Journal of Voice

Vol. 7, No. 4, pp. 3.01-310

© 1993 Raven Press, Ltd., New York

Comparisons of Pharynx, Source, Formant, and Pressure

Characteristics in Operatic and Musical Theatre Singing

Johan Sundberg, *Patricia Gramming, and 1"Jeanette Lovetri

Department of Speech Communication and Music Acoustics, KTH (Royal Institute of Technology), *Department of
Phoniatrics, Karolinska Hospital, Stockholm, Sweden; and ?The Voice Workshop, New York, New York, U.S.A.

Summary: Belting, a vocal technique typically cultivated in musical theatre

singing, differs timbrally from operatic singing in many interesting respects.
The underlying phonatory differences have not been previously investigated in
detail. Yet, belting is frequently associated with disturbances of voice function.
Articulatory and phonatory characteristics are investigated in a female subject
who is a professional singer (co-author JL) trained in both the operatic and
belting styles and in an intermediate vocal technique ("mixed"). This article
presents data obtained from this subject by video-fiberoptic observation of the
pharynx, inverse filtering of airflow, and measurement of subglottal pressure.
The results reveal that belting was characterized by very high subglottal pres-
sures and sound levels, and apparently also by a comparatively high degree of
glottal adduction. Comparisons with other investigations of related aspects of
belting and operatic singing support the assumption that the data obtained from
our subject are representative for these vocal techniques. Key Words: Belt-
ing-Inverse filtering--Subglottal pressure--Formant frequencies--SPL--
Operatic singing--Pharynx shape.

Most voice research in the past has focused on atively long closed phase and a high EMG activity
operatic singing, while the equally phonatorily in- in the geniohyoid muscle. Yanagisawa et al. (3)
teresting type of singing cultivated in the perfor- used fiberoptics to examine the aryepiglottic
mance of popular songs and musicals has been sphincter in a number of different singing styles.
largely neglected by voice scientists. As a conse- They found that the aryepiglottic area was con-
quence, the vocal techniques used in nonoperatic stricted in both the operatic and the belting styles.
singing are poorly understood. A review of the research on belting was recently
One type of singing that is commonly used in mu- published by Miles and Hollien (4).
sical theatre is called belting. It has been described According to the general opinion among voice
by Lawrence (1), who reviewed laryngological ob- teachers and laryngologists, belting may be detri-
servations from 27 "belters." He found laryngeal mental to the voice. Therefore, singing teachers
and pathological similarities between belting and have strived to find alternative types of voice pro-
hyperfunctional speech. Estill et al. (2) studied belt- duction that are healthy and yet meet the demands
ing, as well as five other modes of singing, with raised in musical theatre singing. The purpose of the
regard to spectrum, electromyography (EMG), and present investigation was to compare such an alter-
electroglottography. They showed that belting had native singing style with both belting and the oper-
very strong overtones and was produced with a rel- atic style of singing. The alternative vocal technique
was developed by co-author JL, who calls it mixed.
Accepted July 15, 1992. According to her long-term pedagogical experience,
Address correspondence and reprint requests to Dr. J. Sund-
berg at Department of Speech Communication and Music Acous- it is appropriate for musical theatre singing and yet
tics, Box 70014, KTH, S-10044 Stockholm, Sweden. not harmful to the voice. It sounds somewhat sim-

301
302 J. S U N D B E R G E T A L .

