi u
(/) 4->
< Oa
1
i <D
as &
-P
•a
tn o
•H
fc-4 CO
CD
a
•H
o
K*
u
156
•us
CD
«—1
rd
6
—
i !
0
0
A
o
CO
If
&
tn
•H
m•
•£'
H3
iff'S
51
V 11
:1
i. !'!
% ! )JI
•;hl
u
00 0 •
(Ts •rH LO
—
i t a •
2 CO
[
-n Ch
—
i i
o
rH £) 4-4
0
tr>
£M
o
0 a
03 &
fd
1
rd •
O —'
I
& V—
Ns
XI rd ffi
rd
—
i 1 rH
1 a) oo
£ «H
CQ 0
> >1
H
rH
CD CD
1 X! -P
-P g
rd
H
CO <4-4 »H
—
i1 0 X
^ o
e M
I
rd Oi
H
^ Pu
tn rd
0—
M
-P U
O
CD XI
Ol O
CO -P
•H
"*$ 04
|
m5i»:
i •) w& if
o1 0
m 4J
• a
tr> o
-H
PM CO
CD
V
•H
0
>
157
<D§
i i
rtf
e
m 3:
t
i—!
0
0
roCj
w
•'&
•H
A
r a * £$•
'I | | ;||
If''. 'fI:
i
Ii %
f
1
i '*
i£
k';f*
4 1 I:
:|
LO u
00 0 •
-h in
rH £ •
13 00
1 •ncr*
i—1
o
1—1 JQ M~i
0
aM
o
3 £
w &rd
I
CQ rcJ
1 XI N
•
/—%
^
,£
rti
1
PQ
•—1
h>
i
vo
r^»
i—1
1
r>
I f-'r Jf 1
J
$ I- l£' » ilx 1
Ill I I
ill! 3
8 1
illil
•
(fl ffi
i—i rD *3"
£ CN
O
> >H
rH
(3) <D
JCI 4-1
4-J <d
e
m -H
0 K
_
e0
n
rd a
H Oi
d
wr
0
u
4o-> P3
<d ^
04 O
W 4->
•H
< O-i
» CD
rH rC
in 4-i
•
£
tx» 0
•H
CM Q)
W
O
•H
O
>
158
» I / /; I
i?;y
'* •; :\
%
0)
,a
•p
: k \ ; 1 \
a
o
U)
<D
O
•H
O
>
i\
6-1
i 5
:}<
I m M s
ftfefe'-i?". | v i
?
"'':
' ^ *' V
4'
4il
Q)
rH
fd
e
LO
•
CD
fcp
CXD Q)
C*
rH H
«—1
1
0
O
o
r—I >i
;:|ni-l!i
\'i
. ' %} £
1 .-.• fr- f
• - z-wl '•• '4
: •" %fy->
' : '• J i ; K I
/ ; l? • >; i
' l' i f „<•?. j-1
| O i\
1
;4r r.HJ
< £*-? ;'i s I
• M$ni? ?
!11H <!
tr.t.? _• •;rj lSfI i»fr-s
f i?l#
:s = 5
:va; ;>i||j|||!
" >•: • r>S/ I s» ?
'U- # -'Sl;IIII
.£;,:•: - ..L-fJiv#I
i, i
< T
,- ! '/ i\ I j
<k •#3iJsfirf
I • • ; .XfcJiWI
- M :
m
m
lilftlf
i VI
"5
J r-:i;--r t
*
£:-V '•• v Vf £
{0§jf [r->
I M m
U'4'
O
I
tF>
a
2
w
a
i
.d
0)
I
«H
O
1
o00
LO LO
N«-«» #
00
r£ G\
Q) rH
Hm
<y o
£
o M
> a
<u &
X!
-p
•
CNJ
1
CO
rH
0 N
ffi
6
fd rM^
tr»H
0
M >i
4J rH
o a)
PLI grd
w
< •H
X
1 0
» M
<N Oi
m a«
m
# v»^
tn
•H a
fa
o
•H
Pi
159
0)
xj
4J
i,!II.
£
0
I'll,
.
0)
0
o
•H
O
>
4
, V ^
lilt
<D
iH
rd
'hi
B
0)
«—I
O
CM
tr>
<u
o
o
M
.
rH
>i
JOt
<N
i—{
:
;Wi,
£
3
w
I
0
v
-Mm
1
rd
~-*
•
rH
rC
XJ
o
rd
fd CM
r H *4-4
m
<D o
£
O rM
>
£
r~i
u
rd
Q) &
VD
4-1
i—I
m
oo
•—i
I
iilil
*
*
O
N
B
w
fd r H
M CO
tJ»rH
o
M >i
- P i—I
0
CD
<L> - P
O,
rd
0)
g
1
I
o
M
Oi
a,
l ll
i*<*>
co
m
rd
•H O
Pm
•C
0
4J
*H
01
160
j;!
a
d)
.
