ACTA UNIVERSITATIS OULUENSIS FINLAND
Series D Medica 401
METTE PEDERSEN
BIOLOGICAL DEVELOPMENT AND THE NORMAL
VOICE IN PUBERTY
Academic Dissertation to be presented with the assent of the Faculty of Medicine, University of Oulu,
for public discussion in Auditorium 7 of the University Hospital of Oulu, on May 22nd, 1997, at 12
noon.
OULUN YLIOPISTO, OULU 1997
1
Copyright © 1997
Acta Univ. Oul. D 401, 1997
Manuscript received 24 January 1997
Accepted 20 March 1997
Communicated by
Professor Aatto Sonninen
Docent Lucyna Schalen
ISBN 951-42-4592-X
ISSN 0355-3221
OULU UNIVERSITY PRESS
OULU 1997
2
To Thomaner
Choir, Leipzig
3
4
Pedersen, Mette, Biological development and the normal voice in puberty
Department of Otolaryngology KAS Gentofte University Hospital. 2900 Copenhagen, Hellerup,
Denmark, Statens Serum-institut, Hormone Department 2300 Copenhagen S., Denmark and
Department of Otolaryngology and Phoniatrics, University of Oulu, FIN
90220, Oulu, Finland
Acta Univ. Oul. D 401, 1997
Oulu, Finland
(Manuscript received 24 January 1997)
Abstract
The pubertal voice in musically trained (voice conscious) boys and girls was investigated with voice
range profiles and fundamental frequency (F0) in running speech during reading of a standard text. The
methods were based on: 1) development and evaluation of the function of phonetograph 8301 made by
the firm Voice Profile, 2) combined electroglottographic and stroboscopic examination of the
movements of the vocal folds in speech. The voice analysis was compared with measurements of: 1)
pubertal stages in youngsters and 2) hormonal analysis of all androgens and in girls also estrogens.
The voice range profile measured total pitch and loudness range and an area calculation was made
of measured semitones x dB(A).
An evaluation was made of the electroglottographic curve, combining it with a marking of the
stroboscopic phases of the vocal folds on the curve with a photocell. The electroglottographic single
cycles were stable and 2000 consecutive electroglottographic cycles were measured in 48 boys and 47
girls, aged 8-19 years, in order to measure fundamental frequency in a reading situation.
Individual and average voice range profiles for sopranos, altos, tenors and bassos were examined.
The yearly change of voice range profiles showed a correlation to total serum testosterone of r = 0.72 in
the boys, and serum estrone of r = 0.47 in the girls.
The change in fundamental frequency in puberty was analyzed in 48 boys. Single observations of
the fundamental frequencies showed that total serum testosterone over 10 nmol/l serum probably
represented values for a boy with a pubertal voice. The voice parameters were analyzed in 47 girls.
However, hormonal analysis and pubertal examination was only possible in 41 girls.
F0 was only related to estrone r = -0.34 (p < 0.05). The increase of estrone and of fundamental
frequency range (F0 range) had a predictive value (p < 0.05) for the fall of F0 from 256 to 241 Hz in
puberty.
Keywords: voice range profile, fundamental frequency, normal puberty, estrogens, androgens
5
6
ACKNOWLEDGEMENT
Within Scandinavia Finland has a long tradition of being at a frontier of medical voice research. This is
the invaluable basis for me to apply for a degree here. I am indebted to Professor Aatto Sonninen M.D.,
and Professor Erkki Vilkman, M.D., for their extraordinarily thoughtful and wise advice.
Professor Sonninen and docent Lucyna Schalen, M.D., have evaluated this manuscript kindly.
Choosing them was based on an admiration for their research and ability to structure phoniatrics
practically in Scandinavia as well as the European Union.
More than ever, consciousness of the voice is in focus in many ways. The voice is a very
"effective" instrument if you have a message. Consciousness of the voice in puberty was the main
subject of this work, which was carried out at the Copenhagen singing school and in the ENT clinic in
the centre of Copenhagen, because the subject of phoniatrics seemed difficult to incorporate into the
traditional ENT departments in Denmark. My co-workers were never in doubt of the importance of our
working field.
Many have understood the necessity of undertaking this research, especially the team at Statens
Seruminstitut, Copenhagen and the pediatricians, including dr.med. Søren Krabbe, M.D., who helped in
the research. Several kind secretaries have assisted me, especially Aase Ørslev who was always there
when I needed her, as were Mr. Christian Villumsen and Mr. Kim Rosentoft when I needed technical
assistance.
My daughter Michala was in puberty during the research period and now herself is receiving
excellent scientific training from the Dept. of Physiology at University of Oxford, from where she gave
me many good suggestions. Jolanda Rodio, my singing teacher in Switzerland is also remembered here,
as my adviser. Supporting friends and colleges are kindly thanked for their research suggestions,
among others Lene Høj, M.D., Jytte Udsen, M.D., Ms. cand. pout. Hjørdis Dalsgaard and Mr.
cand.polit. & exam. art. Martin Harms.
This study was partly supported by P. Carl Pedersen Foundation, Copenhagen Education
Authorities, Scientific Research Council of the Danish State and Phoniatric Fund of the Finnish
Foundation of Ear Diseases.
The conversion of the thesis to internet was made by Lars Paaske.
7
8
ABBREVIATIONS AND DEFINITIONS
DHEAS
EGG
El
E2
E1S04
F0
F0-range
LTAS
Octaves
SHBG
ST
SPL
Total pitch range
dihydroepiandrosterone sulphate
electroglottography
estrone
estradiol
estrone sulphate
mean fundamental frequency in running speech in a reading situation
of a standard text
frequency variation in semitones in running speech in a reading situation
of a standard text (= voice range)
long term averaged spectrogram
German discription: C-c-c l-c2-c3,
American: C3 = c, C4 = c1, C5 = c2 etc.
sex hormone binding globulin
semitone in the octave, defined from the phonetogram
sound pressure level
the range from the lowest to the highest semitone in the phonetogram
9
10
LIST OF SEPARATE COMMUNICATIONS
This thesis is based on the following publications which will be referred to by their Roman numerals.
I
Pedersen, M.F. (1977) Electroglottography compared with synchronized stroboscopy in normal
persons. Folia Phoniatr. 29: 191-199.
II
Pedersen, M.F., Møller, S., Krabbe, S., Bennett, P. & Svenstrup, B. (1990) Fundamental voice
frequency in female puberty measured with electroglottography during continuous speech as a
secondary sex caracteristic. A comparison between voice, pubertal stages, oestrogens and
androgens. Int. J. Pediatr. otorhinolaryngol. 20: 17-24.
III
Pedersen, M.F., Møller, S., Krabbe, S. & Bennett, P. (1986) Fundamental voice frequency
measured by electroglottography during continuous speech. A new exact secondary sex
characteristic in boys in puberty. Int. J. Pediatr. otorhinolaryngol. 11 21-27.
IV
Pedersen, M.F., Møller, S., Krabbe, S., Munk, E. & Bennett, P. (1985) A multivariate statistical
analysis of voice phenomena related to puberty in choir boys. Folia Phoniatr. 37: 271-278.
V
Pedersen, M.F. (1993) A longitudinal pilot study on phonetograms / voice profiles in prepubertal choir boys. Clin. Otolaryngol. 18: 488-491.
11
12
CONTENTS
ABSTRACT
ACKNOWLEDGEMENTS
ABBREVIATIONS AND DEFINITIONS
LIST OF SEPARATE COMMUNICATIONS
1. INTRODUCTION ………………………………………………………………………………15
2. REVIEW OF THE LITERATURE ……………………………………………………………. 16
2.1. Voice range profile ………………………………………………………………………... 16
2.1.1. Voice range profiles / phonetography, the method (V) …………………………….. 16
2.1.2. Voice range profiles/phonetography in puberty (II-IV) ……………………………. 17
2.2. Stroboscopy / Electroglottography (I) …………………………………………………….. 17
2.3.1. Fundamental frequency (F0) (I) …………………………………………………….. 18
2.3.2. Fundamental frequency in puberty (II-IV) …………………………………………. 19
2.4. Pubertal development (II-IV)………………………………………………………………. 19
3. PURPOSE OF THE STUDY …………………………….…………………………………….. 21
4. METHODS …………………………………………………………………………………….. 23
4.1. Voice range profiles / phonetography (II-V) ……………………………………………… 24
4.1.1. Instrumentation ……………………………………………………………………... 24
4.1.2. Analysis …………………………………………………………………………….. 25
4.2. Fundamental frequency (F0) in running speech in a reading situation (I-IV) …………….. 25
4.2.1. Stroboscopy with synchronous electroglottography (I) ……………………………. 25
4.2.2. Electroglottographic measurement of running speech in a reading
situation (II-IV)………………………………………………………………………25
4.3. Hormone and pubertal stage analysis (II-IV) ……………………………………………… 26
4.4. Specific statistical calculations (II, IV) ……………………………………………………. 26
4.5. Subjects ……………………………………………………………………………………. 27
5. RESULTS ……………………………………………………………………………………… 28
5.1. Voice range profile measurements (see appendix) …………………………………… 28
5.2. Measurement of vocal fold movements (I) …………………………………………… 29
5.3. Fundamental frequency in running speech in a reading situation in boys
in puberty (III) ………………………………………………………………………... 29
5.4. Fundamental frequency in running speech in a reading situation in girls
in puberty (11) ………………………………………………………………………... 30
5.5. Specific results (II, III, IV)……………………………………………………………. 30
13
5.6. Longitudinal stud V ……………………………………………………………………..... 37
DISCUSSION ………………………………………………………………………………..... 38
6.1. Electroglottography (I-III) ………………………………………………………………… 38
6.2. Fundamental frequency (II-IV) ……………………………………………………………. 38
6.3. Voice range profiles / phonetograms (V) ………………………………………………..... 39
6.4. Voice range profiles in puberty (II, IV, V) ………………………………………………... 40
6.5. Fundamental frequency (F0) in puberty (II-IV) …………………………………………… 41
6.6. Perspectives ………………………………………………………………………………... 42
7. CONCLUSION ………………………………………………………………………………… 43
8. REFERENCES ………………………………………………………………………………… 45
APPENDIX
ORIGINAL PAPERS
6.
14
1. INTRODUCTION
During my work as a laryngologist for 10 years in the Copenhagen singing school under Mogens
Wøldike, questions about the biological development of the pubertal voice often arose. The most
important question was always: does a boy lose the high notes of his voice due to infection, allergy,
misuse or puberty.
Because these causes of voice function failure interfere with each other, it was difficult to give an
exact answer. Still it is our conviction that the main difficulty lies in judging the loss of high notes due
to puberty.
Measuring the voice is necessary for youngsters to understand the pubertal development of their
voices. Here, stroboscopy alone seemed insufficient because registration of pubertal pitch and loudness
changes were not taken into account. For this reason the fundamental frequency in running speech in a
reading siatuation was computed. In addition voice range profiles / phonetography in phoniatrics, a
corresponding measurement to audiometry in audiology, allow a more exact definition of the pubertal
voice.
The mean fundamental frequency measurement in running speech (F0) is an international topic of
discussion. The electroglottographic method has been preferred because it is simple and well defined in
a reading situation. The exact objective measurement of voice was a basis for discussion with
pediatricians and researchers of androgens and estrogens, thereby raising their interest in voice research
as a basis for further understanding of the effect of hormones on voice.
15
2. REVIEW OF THE LITERATURE
2.1. Voice range profiles
2.1.1. Voice range profiles /phonetography, the method (V)
Phonetography is a method of illustrating pitch and loudness range (Calvet et al. 1952). The concept of
standardized phonetography has since 1981 been based on the proposal of the Union of European
Phoniaticians (UEP). Sixteen researchers in Europe from UEP were at that time questioned by Seidner
and Schutte. The results were confirmed by Titze (1994). Some discussion on the set-up of
phonetography has yet to be resolved.
The UEP-standard included a manual measurement with a sound pressure level meter from the
Danish firm Bruel and Kjaer and with a directional microphone placed at a 30-centimeter distance and
a 90-degree angle from the mouth. This equipment depends on manual operation with consideration to
registering moment and speed measurement. Thereafter the loudness measurement research carried out
by many phoniatricians developed rapidly, and resulted in computerized equipment (Bloothooft 1981,
Klingholz & Martin 1983, Pedersen et al. 1984, Seidner et al. 1985, Vilkman et al. 1986, Hacki 1988,
1989, Gramming 1988, Pabon 1991, McAllister et al. 1994). The measurement of loudness is, of
course, dependent on the total sound. Seidner and Wendler (1982) and Sturzeberger et al. (1982)
discuss phonetograms combined with the singing formant (Sundberg 1987) in trained and untrained
singers. A comparison was made in 1988 between their manual method and the computerized set-up
with phonetograph 8301 (Pedersen et al. 1988) which showed good agreement between the two
methods of loudness measurement. Byrne et al. (1994) present the similarity of LTAS in 12 languages,
in this way showing phonetography to be language independent. Loudness variations in pathological
cases were discussed by Gramming (1988), who claimed that the same SPL values were obtained with
different modes of phonation.
Pedersen et al. (1985) showed reduced loudness and semitone range measurements in
phonetograms of infected and allergic school children, compared to trained child singers.
Standardized refinement possibilities of the measurement were made with the newest sound
pressure level meters by Bruel and Kjaer, supplementing the SPL with a maximum peak measurement.
Sustaining a given tone is not possible for unmusical people as discussed by Hirano (1989). However,
if the patient cannot reproduce the indicated tone, the problem
16
can be solved with a supplementary instrumentation with an open 1/2 octave measurement and short
time intervals (Pedersen & Lindskov Hansen 1986, Pedersen 1995).
Since 1993 on-line uptake of loudness and pitch range has been possible with Kay Elemetrics
model 4326, although it is not yet possible to average results with their equipment.
2.1.2. Voice range profiles/phonetography in puberty (II-IV)
In the first years of early phoniatrics, careful studies of semitone ranges of normal school child were
carried out by Flatau and Gutzman (1905). A survey of research on children's voices was proposed at
the international conference for logopedics and phoniatrics in Copenhagen in 1936 and later carried out
by Weiss.
The survey has a time span of 4000 years and focuses almost exclusively on boys' and eunuchs'
voices. The last eunuch died in 1922 (Siegel 1994). The average age of the mutation was recognized to
be around 14.5 years, with one octave of the fundamental frequency lowering in boys, and in girls 1/3
octave (Weiss 1950).
Many others have reached the same conclusions (e.g. Frank and Sparber 1970, Wendler and
Seidner 1987). Blatt (1983) discusses all aspects of educating voices in puberty. In 1984 Komyama et
al. analyzed normal children during puberty, but unfortunately without comparison to other phenomena
of puberty and only with a lower borderline of 60 dB(A). It is possible to further lower the sound level
measurements, therefore their results are not comparable to our line of 30-40 dB(A).
Meuser and Nieschlag (1977) found voice categories (i.e., bass - baritone - tenor) in adult men to
be related to the concentration of serum testosterone. Large and Iwata found differences in formants
related to voice categories in adults as early as in 1972. We also found that it is necessary to
differentiate between voice categories, if an exact evaluation of the pubertal development of the voice
is to be obtained.
Since we started our examinations, Klingholz et al. (1989) in Tölzer Knabenchor have studied
phonetograms with an automatic classification of the sopranos and altos. Konzelmann et al. (1989)
have examined the phonetograms and singing formants in singers and non-singers. Grit Bühring (1990)
has a survey of recent literature in her dissertation and tries to set the circumscripts of phonetograms on
mathematic formulas to compare measurements between the Copenhagen boys choir and the Thomaner
choir.
Behrendt (1989) has followed a fistula register in phonetograms into the adult life of the Thomaner
choir singers. The observations were not compared to other biological parameters. Walker et al. (1993)
found that testosterone increased in pubertal stage III and voice decreased in stage IV in boys. They
showed that a critical vocal cord length and testosterone value was required to initiate voice breaking.
Testicular volumen was a poor predictor of voice brake and saliva testosterone concentrations.
2.2. Stroboscopy / Electroglottography (I)
Phonetography in conjunction with stroboscopy are by J.Wendler (1989) suggested as basis
investigations for voice disorders in the International Federation for Ear-Nose-Throat So-
17
cieties. Stroboscopy has not been applied to quantitative evaluation of the voice but to differentiations
between normal and pathological vocal folds. Videostroboscopy is an excellent means of visualizing
the movements of the vocal folds and of presenting the many questions on the vocal folds (Wendler et
al. 1987).
Pedersen et al. (1988) have prepared a video film with the stroboscopic equipment from Timcke at
the phoniatric department of the medical university in Hannover. The film on Danish choir boys has,
among others, been presented at the Voice Symposium in New York, Manhattan School of Music
1988. It was difficult to visualize the quick register shifts of movements of the vocal folds in puberty.
The combination of stroboscopy and electroglottography seemed relevant at the time, when the
discussion of the interpretation of the glottography curve again was blooming at the international
conference of logopedics and phoniatrics in Interlaken in 1974. Schönhärl had performed a systematic
study of stroboscopies in patients with voice disorders, but a statistical quantitative analysis of the
treatment results was not possible (Schönhärl 1960, Pedersen, 1974, Pedersen et al. 1980).