ilar to belting, but less hyperfunctional. (It should made in a sound-treated booth that was approxi-
be observed that the term mixed does not mean to mately 4 × 2 x 3 m, while recording 2 was made in
suggest any relationship with the so-called voix an ordinary room using a pressure gradient micro-
mixte, as described by, e.g., Vennard (5) and Reid (6).) phone mounted in a flow mask ad modum Rothen-
Our subject was co-author JL, who is active as a berg (7). The microphone recorded the pressure
voice teacher of a nonoperatic singing style, but drop across an acoustic resistance, represented by a
professionally trained in all three styles of singing. screen with fine mesh covering a number of holes in
She can easily switch between these different the mask. In this way, the oral sounds could be
styles, although she finds it difficult to change style captured in terms of the air-flow signal.
after having sung in the belting style for a while. Subglottal pressure was recorded as the oral pres-
This ability actually served as a unique and funda- sure during the occlusion for the consonant/p/. The
mental condition for the entire experiment; our aim pressure was captured by using a small plastic tube
was to describe and explain the timbral and physi- with an inner diameter of - 2 mm. This tube was
ological characteristics of mixed, belting, and oper- mounted in the flow mask in such a way that the
atic styles as produced by one and the same subject. subject had it in the corner of her mouth. In record-
The basic strategy was to compare tones produced ing 1, sound pressure level (SPL) was calibrated by
at identical pitches in each of these three singing recording a set of reference vowels produced by
styles. one of the experimenters at constant SPL values, as
The acoustic and phonatory characteristics of the measured by means of a B&K sound level meter
three singing styles can be assumed to derive from (model 2215), which was held just near the tape
both voice source and formant frequency character- recorder microphone. During recording 2, the DC
istics. These are controlled by subglottal pressure, output signal from the sound level meter was used
laryngeal adjustment, and vocal-tract shaping. Our for measuring the SPL.
investigation included, directly or indirectly, all The recording was made on a TEAC (Tokyo, Ja-
these parameters. We carried out two different ex- pan) multitrack FM tape recorder. The following sig-
periments, which we present separately in this ar- nals were stored: (1) flow mask output, (2) SPL as
ticle. Experiment I concerned the phonatory and picked up by the sound level meter, and (3) subglot-
acoustical aspects, while Experiment II tried to de- tal pressure.
scribe the pharyngeal shape by video-recording a The voice-source analysis was realized using a
fiberoptical inspection. Glottal Enterprises' inverse filter containing two fal-
ters, one for the first and one for the second for-
EXPERIMENT I mant. Using a transient recorder (BT-1; Glottal En-
Method terprises, Syracuse, NY, U.S.A.), a selected por-
Vowel sounds produced in the three singing tion of the signal was repeated and fed to the
styles were analyzed using inverse filtering and reg- inverse filter. The output from the fdter could be
ular spectrogram analysis. In addition, we also de- observed on an oscilloscope.
termined subglottal pressure and sound level. Inverse filtering results in so-called flow glot-
The subject sang comparable tones in the pitch tograms showing transglottal airflow versus time. In
range common to the three singing styles--the the analysis, the inverse filter settings were ad-
pitches G4, Ab4, and A4. She sang the vowel [~e:] at justed until an oscillation-free quasi-closed phase
an intermediate degree of loudness---first in oper- was obtained. However, as expected, it turned out
atic style of singing, then in mixed, and finally in to be extremely difficult to find a unique setting of
belting style. In addition, she sang excerpts of two the inverse filters that gave a nonoscillating quasi-
songs--Gershwin's "Summertime" from "Porgy closed phase. Rather, several different settings
and Bess" and Giordani's "Caro mio b e n " - - i n the seemed possible, and it was difficult to decide
same key in the three singing styles. Each sample which one was correct.
was first produced in the operatic and then in the In such cases, independent estimates of the for-
mixed styles. Thereafter, all samples sung in belting mant frequencies were needed. These were ob-
style were run in one sequence. In this way, a num- tained from glide tones, approximately one fifth
ber of phonations comparable with regard to vowel wide, performed by the subject. These glide tones
and pitch were assembled. were analyzed by means of a Kay Elemetrics spec-
Two recordings were made. Recording 1 was trograph (Pine Brook, NJ, U.S.A.). As the ampli-