•
ii> \
6
\ c
!* !;
UJ
.?4'
\e
rH
0
0
i- I
XJ
0
•'.V>S5-.!
^ :'v
/ R
i
i
fd
CO
;
XI
tj»
•H
XI
0
- H
O
a
•
C\1
•
n
o
CN
r
rH
i—(
>1
XI
0
t n
c o
0
o
a
fd
co
&
1
0
1
G)
«W
J
•—V
X !
N
CD
W
x ;
CD
I
m
rH
VJD
(D a *
^
rH
0
> >i
rH
rH
(1)
X !
0)
- P
- P
VD
m
r
I
r00
3
r
mi
Ji
i i i
.
•
\
4
1«
U
i
d
E
M-4 - H
0
X
^
0
E M
fd
a »
U Qa
tn fd
0—
U
-p e>
o
Q) X?
Oi o
CO -P
-H
1
1 0)
^
XI
m -P
•
&>
£
o
•r-i
PM CO
Q)
o
•H
o
>
161
i
I!..
I ; :>
i iI
iu
i
<D
1—
fd1
£
H
0
0
JZ
O
w
X!
•H
i i
^ • i i i
!
o
CN
1
—
ii 1I
—
CO
o
i
CQ
u
0
•H
a •
•no
M
>iC
rQ
0
0a ^a
w&
fd
a)
rd/-s
N
a) sc
—
i(Di ^
r£C
M
0
<ur~|
x:
-Po)
-P gfd
m
-H
0X
0
e^ m
fUd &
tr> QA
fd
0u ^
-p m
o
<
aD
» o^
W +J
•H
<!
1 04
l xi0
in
m -P
•c
tn o
•P
H
MQ)W
•*
i !
i
:
1
A
<D
1
PQ
—
ii
r
vo
o
«H
1
0\
00
•
O
•H
O
>
162
c a
fd fd
a)
vif«:
i
l
g ~
ISJ
rH W
0
0 ^
& <3\
0 CM
W
*0
l ,
;•/-,} •} >\ -;v\
& £
fd
-H
^3 r -
o
CM
1
^\ i
nv
:1I
i—1
iH
ro
o
1
Q
ti
pCj
<D
I: IKS
i*
i<
; 5 '
; i .
v
i
;
'
i
':
•' •!
;
:
ffe|:tl
s
a
I1 \ ' 11
Yr'AV^r
•
CN|
r~i
i—1
1
rC »d
CD CD
*3
H 0
CD O
£
0
>
i
, f >">
i . «J'kx
w
a)
a) o
rC| -M
«P «r-f
Ou
M-f
0 CD
_ ,£
S -P
fd
M a
tT> 0
0
U V
4-> 0)
o a
Q) - H
QaA
w
C
1
r
vo
LO
f iIiM
V '
* 5
,!
1
00
LO
i
; • • " !
!! '
i
s
H
*0
M <H
0
*H > i
£ H
a)
-r*»+J
rd
>i S
XI -H
X
tn 0
a M
3 a.
w Pu
fd
-^-s.
a)
— P
6
0
a
w
a)
o
•H
• o
tn >
•H
(1)^
rH O
rd <N
6
0) m
M-l O
163
<D
H
rd
B
! i ' l !
i—i
o
o
rCj
O
CO
.1
*H
,3
o
CN
O
£ •
a)
co o
CN3
H
>i
*
:
00
o
tr»
a r*
p a
co rd
< i'
i,
P
•r
?