Among others, Frokjaer-Jensen and Thorvaldsen (1968), based on the results by Svend Smith
(1954) and Fabre (1957), sent a high frequency current through the throat at the level of the vocal folds
and found the impedance variation representing the real time changes in the movements of the vocal
folds in the electroglottograph. This research area culminated in, for instance, the research by Lecluse
(1977) and Smith (1981). It has always been a good idea to combine several methods when one of them
has an uncertain interpretation, which was the case till now with the amplitude and to a certain extent
with the single parts of the electroglottographical curve. Anastopolo and Karnell (1988) and Karnell
(1989) have made a videostroboscopy equipment combined with electroglottography. In the future it
will, of course, be possible to combine several pictures of the videoscreen by freezing some of them or
comparing them with average pictures, as a basis for still more precise quantitative videostroboscopies
(Colton et al. 1989, 1995, Hirose et al. 1988). The equipment will probably be clinically usable. The
method seems optimal for illustrating the mucosal movements, as first described by Svend Smith in
1954.
2.3.1. Fundamental frequency (F0) (I)
The electroglottographical signal was found suitable for giving information on the single movements of
the vocal folds as well as for precisely registering contours of the fundamental frequency in running
speech (Loebell 1968).
The firm of Frokjaer-Jensen, (1968), FJ Electronics, developed an electroglottograph. Based on
this, Kitzing (1979) in his thesis analyzed fundamental frequencies. Guidet and Chevrie-Muller (1979)
have used an electroglottograph for fundamental frequency analyses in neurological patients as well as
voice registers (Roubeau et al. 1987). The different configurations of the electroglottograms in singers
compared to non-singers were discussed by Pedersen (1978). Also, the variation in fundamental
frequency by measuring a histogram configuration has been carried out by constructers, particularly
Fourcin and Abberton (1971) and discussed by Kitzing (1990). Askenfelt et al. (1980) compared vocal
fundamental frequencies measured with a contact microphone and electroglottography.
18
But the historically interesting studies of fundamental frequency measurements in youngsters up
until the last 15 years have been carried out primarily with pointer instruments and often with readings
from of oscillograms (Baken 1987). Baken presents a survey of measurements of fundamental
frequency studies as do Schulz-Coulon and Klingholz (1988). In the future new directions may appear.
The measurement using neural networks for pattern recognition could very well be developed for
clinical use analyzing patterns of fundamental frequency (e.g. Elman & Zipser 1988, Rihkanen et al.
1994).
2.3.2. Fundamental frequency in puberty (II-IV)
A survey of measurements of fundamental frequency in children was given by Baken (1987) and also
by Schulz-Coulon and Klingholz (1988), Hollien and Malcik (1967), Michel et al. (1966), McGlone
and McGlone (1972), Hollien and Shipp (1972), Hollien et al. (1994).
Fairbanks et al. (1949) and Fitch and Holbrook (1970) have made basic studies on the
development of fundamental frequency in children, but regrettably without pediatric evaluation.
Vuorenkoski et al. (1978) have measured the pubertal mean fundamental frequency in endocrinological
patients compared with pubertal stages. Heineman (1976) gave an overview of hormonal voice
disorders, listening evaluation, stroboscopies, sonograms and semitone ranges of the voices of patients
with gonadal diseases. He describes the vocal cords in precocious puberty as thickened with reduced
marginal movements at stroboscopy, a bass voice and a semitone range of 1 octave. Bastian and Unger
(1980) have compared the fundamental frequencies with pubertal stages.
Unfortunately, the literature does not contain many studies where sexual maturation measured by
hormones is shown to be related to the voice. Considering the voice as an interesting biological
parameter is, of course, not limited to puberty, but also voice in menopause as compared to puberty in
girls has had interest (Truuverk & Pedersen 1992).
Calculating work in future will probably be much swifter and new perspectives for clinical use will
open up (Bühring & Pedersen 1992, Krusnevskaja & Pedersen 1992). The biological effect of
testosterone treatment in the senescent was discussed by Behre and Nieschlag (1995). It is continually
necessary for the biological voice research to renew itself in connection with the best research in other
fields (e.g. in acoustical science in music (Siegel 1994) as well as neuroendocrinology of estrogens in
puberty (Rodrigue-Sierra 1986, Blaustein 1986).
2.4. Pubertal development (II-IV)
Puberty is defined as the achievement of reproductive function, clinically reflected in obtaining
secondary sex characteristics. Normal pubertal development is complicated. Brook (ed.) provides a
survey (1995), based in part on Tanner and Whitehouse (1976).
In Western European males a peak height velocity is seen from the age of 12-16 years. The pubic
hair stage II ranges from 9.5-14 years and pubic hair stage III from 11.5-16 years. Stage IV ranges from
12-16.5 years and pubic hair stage V from 13-17.5 years of age measuring the growth of the hair area
in the pubic region.
19
The genital stages related to a testicular enlargement measured with an orchidometer comparing
the testis to a standard, changes from genital stage II at 10-14 years and genital stage V from 13-17
years of age (Brook 1995).
In Western European females the time of menarche is at 11-15.5 years, the peak height velocity at
10.5-14 years. The pubic hair stage II ranges from 9-14 years, pubic hair stage III from 10-14.5 years
and pubic hair stage IV from 10.5-15 years. Pubic hair stage V (young adults) ranges from 11.5-19
years. Breast stage II ranges from 9-13.5 years, stage III from 10-15 years, breast stage IV from 10.515.5 years and breast stage V from 12-19 years of age (Brook 1995).
Mean plasma testosterone in normal boys by the stage of maturation gradually changes from stages
I to V, and the same is the case for estrogens in adolescent females (Brook 1995).
Brook suggests that voice development only begins with a voice change beginning at 13 years of
age, and the adult voice is reached by 15 years of age. A reference is given to Karlberg and Taranger
(1976).
Calcium metabolism is more related to sexual hormones than to age. This situation could be the
same for the voice (Krabbe 1989). The heights od Danish children were examinated by Andersen
(1968), later by Roed et al. (1989) and Hertel et al. (1995), they correlate well to our studies.
Brook mentions that knowledge on the heart and respiratory organs in puberty is limited, and that
these parameters till now have been related to height and development of secondary sex characteristics
(1995). This seemed to have been the case for voice phenomena in the pediatric literature as well. Hagg
and Taranger (1982) describe the voice as childish, pubertal or adult. Walker et al. (1993) found the
significant increase of testosterone in pubertal stage III and vocal cord length inversely related to vocal
frequency decrease in pubertal stage IV.
Kahane (1982) analysed the development of the thyroid cartilage in puberty in relation to height.
Kurita et al. (1980) have measured the enlargement of the vocal folds in puberty. The entire length of
the vocal folds gradually develops until the age of 20 years in both sexes, and in adults the vocal length
in males is 17-21 mm, and in females 11-15 mm. The length did not differ between the sexes under the
age of ten, above the age of 15 the vocal fold is longer in males than in females. In the vocal folds of
12-year-olds there is a slight differentiation of the layer of elastic fibers from that of collagenous fibers.
In the 16-year-olds there is a layer structure as observed in adults (Hirano et al. 1983). Hirano et al.
(1988) found a greater tension effect of the cricothyroid muscle in females than in males, which
together with a greater dorsal component in the abduction of the tip of the vocal process of the
arytenoid cartilage in females might have important physiological and clinical implications. Kersing
(1983) found no evidence that the vocal musculature differs in any important respect from other
skeletal muscles. His findings tend to support the hypothesis that the primary function of the vocal
cords is as that of a sphincter and that sound production is a secondary functional development due in
part to the development of type I muscle fibers. This corresponds to Sato and Hirano (1995) who
studied the development of maculae flavae at the ends of the vocal ligaments which are immature at
birth. It is suggested that the tension during phonation (vocal fold vibration) causes the fibroblasts to be
in an active phase and to synthesize fibers. Hollien (1983) comments that the control of vocal
frequency appears to be mediated by variation in vocal fold mass and stiffness plus changes in the
subglottal air pressure.
Many studies have been carried out on rats and dogs etc., however, singing canaries may be of
greater interest, as androgen induced behavioral effects on singing were found as a
20
probable secondary or independent result of androgens' primary and immediate action on target gene
transcription (Nastiuk & Clayton 1995). Frogs have also been studied, and it has been suggested that
estradiol may increase transmitter release at laryngeal synapses during juvenile development (Tobias &
Kelley 1995).
In male frogs a rapid fiber addition was seen in the male laryngeal muscles at time of sexual
differentiation (Marin et al. 1990, Tobias et al. 1991).
Normal sexual endocrinological development is regulated by gonadotropin releasing hormone
from the hypophysis, a decapeptide (Brook 1995). Thereafter, the pulsating secretion of luteinizating
and follicle stimulating hormones from the hypophysis commences, which regulates the growth of the
testes and the ovaries, where the sexual hormones are produced. Our measuring methods are described
by Lykkesfelt and co-workers (1985). The measurements were comparable to others (Tanner &
Whitehouse 1976, Apter & Vihko 1985).
Strel'Chyonok and Vihko (1985) give a survey of the function of the sex hormone binding globulin
(SHBG). With the measurements of steroids in saliva new possibilities open up in pathology for voice
analysis, comparing voice and hormonal pathologic development in boys as well as girls (Young et al.
1988).
In humans most pubertal pathological histories are case stories of relations between testosterone
and the lowering of the voice (e.g. Yukizane et al. 1994), as well as a longitudinal study of treatment of
girls with Turner's syndrome, with growth hormone, oxandrolone and ethinyl estradiol showing a
distinct decline in speaking fundamental frequency during the first year (Andersson-Wallgren &
Albertsson-Wikland 1994).
With the new understanding of brain regulation of sex hormone binding globulin (SHBG), a
possibility for bringing pubertal voice studies into brain research has opened up, e.g. Blaustein (1986)
and Miranda et al. (1996), who measured steroid receptors and hormone action in the brain.
21
3. PURPOSE OF THE STUDY
The aim of the study was to discuss the following questions:
1.
2.
3.
4.
5.
6.
What are the technical choices for measuring the biological voice phenomena pitch and
loudness in voice range profiles in puberty?
Which method reliably reflects the vibratory frequency of the vocal folds usable for measuring of
fundamental frequency in puberty? How does electroglottography correspond
to vocal fold vibrations?
How do the pitch and loudness ranges of the voice in musically trained boys and girls develop
in puberty?
In which ways does the fundamental frequency (F0) in running speech in a reading situation
change in puberty?
In which ways is the voice related to hormonal development?
How is pubertal maturation connected to voice change in puberty?
22
4. METHODS
Table 1. Overview of the instruments and parameters sed in the present study.
23
4.1. Voice range profiles / phonetography (II-V)
The pitch and loudness ranges of the singers were measured according to the standard proposed by the
Union of European Phoniatricians (1981) and stored in our phonetograph. In phonetography, the
highest and the lowest sound level the subject is able to produce at a given fundamental frequency (F0)
is recorded. Repeating this measurement over the total F0 range of the subject provides a
documentation of the total biological capability of a given person. Standard semitones can be chosen.
The equipment, developed by H. Lindskov Hansen, stores the measurements, which are then at one's
disposal for statistical calculations.
Computed area measurements of the total distance between the curves of soft and loud selected
semitones for the total frequency range were made with interpolation of the, for practical reasons,
omitted semitones.
In the audiogram, only the softest loudness measurement is used. In the phonetogram, the lowest
sound pressure level which the singer/speaker is capable of producing is of interest, but also the highest
SPL. If the person has the capability to change the SPL from very weak to very loud in a large
frequency area, then the voice has a capacity for a higher degree of EXPRESSION.
The measurement method shows the total biological capability range of a voice. Other examination
principles should be related to this measurement. for example will a stable examination showing high
musical qualification in a large frequency and SPL range serve as a documentation of better singing
quality than in a small SPL and frequency range. The measurement of phonetogram areas is made in
Hz x dB(A). As in traditional audiometry certain frequencies were selected: the frequencies of a piano the semitones. Hence the phonetographic area measurement was modified to: semitones x dB(A),
independent of the scales on the graph.
At the unset of our studies planimetric calculation was used, which is why cm2 for the
phonetogram area is used in our material. Still a conversion factor of 1 cm2 being 32 dB(A) x
semitones seemed relevant to use, taking the standard paper of the UEP standard measurement proposal
into account. After we had developed computerized equipment, including software for phonetography
calculations this was no longer necessary. A discussion of the phonetographic measurement is given in
the appendix to 5.1 (Pedersen et al. 1995).
4.1.1. Instrumentation
The phonetograph was developed with possible consecutive RMS measurements of 1/2 seconds or 1, 2,
3, 4 or 5 seconds by minimal or maximal intensity. Each desired semitone was measured with rootmean-squares (RMS) and stored. 1/2 second was used. The measurement method was compared with
the manual method and an agreement of 96% was found (Sturtzebecker et al. 1982, Pedersen et al.
1984 and Pedersen et al. 1988).
As the phonetograms were in any event stored, it seemed relevant to develop software for
statistical treatment wherein not only averages of phonetograms in semitones x dB(A), but also 95%
intervals could be calculated. The number of averagings is only limited by computer capacity (Pedersen
and Lindskov Hansen 1986).
24
4.1.2. Analysis
The total semitone range, the lowest semitone, the middle semitone and the area in semitones x dB(A)
were the preferred statistical calculations to be used in the phonetograms. The highest semitone is
variable, and depends on training more than biology. In the beginning, before the averaging program
was made the areas were calculated through planimetric measurement with a conversion factor of 1
cm2 = 32 semitones x dB(A). The first planimetric calculations were done in cooperation with Patricia
Gramming (1983).
The total semitone ranges were given in semitones, not in hertz, in this way the statistical
possibility for errors is taken into account that ranges of semitones are logarithms and hertz are linear
calculations for the frequencies (geometric means used). DBs are logarithms. The measurements were
carried out in the singing school (for II-IV) and in the ENT clinic (V). The phonetograph stored the
data connected via a portable IBM computer.
4.2. Fundamental frequency (F0) in running speech in a reading situation (I-IV)
The fundamental frequency was registrated electroglottographically, measuring the cycles: the curve
from zero to zero in the electroglottogram. Mean values (after calculation in the software program for
2000 cycles) were expressed in hertz, 95% intervals of the fundamental frequency were expressed in
semitones: the tone range in continuous speech. A portable computer, an electroglottograph from FJ
electronics, type 830, a sound pressure level meter type 2208 and a stroboscope type 5066 from Bruel
and Kjaer, and Timcke, KS 3 were used, a Tectronix oscilloscope controlled the curve position.
The text read was from a book edited by the International Phonetic Society (1964) and precisely
translated into Danish by a phonetician ("The Northwind and the Sun" translated by Birgit Hutters from
Copenhagen University, Phonetic Dept.).
4.2.1. Stroboscopy with synchronous electroglottography (I)
We have conducted a synchronized examination of stroboscopy and electroglottography on untrained
hospital employees for analysis of the single curves in the electroglottogram with the Danish
electroglottograph (FJ Electronics). The set-up is illustrated in I Fig. 1.
4.2.2. Electroglottographic measurement of running speech in a reading situation (II-IV)
The software based on 2000 duty cycles was used during the reading of the standard text with the mean
fundamental frequency in Hz and the 95% intervals in semitones, the signals being grouped in
semitones from 60-684 Hz.
25
In this way, an exact measurement of the fundamental, the lowest partial in the spectrogram is
made, by registering the continuous changes of the basic voice sound with all its nuances of frequency.
4.3. Hormone and pubertal stage analysis (II-IV)
The hormonal analyses were based on our knowledge of probable changes in puberty (Brook 1995),
and carried out in cooperation with the hormonel department, Statens Seruminstitut. The children were
evaluated down to 8 years of age because the adrenarche (increased adrenal function of that age) could
perhaps show relations that would aid in understanding of the puberty. The following values were
analyzed: serum testosterone (free and total, which are closely related), dihydroepiandrosterone
sulphate (DHEAS), delta 4 androstendione and the transport globulin for testosterone: sex hormone
binding globulin (SHBG). For the girls the analysis also included: estradiol (E2), estrone (E1) and
estrone sulphate (El S04). The relations and formation and time of function for androgens and estrogens
are complicated, all the androgens and not only the gonadal hormones were measured. The results of
analyses, including the tables and figures, give more details on the relationship between sexual hormones, growth regulation of the body and the specific voice-related changes, providing a better
understanding of their possible interaction with growth hormones. Height, weight, testis volume
measured with an orchidometer, pubic hair stage and breast development were evaluated in stages of
puberty, as suggested by Krabbe, the pediatrician on our team. Puberty was defined as achievement of
reproductive function, in practice, defined as obtaining secondary sex characteristics, pubic hair stage
V-IV (Brook 1995).
4.4. Specific statistical calculations (II, IV)
The logarithmic criteria used here are based on geometric means and as such substantially stronger than
linear ones. A one-way multivariate analysis was carried out, classified by fundamental frequency (F0)
in running speech in a reading situation, in order to find predictions for the decline of puberty.
Correlation coefficients have been calculated for all variables, in order to find connections between
them, to calculate age dependency and to show age independent parameters, all with the assistance of
partial correlation coefficients (BMDP).
26
4.5. Subjects
Table 2. Overview of the single studies.
In (I) 20 hospital employees were used. In articles (III) and (IV) 48 boys were examined, but in (II)
47 girls were tested for voice parameters, the success rate for blood tests being 41 girls. In (V) three
boys were examined six times, beginning at the end of the 7th grade.
The 48 boys and 47 girls were part of the voice study (II-IV) which was cross sectional with 4-5
pupils in each school class, all being between 8 to 19 years of age. All were musically gifted judged by
professional evaluation (Seashore 1938, Pedersen 1992, Dejonckere et al. 1995) with evaluation of
vibrato, test for rhythm and tone jumps (ref. by Pedersen 1991 b) and presentation of a standard song
including high notes: "I Østen stiger solen op", by the composer Carl Nielsen. Stroboscopically nothing
abnormal was found, the pubertal development was evaluated as normal by the pediatrician in our
team, based on the standards referred by Brook (1995).
All were placed in traditional voice categories by singing teachers who knew the pupils well
(1. and 2. sopranos, altos, tenors and bassos).