Journal of Voice, Vol. 7, No. 4, 1993

 A R T I C U L A T O R Y A N D P H O N A T O R Y C O M P A R I S O N S OF S I N G I N G S T Y L E S 303

tude of a spectrum partial increases when it trans- styles. The results are quite consistent within and
verses a formant, estimates could be made of the neatly separated between the styles. For all pitches,
two lowest formant frequencies from the spectro- operatic style gave the highest peak amplitudes, and
grams. Using these more precise estimates of the mixed style gave the smallest. The very high peak
two lowest formant frequencies, an inverse filter amplitude implies that the source-spectrum funda-
setting was again tried. It turned out that, with mi- mental is strongest in operatic singing. Further-
nor adjustments, these formant-frequency values more, the closed phase is clearly longest in belting.
resulted in convincing glottograms. This indicated A high glottogram peak amplitude can result both
that the subject reproduced the formant frequencies from a high subglottal pressure combined with a
rather accurately in repeated renderings. comparatively high degree of glottal adduction, and
from a lower subglottal pressure combined with a
Results lesser degree of glottal adduction. Therefore, it is
Figure 1 shows two sets of spectra of the same interesting to examine the relationship between the
vowels sung at identical pitches. These vowel peak amplitude of the flow glottogram and the sub-
sounds were taken from the songs as sung in each of glottal pressure. This comparison can be made in
the three styles. Several observations can be made. Fig. 6. It can be seen that operatic and mixed styles
First, there are differences in the relative ampli- both were generated with moderate subglottal pres-
tudes of the two lowest spectrum partials. In oper- sure, but operatic style results in considerably
atic singing, the fundamental tends to be strong, higher peak amplitudes of the glottogram. This sug-
while in belting it is weak. Indeed, in belting the gests that glottal adduction is lower in operatic than
fundamental disappeared almost entirely in the in mixed style. Belting was produced with high sub-
spectrogram of the first vowel in "Summertime." glottal pressure, but this does not lead to particu-
This may reflect both voice-source and formant- larly high peak amplitudes. This suggests that glot-
frequency differences. tal adduction was comparatively high in belting, and
The amplitudes of the partials in the upper part of it seems relevant to examine the ratio between these
the spectrum, i.e., in the region of the singer's for- amplitudes and pressures. As glottal adduction obvi-
mant, were higher in operatic and mixed singing ously has a great influence over phonation, it is not
styles than in belting. In addition, the strongest par- surprising that an acoustic analysis of phonation
tial in this region appeared at a lower frequency in can reveal the degree of glottal adduction. Sundberg
the operatic than in the mixed styles. et al. (8) recently demonstrated that the ratio be-
Figure 2 shows SPL at 0.5 m for two performances tween the peak amplitude of the flow glottogram and
of the first phrases of one of the songs in operatic the subglottal pressure seems to vary consistently
and mixed singing. It can be seen that the SPL var- with the mode of phonation. Sundberg and Rothen-
ied only by small amounts between the renderings. berg (9) called this ratio peak glottal permittance.
Thus, the SPL seemed reasonably reproducible. Peak glottal permittance is shown in terms of a
Figure 3 compares SPL in one of the perfor- correlogram in Fig. 7. It can be seen that this per-
mances of the same phrases in the three singing mittance was much higher in operatic than in belting
styles. The figure shows that SPL was quite similar and mixed singing. This suggests that glottal adduc-
in operatic and mixed styles, while it was at least 10 tion was weaker in operatic style than in mixed
dB louder in belting. style and belting.
Figure 4 shows the relationship between subglot- Figure 8 shows the formant-frequency values ob-
tal pressure and SPL at 0.5 m in the sustained pho- tained from inverse filtering. It can be observed that
nations on the pitches G4, Ab4, and A4. It can be the three singing styles were also well separated
seen that operatic and mixed styles were produced with respect to formant frequencies. In the sus-
with lower subglottal pressures than belting, and tained vowel shown, the first and second formants
that the SPL values were accordingly lower. In fact, were lowest in operatic style. In mixed style and
for these tones the SPL seemed to be a linear func- belting, both the first and second formants were
tion of the log of the subglottal pressure. In other much higher, and the second was highest in mixed.
words, it seems fair to assume that the SPL in belting
is louder because the subglottal pressure is higher. EXPERIMENT II
Figure 5 shows flow glottograms for sustained The aim of this experiment was to examine the
tones sung at the same three pitches in the three typical pharyngeal shapes characterizing operatic,