&
Q)
PQ
«—I
CO
in
<r>
CM
Pi
u)
X! N
<D «
i—I r a) ^
£ rH
0
> >1
fH
0 ) <D
^ -P
- P rd
g
-rl
O
vo
o
rH
68
fd
M
tn
O
a,
Pu
rd
—
•P
0
0)
01
CO
Q
U
f,
X
o
-P
•H
<5 a *
l
f <D
r- &
in -P
h i
.
i
•
'
ii
•r-f
1
I
s
X
£
tx> O
CO
0)
o
•H
O
>
164
]'
:
!
r f j tilllfn:
a) m
H O
fd
S
a) a
m rtf
,M ; iCf :{i ;«j
•J?;,'
• r i !\ in- ^:l!r"
.''?, .i jji'ij
' ! '
ll4!
a
fd •
'
-I" rs if S -1
<L> N
Hfd «
E o
00
CD 00
fcn
1 1
M I L . .
iii
Pi r u1 n
^ vL hi ; 1
v:
a> ^
h a
o
CM
h d
0
o in
vo
1
i—1
«H
tn rH
00
^rKlrlsSflSttl
r Lji I i1 y,!t5 tfi!H
<
I ^ M t ft
'•; •;• ^ r L if i :
o
1
w
1
X!
<L)
ii! f i !;
S
rH
1
' ; j '! |
1
f?
' J i Ir I I
Ml '!
.1i
a)fda
H<U
I
u
•
a a)
3 -p
c/> fd
B
^ *H
a) X
— 0
U
ro
o
CN
£ w
0
> ^
<u
-C 0
-P o
0) n3
m w
o a)
_ x;
e
fd -P
^ -H
tn cu
0
U (D
«P rCj
o
O -P
<D
£h
£
W 0
<
Tf
i
a
* 0)
l
hi
i.i
f
I i
00 -H
in n
• 0
tJ> O
PM CO
a)
o .
O
O
> <N
165
d
0
0)
0)
o
*oH
>
CD
rH
nj
6
cd
1
ii
'
i•
i?
91
if !
[,] <
(D
tn
(
o H1)
CNJ rH
0
i u
CM >1
i—!
rH tn
d •
«—i 3 rCO o
f
CN|
U -rl U-l
J — 0
|Q) rM
0) 10 a
CD
rd
H P4
I CD
£
0 0 •
>—
H
O
N
CD W
r
o
•
vo 4~> t
rH m rH
0
g CM
in cd >i
u
tr> CD
H
0 4->
U rd
-P 6
U -H
<D X
Ot O
CO U
Pa
< 04
if
Jwrd
1
in u
• X!
tr» o
•H -P
p4 -H
Oi
0
-P
166
fI
i110f
: 'i|'iv!'
' iii'L
jfl'f
I tf'
H{ l| :f
•M ES
t?!f .'.f.ii
rti
Iliil
||
!1 II
W
: llHi
o
CM
I
CM
i—i
(N
O
f
u
i
CD
CD
I
s
1—{
h)
i
o
vo
1
on
KD
<D
i—1 in
rd .
SH
VQ
rH CM
0
OU
.a a
O fd
W
rH
X! oo
tn>H
•H
Xi >1
H
M CD
0 -P
*r-| fd
£ g
•n X
0
>i U
&
04
04
tr* fd
2
in a
•Hr CD
w
d
, o
I CD O
1 CD
in
rH CD
CD J2
£ 0
0 -P
t> *H
a.
CD
XI CD
-P XJ
-P
o a
^
g o
fd nd
U CD
tP a
0 -H
M^
-P S
O 0
<d o
•
W W <J\
'• 11; I
l:|l
I n
CD O
O CN
1 -ri
1 0 MH
o > o
VD
<D X
• H £j
tp (d fd
•H £
Cm CD
M-t
•
TJ —
£ N
cd
167
CD
I—I
i' - JK li
ill
fd
e
81
o
o
Xi
o
ill
w
&
i\i;:
&
•H
Xi
t t ii
Ch
o
CM
U
0
•H
a
CD
•
<Js
If) O
04
CNJ
>1
&
<N
M-l
0
tr>
fi ^
w
fd
W
fd
-a
fd
f
'
PQ
H
0}
. r
LO
CNJ
&
N
fd ixj
r—I i n
CD VD
£ H
o
>
>i
r-f
Q) a )
^
-p
- P rd
e
IH -H
O
LO
X
efd a.
2
m a,
tr> f d
o —
u
-P
0
Q)
04
CD
W
,£
U
-P
-H
04
<
1
I
CD
r-i rd
VD - P
•
•H
Pm
G
O
W
(1)
a
•ri
o
>
168
-1
iiOfii'S'I i 4
I
1
\ t% K."