27
5. RESULTS
5.1. Voice range profile measurements (see appendix)
On the abscissa of the voice range profile piano semitones are selected, and the corresponding loudness
intervals of a voice are measured. The resulting figure, calculated as the phonetographic area shows the
vocal capacity of a voice, including register shifts and possible unstable parts of loudness regulation.
During the past decades the method has proved its value in evaluating voices.
One of the problems with the method has been connected with the frequency weighting of the
phonetograph 8301. Some researchers use a linear frequency response or a C-weighting, as they claim
it is only possible to accurately register the lowest tones by this. Others prefer an A-weighted frequency
response. The reason for this is that it makes it much easier to get a relevant signal-to-noise ratio in
noisy surroundings, for the A-curve cuts away most lowfrequency disturbances. The A-curve also
results in voice range profiles which correspond to the audible sound impression of the ear due to the
reduced low-frequency sensitivity of the ear.
The spectral composition influencing the contour of the voice range profile was analyzed in the
instrumental set-up of an A-curve and a C-curve. The tone generator in the phonetograph 8301 was
constructed with tones having a high harmonic content. A series of tones from 58-1865 Hz with a
spectrum which resembles a soft voice quality, were supplied to a loudspeaker. The sound pressure
level at the microphone location was kept constant at two levels: 100 dB and 60 dB respectively, by
means of a simultaneous SPL monitoring with a calibrated Bruel and Kjaer sound pressure level meter
type 2236, which has built-in frequency weightings according to curve A and C frequency response.
Fig. 2 (appendix) shows the frequency weighting of curve A (the lower line) and the output from
phonetograph 8301 (the upper line) with the above mentioned input signal. It is seen that the signal is
not dampened by 30 dB at 50 Hz relative to 1000 Hz, but only by 10 dB.
The reason is that it is not the level of the fundamental frequency alone which determines the
output level, but the sum of all the harmonics (which are not so strongly dampened as the fundamental
frequency).
In Fig. 3 (appendix) the influence on a phonetogram of a type A-weighting relative to a type Cweighting is shown. It was drawn on the basis of 11 generated tones at the 100 dB
28
and 60 dB levels. It is seen that the type C-weightings (the upper lines) are quite near the 100 dB and
60 dB lines, whereas the A-weighting (the lower lines) are reduced with 10 dB at the lowest frequency.
However, as the strongest energy at low frequencies is placed in the first formant, the discrepancy is
less pronounced in real speech and singing especially if the test tone used for phonetograph 8301 is an
"ah".
The results indicate that it is preferable to use the A-weighting for recording voice range profiles,
partly because the error made is not serious, and partly because of the great advantage that it is possible
to make the recordings in noisy surroundings (up to 30-35 dB SPL (A) noise).
5.2. Measurement of vocal fold movements (I)
In the study (I), the electroglottographic curve was used to register the stroboscopic picture. The phases
were marked with a photocell attached to a larynx mirror and connected to an oscilloscope and a tape
recorder in 20 normal non-singers. This makes it possible to plot the variations in the curves.
Measurement of coefficients for opening and closing phases was carried out, also in those cases where
the glottographic curve alone could not be used to define these coefficients. The variations were seen in
the form of the electroglottographic cycle. The cycles were dependent on frequency variations alone.
On the basis of this experience, a computer program was developed measuring 2000 consecutive
electroglottographic cycles during reading of a standard text.
The average values of signals were given in Hz and the standard deviations in semitones. A range
from 60-784 Hz was analyzed. The analysis was made with the electroglottograph attached to an
oscilloscope connected to a computer calculating the signals.
In comparing the signals with the stroboscopic inspection (I, Fig. 3-5), the earlier discussion of the
accuracy of the electroglottographical signal for measurement of fundamental frequency was brought to
an end.
5.3. Fundamental frequency in running speech in a reading situation in boys in puberty (III)
The variation of fundamental voice frequency measured by the average of 2000 consecutive
electroglottographic cycles in a reading situation was examined in relation to pubertal development and
androgens. Fundamental voice frequency, among other parameters, was related to height (r= -0.82),
pubic hair stage (r= -0.87), testis volume (r= -0.78) total testosterone (r=-0.73) and serum hormone
binding globulin (r=0.75).
Single observations of fundamental frequency showed a clear grouping of results. Serum
testosterone of more than 10 nmol/l serum probably represents values for a boy in puberty. There seem
to be comparable relations with other androgens.
In Fig. I (III) the voice parameters in 48 boys, aged 8-19 years, are shown, including F0, the lowest
biological semitone, the middle biological semitone in the voice range profile, the F0 range in running
speech in a reading situation and at the top the voice range profile area (phonetogram area).
29
In Fig. 2 (III) the connection between F0 in running speech (abscissa) and SHBG, serum
testosterone, testis volume and pubic hair stage (the top) are demonstrated.
In Table 1, the measurements were divided according to the pediatric tradition for pubertal
measurements: three age groups - the first considered pre-pubertal, the second pubertal and the third
post-pubertal. The voice analyses were compared with the pubertal stages and the androgens, and a
correlation to F0 was calculated with the BMDP statistical program. Some of the results were
mentioned. The serum testosterone, testis volume and pubic hair stages are closely connected to SHBG
and significantly related to the F0 in running speech in a reading situation.
5.4. Fundamental frequency in running speech in a reading situation in girls in puberty (II)
As with the boys, the change in fundamental voice frequency in running speech in a reading situation in
female puberty was analysed in 47 voice-trained girls, through a comparison of 2000 consecutive
electroglottographic cycles. The results were compared with serum concentrations of androgens and
estrogens (estradiol, estrone and estronesulphate) and somatic puberty (weight, height, breast stages
and pubic hair stages). Fundamental frequency (F0) was related only to estrone r = -0.34 (p<0.05).
However, the tone range in running speech in a reading situation and the lowest biological semitone in
the voice range profiles were found to be significantly correlated with several of the puberty and
hormone parameters.
In Table 1 (II), a traditional division into a pre-pubertal, pubertal and post-pubertal group was
made for voice parameters, pubertal stages and hormones. A significant change was seen for estrone,
estrone sulphate, DHEAS and androstendione between the 3 groups. This was also the case for the
voice parameters F0 range, the voice range profile area (phonetographic area) and the lowest semitone.
Fig. 1 (II) is an illustration of the measured values of voice parameters in 47 girls (from the bottom: F0,
lowest biological semitone, total semitone range in the voice range profile, F0 range and voice range
profile area (top), also called the phonetogram area).
5.5. Specific results (II, III, IV)
The observations were transformed logarithmically in order to obtain normal distribution. The
differences between age groups were analysed through one-way analysis of variance (ANOVA), and
correlation coefficients were calculated for all pairs of variables to illustrate the connection between
them. The influence of a common age-dependency was studied by means of partial correlation
coefficients. A multidimensional linear regression analysis was carried out with F0 in running speech
in a reading situation as the dependent variable and puberty stage and hormone levels (in all 12
variables) as descibing factors. BMDP programme (9R) with Mallow's CP as a test criterion was used.
This test was carried out for 48 boys and 41 girls and for the two subgroups of 41 girls, pre- and post
menarche.
For the 48 boys, the results are presented in Table 2 (IV). A significant relation was found between
11 boys in the pubertal stages II-IV, age 13.5, F0 219 Hz and SHBG, indicating an age independent
relation connected to falling F0.
30
For the 41 girls, where hormonal analysis and voice analysis were available, the statistical results
are presented in Table 3 (II).
For the 41 girls, who were measured fully, the best set of describing variables were F0 range and
estrone sulphate. Before menarche, height and estronesulphate had a p value of p<0.001, pubic hair
stage p<0.05. After menarche F0 range had a value of p<0.001 and time after menarche p<0.01. Age
showed a relation to F0 of p<0.05.
In (IV), Fig. 1 voice range profiles of an untrained small boy (Fig. I a) with a small semitone and
loudness range especially for soft notes is seen. For the trained singer shown in Fig. I b, the highest
notes, in particular, changed from a2 to g3. The loudness is better for soft as well as loud SPL. Two
register shifts are seen in the middle at different places for loud and soft intonation.
In (Fig. lc) the typical pubertal register shift is defined, and a lowering of the lowest semitone and
a high loudness range for the lowest register shows up. The upper register is less defined. For a post
pubertal young man (Fig. I d), the lowest semitone and loudness range is preserved and new stable
semitones are built in a higher register.
In the figures (see below) examples of voice categories in trained choir girls are presented (Fig. 1).
Greater loudness range is seen in the upper part for sopranos and in the lower part for altos. On the
abscissa the semitone ranges which were usable in a choir situation were marked. (1st soprano, 2nd
soprano, child alto, post-pubertal 1st soprano, 2nd soprano and alto). It is noted that the difference
between the pre- and post-pubertal first sopranos and altos for the lowest semitone is f->e and d->c
respectively. The register shift is stable for the highest loudness in the post pubertal girls (= or > 90
dB(A)). Below in the same figure a pubertal girl aged 14.8 years is presented with a lower loudness
range in the upper register. Compared with the pubertal phonetogram of the boy in (IV) Fig. 1, the
reduction of loudness at the register shift and in the upper register gives an impression of a pubertal
change that, in principle, is of the same kind.
Fig. 2 presents computed averages of the yearly differences in the voice profiles of 48 boys (Fig.
2a) and 47 girls (Fig. 2b).
The abscissa marks the given semitones and also the frequencies of the octave shifts. The ordinates
present dB(A). The statistical 95% deviations were reduced with age in boys (Fig. 2a) as given for the
measured loudness. For the lowest semitones the child measurements have small deviations. From 1217 years, the variation in pubertal lowering of the lowest semitone corresponds to the statistical results
showing that the lowering was related to the increase in testosterone values more than to age. In Fig. 2b
the main result is that the lowest tone only drops by 2 semitones. The areas in semitones x dB(A) in 810 year old girls, especially the highest loudness is smaller than in boys. But in the post-pubertal young
women, the size of the semitones x db(A) is equal to the size in young men, with a higher location in
relation to octaves.
Fig. 3 shows the classifications of the phonetograms in the 48 boys (Fig. 3a) and 47 girls (Fig. 3b)
in voice categories as defined by the singing teachers. The set-up illustrates the need for voice category
analysis, or possibly the same lowest tones in averaged phonetograms, perhaps in untrained singers
also, when phonetograms are used for statistics.
In Fig. 3a the computed 95% intervals in puberty are reduced for the lowest tone in relation to Fig.
2a and the pubertal reduction of the loudness for high notes is clearly seen.
It can be seen that especially for the post pubertal young men the maximal loudness range
gradually lowers toward the 2. bassos, meaning that not only the semitone ranges but also maximal
loudness is decisive for the position in a choir.
The same tendency is seen in Fig. 3b for post pubertal young women. The elite phenomenon of
singing is illustrated for 1. sopranos (Fig. 3a and Fig. 3b) where the computed standard
31
deviations are high at c3 (1047 Hz). (Pedersen et al. 1982, 1983, 1986, 1990, 1991.3, Behrendt &
Pedersen 1989).
Fig. 1. Examples of voice categories in girls. Greater loudness variation is seen in the upper part
for sopranos and in the lower part for altos. (1. soprano, 2. soprano, child alto, post pubertal 1.
soprano, 2. soprano and alto). Below a pubertal girl age 14.8 years is presented with lower
loudness range in the upper register.
32
The abscissa marks the given semitones, and the frequencies at the octave shifts. The ordinates
present dB(A).
33
Fig. 2b. The averaged yearly change of phonetograms with 95% deviations in puberty in girls.
The abscissa marks the given semitones, and the frequencies at the octave shifts. The ordinates
present dB(A).
34
Fig. 3a. The averaged phonetograms and 95% deviations of categories of boys in a boys and
young menschoir defined by the singing teachers. A pubertal group is also given. The abscissa
marks the given semitones, and at the octave shifts the frequencies are also given. The ordinates
present dB(A).
35
Fig. 3b. The averaged phonetograms and 95% deviations of categories of girls and young women
in a girl schoir as defined by the singing teachers. A division between pre-pubertal and postpubertal girls was made. The abscissa marks the given semitones, and at the octave shifts the
frequencies are also given. The ordinates present dB(A). No pubertal group was found by the
singing teachers.
36
5.6. Longitudinal study (V)
We have made a longitudinal study of three boys from the end of the 7th school year before 6 weeks
summer-holiday till the end of the 8th school year with phonetograms and hormone analysis every
second month, in all 6 examinations for every boy. The six phonetograms of one boy (V, Fig. 1) were
compared with the average of the time in the choir and the time after for all 3 boys (V, Fig. 2).
This figure is made with our computerized averaging program. Boy l was measured with
phonetography and blood sample five times before he was excluded from the choir, and the averaged
phonetogram represents his pitch and loudness range during the singing period at the upper left picture
in (V) Fig. 2. At the lower left in (V) Fig. 2 a very typical pubertal phonetogram shows up reminiscent
of the higher childish tones, but with reduced loudness range ("reduced strength") and clear changes in
register positions.
The case in boy 2 is the opposite of boy 1 (the middle part in V Fig. 2). Already in the one
phonetogram prior to exclusion from the choir a reduced loudness range is seen for high notes and a
register shift is defined (the middle picture in V Fig. 2, upper phonetogram). After exclusion 5
averaged phonetograms with a typical pubertal configuration are presented, reminiscent of high child
registers and two beginning adult registers.
Boy 3 is represented in V Fig. I with three phonetograms from the last time in the choir (upper
figure) and three following exclusion from the boys choir (lower figure). V Fig. 2 illustrates the large
amount of information given by voice range profiles for semitone range, position of register change and
loudness variations in their relation to frequency etc.
The exclusion from the choir occurred during the eighth school year in all three boys. The changes
in this year were not age related (V Table 1).
37
6. DISCUSSION
6.1. Electroglottography (I-III)
It is noted that Titze (1994) does not recommend electroglottography as a standard measurement
method. The reason is that the method can be developed further (Herzel et al. 1994). Still
electroglottography is considered a useful test in voice evaluation. Dejonckere (1995 at the 1. World
Conference of Voice) suggests that for extracting fundamental frequency (FO) the waveform in most
cases is considerably simpler to use than the acoustical waveform: intonation contours and especially
distributions of FO can be used to describe normal and abnormal voice characteristics.
We also examined the positive parts of the electroglottographic cycle divided by the whole cycle
as suggested by Frokjaer-Jensen (1983), and Hacki (1989) in our choir studies. A separate work of its
change in puberty was under consideration, confirming other results of differences in impedance
between the registers in these normal voices (Lecluse 1977). But even if electroglottography is the most
stable and simple measurement for fundamental frequency in running speech in a reading situation e.g.
as illustrated by Chan (1994), there seems to be an agreement with Titze (1994) that standards for the
electroglottographic curve now could be a threat to future research. His example of the long-winded
discussion of the Frokjaer-Jensen apparatus (1968) turning the curve of the closing phase down and the
laryngograph (Fourcin & Abberton 1971) up recalls the humorous aspect. The discussion of the curve
ended with the multichannel phonetograph by Rothenberg (1992).
The use of the electroglottographic curve for understanding nonlinear chaoslike changes in voice
disorders shows that the method will follow voice research into neurobiology. The synchronization
with stroboscopy will, of course, ensure the relevant concepts (Hertegard & Gauffin 1995).
6.2. Fundamental frequency (II-IV)
The discussion of fundamental frequency is interesting. To some extent the different approaches to the
phenomenon depend on the researchers' background.
38
Pitch is a perceived tone property of central importance in both speech and music (Sundberg
1994). Sundberg continues by defining pitch physically by the frequency of the fundamental, the lowest
spectrum partial, being mainly controlled by the vocal folds but influenced by subglottal pressure
(Sundberg 1994).
The pitch measured with electroglottography is well defined, based on the measurement of the
primary larynx sound, at best combined with flowglottography.
The discussion with the physicists, especially the critics among them, for my part resulted in an
exclusion of the pitch in a study of patients with brain damage (Pedersen 1995). This illustrates the
need for interdisciplinary standards discussions in clinical pathology.
The laryngeal perturbation analysis suggested, among others by Karnell (1991), that jitter and
shimmer seems necessary in less well defined groups than the singers in this material. Another point of
discussion is the test utterance. Titze suggests that clinics and laboratories could benefit from some
consensus.
Titze suggests several non-speech and speech utterances (1994). From a medical point of view, the
comment could be that voice is only one of many biological parameters, and that the use of test
utterances in a reading situation is well established (Kitzing 1979).
Therefore, we chose only to use two parameters: the mean fundamental (FO) and its range (FO
range) during reading of a standard text, as suggested by the International Phonetic Association, (1964)
in this study.
6.3. Voice range profiles / phonetograms (V)
The international voice range profile has been chosen for quantitative measurement of voices (Airainer
& Klingholz 1993, Pedersen 1991 a,b). The method has recently been accepted as the method to be
used as a basis of further voice testing (Int. Fed. Oto-Rhino-Laryngol. Societies (Wendler 1989)), Int.
Ass. of Logopedics and Phoniatrics, and first World Conference of Voice (Schutte 1995)).
(synonymous: phonetogram, Stimmfeld, voice profile).
The reason for the basic status of phonetography is the graphical presentation of a person's
maximal and minimal vocal capability concerning fundamental frequency range and dynamic range on
several standard pitches chosen (Schutte 1995).
Referring to the problem of what fundamental frequency really is, the best description is achieved
by comparing the measurement set-ups on the market. This comparison was carried out in a group of
researchers who themselves are constructers. The apparatus discussed were the Key-elemetric model
4326 (1993), the Phonomath (1988) by Homoth, and the Phonetograph 8301 by Voice Profile
(1984/1995).