Journal of Voice, Vol. 7, No. 4, 1993

304 J. SUNDBERG ET AL.

SUMMERTIME

OPERA MIXED BELTING

I
3
I
4 0
I
J t
2
t
3
1
4
I
0
,I

1 2
. ,fl
3
4-;
F4 -R E Q U E N C Y (kHz) 4-
ii " - r~,/,
3- ~'a'~',P~ '~ 3-

2- 2- 2-
rJ
Z

1- ~ 1-

0-~ 0- 0-

I I
Is
COR

opERA
BELTING

I I t I I I I I I I I I I I I
0 1 2 3 40 1 2 3 4 0 1 2 3 4
4- FREQUENCY (kHz)
I,

2-.~ ~, .~

2 ~ _ - " ~ T - ~ _ ?iT "

- . . . . . I
i

ls
Spectrograms and spectrum sections from a Kay Elemetrics Sona Graph of indicated vowels sung at identical pitches in operatic,
F I G . 1.
mixed, and belting style.

mixed, and belting singing styles using fiberoptics. ior with behaviors described in other investigations
This allowed us to find out if the acoustic charac- of belting.
teristics found for the different singing styles in Ex-
periment I were consistent with the pharyngeal Method
characteristics. Many previous investigations of An Olympus fiberscope (Tokyo, Japan) was in-
belting have also used fiberoptics. This allowed us troduced paranasally into our subject, who pre-
also to compare our single subject's belting behav- ferred to refrain from using topical anaesthesia. The

Journal of Voice, Vol. 7, No. 4, 1993

 A R T I C U L A T O R Y A N D P H O N A T O R Y COMPARISONS OF SINGING STYLES 305

110~

~100 l
.......u~
"~ 9 0 1 + ~ . . . . . . . . . . .

+l++o +..........
+ +....... '+++++++III + +l+l
6o I +~--,++ :~: + ~ : ~ + ........ -:::+ : = - - + ~ - V +::~+:- -II # +u u .... :
FIG. 2. Sound pressure level (SPL) at 0.5 m for
two performances o f the first phrases of one o f the
songs in operatic and mixed styles.
llOil OPERATIC _ _ ...
2

l°° l

60

image was recorded on a video-cassette recorder. belting styles. Within each of the three styles, how-
The monitor was placed so that both the investiga- ever, there was a considerable variability depending
tor and the subject could watch the image. Pitch on the vowel and the pitch. Therefore, in comparing
references were provided by means of a simple Ya- the three singing styles, it is necessary to select the
maha synthesizer. same vowel and pitch. A 3.5-min edited copy was
Wherever possible, the subject sang the entire made of the video recording to facilitate compari-
program in operatic, mixed, and belting styles. The son. In the tape, the same vowel sung at the same
program consisted of sustained vowels in the pitch pitch in all three singing styles was copied in se-
range A3 to E5 and the first line of the same songs quence on the tape. At the highest pitches, no sam-
sung in the same key, as in Experiment I. In addi- ples from belting were available, since the subject
tion, the subject also demonstrated a very special preferred not to produce these pitches in that sing-
kind of phonation that she called " c h i p m u n k , " ing style.
which appeared phonatorily interesting, as it As fiberoptic observations are difficult to quan-
seemed to represent a phonatory extreme. tify, the edited copy was presented to a panel of six
expert observers, five phoniatricians, and one
Data processing laryngologist. They were given a copy of the cas-
The pharyngeal landscapes seemed to differ in sette that they could watch and listen to repeatedly.
several respects between the operatic, mixed, and Their task was to describe, on the basis of what

110

100

9o

..+'- .................... + .-- -+- ....... iu+ x -L,+.+-~ If_-Xg/.': + ~+-

lJ+ I + -+-+, --v.~m,,. .... ~ ~-m+~+. . . . . . . . . . . . . . . . . . .
70 I . .'. . . . . . . . . ........ . . . .... +. .+.".~. . . ----+'-
. . . .- . .' + '. . . ++ , " + ++'l +~--+.....++--,~,+t-~\
++-b+++....... i.+.+++ ~+++++.+~+---+++~+~- ~I ..... _+ ++.+.I-~-++~ - , - ; ++

60;+-;+--++L'--+-+'
....... ++++~ ............... + ....++P +-+ ++ +;++-+++ ~ ~I
FIG. 3. Sound pressure level in one of the performances of the same phrases in operatic, mixed, and belting styles.