'.-,s $ 1
,i
ro
CM
1
CM
rH
CM
O
J
o
»
JZ
fd
i
a
CNJ
U
r
00
00
1—1
1
! Ch
o
CM
i
"ii
l1 ^i
W
1 *&
I
iK|J
1I * S
« n ]••!:HI
11 i <i|y
;
m
o
a)
rH ^
fd £
e
cu n;
M-i
T$ •
£—
fd N
«
CD
rH IT)
fd •
£H
VD
<D CM
<D
rH jZj
i—i fd
0
OH
ro
>iH
.Q
>i
tj» rH
a a)
2 -p
is) fd
e
-H
cd x
— o
u
& Ck
fd Qa
fd
i—i
<D
£u
0
> TJ
CD
CD ^
A 0
-P o
4-4 CO
0 0)
S o
fd 4J
M -H
tn a.
O
U 0)
-P A
O -P
0)
cm a
W o
<! ^
i <y
ra
CM -H
V£> £}
«O
tr» U
•H
P>4 0)
a)
o *
•H Cft
OO
> CM
169
«? i
a) N
rd
g o
f-i
0
0 15
xi a
o rd
to
o
CNJ
£r>c\j
•H
r£ > i
rH
U Q)
0 -P
rH
H
(N
f
00
1—1
•! i
O
•H
a
rd
e
0 -rH
•m X
0
>i M
Xi
Oa
a*
tr> rd
w
C
$
rH
IB • k
W f<
i
rd a)
d
0
^
o
rd
w
rH <D
<D eTj
£ o
0 -P
£> *rH
PH
0)
^
0)
-P
-P
MH
o a
_ 0
6
rd H3
M 0)
tr» a
0 -rH
w r
r
n't
(d
i
Pill
S
rH
h)
I
!>
CN
11"
1
O
r-
te n
|;{ . f $
•P 6
O O
CD U
w
<
to
Q)
U
I •H
O
CO >
V0
.
&
t Af
4
§
0) i—I
• i—I ( N
t n fd
• H g mh
0) O
*a £
£ fd
fd Ph
170
a)
•—i
rd
6
rH
0
0
o
U)
•H
U
0
CN
rH
CN
I
CO
1—I
rH
O
1
W
1
0)
a)
i
Hit-
^ I
•H
£ •
a) co
rH
CM
>1
If)
M-l
0
tn
a rM
0
a
(/} fd
rQ
-H
s
— «
1 CD N
1 CD m
rH LO
CD VO
£ rH
0
> >1
rH
rH
CO CD CD
& -P
fd
1 -P g
CM m -h
LO 0 X
_0
n
1 g
cd Cu
G\
M 0*
tr» fd
o—
u
-P W
O
a> ra
04 o
W -P
<
•H
P-4
1 CD
^ rG
VO -P
? . 'I .
I4 I1
• a
o
•H
0)
0)
a
•H
O
>
171
*I
1
l Y.' iy r
f
1
. t. A, (• '
0
&
-P
a
o
,
:
\
03
a)
o
•H
O
>
:Uh
CO
I—1
CM
CM
f
'i
:
\ i '•
CD
i—1
fd
E
cu
ty*
a)
rH
i—i
o
o
>1
O
:-i
•• '•
I
,
; *
W
A
tr>
a
3
to
1
fd
"w
\ I
m
' W •
,,
; 4
*
;.f: ,
r-c
'
O.p.
fd
«—i
u
*>
vo
CN
• V - i -
r•
•X
<
00
rH IH
cu o
£
O M
> c
fd
a; «
-p
o N
6 W
fd lo
M VD
tJ»rH
O
U
I.J
t
00
rCj rH
(d CN
>i
-P rH
O CD
0 -P
a* fd
CO g
•H
» o
1 u
m On
v o
a.
sfd
• "—
•"*
tx»
*H W
P4
O
-P
Of
172
1
1
i
1
j
i
•V i^-
|
" : m
S:*!j pii, E B . !
?
; J
, J
11 •
| v S § i ! @ s y : | i : I 1
.l,i!iUj
•••:
Ws*
-
I W
\i
V* %-':Vi
1
h1
) ~i "
.I *
r.
r : < i i l
[;{
^
!
!
00
r~|
(N
f
00
a—1
H
Q) T3
CP a
QJ rtf
«—1
i—1 Lf)
0 .
a^
r*»
£>1 rH
tPr>Hi
£ Q)
0 -P
CQ grd
i
jf
f Jxj
OJ X
1 — u0
XI Xi Qa
G) 0) QA
cd
1 rH —
(1)
& CM
!
I CM O
o > TJ
CD
1 Xi(D 0^
•H «P U
r*H M-4 U)
0 0)
1 _
S X!
o
rd -P
o
M "H
CN tn 04
0
m a)
•P rCS
O -P
a) a
a.