They all measure RMS with a resolution of 1 semitone based on an inbuilt tone generator during
intonation of a sustained "ah" with a distance of the microphone from the mouth of 15 cm (Keyelemetrics) to 30 cm (Phonomath and Phonetograph 8301).
The loudness measurement apparatus available on the market are supplemented with Stimmkonditionsmesser by Meditronic SMO3, Bruel and Kjaer sound pressure level meter type 2236 and to
some extent Soremed, (EVA work station). Earlier Vilkman et al. (1986) had an apparatus available for
a short time, and Schutte is now trying to adapt software for sale based on advanced Windows
software.
39
The intensity measurement was only dB(A) weighted in the Phonetograph 8301 equipment
available in 1984, but a new one with both dB(A) and dB(C) has been available since 1995. Keyelemetric and Phonomath have built-in possibilities for dB(A) and dB(C). Soremed, EVA workstation
uses dB(A) or dB(C). Stimmkonditionsmesser, SMO3, from Meditronic uses dB(A) and the most
advanced sound pressure level meter type 2236 from Bruel and Kjaer uses dB(A) and dB(C) (+/- peak)
connectable to a computer for calculations.
To some extent, the above-mentioned set-ups illustrate the problems of phonetograpy and their
solution: dB(A) and dB(C) can be used if information is given of the measurement method. Differences
are seen in the lowest frequencies. Supplementary measurement of peak values and singing formants is
desirable. The rescaling method for phonetography was compared with the absolute frequency
measurement method by Sulter et al. (1995), this shows the superiority of the latter in relation to e.g.
Coleman et al. (1977), due to the increased absolute frequency related information e.g. for registers.
They also emphasize the use of the phonetogram area as a value independent of plotting factors. A
discussion of registers was made by Svec and Pesak (1994) and Vilkman et al. (1995). Measurements
based on a set-up which included high-speed video-filming might elucidate the changes in registers in
puberty.
6.4. Voice range profiles in puberty (II, IV, V)
The phonetograms (voice range profiles) give a more stable basis for comparison with other
phenomena of puberty than the range of semitones alone. This was illustrated in our comparison
between Copenhagen singing school and Thomaner Choir, Leipzig, showing high SPL ranges for high
notes in Thomaner choir sopranos (Behrendt & Pedersen 1989).
The highest semitone is to our knowledge dependent on the training situation and singer selection,
though probably not the lowest semitone, which was significantly related to SHBG (IV,V) (Pedersen &
Moller 1987).
In their examination of voice range profiles in ten year-old children the references include few
analyses by other researchers, (McAllister et al. 1994). The study included the one by Flatau and
Gutzmann from 1905 which still is true for semitone ranges in school children. But our study has
resulted in a possible standard for voice range profiles for normal children. Bohme and Stucklic (1995)
conducted a study of untrained children from 5-14 years of age where a standard voice range profile for
7-10-year-old girls and boys was established. The difference of around 5 dB(A) maximal loudness in
favour of the boys was the same as in our material. No area calculation was made. In puberty they
refrained from averaging voice range profiles due to the inter individual variations, which serves to
illustrate our use of trained (well-defined) children to understand the difference between the genders in
puberty. Bohme and Stucklic regret that the voice range profiles were unstable in 5-6 year olds.
Methods (like child audiometry) still need to be developed for rehabilitation of children with hearing
disabilities.
The musicality of the children in the Copenhagen singing school is not especially high
(Copenhagen city area). Still, the pupils have a consciousness of their voices which is dependent on
voice education. Pahn and Pahn (1991) made a relevant program both for testing vocal musical
conscience, corresponding to ours, and for speech.
40
From a pediatric point of view, and even in the ENT / phoniatric clinic, the survey of yearly
change in voice range profiles, can help in differentiating pathological from normal voices as described
by Pedersen et al. (1985) and Bonet and Casan (1994). Voice range profiles can be used as an objective
tool for documentary of healing of voices. The division into singing voice categories can be used by
singing teachers to evaluate the future position of a child based not only on semitone range in singing,
especially the lowest tone but also on loudness range. The greatest loudness range should also be
decisive for the category of the child.
It is an easy task to analyze testosterone and estrone as a routine in hormonal and pubertal stage
disorders and compare them with voice range profiles, especially the lowest semitone and the area in
semitones x dB(A), before possible voice therapy for changing of voices is planned.
6.5. Fundamental frequency (FO) in puberty (II-IV)
The statistical calculations for prediction of voice changes not only with testosterone and estrone but
also SHBG in puberty seem to be of interest. The presented calculations are based on the BMDP
statistical program used in a stratified study confirmed in boys by a longitudinal study. Although the
results in our study seem obvious, a larger longitudinal study might refine our understanding of the
function of SHBG as a carrier of testosterone. The carrier globulins for hormones are currently being
intensively examined, and the results can be a basis for further studies (Strel'Chyonok & Vihko 1985).
Tobias et al. (1991) found androgen induced masculinization of tension and fiber type of the larynx in
frogs differing from fiber recruitment. No such studies seem to have been made in humans. But the
results indicate that the development is still more complicated. Longitudinal studies are immensely
labor intensive (citation: Hollien et al. 1994). Hollien et al. made a study on adolescent voice change in
males (1994). The lack of hormonal analysis is a problem and this is from a pediatric point of view also
the case for lack of pubertal staging. The results are interesting, but therefore not usable in a medical
setting, also because the study ceases at the age of 15.5 years. The results are relevant for our analysis
and should be set in a connection: 48 boys were followed from 10.5-11.5 years of age. The
measurement of fundamental frequency was made electronically by reading of a text with a standard
procedure as in earlier done in our cases.
Hollien et al. refer that the voice change beginning at mean 233 Hz at 13.4 years (SD 10.4 months)
was generally preceded with a varying pattern of fundamental frequency for at least 4 months
succeeded by a decreasing slope for at least 6 months. In 34 boys with complete analyses (Hollien et al.
1994), the end of the change at mean 122 Hz could be defined as that point occurring at the (lowest)
fundamental frequency immediately preceded by a period of frequency stabilization - lower than 140
Hz on average 18 months later.
The study by Hollien et al. (1994) can also be used to confirm the results showing the voice to be a
necessary part of defining puberty.
Comparison with other studies is a problematic task. Even Hollien (1994) uses for the first months
in 1990 manual measurements of oscillographic traces obtained from a phonellograph. Herein lies the
problem of exact comparison with earlier studies especially in females where our exact electroglottographic results show the fundamental frequency range (FO range) in running speech in a reading
situation as the new interesting parameter.
41
Still there seems to be an consensus now that in speech fundamental frequency (SFF) must be
measured continuously for several seconds, and that the fundamental frequency variation in on-going
speech is a necessary parameter in females. Russell et al. (1995) find standard deviations of SFF in
adult young females in the same range as in our post-pubertal girls. In terms of hertz, the results found
by Hollien et al. (1994), possibly geometrical means, are interestingly higher in boys during puberty
than later on, which of course can be related to the non-singer background. A coordination with
examination of females would have been of great value in his examination.
6.6. Perspectives
Our study can be used as an updated version of earlier voice research in puberty based on technical
advances and new hormonal knowledge. It is a necessary task which hopefully will be never-ending
(Weiss 1950).
The future perspectives are technical: using the programs for example referred to at the
International Computer Music Conference (Siegel 1994) in our biological voice research, and genetic:
studies of musical vocal gifts (Sataloff 1995). A combination is for example ongoing research of
voiceprint and genetic studies, perhaps also in puberty.
For the biological understanding of voice, it remains necessary to discuss results with experienced
singing teachers, a relevant place is the European Voice Teachers Conferences (Pedersen 1991), and
Collegium Medicorum Theatri (Dejonckere 1995) because the musical quality of the best trained
voices remains a guideline for biologists.
It has, however become easier to separate the technical/ biological side of voice training from the
musical side than has previously been the case, making it a simpler task to define pathology, also in
puberty.
42
7. CONCLUSION
Phonetograms / voice range profiles based on db(A) give a stable knowledge of total semitone ranges,
supplemented with exact measurement of maximal and minimal loudness. This can be used to compare
the voice function in puberty with other biological phenomena.
Based on the stroboscopical findings, the electroglottographical waveform reliably gives the
vibratory frequency of the vocal folds for fundamental frequency in running speech in a reading
situation.
The voice range profiles / phonetography and electroglottography for voice fundamental frequency
analysis compared with other biological parameters gives supplementary understanding of the voice
development during puberty. The introductional questions can be elucidated in the following way:
1.
2.
3.
How do pitch and loudness of the voice develop in puberty in musically trained boys and girls?
The voice range profile develops during puberty in the two sexes in nearly the same way, with a
narrowing of the voice range profile area located at the pubertal register shifts, especially from
13.5 to 14.5 years of age, the greatest shift occuring in boys.
In which way does the fundamental frequency (F0) in running speech in a reading situation change
in puberty?
The decrease in fundamental frequency in both sexes is related to testosterone and estrone. For
boys the mean fundamental frequency in a reading situation (FO) is related to testosterone, and for
girls the tone range, in particular, in a reading situation (FO range) is related to estrone. This
parameter should always be combined with mean fundamental frequency measurements in both
sexes, but especially in girls.
In which way is hormonal development related to voice?
The development of the voice in puberty is significantly related to the rise of testosterone in boys
and estrone in girls with extraction of age. The drop in sex hormonel binding globulin for
testosterone (SHBG) predicts the fall in the mean fundamental frequency (FO) in a reading
situation (when under 219 Hz) in boys in puberty stage II-IV, where testosterone is increasing. In
girls the fall in the mean fundamental frequency is predicted by both the increase of estrone,
estronesulphate (E1SO4) and the semitone range in running speech in a reading situation.
43
4.
How is pubertal maturation connected to the voice change?
Sexual ripening varies, qualitatively and quantitatively, in boys and girls, but occurs for voices at
13.5 to 14.5 years of age. Lack of adequate apparatus for measuring the differences and similarities
may account for the lack of girls choirs over the decades. A voice identity based on phonetograms
and fundamental frequency (FO) as well as biological range (FO range) in running speech
probably constitute an implicit knowledge for an optimal development of the musical expression in
girls as well as in boys (Collin & Koppe 1995).
44
8. References
Airainer, R. & Klingholz, F. (1993) Quantitative evaluation of phonetograms in the case of functional
dysphonia. J. Voice 7: 136-141.
Anastopolo, S. & Karnell, M.P. (1988) Synchronized videostroboscopy and electroglottography. J.
Acoust. Soc. Am. 83: 1883-90.
Andersen, E. (1968) Skeletal maturation of danish school children in relation to height, sexual
development and social conditions. Acta Paediatr. Scand. Suppl. 185, p. 64 72 and 98-101.
Andersson-Wallgren, G. & Albertsson-Wikland, K. (1994) Change in speaking fundamental frequency
in hormone-treated patients with Turner's syndrome - a longitudinal study of four cases. Acta
Paediatr. Scand. 83: 452-5.
Apter, D. & Vihko, R. (1985) Premenarcheal endocrine changes in relation to age at menarche. Clin.
Endocrinol. 22: 753-60.
Askenfelt, A., Gauffin, J., Sundberg, J. & Kitzing, P. (1980) A comparison of contact microphone and
electroglottography for the measurement of vocal fundamental frequency. J. Speech Hear. Res. 23:
258-73.
Baken, R.S. (1987) Clinical measurement of speech and voice. College Hill. Boston: 197-240.
Bastian, H.-J. & Unger, E. (1980) Untersuchungen des Zusammenhangs von Akzeleration, Mutation
und Dysphonie anhand von Langsschnitt Untersuchungen. Arztl. Jugendkd. 71: 205-11.
Behre, H.M. & Nieschlag, E. (1995) Biological effects of testosterone: new aspects. Eur. J. Clin.
Invest. 25 (suppl. 2): A 49.
Behrendt, W. (1989) Auditive Beurteilung des Stimmklanges mannlicher Fistelstimmen bei ehemaligen
Chorknaben. Sprache, Stimme, Gehor 13: 60-63.
Behrendt, W. & Pedersen, M.F. (1989) A comparative pubertal study of Thomaner choir, Leipzig and
Copenhagen boys choir. Oral presentation at Voice Symposium, Philadelphia.
Blatt, J.M. (1983) Training singing children during the phase of voice mutation. Ann. Otol. Rhinol.
Laryngol. 92: 462-68.
Blaustein, J.D. (1986) Steroid receptors and hormone action in the brain. Ann. N. Y. Acad. Sci. 474:
400-13.
Bloothooft, G. (1981) Voice profiles, vowel spectra and vocal registers. Proc. IXth Congr. Union of
European Phoniatricians. Amsterdam. 1981: 83-85.
Bohme, G. & Stuchlik, G. (1995) Voice Profiles and Standard Voice Profile of Untrained Children. J.
Voice 9: 304-307.
45
Bonet, M. & Casan, P. (1994) Evaluation of dysphonia in a childrens choir. Folia. Phoniatr. 46: 27-34.
Brook, C.G.D. (ed.) (1995) Clinical Endocrinology. Blackwell scientific publications. 3nd. ed. Oxford.
Buhring, G. (1990) Die Anwendung der CAD-System: CADy bei der Computer - gestutzten
Stimmfeldwertung. Dissertation A. Leipzig.
Buhring, G. & Pedersen, M.F. (1992) Fractal dimensions of phonetograms, a stratified and longitudinal
study during development of male voices.Proc. XXIlnd Congr. Int. Ass. Logoped. Phoniatr. 1:1-9.
Byrne, D., Dillon, H. & Tran, K. (1994) Aninternational comparison of long-term average speech
spectra. J. Acoust. Soc. Am. 96: 2108-20.
Calvet, J & Malhiac, G. (1952) Courbe vocales et mue de la voix. Congres Bull. Soc. Fr. Foniatr.
(1950) and J. Fr. Otorhinolar. 1:115-124.
Chan, R.W.K. (1994) Does the voice improve with vocal hygiene education? A study of some
instrumental voice measures in a group of kindergarten teachers. J. Voice 8: 279-291.
Coleman, R.F., Mabis, J.H. & Hinson, J.K. (1977) Fundamental frequency - sound pressure level
profiles of adult male and female voices. J. Speech. Hear. Res. 20: 197-204.
Collin, F. & Koppe, S. (ed.) (1995) Humanistisk videnskabsteori. The Danish Radio publications.
Copenhagen.
Colton, R.H., Casper, J.K., Brewer, D.W. & Aronson, D.G. (1989) Digital processing of laryngeal
images: A priliminary report. J. Voice 3: 132-42.
Colton, R.H., Woo, P., Brewer, D.W., Griffin, B. & Casper, J. (1995) Stroboscopic Signs Associated
with Benign Lesions of the Vocal Folds. J. Voice 9: 312-325.
Dejonckere, PH. (1995) Electroglottography:an useful test in voice evaluation. Ist. World Conf. of
Voice, Portugal. Abstract book: 136.
Dejonckere, PH., Hirano, M. & Sundberg, J. (1995) Vibrato. ed. Singular publishing group, San Diego.
Elman, J.L. & Zipser, D. (1988) Learning the hidden structure of speech. J. Acoust. Soc. Am. 83: 161526.
Fairbanks, G., Wiley, J.M. & Larsman, F.H. (1949) An acoustical study of vocal pitch in seven and
eight-year-old boys. Child. Dev 20: 63-9.
Fabre, P. (1957) Un precode electrique percutane d'inscription de 1'accolement glottique au cours de la
phonation glottographie de haute frequance. Premiers resultats. Bull. Acad. Natl. Med. 121: 66.
Fitch, J.L. & Holbrook, A. (1970) Modal vocal fundamental frequency of young adults. Arch.
Otolaryngol. 92: 379-82.
Flatau, T.S. & Gutzmann, H. (1905) Die Singstimme des Schulkindes. Archiv Laryngologie 20: 32748.
Fourchin, A.J. & Abberton, E. (1971) First application of a new laryngograph. Med. Biol. Illus. 21:
172-182.
Frank, F. & Sparber, M. (1970) Stimmumfange bei Kindern aus neuer Sicht. Folia Phoniatr. 22: 397402.
Frokjaer-Jensen, B. (1983) Can electroglottography be used in the clinical practice. Proc. XIXth Congr.
of IALP, Edinburgh. vol II: 849-851.
Frokjaer-Jensen, B. & Thorvaldsen, P. (1968) Construction of a Fabre Glottograph. ARIPUC.
Copenhagen University, Denmark 3: 1-8.
46
Gramming, P., Pedersen, M.F. & Kitzing, P. (1983) Datoriserade fonetogram for objectivering av rost
funktions storninger. Lakar stamman / Stockholm. Otolaryngologisk section (oral presentation).
Gramming, P. (1988) The phonetogram. Thesis. University of Lund. Sweden.
Guidet, C. & Chevrie-Muller, C. (1979) Computer analysis of prosodic and electroglottographic
parameters in diagnosis of pathologic voice. In: Winkler, P. (ed.) Investigations of the speech
process. Quantitative linguistics. Dr. Brockmeyer Studienverlag, Bochum. 19: 233-61.
Hacki, T. (1988) Die Beurteilung der quantitativen Sprechstimm - leistungen. Das Sprechstimmfeld.
Folia Phoniatr. 40: 190-6.
Hacki, T. (1989) Klassifizierung von Glottisdysfunktionen mit Hilfe der Elektroglottographie. Folia
Phoniatr. 41: 43-48.
Hagg, U. & Taranger, T. (1982) Maturation indication and the pubertal growth spurt. Am. J. Orthod.
October: 299-309.
Heinemann, M. (1976) Hormone und Stimme. J. Ambrosius Barth Verlag. Leipzig.