Journal of Voice, Vol. 7, No. 4, 1993

306 J. S U N D B E R G E T A L .

95 or two members. These were the pharyngeal side

[] walls, the sinus piriformes, the larynx height, and
90 []
[] the vocal folds. Our panel found that in belting, the
85 [] side walls of our subject were clearly advanced,
te~
,=;
OA thus constricting the pharynx; the constriction was
,~ 80
less in mixed and nil in operatic style. The sinus
,~/Y", 00PERA~c
75 /~ ~ MIXED piriformes seemed wide in operatic style, medium
[] BELTING in mixed, and narrow in belting.
70 The interpretation of the fiberscope view with re-
1.0 1.5 2.0 2.5 3.0 3.5
spect to larynx height was facilitated by one transi-
SUBGLOTTALPRESSURE Ps (kPa)
tion from mixed to operatic style, and during pho-
FIG. 4. Sound pressure level at 0.5 m as function of subglottal nation, when the subject firmly pinched the fiber-
p r e s s u r e in sustained phonations of the vowel [ae:] sung on the
indicated pitches in operatic, mixed, and belting styles. scope tube with her velum so that the vertical
position of the fiberscope tip remained constant.
This facilitated the interpretation of the video re-
they saw, what they believed was typical for the cording. The larynx was low in operatic style, high
four singing styles. No restrictions were imposed in belting, and in an intermediate position in mixed
regarding terminology. style. The visibility of the vocal folds was best in
operatic and mixed styles, and less in belting. These
Results differences were probably caused by changes in the
The answers from our panel have been combined larynx tube and also in the position of the ventric-
in Table 1. In their descriptions, the panel referred ular folds.
to 11 different structures or features. In Table 1,
synonymous terms have been changed so as to in- GENERAL DISCUSSION
crease uniformity. There were no instances of con-
flicting observations among the panel members. By and large, the typical pharyngeal characteris-
Some variables were mentioned by most members, tics for the different singing styles were consistent
while other variables were mentioned only by one with the acoustic characteristics found in Experi-
OPERATIC MIXED BELTING

PITCH A4 (440 Hz)

PITCH Ab4 (414 Hz)

# PITCH G4 (397 Hz)

TIME TIME TIME

FIG. 5. Flow glottograms for sustained t o n e s sung on the vowel [~e:] sung at the indicated pitches in operatic, mixed, and belting styles.