W 0
< TJ
1
<D
ra
<X> -H
kD
•O
tn o
•H
Pm to
CP
U •
•H 00
0 rH
> CM
s
i
•
a—
fd n
w
0) •
rrH
CT\
ed r^
o
!°
:
M-4
0
CD
H ^
fd a
£ fd
a) &
H-t
173
• !
•
-
•
m
•'
CD
vK-
? / •:
. v'l . :
£i
-P
-v • '
i
a
o
\k
w
0)
o
•H
O
•:'r "
ftliiifSiliiilfiii
>;
mm
>
r>:k '.>•
0)
rH
fd
s
•fir '
•?-
. M l
IT)
«
.115 •
® ; •
•
L
'
j i \
'
.'f
*
<D
tr>
a)
rH
H
0
to
rH
CM
1
^
>
1 1
£ - *'•
r W h
H
l
>1
rQ
- M ' P
O
r . w - " • : } } >
••
tx>
a
3
w
o
)v;>-
1
s
U
3 "
I
i-« -
' ••v
'! f P
:•
',V
'"iV •
I
H
I. «
'
j ;
4-4
£(1) O
O ^
> a
rd
0) &
-M
IM ^
O N
W
i
m
I Q) i—I
m
m
LO
cr>
vo
H
E
N
rd H
M oo
tr>H
o
Vr
4J rH
O 0)
:V
'*>* i
•
.
IT)
I Q) CN
: U 'v; !
.•
—
cv
(D
«H
u
•:
•
:
t-'jifV..
m
—
•H i f )
M >1
-
U
? ";
•V . i
-
1 1
•
*a
i 7-
r
CD -P
a* fd
w
<
r
r
rVD
e
•H
X
o
M
a
Q,
fd
•H U
fc,
JQ
0
4J
•H
01
174
-
^
-
?
T
T
r
^
r
r
.
?
r
„
w m ; '
•"* vi
:
V
• I'll
* \i * j
'<$*•: 0 * K
i
St UKP)
Y'*.r->:A *$;V
W " < 3V:
Vvjjhx:-i
'
:
mw,k
n-m i. -t
mm
y ' -I". '"' '-!!•
rH U-J
!$'•
" >'4®jU-ma
• JW-vt
;t: WW
Zffl *!U
•j
•^;j: V f ;
.i
00 Q,
VD Q,
BIBLIOGRAPHY
Books
Borg, Walter R., Educational Research, An Introduction, New
York, David McKay Company, Inc., 1963.
Mouly, George J., The Science of Educational Research, New
York, American Book Company, 1963.
Potter, Ralph K., George A. Kopp, and Harriett Green Kopp,
Visible Speech, New York, Dover Publications, Inc., •
1966.
Articles
Delattre, Pierre, "Vowel Color and Voice Quality," The
Bulletin of National Association of Teachers of Singing,
XV (October, 1958) , 4-7.
Dudley, Homer and Otto 0. Gruenz, Jr., "Visible Speech
Translators with External Phosphors," The Journal of
the Acoustical Society of America, XVIII (July, 1946),
62-73.
Flechtner, Adalene Smith, "Low Vowel Formant in Soprano
Voices," The Bulletin of National Association of
Teachers of Singing, XXVI (December, 1969) , 23-26.
Howie, John and Pierre Delattre, "An Experimental Study of
the Effect of Pitch on the Intelligibility of Vowels,"
The Bulletin of National Association of Teachers of
Singing, XVIII (May, 1962), 6-9.
Koenig, W. , H. K. Dunn, and L. H. Lacy, "Sound Spectrograph,"
The Journal of the Acoustical Society of America, XVIII
(July, 1946), 19-49.
Kopp, G. A. and H. C. Green, "Basic Phonetic Principles of
Visible Speech," The Journal of the Acoustical Society
of America, X V I I I ^ u I y T T W 6 T 7
Large, John W., "An Acoustical Study of Isoparametric Tones
in the Female Chest' and Middle Registers in Singing,"
The Bulletin of National Association of Teachers of
Singing, XXV (December, 1968), 12-15.
175
176
Large, John and Thomas Shipp, "The Effect of Certain
Parameters on the Perception of Vocal Registers," The
Bulletin of National Association of Teachers of Singing,
XXVI (October, 1969), 12-15.
Perkins, William H., Granville Sawyer, and Peggy Harrison,
"Research on Vocal Efficiency," The Bulletin of National
Association of Teachers of Singing, XV (December"^ 1958) ,
4-7.