Hertegard, S. & Gauffin, J. (1995) Glottal area and vibratory Patterns studied with simultaneous
stroboscopy, flow glottography and electroglottography. J. Speech. Hear. Res. 38: 85-100.
Hertel, N.T., Scheike, T., Juul, A., Main, K., Holm, K., Bach-Mortensen, N. Skakkebaek, N.E. &
Muller, J.R. (1995) Kropsproportioner hos danske børn. Ugeskr. Laeger. 157: 6876-81.
Herzel, H., Berry, D., Titze, I.R. & Saleh, M. (1994) Analysis of vocal disorders with methods from
nonlinear dynamics. J. Speech. Hear. Res. 37: 1008-1019.
Hirano, M. (1989) Objective evaluation of the human voice: clinical aspects. Folia Phoniatr. 41: 89144.
Hirano, M., Kurita, S. & Nakashima, T. (1983) Growth, Development and Aging of Human Vocal
Folds. In: Bless, D.M. & Abbs, J.H. (eds) Vocal Fold Physiology, College Hill Press, San Diego
California: 22-43.
Hirano, M., Kiyokawa, K. & Kurita, S. (1988) Laryngeal Muscles and Glottic Shaping. In: Fujimura,
O. ed. Vocal Physiology: Voice Production, Mezhanisms and Functions. Vol. 2. Raven Press, Ltd.
N.Y: 49-65.
Hirose, H., Kiritani, H. & Imagawa, H. (1988) High-Speed Digital Image Analysis of Laryngeal
Behavior in Running Speech. In: Vocal Physiology: Voice Production, Mechanisms and Functions.
Vol. 2. Fujimura, O. (ed.) Raven Press, Ltd. N.Y.: 335-45.
Hollien, H. & Malcik, E. (1967) Evaluation of cross-sectional studies of adolescent voice change in
males. Speech Monographs 34: 80-84.
Hollien, H. & Shipp, T. (1972) Speaking fundamental frequency and chronologic age in males. J.
Speech. Hear. Res. 15: 155-59.
Hollien, H. (1983) In Search of Vocal Frequency Control Mechanisms. In: Bless, D.M. & Abbs, J.H.
(eds) Vocal Fold Physiology. College - Hill Press, San Diego, California: 361-67.
Hollien, H., Green, R. & Massey, K. (1994) Longitudinal research an adolescent voice change in
males. J. Acoust. Soc. Am 96: 2646-54.
International Phonetic Assoziation (1964) reprint from 1949: The Principles of the International
Phonetic Assoziation. ed. Dept. Phonetics, Univ. College, London, W.C.1.
Kahane, J.C. (1982) Growth of the human prepubertal and pubertal larynx. J. Speech. Hear. Res 25:
446-55.
Karlberg, P. & Taranger, J. (1976) The somatic development of children in a swedish urban
community. Acta Paediatr. Scand. Suppl.: 256, 1-148.
47
Karnell, M.P. (1989) Synchronized video stroboscopy and electroglottography. J. Voice 3: 68-75.
Karnell, M.P. (1991) Laryngeal pertubationanalysis: Minimum length of analysis window. J. Speech.
Hear. Res. 34: 544-548.
Kay Elemetrics Corp. (1993) Voice range profile model 4326. Operation manual. Pine Brook NJ.
Kersing, W. (1983) De Stimmband Musculatur. Een Histologischeen histochemische
Studie. Thesis. Utrecht. The Netherlands.
Kitzing, P. (1979) Glottographisk Frekvensindikering. Thesis. University of Lund. Sweden.
Kitzing, P. (1990) Clinical application of electroglottography. J. Voice 4: 238-249. Klingholz, F. &
Martin, F. (1983) Die quantitative Auswertung der Stimmfeld Messung. Sprache, Stimme, Gehor 7:
106-110.
Klingholz, F., Jolk, A. & Martin, F. (1989) Stimmfeld Untersuchungen bei Knabenstimmen
(Tolzer Knabenchor). Sprache, Stimme, Gehor 13: 107-111.
Komyama, S., Watanabe, H. & Ryu, S. (1984) Phonetographic relationship between pitch
and intensity of the human voice. Folia Phoniatr. 36: 1-7.
Konzelmann, U., Moser, M. & Kittel, G. (1989) Stimmfeldmessungen bei Chor Sangern
vor und nach Stimmbelastung unter besonderer Beruchsichtigung des Sangerformanten.
Sprache, Stimme, Gehor 13: 112-118.
Krabbe, S. (1989) Calcium homeostasis and mineralization in puberty. Dan. Med. Bull. 36: 113-24.
Krusnevskaja, I.I. & Pedersen, M.F. (1992) Phonetograms of radiated voices in Chernobyl.
Proc. XXI congr. Int. Ass. Logoped. Phoniatr. Hannover 16: 1-9.
Kurita, S., Hirano, M., Mihashi, S. & Nakashima,' T. (1980) Layer structure of the vocal
fold, age-dependant variation. Proc. XVIIIth congr. Int. Ass. Logoped. Phoniatr.:
537-39.
Large, J. & Iwata, S. (1972) The male operatic head registers versus falsetto. Folia Phoniatr. 24: 19-29.
Lecluse, F.L.E. (1977) Electroglottographie. Thesis. Utrecht. The Netherlands.
Loebell, E. (1968) Uber den klinishen Wert der Electroglottographie. Arch. Klin. Exp.
Ohren Nasen Kehlkopfheilkd. 191: 760-764.
Lykkesfeldt, G., Bennett, P., Lykkesfeldt, A.E., Micic, S., Moller, S. & Svenstrup, B.
(1985) Abnormal androgen and oestrogen metabolism in men with steroid sulphatase
defficiency and recessive x-linked ichthyosis. Clinical Endocrinol. 23: 385-93.
Marin, M.L., Tobias, M.L. & Kelley, D.B. (1990) Hormone-sensitive stages in the sexual
differentiation of laryngeal muscle fiber in Xenopus laevis. Development 110: 703-11.
McAllister, A., Sederholm, E., Sundberg, J. & Gramming, P. (1994) Relations between voice range
profiles and physiological and perceptual voice characteristics in ten-yearold children. J. Voice 8:
230-239.
McGlone, R.E. & McGlone, J. (1972) Speaking fundamental frequency of eight-year-old girls. Folia
Phoniatr. 24: 313-317.
Meuser, W. & Nieschlag, W. (1977) Sex hormones and vocal register in adult men. Acta
Endocrinol. Suppl. (Copenh.) 208: 61.
Michel, J.F., Hollien, H. & Moore, P. (1966) Speaking fundamental frequency characteristics of 15, 16
and 17-year old girls. Language and Speech 9: 46-51.
Miranda, R., Sohrabji, F., Singh, M. & Toran-Allerand, D. (1996) Nerve growth factor
(NGF) regulation of estrogen receptors in explant cultures of the developing forebrain.
J. Neurobiol. 31: 77-87.
48
Nastiuk, K.L. & Clayton, D.F. (1995) The canary androgen receptor mRNA is localized in the song
control nuclei of the brain and is rapidly regulated by testosterone. J. Neurobiol. 26: 213-224.
Pabon, J.P.H. (1991) Objective acoustic voice-quality parameters in the Computer phonetogram. J.
Voice 5: 203-16.
Pahn, J. & Pahn, E. (1991) Formblatt, Eigenschaften, Ablauf und Bedeutung des Tests der Sensibilitat
formaler sprachlicher Elemente im Hinblick auf Perzeption und Produktion. Sprache, Stimme, Gehor
15: 19-23.
Pedersen, M.F. (1974) A clinical examination of patients with benign tumors of the larynx, before and
after microlaryngoscopy. Proc. XVIth congr. Int. Ass. Logopedics and Phoniatr.: 378-83.
Pedersen, M.F. (1978) Electroglottography compared with synchronized stroboscopy in students of
music. The study of sound, Tokyo 18: 423-434.
Pedersen, M.F., Eriksen, K. & Heramb, S. (1980) A follow up study of patients with voice disorders.
Proc. XVIIIth congr. Int. Ass. Logopedics and Phoniatr. Washington 1: 621-26.
Pedersen, M.F., Kitzing, P., Krabbe, S. & Heramb, S. (1982) The change of voice during puberty in 11
to 16 year old choir singers measured with electroglottographic fundamental frequency analysis and
compared to other phenomena of puberty. Acta Otolaryngol. Suppl. (Stockh.) 385: 189-192.
Pedersen, M.F. & Munk, E. (1983) Register examination during mutation in the Copenhagen Boys
Choir.Proc. XIXth congr. Int. Ass. Logopedics and Phoniatr.: 686-87.
Pedersen, M.F., Lindskov Hansen, T., Lindskov Hansen, H. & Munk, E. (1984) A Phonetograph for
Use in Clinical Praxis. Acta Otolaryngol. Suppl. (Stockh.) 412: 138.
Pedersen, M.F., Moller, S., Eriksen, K. & Sondergaard, U. (1985) Quantitative and qualitative
diagnoses of children with voice disorders. New Dimensions in Otorhinolaryngology - Head and
Neck Surgery vol 2: 470-71.
Pedersen, M.F., Moller, S., Krabbe, S., Bennett, P. & Munk, E. (1986) Phonetograms in Choir Boys
Compared with Voice Categories, Somatic Puberty and Androgen Development. J. Research Singing
9: 39-49.
Pedersen, M.F. & Lindskov Hansen, H. (1986) Computerized phonetograms for clinical use. Folia
phoniatr. Proc. XXth Congr. of IALP, Tokyo: 170-171.
Pedersen, M.F. & Moller, S. (1987) A transport globulin, serum hormon binding globulin, as a
predicting factor of voice change in puberty?. Proc. Xlth int. Congr. Phonetic Sciences, Tallinn 4:
296-299.
Pedersen, M.F., Seidner, W. & Wendler, J. (1988) Collection and processing of voice field data.
Methodical approaches. Proc. XVth congr. Union of European Phoniatricians. Erlangen: 107.
Pedersen, M.F., Hacki, T. & Loebell, E. (1988) Videostroboscopy of choir boys in puberty. Video
presentation. Medizinische Hochschule. Hannover.
Pedersen, M.F., Moller, S. & Bennett, P. (1990) Voice Categories Compared with Phonetograms,
Androgens, Estrogens and Puberty Stages in 8-19 Year Old Girls. J. Research Singing 13: 1-4.
Pedersen, M.F. (1991a) Computed phonetograms in adult patients with benign voice disorders before
and after treatment with a non - sedating antihistamine (Loratadine). Folia Phoniatr. 43: 60-67.
Pedersen, M.F. (1991b) Pilotstudie der Stimmfunktion vor und nach Behandlung von
Hirngeschadigten. In: H. Gundermann (ed.) Die Krankheit der Stimme, die Stimme der Krankheit.
Fischer Verlag, Stuttgart: 162-71.
49
Pedersen, M.F. (1991c) Die biologische Entwicklung der Stimme in der Pubertat. Bundesverband
Deutscher Gesangpadagogen. Dokumentation ed. Detmold Hochschule für Musik: 28-37.
Pedersen, M.F. (1992) Bookreview. C.E. Seashore. Psychology of Music. Folia Phoniatrica 44:312.
Pedersen, M.F. (1995) Stimmfunktion vor und nach Behandlung von Hirngeschadigten, mit
Stroboskopie, Phonetographie und Luftstrom - analyse durchgefuhrt. Sprache, Stimme, Gehor 19:
84-89.
Roubeau, B., Chevrie-Muller, C. & Arabia-Guidet, C. (1987) Electroglotto - graphic study of the
change of voice registers. Folia Phoniatr. 39: 280-9.
Rihkanen, H., Leinonen, L., Hiltunen, T. & Kangas, J. (1994) Spectral pattern recognition of improved
voice quality. J. Voice 8: 320-326.
Rodriguez-Sierra, J. (1986) Extended organizational effects of estrogen at puberty. Ann. N. Y. Acad.
Sci. 474: 293-306.
Roed, J., Larsen, RB. & Ibsen, K.K. (1989) The heights of a population of children in greater
Copenhagen aged 7-18 years in 1981 and 1985. Ugeskrift Laeger 159: 895-897.
Rothenberg, M. (1992) A multichannel electroglottograph. J. Voice 6: 36-43.
Russell, A., Penny, L. & Pemberton, C. (1995) Speaking fundamental frequency changes over time in
women: a longitudinal study. J. Speech. Hear. Res. 38: 101-109.
Sataloff, R.T. (1995) Genetics of the voice. J. Voice 9: 20-26.
Sato, K. & Hirano, M. (1995) Histologic investigations of the macula flava of the human vocal fold.
Ann. Otol. Rhinol. Laryngol. 104: 138-143.
Schonharl, E. (1960) Die Stroboskopie in der praktischen Laryngologie. Thieme. Stuttgart.
Schulz-Coulon, H.J. & Klingholz, F. (1988) Objective und semiobjective Untersuchungen der Stimme.
In: Proc. XV Congr. Union of European Phoniatricians Erlangen: 3-88.
Schutte, H.K. & Seidner, W. (1983) Recommendation by the Union of European Phoniatricians (UEP):
Standardizing voice area measurement / phonetography. Folia Phoniatr. 35: 286-8.
Schutte, H.K. (1995) Phonetogram: Voice capacities and clinical value. Abstracts, Ist. World Conf. of
Voice, Oporto, Portugal: 265.
Seashore, C.E. (1938) Psychology of music. McGraw-Hill. 1.Republication Dover 1967.
Seidner, W. & Schutte, H.K. (1981) Standardisierungsvorschlag Stimmfeld Messung / Phonetographie.
Proc. Ixth Congr. Union of European Phoniatricians. Amsterdam: 88-94.
Seidner, W. & Wendler, J. (1982) Die Sangerstimme. Wilhelmshafen, Heinrichshofen: 174-78
Seidner, W., Kruger, H. & Wernecke, K.D. (1985). Numerische Auswertung spektraler Stimmfelder.
Sprache, Stimme, Gehor 9: 10-13.
Siegel, W. ed. (1994) Proceedings, Int. Computer Music Conference, Århus.
Smith, S. (1954)Remarks an the physiology of the vocal cords. Folia Phoniatr. 6: 166-79.
Smith, S. (1981) Research an the principle of electroglottography. Folia phoniatr. 33: 1-10.
Strel'Chyonok, O.A. & Vihko, R. (1985) Specific steroid-binding glycoproteins of human bloodplasma.
Novel data an their structure and function. J. Steroid. Biochem. 35: 519-34.
Sturtzeberger, E., Wagner, H., Becker, R., Rauhut, A. & Seidner, W. (1982) Einrichtung zur
simultanen Registrierung von Stimmfeld und hohem Sangerformant. HNO-Praxis Leipzig; 7: 223226.
50
Sulter, A.M., Schutte, H.M. & Miller, D.G. (1995) Differences in Phonetogram Features Between Male
and Female Subjects With and Without Vocal Training. J. Voice 9: 363-377.
Sundberg, J. (1987) The science of the singing voice. Northern Illinois Univ. Press. Sundberg, J. (1994)
Perceptual aspects of singing. J. Voice 8: 106-122.
Svec, J. & Pesak, J. (1994) Vocal breaks from modal to falsetto register. Folia Phoniatr. 46: 97-103.
Tanner, J.M. & Whitehouse, R.H. (1976) Clinical longitudinal standards for height, weight, height
velocity and stages of puberty. Arch. Dis. Child. 51: 170-9.
Titze, I.R. (1994) Toward standards in acoustic analysis of voice. J. Voice 8: 1-7.
Tobias, M.L. & Kelley, D.B. (1995) Sexual differentiation and hormonal regulation of the laryngeal
synapse in Xenopus laevis. J. Neurobiol. 28: 515-26.
Tobias, M.L., Marin, M.L. & Kelley, D.B. (1991) Temporal constraints an androgen directed laryngeal
masculinization in Xenopus laevis. Dev. Biol. 147: 260-70.
Truuverk, C. & Pedersen, M.F (1992) A pilot study of phonetograms compared with menopausal
estrogens and androgens in 4 sopranoes in World Festival Choir. Abstracts XXInd Congr. Int. Ass.
Logopedics Phoniatr. Folia Phoniatr. 1-2: 85.
Vilkman, E., Sonninen, A. & Hurme, P. (1986) Observations an voice productions by means of
Computer voice profiles. Proc. XXth congr. Int. Ass. Logopedics and Phoniatr, Tokyo: 370-1.
Vilkman, E. AIku, P. & Laukkanen, A.-M. (1995) Vocal-fold collision as a differentiator between
registers in the low-pitch range. J. Voice 9: 66-73.
Vuorenkoski, V., Lenko, H.L., Tjernlund, P., Vuorenkoski, L. & Perheentupe, J. (1978) Fundamental
frequency during normal and abnormal growth, and after androgen treatment. Arch. Dis. Child. 53:
201-9.
Walker, J.M., Williams, D.M., Harries, M., Hacking, J. & Hughes, I.A. (1993) A study of the
mechanisms of normal voice maturation in pubertal boys. Pediatr. Res. 33, (5 suppl) S 84.
Weiss, D.A. (1950) The pubertal change of the human voice.Folia Phoniatr. 2: 126-59.
Wendler, J. & Seidner, W. (1987) Entwicklung der Stimme, in Lehrbuch der Phoniatrie. Thieme
Verlag: 164-69.
Wendler, J., Koppen, K. & Fischer, S. (1988) The validity of stroboscopic data in terms of quantitative
measuring. Folia Phoniatr. 40: 297-302.
Wendler, J. (1989) Basic equipment for voice diagnosis. Newsletter. Int. Fed. Otorhinolaryngol.
Societies. Febr.-March: 3.