Journal of Voice, Vol. 7, No. 4, 1993

 A R T I C U L A T O R Y A N D P H O N A T O R Y C O M P A R I S O N S OF S I N G I N G S T Y L E S 307

7.0 1.4
m
® A
6.0 Z" A A
m
< 5.0 8 L~
z 1.3

4.0
k~
o Z
<
3.0
A 1.2
o
AA I vowEL [~:]
F O 0:393 Hz
=

Z Ab:415 Hz
2.0 ~ - - 0 A: 440 Hz
1.0 2.0 3.0
O OPERATIC
SUBGLOTTALPRESSURE(kPa) (~) IA MIXED
1.1 ®
FIG. 6. Flow glottogram peak amplitude versus subglottal pres-
sure for the vowel [~e:] sung at the indicated pitches in operatic, 700 800 900
mixed, and belting styles.
FIRST FORMANT FREQUENCY (Hz)
FIG. 8. Formant frequency values obtained from inverse filter-
ment I. The main acoustic differences were that in ing of the vowel [~e:] sung at the indicated pitches in operatic,
mixed, and belting styles.
belting, subglottal pressure and SPL were high,
while glottal permittance was low. These character-
istics seem to suggest a hyperfunctional type of associated with phonatory strain, which would cor-
voice use. In addition, formant frequency differ- respond to a high larynx position, a high degree of
ences were found. Now we can ask if the pharyn- glottal adduction, and a relatively high subglottal
geal characteristics observed in lateral pharyngeal pressure. In this qualitative sense, the observations
walls, sinus piriformes, and larynx height agree made from our fiberscope investigation agree with
with these acoustic differences. those of the acoustic analysis.
Unfortunately, the relationships between phar- The first formant frequencies in particular, but
ynx shape and acoustic voice characteristics are not also the second, were found to be higher in belting
understood in detail; we do not know explicitly and mixed than in operatic styles. A raised larynx
which pharyngeal landscapes are associated with has been shown to increase the frequencies of these
specific types of phonation. However, it is gener- formants considerably in the vowel [~e:] (10,11).
ally assumed that a constricted pharynx is typically Thus, the acoustic observations seem to be consis-
tent with the observations of pharynx shape. In this
PEAKGLOTTALPERMITTANCE investigation only one single subject was used.
(PEAK FLOW
~ GLOTTOGRAM AMPLITUDE) Therefore, it is particularly interesting to find out to
0.4 0.4 what extent her voice behavior was similar to that
A MIXEO / of other singer subjects that have been analyzed in
[] BELTING
previous investigations.
Using fiberlaryngoscopy and acoustic analysis,
0.3 0,3
Yanagisawa et al. (3) studied aryepiglottic constric-
tion in the vowel/i:/, as sung by five subjects in a
z z number of different singing styles, including belting
0.2 0.2 and operatic. The investigators found similarities in
this respect between these two styles. However,
none of our observers mentioned this parameter. It
0.1 ~ ~ 0.1
is possible that this difference is typical only for the
vowel /i/; a number of different vowels were in-
0.1 0.2 0.3 0.4 cluded in Experiment II, so a difference that ap-
IN OPERATIC
peared only in the v o w e l / i / m a y have escaped the
FIG. 7. Correlogram of peak glottal permittance values ob- panel.
served in mixed and belting styles for the same vowel [~e:] sung
at the indicated pitches; the values are plotted as function of the Other observations collected from the panel in
corresponding values in operatic style. Experiment II were in agreement with those re-

Journal of Voice, Vol. 7, No. 4, 1993

 308 J. S U N D B E R G ET A L .

T A B L E 1. Summary o f the characteristics o f the four singing styles shown on the video recording a

Parameter Observer Opera Mixed Belting Chipmunk

Sidewall (16) 1 Passive Some active Active
2 Nil Pressed Narrow
3 Vibrato Vibrato Less vibrato
4 Separated Constricted Very constricted Extra narrow
5 At low pitch:wide Very constricted Constricted
6 Separated Slightly constricted Constricted Narrow
Summary: Separated Medium Constricted
Sinus piriformes (14) 1 Wide Less visible Narrow
2 Nil Smaller Narrow
4 Wide Narrow
5 Wide, pitch dependent Rather wide Deep Very narrow
Pitch dependent Constricted
6 Wide Wide Wide?
Summary: Wide Medium Narrow
Larynx height (11) 1 Low Higher High
2 Low Lower? High
5 Varied Raised
6 Low Medium Elevated High
Summary: Low Medium High
Slightly visible,
Vocal fold (9) 1 Visible Visible vowel dependent
2 Stretched
3 Stretched
4 Visible Visible Less visible
5 Stretched, thin
6 Visible
Summary: Visible Visible Less visible
Ventricular fold (8) 2 Open Compressed Compressed Narrow
5 Slightly adducted Very adducted
6 Passive Slightly active Adducted Compressed
Summary: Separated Medium Adducted
Arytenoid (7) 1 Adducted
3 Top tight
4 Top split Top split Top split Top tight
Slightly tilted Tilted
5 Top split
6 Adducted Compressed Compressed
Summary: Top split Top split Top split
Tilted
Larynx tube (6) 1 Passive Not fully open Slightly narrow
4 Wide Constricted Narrow
5 Stable Slightly narrow Very constricted
Summary: Wide Medium Constricted
Epiglottis (5) 2 Straight Tilted
3 Vibrato Vibrato Close to arytenoids
Glottis (5) 2 Wide Narrow
3 Incomplete closure Posterior chink No chink
Esophagus entrance (3) 2 Nil Compressed Compressed Narrow
Aditus (3) 3 Wider than operatic Narrow Very narrow
5 Very constricted
6 Slightly constricted
a Data are descriptions of six expert observers. The numbers within parentheses indicate the total number of times that the parameter
was mentioned in the judges' characterizations of opera, mixed, and belting.