Peterson, Gordon E. and Harold L. Barney, "Control Methods
Used in a Study of the Vowels," The Journal of the
Acoustical Society of America, XXIV (March," 1952) , 175-
t§T:
Potter, Ralph K. , "Introduction to Technical Discussion of
Sound Portrayal," The Journal of the Acoustical Society
of America, XVIII (July, 1946) , 1-3.
Riez, R. R. and D. Scott, "Visible Speech Cathode-Ray Translation," The Journal of the Acoustical Society of
America, XVIII (July, 1946), 50-61.
Scott, Anthony, "A Study of the Components of the Singing
Tone Utilizing the Audio Spectrum Analyzer," The
Bulletin of National Association of Teachers of Singing,
XXIV (May, 1968), 40-41.
Steinberg, J. C. and N. R. French, "Portrayal of Visible
Speech," The Journal of the Acoustical Society of
America, XVIII (July, 1946), 4-19.
Taff, Merle E., "An Acoustical Study of Vowel Modification
and Register Transition in the Male Singing Voice," The
Bulletin of National Association of Teachers of Singing,
XXII (December, 1965), 8-11, 35.
Triplett, William M., "An Investigation Concerning Vowel
Sounds on High Pitches," The Bulletin of National
Association of Teachers of Singing, XXIIl" (February,
1966), 6, 8, 50,
van den Berg, Janwillem and William Vennard, "Toward an
Objective Vocabulary for Voice Pedagogy," The Bulletin
of National Association of Teachers of Singing, XV
(February, 1959), 10-15, 24.
Vennard, William and James W. Irwin, "Speech and Song Compared in Sonagrams
The Bulletin of National Association of Teachers of Singing, XXIII (December, 1966)T~
18-23.
177
Unpublished Material
Appleman, Dudley Ralph, "A Study by Means of Planigraphy
Radiograph and Spectrograph of the Physical Changes
Which Occur During the Transition from the Middle to
the Upper Register in Vocal Tones," unpublished
doctoral dissertation, Indiana University, Bloomington,
Indiana, 1953.
Arment, Hollace Elbert, "A Study by Means of Spectrographic
Analysis of the Brightness and Darkness Quality of
Vowel Tones in Women's Voices," unpublished doctoral
dissertation, Indiana University, Bloomington, Indiana,
1960.
Harper, Andrew Henry, Jr., "Spectrographic Comparison of '
Certain Vowels to Ascertain Differences Between Solo
and Choral Singing, Reinforced by Aural Comparison,"
unpublished doctoral dissertation, Indiana University,
Bloomington, Indiana, 1967.
Helvey, Kenneth W., "A Study of the Methods of Choral Tone
Production of Selected Choral Directors," unpublished
master's thesis, University of Southern California,
Los Angeles, California, 1952.
Hohn, Robert W., "A Study of the Relationship Between Vowel
Modification and Changes in Pitch in the Male Singing
Voice: A Spectrographic Analysis, Reinforced by the
Judgment of Phonetics Experts," unpublished doctoral
dissertation, Indiana University, Bloomington, Indiana,
1959.
Johnson, Merion James, "An Investigation of the Effect of
Training in the Articulation of Vowels by the Speaking
Voice Upon the Articulation of Vowels by the Singing
Voice," unpublished doctoral dissertation, Indiana
University, Bloomington, Indiana, 1966.
Scott, David Watters, "A Study of the Effect of Changes of
Vocal Intensity Upon the Harmonic Structure of Selected
Singing Tones Produced by Female Singers," unpublished
doctoral dissertation, Indiana University, Bloomington,
Indiana, 1960.
Sullivan, Ernest G., "An Experimental Study of the Relationships Between Physical Characteristics and Subjective
Evaluation of Male Voice Quality in Singing," unpublished doctoral dissertation, Indiana University,
Bloomington, Indiana, 1956.
178
Whitworth, James Ralph, "A Cinefluorographic Investigation
of the Supralaryngeal Adjustments in the Male Voice
Accompanying the Singing of Low Tones and High Tones,"
unpublished doctoral dissertation, State University of
Iowa, Iowa City, Iowa, 1961.
Wooldridge, Warren Buchanan, "The Nasal Resonance Factor in
the Sustained Vowel Tone in the Singing Voice," unpublished doctoral dissertation, Indiana University,
Bloomington, Indiana, 1954.
Wyatt, Larry W., "Blend in Choral Sound," unpublished
master's thesis, School of Music, North Texas State
University, Denton, Texas, 1967.