Young, M.C., Robinson, J.A., Read, G.F., Riad-Fahmy, D. & Hughes, I.A. (1988) 170 H-progesterone
rythms in congenital adrenal hypoplasia. Arch. Dis. Child 63: 617-23.
Yukizane, S., Yamakawa, R., Murakami, T., Kato, H. & Niikawa, N. (1994) A 15-year-old girl with
pubertal masculinization due to bilateral gonadoblastoma and 45, X/46, X, +Mar Karyotype. Kurume
Med. J. 41: 155-9.
51
52
APPENDIX
53
Proceedings
of the
XXIII World Congress of the
International Association of Logopedics and Phoniatrics
Cairo, 6-10, August, 1995
M. Nasser Kotby, Editor
54
STANDARDIZING VOICE RANGE PROFILE MEASUREMENTS
(Phonetography / Stimmfeldmessung)
M.F.PEDERSEN, B. FROKJAER - JENSEN, F. PABST, H.K.SCHUTTE,
T.HACKI, H.L.HANSEN
ABSTRACT
Approch to specifications for standardized phonetography is suggested. The influence from the
frequency response of the phonetograph is discussed in details.
GENERAL PRESENTATION
Phonetography is a method used for registration of the total frequency range and the total dynamic
range of the voice. The result is shown in a graph called a phonetogram, a voice area, or better a voice
range profile (Stimmfeld in German, courbe vocale in French). The fundamental frequencies of several
arbitrarily chosen tones are shown along the horizontal axis, and the corresponding sound level pressures are shown along the vertical axis. In some cases phonetograms have been 'modulated' with
information about registers and spectral properties (1, 2].
During the past years, phonetography has proved to be a very useful tool for assessment of voices,
normal [3] as well as. pathological [4, 5], for assessment of the dynamic and tonal range in speech, and
useful for testing singers and their potentialities [6, 7].
In 1981 Schutte and Seidner recommended a "Standardization of Voice Area
Measurements/Phonetography" which was presented at a meeting in the Union of European
Phoniatricians (8]. Most of these proposals are still valid, but new technical possibilities and the
practical experience with phonetography during the past 15 years have revealed the fact that some of
these specifications have to be reevaluated.
The basic specifications have not changed since 1981: The microphone distance is standardized to 30
cm, or if another distance is used, the measurements are recalculated so that they reflect the levels at a
distance of 30 cm. In most cases the dynamic range is fixed at 80 dB from 40 to 120 dB. The frequency
range is normally put at five octaves, and normally both the frequencies and the appurtenant tone
names are shown. Lowest and highest frequency vary somewhat, but in general a standard pitch range
from 50 to 2000 Hz seems to be in use.
RECOMMENDATIONS
One of the main problems in phonetography has been the problem associated with the frequency
weighting of the phonetograph. Some researchers use a linear frequency response or a C-weighting, as
they claim that only by using that sort of frequency weighting will it be possible to make a fair
registration of the lowest tones. Other researchers prefer an A-weighted frequency response. In this
way, they claim, it is much easier to get a good signal-to-noise ratio in noisy surroundings, e. g. in a
laboratory or in a clinic, as the A-weighted response cuts away most low-frequency disturbances.
Furthermore, the A-curve results in phonetograms which correspond visually to the audible sound
impression of low. tones because of the reduced lowfrequency sensitivity of the ear.
Thus, it is clear that a phonetogram which is based upon sine tones will depict the frequency response
of the phonetograph, which means that a 50 Hz tone will be damped 30 dB relative to a 1000 Hz tone,
and this is not acceptable. But the human voice does not produce sine tones, the human voice has a
complex spectrum often with very strong harmonics. The question is then: How does the spectral
composition influence the contour of a voice range profile ?
55
suplementary fig.5.
Measurements of voice range profiles:
a; dB(A), b; dB(C) with reverse corrections
56
In order to suggest the best recommendation for the frequency weighting in the phonetograph, an
investigation of the influence of the A-curve and the C-curve has been carried out:
The instrumental set-up consisted of a phonetograph which was altered so that it was possible to switch
between curve A and curve C. The input was generated by a tone generator with tones having a high
content of harmonics. A series of tones from 58 to 1865 Hz, with a spectrum which sounded as a soft
voice quality, were supplied to a loudspeaker. The sound pressure level at the place of microphone was
kept constant at two levels: 100 dB and 60 dB respectively, by means of a simultaneous SPL
monitoring with a calibrated Brüel & Kjaer sound pressure level meter type 2236, which had built-in
frequency weightings according to curve A, C (IEC 651 type 1), and linear frequency response.
Fig. 1 shows the acoustic analyses of two low-pitched tones at approx. 50 Hz and 100 Hz where the
discrepancies are most pronounced. It is noticed that the spectral energy is concentrated in the 4-6 first
harmonics (soft voice quality), and that the fundamental frequency and the second (and third)
harmonics are of equal level.
Fig. 2 shows the frequency weighting of curve A (the lower line) and the output from the phonetograph
(the upper line) with the above-mentioned input signal. It is seen that the signal is not damped 30 dB at
50 Hz relative to 1000 Hz, but only 10 dB. The reason is that it is not the level of the fundamental
frequency alone which determines the output level, but the sum of all the harmonics (which are not so
strongly damped as is the fundamental frequency).
The influence on a phonetogram of a type A weighting relative to a type C weighting is shown in Fig.
3, which is drawn on the basis of 11 generated tones with 100 dB and 60 dB levels. It is seen that the
type C weightings (the black lines) are quite near tile 100 dB and 60 dB lines, whereas the A weighting
(the grey lines) are reduced with 10 dB at the lowest frequency.
The greatest discrepancy between the real and the measured levels (-10 dB) happens at the lowest
frequencies between 50 and 100 Hz, and is more pronounced the stronger the fundamental frequency
is. However, as the strongest energy at low frequencies are placed in the first formant (Fl is 250-300 Hz
for an [i:] and 700-850 Hz for an [a:], the discrepancy is less pronounced in real speech and singing,
especially if tile vowel used for the phonetograph recording is an [a:].
The conclusion must be that it is acceptable to use the A weighting for recording of phonetograms,
partly because the error made is not serious, and partly because of the great advantage that it is possible
to make the recordings in noisy surroundings (up to 30-35 dB SPL (A) noise is acceptable).
Another main problem is the size of the 'open frequency range'. Some persons are not able to reproduce
a given tone with the correct frequency. It is very difficult to make valid recordings with tone-deaf
subjects. The investigator must have a really good tone perception (absolute pitch perception), if the
phonetograms are made manually. In this case the best thing to do is to make an automatic registration
of both frequency and level. When recording phonetograms of tone-deaf persones, the phonetograph
must have a broad-band 'open frequency range' with at least halfoctave filters. Such broad filters are
also necessary when making voice range proriles of reading. For singers it is sufficient to set up
frequency windows of a semitone each. We would recommend an 'open frequency range' which could
be switched between a semitone, a tone, two tones, half octave, and no limitation.
Information on the accuracy of frequency measurements is seldom given by
57
58
the various manufacturers of phonetographs, but we would recommend a 3% as a standard, because it
corresponds to the distance between the semitone steps.
It is also an advantage to be able to adjust the recording time. The responses from some types of
patients are often very short, and therefore, a short recording time is necessary. On the other hand,
singers have often a pronounced frequency and/or intensity vibrato; in such cases it is an advantage that
the phonetograph is able to integrate the measurements over a few seconds. Finally, if phonetograms
are recorded when reading a text aloud (in order to find the gravity point and distribution of the voice
area profile in reading), it is necessary to have a recording time of at least 1 minute. We would
recommend a 'record duration' selectable from 0.5 second over 1, 2, 3, 5, 15, 30, to 60 seconds.
The registration of the sound pressure level is integrated and is, therefore, often called sound intensity
even if the measurements are specified in the measured dB values. Most researchers and builders of
phonetographs use a fast logarithmic RAMS measurement, which we also would recommend. The
frequency range of the electronics varies in different phonetographs. We would recommend a limited
range of 40-15,000 Hz winch is sufficient as no voices have much energy above 15,000 Hz. The digital
solution is in most instruments set at 0.5 dB before calculation and 1.0 dB after calculation, which we
also can recommend as a sufficiently exact measure for practical use.
Some phonetographs have facilities for curve editing. Simple calculations on the recorded
phonetograms can also be made. We consider it important that all phonetographs are able to calculate
the voice profile area. The unit used must be semitones times decibel. A print-out of a voice range
profile must also contain information about the lowest and the highest tone, the pitch range, the
weakest and the strongest tone, and the dynamic range. Some phonetographs are equipped with
specialities such as simultaneous information in the voice range area of the singer's formant,. of jitter,
shimmer, or voice quality. It is convenient to have this information, but we do not consider these
parameters important for the voice range profile. It is also useful to have an averaging program which
can average a series of phonetograms. Fig. 4 shows such an average of 18 good female voices (all
trained logopeds). Notice the dotted lines on both sides of the solid curve. These lines indicate the
standard deviation on the average. The lowest and the highest frequencies are shown with standard
deviations.
REFERENCES
[1] Bloothooft, G. (1981), Voice profiles, vowel spectra and vocal registers, CoMeT Conference,
Amsterdam.
[2] Pabon, J. P. Ii. (1989), Automata phonetogram recording supplemented with acoustical voicequality parameters, Proc. from Intern. Voice Symp., Edinburgh.
[3] Seidner, W. (1985), Objektive Qualitätsbeurteilung der Stimme mittels Dynamikmessungen,
Zeitschr. klin. Medizin 40.
[4] Hacki, T. (1988), Die Beurteilung der quantitativen Sprechstimmleistung, Folia Phoniatr. 40, pp
190-196.
[5] Gramming, P. (1988), The Phonetogram, an experimental & clinical study, Malmö, Sweden.
[6] Pedersen, M. F. et al. (1986), Computerized phonetograms for clinical use, Proc. X.X. , IALP
Congress, Tokyo, Japan.
[7] Ohlsson, A.-C. et al. (1986), Phonetograms of normal and pathological voices, Working Papers in
Logopedics & Phoniatrics 3, Lund, Sweden.
[8] Schutte, H. K. et al. (1983), Recommendation by UEP: Standardization of Voice Area
Measurements, Folia PhoniaTrica, 33.
59
60
ORIGINAL COMMUNICATIONS
The permission of the following copyright owners to reproduce original articles is gratefully
acknowledged:
Karger (Folia Phoniatricia, Reports of IALP Conferences)
Elsevier (Int. J. Pediatric Otorhinolaryngology)
Blackwell (Clinical Otolaryngology)
61
62
I
Folia phoniat. 29: 191-199 (1977)
Electroglottography Compared with Synchronized Stroboscopy in Normal Persons 1, 2
M. Fog Pedersen
ENT Department, Gentofte University Hospital, Hellerup
Introduction
During the past years intensive studies have been performed to develop the objective registration
of the movements of the vocal chords (Titze and Strong, 1975; Sondhi, 1975).
Fabre (1957), Michel (1967), Holm (1970), Reinsch and Gobsch (1972), Fourchin (1974),
Pedersen and Boberg (1973), and Pedersen (1974) have performed studies with electroglottography.
Kitamura et al. (1967), Hertz et al. (1970) and many others tried to describe the ultrasonic recordings
of vocal fold vibrations. Among others, Sonesson (1960) and Kitzing and Sonesson (1974) described
coefficients for the photoelectric picture of the movements of the vocal chords.
Köster and Smith (1970) concluded that electroglottography is a most relevant method to describe
the phases of the vocal chord in the chest register. Michel er al. (1970) have compared the opening and
the closing phases of the vocal chord in electroglottography with high-speech films. Fourchin (1974)
compared the electroglottography with photos of stroboscopic movements and discussed the problems
of using this method in a clinical setting. Lecluse et al. (1975) tested the glottographs on excised human
larynxes and found correspondence between glottograms and the movements of the vocal chords.
1
2
Paper read at the 8th World Congress of Phonetics, Leeds 1975.
Statistics performed with assistance from the Danish Research Fund.
Received: May 12, 1976; accepted: June 18, 1976.
63
I
Pedersen
192
Fig. 1. Experimental set-up: stroboscope, larynx mirror with attached photocell, electroglottograph
and oscilloscope.
The intention of this work was an objective registration of the stroboscopic examination
(Schönhärl, 1960). The evaluation of the glottographic curve is difficult. In this study, the
electroglottographic curves were used only for registration of the stroboscopic picture, and marking of
the phases was done with a photocell. It was then possible to plot the variations in the curves.
Measurement of coefficients for opening and closing phases was possible in those cases where the
glottographic curve could not be used alone to define these coefficients. Thereafter, quantitative
measurements of hyper- and hypofunctional diseases (Wendler et al. 1973) as well as organic disorders
of the vocal chords should be possible.
Method
The examination was carried out with a glottograph (FJ Electronics, type 830), a stroboscope
(Thiemcke, type KS 3), a sound level meter (Brüel & Kjaer, type 2205), and an oscilloscope with
polaroid camera. A photocell attached to a larynx mirror was connected to the oscilloscope (fig. 1).
64
I
Electroglottography Compared with Synchronized Stroboscopy
193
Fig. 2. Point I is the maximum opening phase and point II is the maximum closing phase in the
stroboscope, marked on the electroglottography. III and IV are the changes from. the plateau to the
slope on the glottographic curves. Various quotients are defined (see text).
At first, pictures were made with different time scales to record suitable electroglottographic
curves when received from the electrodes on the throat during intonation. The picture on the
oscilloscope was shown as sweeps with appropriate amplitude. Hereafter, the intensity of the sound
pressure in front of the throat was measured in relation to the background noise. The examination was
carried out in a standard examination room of the ENT department with a background noise of 38-42
dB. When the stroboscope was in use, the background noise was around 50 dB. During stroboscopy the
light glimpses form the photocell were shown in the maximum closing phase. The Thiemcke
stroboscope emitted a suitable light for fixation of the phase in the larynx mirror. The mirror with a
photocell marked the maximum closing phase on the glottography. After having photographed the
closing phase, the examination was repeated with the maximum opening phase.
Figure 2 shows the maximum opening (point I) and closing (point II) phases. Points III and IV are
the changes from the plateau to the slope of the glottographic curves.
The time from point III to point II is called 'a'. 'b' is the whole closing phase. 'c' is "the time from
point II to point IV. 'e' is the whole cycle. 'd' is the whole opening phase, and 'b' and 'd' correspond to
the stroboscopic findings. 'f' is the closing phase described by the inverse tangent of the slope.
65
I
Pedersen
194
Fig. 3-5. Examples of the variation of the glottographic curves in persons with normal
stroboscopy. The markings of the stroboscopic picture for maximum opened and closed phases are
seen.
66
I
Electroglottography Compared with Synchronized Stroboscopy
195
Mean values and deviations for the coefficients have been calculated: A (= a/e), B (= a/b). C (=
c/e), and D (= c/d). Calculations have also been computed for the mean values of the whole closing
phase (b), the opening phase (d), and f.
All values and coefficients have been calculated for every person. Mean values and mean
coefficients were calculated from these averaged values. The correlation calculations were obtained
from the co-variance matrix correlation coefficient analysis.
Material
The subjects consisted of 20 employees of the hospital, 14 females and 6 males. They were used to
evaluate the glottographic curve in relation to the stroboscopic picture.
The subjects had never suffered from voice problems. The function of the voice was normal with
respect to the stroboscopic picture. The intensity was around 68-72 dB. The
respiration was abdominal or mixed abdominal/costal in all persons. The frequency of speech was
200-220 Hz for women and 100-110 Hz for men.
The persons intonated an 'a' sound with their usual basic frequency. 'a' was used because it is the
most commonly used vocal in examinations with a larynx mirror.
Results
No differences were seen between the curves for men and women. The examination was very
simple and did not give technical problems, although an average of half of the persons became tired
from the long duration of the laryngoscopy.
The movements of the vocal chords in normal persons gave variations from person to person in the
glottographic curves. Figures 3-5 show examples of the variations. After marking of the closing phase,
70% showed a maximum point in the bottom of the curve of the closing phase. The maximum closing
phase was constant over the duration of the half-hour examination.
To evaluate the clinical use, a statistical analysis was performed among the different glottographic
coefficients to find uniform data. The results of the analysis for mean values and deviations are seen in
figure 6. The coefficients A and C had the smallest variations, with deviations for A and C in the 95%
confidence limits of 2.9-18.1% and 18.7-48.9%, respectively.
For the remaining coefficients, deviations from mean values were approximately 50%. This was
also the case for the calculation of the closing phase defined by the inverse tangent of the slope. No
correlation was found between the different coefficients calculated by comparing all values.
67
I
Pedersen
196
Fig. 6. Result of the analysis for mean values and deviations of the opening and closing phases,
and coefficients (fig. 2). Variations of the closing phases and of the opening phases in relation to the
whole cycle (= A and C) were smaller than those of the other values.
Discussion
The problems concerning electroglottography concentrate on the apparatus, evaluation of the
curves, and the possible practical use.
The use of the glottograph is not so difficult in the practical clinic. We are currently constructing a
foot pedal so that the photographs of the maximum opening and closing phases can be performed more
quickly.
68
I
Electroglottography Compared with Synchronized Stroboscopy
197
Measurement of the AC-coupling distorts the curve so that the closing phase is less well defined. If
the maximum opening and closing phases are marked stroboscopically, no distortion is calculated in
the time of the described coefficients. It is possible with glottography to use DC-coupling although the
curve cannot yet be stabilized on the oscilloscope.
Glottography has not been used in the clinical work for quantitative evaluation of voice disorders
due to the lack of certainty when describing the phases. Sonesson (1960) and Reinsch and Gobsch
(1972) have divided the curves in a way that cannot be utilized for evaluation of pathologic voices or of
normal voices in which the voice parameters were not optimal. When the curves are seen in relation to
the stroboscopic examination, evaluation of the relationship and duration of the phases can be
performed with greater certainty.