ported in previous investigations. The most fre- 12). In addition, Estill (12) found that the EMG sig-
quent observations mentioned by our panel con- nal from the middle constrictor was greater in belt-
cerned side walls, sinus piriformes, and larynx ing than in operatic style; this would lead to
height. Precisely these structures have been men- advanced side walls in belting, which were ob-
tioned in all previous investigations of belting (1,3, served in the present investigation. Using electro-

Journal of Voice, Vol. 7, No. 4, 1993

 A R T I C U L A T O R Y A N D P H O N A T O R Y CO MP A R ISO NS OF S I N G I N G S T Y L E S 309

glottography, Estill found belting to be typically as- may have entailed differences in vocal fold length,
sociated with a long closed phase, while operatic which may reduce the comparability of this mea-
singing had a short closed phase (12). Our inverse- sure. More theoretical work is needed before such a
filtering analysis showed a clear difference of this conclusion can be drawn.
kind between our subject's operatic and belting The formant frequencies of the sustained vowels
styles. sung at identical pitches in the three singing styles
Belting and operatic singing on the vowel/i/have were very similar within style. On the other hand,
been analyzed acoustically and physiologically by no corresponding consistency in the formant values
Estill et al. (2) and Yanagisawa et al. (3). Their sub- could be observed in the songs. This suggests that
ject was able to demonstrate belting and operatic in singing sustained, isolated vowels, our subject
singing within a wider range than our subject. Still, used a stable and neutral set of formant frequencies,
the acoustic characteristics were similar to those whereas in performing songs, she used formant fre-
found in the present investigation. They found that quencies for musical expression, i.e., vowel color-
the mean spectrum envelope of operatic singing was ing. The variability of formant frequencies in sung
dominated by the fundamental, while that of belting performances is an interesting question for future
had a weak fundamental and relatively strong over- investigation.
tones; the spectrum level difference between oper- It is interesting that during the inverse-filtering
atic and belting styles was no less than 35 dB at 3 experiment all data were clearly stratified; Figs. 4-8
kHz for the fundamental of this vowel. Although we indicated that the three singing styles were consis-
found a somewhat smaller difference for the vowel tently differentiated with regard to several phona-
/a/, qualitatively similar spectrum differences were tory parameters: subglottal pressure, SPL, peak
found. This suggests that our subject's phonatory glottal permittance, and formant frequencies. This
behavior in belting was similar to that studied by means that our subject used her voice both consis-
Estill et al. (2) and Yanagisawa et al. (3). tently and differently in the three singing styles. As
Estill et al. (2) also studied the glottal voice a result, we obtained a rather complete description
source by using electroglottography. This offers an- of how she used her voice in the three singing
other opportunity to compare our results with those styles.
of a different subject. The waveforms indicated that
the closed phase was longer in their subject's belt- CONCLUSIONS
ing compared with operatic singing. This agrees
with oar flow glottogram findings. The agreement Belting and operatic styles differed with respect
supports the aforementioned conclusion that our to subglottal pressure, glottal adjustment, and artic-
subject's behaviors in operatic and belting style ulation. Operatic singing seemed characterized by
were similar to those studied by Estill et al. (2) and moderate subglottal pressures and glottal adductive
Yanagisawa et al. (3). forces, and the formant frequencies suggested a
These instances of agreement suggest that our moderate degree of jaw opening and a lowered lar-
subject offered typical examples of belting. Thus, ynx. Belting was produced with high subglottal
although our investigation concerned only one sub- pressures and greater glottal adductive forces, and
ject, it apparently yielded representative results. In the formant frequencies suggest a wide jaw opening
addition, the acoustic analyses and the fiberoptic and an elevated larynx. Mixed singing seemed char-
examination of the larynx produced results that ap- acterized by moderate subglottal pressures, moder-
pear to be compatible. ate glottal adductive forces, and by formant fre-
We determined the peak glottal permittance, i.e., quencies, suggesting a wide jaw opening and a
the ratio between the peak glottal airflow and the raised larynx. Although the pharyngeal landscape
underlying mean subglottal pressure. The result varied considerably with vowel and pitch, belting as
showed that the permittance was clearly highest in compared to operatic style seemed characterized by
operatic and lower and approximately the same in a high larynx position. In addition, the side walls
belting and mixed styles. Peak glottal permittance were advanced and the sinus piriformes were small
must depend on glottal adduction. However, the in many vowels in belting. The subglottal pressure
similarity in the permittance does not necessarily differences seemed to result in considerably higher
imply that glottal adduction was the same in this SPL in belting than in mixed and operatic singing.
case. The great difference in subglottal pressure The amplitude of the voice-source fundamental was