Beach and Kelsey (1968) and Kolmer et al. (1973) have encountered difficulties with the use of
glottography, probably because of the complex movement pattern of the vibrating vocal folds. Large
deviations were found between the phases of the curve in our study, but the quantitative descriptions in
normal persons are still regarded as a progressive step. It is now possible to evaluate pathologic voices
quantitatively from a normal material. Patients with organic voice disorders, i.e., tumors, seem to be
outside the normal ranges. Functional voice disorders lie in the borderline, but quantitative calculations
must be made for the various disorders together with a quantitative division in the variations of the
phases.
It is, of course, expected that a logopedic treatment changes a glottographic curve of the patient
towards an optimum for the different phases. Measurements are therefore desired on singers for the
duration and variation of the phases. Other persons with optimal voices could also be used. It is our
impression that a simultaneous quantitative acoustical analysis of the voice also could be helpful for
determining the borderline cases of normal and/or optimal voice.
Summary
A marking of the stroboscopic picture on electroglottographic curves was performed on 20 normal
persons. The maximum opening and closing phase was marked by a photocell on the glottographic
curve, the photocell following the glimpses from the stroboscope. The duration of the opening and
closing phases were calculated together with the slopes of the opening and closing phases. The closed
phase, defined by the inverse tangent of the slope, was also measured. The calculations were computed,
and no correlation was found between the durations of the phases. The variations in the material were
very large- In order to:
69
I
Pedersen
198
calculate the phases with greater accuracy, an examination on singers would be of value. Calculation of
the duration of the phases together with their variations from curve to curve would also be valuable.
References
Beach, J.L and Kelsey. CA.: Ultrasonic doppler monitoring of vocal-fold velocity and displacement. J.
acoust Soc. Am. 46: 1045-1047 (1968).
Fabre, P.: Un precode electrique percutane d'inscription de 1'accolement glottique au cours de Ira
phonation: glottographie de haute fréquence. Premiers resultats. Bull. Acad. natn., Med. 121: 66
(1957).
Fourchin, A.J.: Laryngographic examination of vocal fold vibration. An International Symposium of
Ventilatory and Phonatory Control System, p. 315 (Oxford University Press,-London 1974).
Fourchin, A.J. and Abberton, E.: First application of a new laryngograph. Med. bioL Must. 21: 172182 (1971).
Hertz, CH.; Lindström, K, and Sonesson, B.: Ultrasonic recording of the vibrating vocal folds. Acta
oto-lar. 69: 223 (1970).
Holm, C: Erste Ergebnisse einer Elektroglottographie im Kindesalter (Kongressbericht). Arch. kiln.
exp. Ohr.-Nas.-KehlkHeilk.196 (1970).
Holmer, N.-G.; Kitzing, P., and Lindström, K: Echo glottography. Ultrasonic recording of vocal fold
vibrations in preparations of human larynges. Acta oto-lar. 75: 454-463 (1973).
Kitamura, T.; Kaneko, T.; Asano, 1L, and Miura, T.: Ultrasonoglottography. A preliminary report Med.
Ultrason. 5: 40-41 (1967).
Kitzing, P. and Sonesson, B.: A photoglottographical study of the female vocal folds during phonation.
Folia phoniat. 26: 138-149 (1974).
Köster, J.-P. und Smith, S.: Zur Interpretation elektrischer und photoelektrischer Glottogramme. Folia
phoniat. 22: 92-99 (1970).
Lecluse. F. L.E.; Beocaar, M.P., and Venchuure, J.: The electroglottography and its relation to glottal
activity. Folia phoniat. 27: 215-224 (1975).
Michel, C van: Morphologic de la courbe glottographique dans certains troubles fonctionnels du
larynx. Folia phoniat. 19: 192-202 (1967).
Michel, C van; Pfister, K.A. et Luchsinger, R.: Electroglottographic et cinematographic laryngée ultrarapide. Folia phoniat 22: 81-91 (1970).
Pedersen, M.F.: Clinical investigations of patients with benign tumors of the larynx before and after
microlaryngoscopy. 16th Int Congr. Phoniatrics Logopedics, Interlaken. Folia phoniat. 26: 184-185
(1974).
Pedersen, M.F. and Boberg, A.: Examination of voice function of patients with paralysis of the
recurrent nerve. Acta oto-lar. 75: 372-374 (1973).
Reinsch, M. und Gobsch, H.: Zur quantitativen Auswertung elektroglottographischer Kurven bel
Normalpersonen. Folia phoniat. 24: 1-6 (1972).
Schonharl, E.: Die Stroboskopie in der praktischen Laryngologie (Thieme, Stuttgart 1960).
70
II
International Journal of Pediatric Otorhinolaryngology, 20 (1990) 17-24
Elsevier
17
PEDOT 00660
Fundamental voice frequency in female puberty
measured with electroglottography during
continuous speech as a secondary sex caracteristic.
A comparison between voice, pubertal stages,
oestrogens and androgens
M.F. Pedersen, S. Moller, S. Krabbe, P. Bennett and B. Svenstrup
Department of Pediatrics, Rigshospitalet, Copenhagen (Denmark) and Hormone and Biostatistic
Departments, Statens Seruminstitut, Copenhagen (Denmark)
(Received 3 February 1990)
(Accepted 24 April 1990)
Key words: Estrogen; Puberty; Fundamental frequency
Abstract
The change of fundamental voice frequency in continuous speech in female puberty was analysed
in 47 girls by comparison of 2000 consecutive electroglottographic cycles in a reading situation. The
results were compared with serum concentrations of androgens (dihydroepiandrosterone, 8-4androstenedione, testosterone, and sex hormone binding globulin), oestrogens (oestradiol, oestrone, and
oestronesulphate), and somatic puberty (weight, height, mamma stages, and pubic hair stages).
Fundamental frequency in continuous speech was related only to oestrone r = - 0.34, (P < 0.05). But the
tone range in continuous speech and the lowest tone in the phonetogram were found to be significantly
correlated with many of the pubertal and hormone parameters. All these correlations could be
explained by a common age-dependency. By multiple regression analysis different sets of variables for
prediction of speaking fundamental frequency were found before and after menarche.
Correspondence: M.F. Pedersen. ENT, consulting phoniatrician. Østergade 18. DA 1100 Copenhagen.
Denmark.
0165-5876/90/$03.50 O 1990 Elsevier Science Publishers B.V. (Biomedical Division)
71
II
18
Introduction
Electroglottographic measurements have been used for several years [2,11]. The systematic
standardization as used by Kitzing [8] for fundamental frequency analysis supplemented with
phonetography as proposed by Seidner and Schutte [16] was our prerequisite for studying the
fundamental voice frequency in male puberty [15], and in this study the procedure was carried out in
female puberty. On the basis of an earlier pilot study [12] we have tried to elucidate the pubertal
development of female fundamental frequency and its variation in continuous speech by electroglottographic registration. By means of multivariate statistical analysis the fundamental frequency in
continuous speech was compared with pubertal stages and hormonal changes of adrenals and gonads
[4]. The stratified study was carried out in a voice-trained group of girls in a singing school in order to
have a well-defined level of voice-training and musicality for the comparison.
Materials and methods
From each of 10 school class levels in a singing school 4-5 girls were selected at random,
altogether 47 girls within the age range of 8-19 years. Informed consent was obtained from the homes
of all girls. By indirect laryngoscopy and stroboscopy no pathological conditions were discovered and
the pubertal development was normal. The fundamental frequency was analysed with an average of
2000 consecutive electroglottographic signals in divisions of semitones from 60 to 784 Hz. The
average values of signals in Hz and standard deviations in semitones were computed. The method is
described in greater detail elsewhere [8,14]. The phonetograms were measured with a Bruel and Kjaer
sound level meter [16], the lowest and middle semitones were extracted and the areas were computed
[13]. The stages of pubic hair and mamma development were examined as well as height and weight
[17]. Information about menarche was obtained from the girls' homes with a written consent. The
serum concentrations of adrenal and gonadal hormones were measured in blood samples taken before
noon [4]. For those girls who had experienced menarche they were taken on the 3rd to 6th day after the
first menstruation day to exclude fluctuations. The hormones examined were total and free serum
testosterone, dihydroepiandrosterone (DHEAS), S-4-androstendione, sex hormone-binding globulin
(SHBG), oestrone (El), oestrone sulphate (E1SO4), and oestradiol (E2). The methods were described by
among others Lykkesfeldt et al. [10] and Pedersen et al. [14].
Statistical methods
The observations were logarithmically transformed in order to obtain normal distribution. The
differences between age groups were analysed by one-way analysis of variance (ANOVA), and
correlation coefficients were calculated for all pairs of variables to illustrate the connection between
them. The influence of a common age-dependency was studied by means of partial correlation
coefficients. A multidi-
72
II
19
mensional linear regression analysis was carried out with fundamental frequency in continuous speech
as dependent variable and puberty stage and hormone levels (in total 12 variables) as describing
factors. BMDP programme (9R) with Mallow's CP as test criterion was used (1981). This analysis was
carried out for the two subgroups of girls pre- and postmenarche as well as for the whole group.
Results
Table I shows for 3 age groups the geometric means of fundamental frequency in continuous
speech in Hz and its variation in semitones compared with some of the parameters in the
phonetograms. The first column represents the age group 8.6-12.9 years which is considered a child
group in connection with the voice phenomena, and the third column 16-19.8 years which is an adult
group as defined by the singing teachers in the music school, (all parameters were found in 18, 12, and
11
TABLE I
The geometrical means in 3 age groups for voice parameters, pubertal stages, and hormones
The relative standard deviations were between 11 and 140% (significance of difference between
groups: ** P <0.01,*P <0.05).
73
II
20
TABLE II
The correlation coefficients (r) between the fundamental frequency in continuous speech (Fo) and its
tone range in relation to female sex hormones, adrenal hormones, pubic hair stage, mamma
development, and voice parameters
The correlations to the lowest tone in the phonetogram and age are also shown. (Significance:
** P <0.01, * P <0.05.)
girls respectively for the 3 groups). Significant differences between all 3 groups were found with oneway ANOVA marked with * * (P < 0.01) and * (P < 0.05). No significant change in serum
testosterone, which is the most potent androgen in girls, was shown. All girls in group 3 had menarche.
Among the voice parameters the tone-range in continuous speech was significantly different in the 3
groups (P < 0.01). This was also the case for the lowest tone in the phonetogram and the
phonetographic area (P < 0.05 and P < 0.01 respectively). However, the fundamental frequency in
continuous speech was not significantly altered from the first to the third group.
Table II shows the correlation coefficients (r) for fundamental frequency in continuous speech in
relation to female sex hormones, adrenal hormones, puberty stage phenomena and voice parameters.
Only the correlation coefficient for El was significant (-0.34, P < 0.05). The correlation coefficient for
E2 was lower, this hormone is related to the menstrual cycle after menarche. For the tone range of
fundamental frequency in continuous speech and lowest tone in the phonetogram many significant
correlations with the hormone and pubertal parameters were found
74
II
21
Fig. 1. Illustration of the connection between age and the voice parameters analysed in Tables I and II,
(phonetogram area, tone range (e.g. voice range) in continuous speech, tone range in singing and the
lowest semitone measured from the phonetogram, and fundamental frequency in continuous speech). z.
Mamma stage 1; , Mamma stage 2-4; U. Mamma stage 5-6.
75
II
22
TABLE III
The best sets of describing variables for the logarithm of fundamental frequency in continuous speech
(Fo) during female puberty, calculated for the whole group and for the two subgroups classified by
menarche
Linear correlation coefficient SHBG, r -0.93 to menarche. * P < 0.05; * * P < 0.01; * * * P < 0.001.
but all could also be described by a common age-dependency. Fig. 1 illustrates the change of voice
parameters analysed in Table II.
Table III shows, that E1SO4 and tone range in speech were the most useful hormonal variables that
could be found for prediction of voice change in puberty if all girls were included in a statistical setting
to examine prediction. The best result obtained showed only significance at the 5% level. If, however,
the material was divided in pre- and postmenarche girls, more significant results were obtained. The
table shows that the variables selected in the two groups were different; before menarche height and
E1SO4 had a prediction significancy of P < 0.001 and after menarche tone range in speech had a
prediction significancy of P < 0.001.
Discussion
This study showed that fundamental frequency in continuous speech changed insignificantly from
256-248-241 Hz in a stratified study of 47 musically gifted and voice-trained girls from 8.6-12.9 years
through 13-15.9 years to 16-19.8 years. The tone range in fundamental frequency changed from 3.74.2-5.2 semitones and the lowest tones in the phonetograms from 166-156-145 Hz, both changes being
significant (P < 0.01 and P < 0.05, respectively). For the girls before menarche, height and E1SO4 were
the most important predictive variables in relation to fundamental frequency in continuous speech,
whereas the tone range of fundamen-
76
II
23
tal frequency and time after menarche were the most important predicting variables for girls after
menarche. Weight and sex hormone binding globulin (SHBG) in girls had a higher correlation
coefficient to pubertal stages than androgen and oestrogen development in studies by Apter and Vihko
[1]. Bartsch et al. [3] found that SHBG < 56.8 nmol was an expression of past puberty in girls which is
in correspondance with Lee et al. [9] who suggest that this protein has a metabolism which is similar in
boys and girls. These results corresponded to ours only in boys where SHBG was the best predictive
factor for the fundamental frequency in continuous speech in the pubertal group [14]. Further studies
on SHBG have to be carried out. A decrease in fundamental frequency in continuous speech of 15 Hz
and the lowest tone in the phonetogram of 21 Hz on average in the 3 groups is no large alteration. But
the correspondance between the two phenomena should be noted even if the measuring methods were
different, electroglottography and phonetography, respectively. The singing teachers, however, agree
with our results: a clear difference could be heard and was also measured by increase in the
phonetographic area from 17.3-21.8-28.3 cm' in the 3 age groups. Studying female voice is gaining
more interest [6], a differentiation of untrained/ trained female voices seems necessary in research on
female voices which is underlined by research on female voice perception where Gelfer [7] found that
experienced listeners relied on fundamental frequency (= pitch) somewhat more than naive listeners
who to a greater degree used the tone range in continuous speech. Supplementary understanding of the
central regulating mechanisms, for example by analysing SHBG and its relation to brain function [5],
might be an interesting way to attain knowledge of the pubertal changes in the human voice [18].
Conclusion
Fundamental voice frequencies in continuous speech in female puberty decrease as in boys, but
more slightly, which is also the case for the lowest tone of the voice.
The change of tone range in continuous speech seems to be a necessary parameter in connection
with female pubertal stages and oestrogen development. The fundamental voice frequency in
continuous speech was found significantly related to oestrone in girls corresponding to the relation to
testosterone earlier found in boys.
References
1
2
3
4
Apter, D. and Vihko, R., Premenarcheal endocrine changes in relation to age at menarche, Clin.
Endocrinol., 22 (1985) 753-760.
Baken, R.J. (Ed.), Clinical Measurement of Speech and Voice, College Hill. Boston. MA (1987).
Bartsch, W., Horst, H.-J., and Derwahl, A: M., Interrelationship between sex hormone binding
globulin and betaoestradiol, testosterone, 5 alpha-dihydrosterone, thyroxine, and triiodo-thyronine
in prepubertal and pubertal girls, J. Endocrinol. and Metab., 50 (1980) 1053-1065.
Brook, C.G.D., (Ed.), Clinical Paediatric Endocrinology, 2nd edn., Blackwell Scientific
Publications. Oxford, 1989.
77
II
24
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Cunningham. SK., Laughlin, T., Culliton, M. and Mclkenna, T.J., Plasma sex hormone binding
globulin levels decrease during the second decade of life irrespectively of pubertal status. J. Clin.
Endocrinol. Metabol., 58 (1984) 915-918.
Fant, G., Gobl, C., Karlsson, I. and Lin, Q., The female voice-Experiments and overview, J.
Acoust. Soc. Am., 82, Suppl. 1 (1987) 90.
Gelfer, M.P., A multidimensional scaling study of normal voice quality in females, J. Acoust. Soc.
Am., 81, Suppl. 1 (1987) 69.
Kitzing, P., Glottographisk Frekvensindikering, Thesis, Univ. of Lund, Malmö. Sweden, 1979
Lee, J.R., Lawder, L. E., Townsend, D.C., Wetherhall, J.D., and Haehnel, R., Plasma sex hormone
binding globulin concentration and binding capacity in children before and during puberty, Acta
Endocrinol., 109 (1985) 276-280.
Lykkesfeldt. G., Bennett, P., Lykkesfeldt, A.E., Micic, S., Moller, S. and Svenstrup, B., Abnormal
androgen and oestrogen metabolism in men with steroid sulphatase difficiency and recessive Xlinked ichthyosis, Clin. Endocrinol., 123 (1985) 385-393.
Pedersen, M.F., Electroglottography compared with synchronized stroboscopy in students of
music. Study of sounds, Phonetic Soc. Japan, 18 (1978) 423-434.
Pedersen, M.F., Moller., Krabbe, S., Munk, E., Bennett, P. and Kitzing, P., Change of voice in
puberty in choir girls, Acta Otolaryngol., Suppl. 412 (1984) 46-49.
Pedersen, M.F. and Lindskov Hansen, H., Computerized phonetography for clinical use, Acta
Otolaryngol., Suppl. 412 (1984) 138.
Pedersen, M.F., Moller, S., Krabbe, S., Munk, E. and Bennett, P., A multivariate statistical analysis
of voice phenomena related to puberty in choir boys, Folia Phoniatr., 37 (1985) 271-278.