Journal of Voice, Vol. 7, No. 4, 1993

310 J. SU ND B E R G E T AL.

higher in operatic than in belting and mixed styles, 3. Yanagisawa E, Estill J, Kmucha S, Leder S. The contribu-
probably because of the greater adductive forces in tion of aryepiglottic constriction to 'ringing' voice qual-
ity--A videolaryngoscopic study with acoustic analysis. J
belting. As our subject seemed to produce typical Voice 1989;3:342-50.
examples of belting and operatic singing, we con- 4. Miles B, Hollien H. Whither belting? J Voice 1990;4:64--70.
cluded that this was a representative description of 5. Vennard W. Singing: the mechanism and the technique.
belting. New York: Carl Fischer, 1967.
6. Reid CL. A dictionary o f vocal terminology. An analysis.
Acknowledgment: The fiberscope equipment was pro- New York: Joseph Patelson Music House, 1983.
vided by the Phoniatric Department of the Malm6 Gen- 7. Rothenberg M. A new inverse-filtering technique for deriv-
eral Hospital. The kind assistance of Grad. Eng. Leif,~k- ing the glottal airflow waveform during voicing. J Acoust
erlund of this department is gratefully acknowledged. Soc A m 1973;53:1632-45.
Grants from the Royal Swedish Academy of Music and 8. Sundberg J, Scherer R, Titze I. Phonatory control in male
from The Voice Foundation financed costs for this inves- singing: a study of subgtottal pressure, fundamental fre-
tigation. quency, and mode of phonation on the voice source. STL-
QPSR 1990;No. 4:59-79.
9. Sundberg J, Rothenberg M. Some phonatory characteristics
REFERENCES of singers and nonsingers. STL-QPSR 1986;No. 4:65-77.
10. Fant G. Nonuniform vowel normalization. S T L - Q P S R
1. Lawrence V. Laryngological observations on belting. J Res 1975;No. 2-3:1-19.
in Singing 1979;2;26-8.
2. Estilt J, Baer T, Honda K, Harris K. Supralaryngeal activity 11. Nordstr6m P-E, Sundberg J. Raised and lowered larynx--
in a study of six voice qualities. In: Askenfelt A, Felicetti S, the effect on vowel formant frequencies. J Res in Singing
Jansson E, Sundberg J, eds. Proceedings o f the Stockholm 1983;VI(2):1-6.
Music Acoustics Conf. 1983, Vol. I, Stockholm: The Royal 12. Estill J. Belting and classic voice quality: some physiological
Swedish Academy of Music, Publ. No. 46:1;1985:157-74. differences. Med Probl o f Performing Artists 1988;3:37-43.

Journal of Voice, Vol. 7, No. 4, 1993