Pedersen, M.F., Moller, S., Krabbe, S. and Bennett, P., Fundamental voice frequency measured by
electroglottography during continuous speech, Int. J. Pediat. Otorhinolaryngol., 11 (1986) 21-27.
Seidner, W. and Schutte, H.K. Empfehlung der UEP: Standardisierung Stimmfledmessung/
Phonetographie, HNO-Praxis, 7 (1982) 305-307.
Tanner, J.M., Growth of Adolescence, 2nd edn., Blackwell Scientific Publications, Oxford, 1962.
Weiss, D.A., The pubertal change of the human voice, Folia Phoniatr., 2 (1950) 126-169.
78
International Journal of Pediatric Otorhinolaryngology. 11 (1986) 21-27
Elsevier
III
21
POR 00351
Fundamental voice frequency measured by
electroglottography during continuous speech.
A new exact secondary sex characteristic
in boys in puberty
M.F. Pedersen, S. Møller, S. Krabbe and P. Bennett
The School Health Service, Copenhagen. Statens Seruminstitut, Biostatistical Department and
Rigshospitalet. Pediatric Department, University of Copenhagen (Denmark)
(Received September 23rd. 1985)
(Accepted December 9th. 1985)
Key Words: androgens - puberty - fundamental frequency
Summary
The variation of fundamental voice frequency measured by the average of 2000 consecutive
electroglottographic cycles in a reading situation has been examined in relation to pubertal
development and androsens. Fundamental frequency among other parameters was related to height (r 0.82), pubic hair stage (r -0.87), testis volume (r -0.78), total testosterone (r -0.73) and serum hormone
binding globulin (r 0.75). Single observations of fundamental frequency show a clear grouping of
results under and over 200 Hz. Fundamental frequency of more than 200 Hz and serum testosterone of
more than 10 nmol/1 probably represent values for a boy in puberty. There seem to be comparable
relations with other androgens and with serum hormone binding globulin.
Introduction
It is only recently that advanced technology, both economically feasible and readily available, for
analysis of fundamental voice frequency (Fo) in continuous speech has been developed [3]. We have
tried in several cases to obtain knowledge
Correspondence: M. F. Pedersen. The School Health Service. Copenhagen. Sjaelør Boulevard 135.
DK-2500 Valby, Denmark.
79
III
22
about the development of Fo [14-17]. In this study we present figures showing the relation between Fo
and (1) androgen development and (2) secondary sex characteristics.
Many medical scientists have worked with puberty development using 3 clinical voice groups:
child, puberty and adult voices [9.20]. Others. e.g. Vuorenkoski et al. [22] have used less accurate
measuring methods of Fo. So far. it has not been possible to rely on the methods used for the study of
voices in puberty in a clinical and pathological situation. Therefore, the voice has not obtained the
place in puberty stage evaluation and pathology which it deserved.
The aim of this work has been to compare the well-defined examination of Fo during continuous
speech [13] with androgen development during puberty and also with secondary sex characteristics in
order to find out if Fo could be used to characterize the pubertal stages.
Materials and Methods
Forty-eight boys from the Copenhagen Singing School with healthy voices and normal puberty
development were stratified with 5 boys from each class of the 3rd to 12th level in cross-sectional
examinations. Examination of androgens: serum testosterone, dihydrotestosterone, free testosterone,
serum hormone binding globulin (SHBG), delta 4 androstendione, and dihydro-epi-androsterone
sulphate was carried out on blood samples taken before noon and secondary sex characteristics were
examined at the same time by an experienced pediatrician. The Fo in Hz was analyzed with our
dataprogram with 2000 consecutive cycles by means of electroglottography during reading of a
standard text [11,13] and the voice range in a reading situation in semitones was also recorded.
Together with an experienced singing master phonetograms were made [19] in which e.g. the lowest
and middle frequency plus the total range in semitones were measured to elucidate the relation between
Fo during continuous speech and the biological resources as regards the voice range and intensity
variation. The boys. all of whom were or had been singing in the Copenhagen Boys' Choir, were
examined by indirect laryngoscopy and stroboscopy and showed no abnormal findings. A logarithmic
transformation of the measurements of all the quantitative variables was performed to obtain normal
distributions. The correlation coefficients between the variables were calculated including all
observations, and a one-way analysis of variance was performed. dividing the boys into 3 age groups.
Results
Table I shows the geometrical means in 3 age groups of voice characteristics. androgens. height.
testis volume and puberty hair stage. The correlation coefficients (r) between Fo and these variables are
also shown. This table has in part been given
80
III
23
Fig. 1. The observations of the voice parameters: phonetogram area. Fo variance in continuous speech.
middle and lowest biological frequency together with Fo. All parameters are related to age as abscissa.
z. age: 8.7-12.9 years: . age: 13-15.9 years; U, age: 16-19.5 years.
(use 150% scale ed. ref. Lars Paaske)
81
III
24
in a previous paper [17] but has now been extended with the pubertal characteristics.
In Fig. I the computed observations of the voice parameters: phonetogram area. Fo variance in
continuous speech, the middle and lowest biological frequency together with Fo are seen with age as
abscissa. All parameters were related to age with P < 0.01.
In Fig. 2 some of the parameters with the highest r values in relation to fundamental frequency:
pubic hair stage, testis volume, total serum testosterone and serum hormone binding globulin were
plotted with Fo as abscissa. The difference between child and adult values is seen. The group in
puberty is well defined by the
Fig. 2. Pubic hair stage. testis volume, total serum testosterone and serum hormone binding globulin in
relation to Fo as abscissa. For meaning of symbols. see Fig. 1.
82
III
25
TABLE I
GEOMETRICAL MEANS IN 3 AGE GROUPS OF SECONDARY SEX CHARACTERISTICS.
HEIGHT AND WEIGHT IN PUBERTY, ANDROGENS AND VOICE CHARACTERISTICS
The correlation coefficients (r) between Fo and these variables are also given.
two parameters, Fo and total serum testosterone, the first still being > 200 Hz and the second < 10
nmol/l serum.
Discussion
We have shown data of Fo during continuous speech related to (1) androgen development, (2)
puberty characteristics and (3) other biological voice phenomena. Naturally, these new results must be
considered as preliminary, and further confirmation, also in less musically trained boys. is needed.
However, the average values in the age groups are in accordance with the findings of others [1,5-7,211,
the total voice range in our study being a little larger in the trained boys.
This study shows changed levels for both serum testosterone and SHBG before and after puberty
as defined by Fo during continuous speech.
We have earlier tentatively stated that a child Fo during continuous speech is rarely found with a
total serum testosterone value > 10 nmol/l [161. This is confirmed in the present work (Fig. 2). Thus, it
seems that boys in puberty have Fo of a child and serum testosterone values of an adult. Child and adult
values for the
83
III
26
other androgens obtained in this work were also correlated to Fo. However, for a proper formulation of
values of limit for child and adult age. it seems reasonable to wait until a larger number of boys with Fo
of around 200 Hz between the two levels has been obtained.
By means of such a material it could be possible to obtain a better understanding of the fall in Fo
which changes the child voice to the male adult voice. However, we are of the opinion that the results
shown in Table I and Fig. 2 can be used. They also agree with those of others with respect to serum
hormone binding globulin [4], where it is found that SHBG might be under control of other influences
than androgens. Serum testosterone and other androgens should probably be analyzed together with
hypothalamo-pituitary hormonal development to get further understanding of Fo in puberty [2]. As
regards the completed growth of the vocal ligament. Hirano [10] has concluded that the adult length
occurs in young males about 20-years-old, the vocal ligaments having reached mature structure
between 12 and 16 years. Our results could also be compared with the results on growth of the pubertal
larynx by Kahane [12] who found an r value of 0.78 between crown heel length and total length of
vocal folds in prepubertal and pubertal periods in 10 Causasian males and 10 Caucasian females.
We anticipate that the development of the girls' voices can be elucidated in the same way. We have
made a pilot study on the relations between Fo development and other female voice phenomena
compared with hormonal and pubertal development [18]. Furthermore, it is our opinion that our studies
can be used in pathological cases of Fo. We agree with Guidet and Chevrie-Muller that Fo
measurement is of importance in patients with e.g. neurological and psychiatric diseases [8].
We hope that this study on Fo in continuous speech related to hormonal development and sex
characteristics will contribute to a better understanding and a practical evaluation of the human voice in
puberty.
References
1
2
3
4
5
6
7
8
Bastian. H.-J. and Unger. E., Untersuchung des Zusammenhanges von Akzeleration. Mutation and
Dysphonie anhand von Längsschnittuntersuchungen. Artzl. Jugendkd. Leipzig. 71 (1980) 205-211.
Brook. C.G.D. (Ed.). Clinical Paediatric Endocrinology. Blackwell Scientific Publications. Oxford.
1981.
Coleman. R.F., Tracking vocal changes in laryngeal pathology patients. In Vocal Fold Physiology.
College-Hill Press. San Diego. 1983. pp. 432-434.
Cunningham. S., Loughlin. T., Culliton. M. and McKenna. T.J., Plasma sex hormone binding
globulin levels decrease during the second decade of life irrespective of pubertal status. J. Clin.
Endocrinol. Metabol., 58 (1984) 915-918.
Frank. F. and Donner, F., Vom Knabenalt zum Startenor. Eine sonographische Analyse. Sprache Stimme - Gehor. 5 (1981) 21-32.
Frank. F. and Sparber. M., Stimmmumfänge bei Kindern aus neuer Sicht. Folia Phoniatr., 22
(1970) 297-402.
Flatau. T.S. and Gutzmann. H., Die Singstimme des Schulkindes. Arch. Laryngol., 20 (1905) 327348.
Guidet. C. and Chevrie-Muller, C., Computer analysis of prosidic and electroglottographic
parameters in diagnosis of pathologic voice. Quant. Linguist., 19 (1984) 233-263.
84
III
27
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Hagg. U. and Taranger. J., Maturation indicators and the pubertal growth spurt. Am. J. Orthod., S2
(1982) 299-309.
Hirano. M., Kurita. S. and Nakashima. T., The structure of the vocal folds. In Vocal Fold
Physiology. Tokyo. 1981, pp. 33-39.
International Phonetic Association (IPA) Book. The Northwind and the Sun.
Kahane. J.C., Growth of the human prepubertal and pubertal larynx. J. Speech. Hear. Res., 25
(1982) 446-455.
Kitzing, P., Glottografisk Frekvensindikering, Thesis. Lunds Universitet. Malmö. 1979.
Pedersen, M.F., Electroglottography compared with synchronized stroboscopy in normal persons.
Folia Phoniatr., 29 (1977) 191-199.
Pedersen. M.F., Electroglottography compared with synchronized stroboscopy in students in
music. Study of sounds. Phonet. Soc. Jpn., 1S (1978) 423-434.
Pedersen, M. F., Kitzing, P., Krabbe, S. and Heramb. S., The change of voice during puberty in 11
to 16 years old choir singers measured with electroglottographic fundamental frequency and
compared to other phenomena of puberty, Acta Otolaryngol., Suppl. 386 (1982) 1S9-192.
Pedersen. M.F., Munk, E., Bennett, P. and Møller, S., The change of voice during puberty in choir
singers measured with phonetograms and compared to androgen status together with other
phenomena of puberty. Proc. 10th Int. Congr. Phonet. Sci., 1984, pp. 604-608.
Pedersen, M.F., Munk, E., Moller, S., Bennett, P. and Krabbe. S., The change of voice during
puberty in girls measured with electroglottographic fundamental frequency analysis and
phonetograms compared with changes of androgens and secondary sex characters, Acta
Otolaryngol. (Stockholm). Suppl. 412 (1984) 46-49.
Seidner. \V, and Schutte. H.K., Empfehlung der UEP: Standardisierung Stimmfledmessung/
Phonetographie, HNO-Praxis, Leipzig. 1982, pp. 305-307.
Tanner. J.M., Growth of Adolescence. 2nd edn., Blackwell Scientific Publications. Oxford. 1962.
Tosi. O., Postan, D. and Bianculli, D., Longitudinal study of children's voices at puberty. Proc.
16th Int. Congr. Logoped. Phoniatr., 1974, pp. 486-490.
Vuorenkoski. V., Lenko. H.L., Tjernlund, P., Vuorenkoski. LL and Perheentupa. J., Fundamental
voice frequency during normal and abnormal growth and after androgen treatment. Arch. Dis.
Childh., 53 (1978) 201-209.
85
86
IV
©1955 S. Karger AG, Basel
0015-5705155/0376-027152.75/0
Folia phoniat. 37: 271-278 (1985)
A Multivariate Statistical Analysis of Voice Phenomena
Related to Puberty in Choir Boys
M.F. Pedersen, S. Moller, S. Krabbe, E. Munk, P. Bennett
Copenhagen Singing School, Sankt Annae Gymnasium;
Department of Pediatrics, Rigshospitalet, University of Copenhagen;
Hormone and Biostatistical Departments, Statens Seruminstitut, Copenhagen, Denmark
The aim of our work has been to treat boys from the Copenhagen Boy's Choir who have pubertal
problems [1]. We have tried to obtain a better understanding of the physiology of voice change in
puberty and the present work is an extension of a previous study [2]. A more specific aim has been to
find a variable which might predict the time when boys lose the high tones during puberty.
In recent years better methods have been developed for the evaluation of hormonal effects on voice
phenomena in puberty. The phases of the electroglottographically measured movements of the vocal
cords in normal persons and in music students have been studied [3, 4]. Among others Kitzing [5] used
electoglottography for the study of fundamental frequency (Fo). Phonetography has been proposed by
Seidner and Schutte [6) as a standard method in phoniatrics. Bloothooft [7] has developed an automatic
method of phonetography including analysis of the quality of singing formants. We have reported
primary results on phonetography in puberty [8]. Other authors have examined puberty from a
phoniatric [9, 10, 11-13] or a pediatric point of view [14-16]. Recently, studies on the growth of larynx
cartilage [ 17], on the finishing of the cover layer of the vocal cords [18], and on the development of
the phonetogram area [19] have been published. These results, however, did not include investigation
of the hormonal development.
In order to obtain a better knowledge of the connection between Fo, phonetograms, age and
secondary sex characteristics together with androgen status we examined a group of choir boys at
various ages of puberty and adolescence in a cross-sectional study.
Materials and Methods
Forty-eight boys aged 8-19 years from the Copenhagen Boys' Choir participated in the crosssectional study. The selection of the boys was stratified with an equal number selected at random from
each school class. Boys who fulfil the criteria of a fine musicality
87
IV
272
Pedersen/Møller/Krabbe/Munk/Bennett
and a good voice enter the Singing School on the 3rd class level and remain in the school until the 12th
class level. All the boys chosen for the study were in good health, their vocal cords had no organic
changes and none of them took drugs. An informed written consent was obtained from their homes.
Blood samplings were carried out in the morning, pubertal development was evaluated by the pubic
hair stage according to Tanner [20] and testicular volume was assessed by an orchidometer.
Voice Analysis
All the boys were stroboscopically examined to exclude abnormal findings. Fo measured by
electroglottography was computerized together with voice range. A program was made by our
laboratory based on an average of 2,000 electroglottographic cycles consecutively measured during
reading of the standard text 'The Northwind and the Sun' in a balanced Danish translation, mean
frequency in Hertz and 95 interval expressed in semitones. The phonetograms were performed by a
phoniatrician and an experienced singing master according to the proposed international standard [6].
Further analysis was carried out after recording on a phonetograph developed together with a Danish
engineering firm (see under 'Equipment').
Androgen Status
The determination of sexual hormone-binding globulin (SHBG) in human serum is based on
incubation of the serum with a fixed amount of radioactive dihydrotesterone and increasing amounts of
nonradioactive dihydrotestosterone, sedimentation of the SHBC-dihydrotesterone complex with
ammonium sulfate counting a part of the supernatant, and calculation of the capacity of serum hormone
globulin binding of steroid with a Scatchard plot. The intra-assay coefficient of variation for SHBG
was 7 %. All steroids were determined by radioimmunoassay. Testosterone, dihydrotestosterone and
androstenedione were determined simultaneously after extraction and chromatography on Celite
columns. Dihydroepiandrosterone sulfate (DHEAS) was converted to unconjugated dihydroepiandrosterone by solvolysis. whereafter the free steroid was measured by radioimmunoassay after
chromatography. The intra-assay coefficient of variation for all steroid determinations was 9%. Free
(nonprotein-bound) testosterone was calculated from the binding capacity of SHBG and the
testosterone value, assuming that the serum albumin was constant.
Equipment
The following equipment was used: electroglottograph FJ Electronics type 830; Microdatamate
Digisys type 2205, Cybernex model LGR-1; spectral analyzer Nicolet 440 A mini ubiquitous; sound
level meter with microphone, Bruel & Kjær type 2205; oscilloscope Tectronix type 510 3n, storage
with polaroid camera; text: IPA book 'The North Wind and the Sun; stroboscope: Bruel & Kjær type
5066; phonetograph: Voice Profile Instruments type 8301.
Statistical Analysis
A logarithmic transformation of the observations of all parameters was required to obtain normal distribution. Data were investigated by one-way analysis of variance dividing the boys into three age-groups,
and the correlation coefficients were calculated by comparing all the parameters. Multiple regression
analysis of F, with the hormone values, age and stage of puberty as independent variables was carried
out.
Results
Figure 1 illustrates phonetograms obtained during singing at different ages: (a) 9.25 years: the boy
starts at 3rd school class level; (b) 11.9 years: the singing quality is maximal; (c) 14.6 years: the boy
can no longer sing properly; (d) 18.2 years: the young man sings again with a good voice.
Table I shows the results in three agegroups. Fo was correlated to all data (p < 0.001). Comparison
with the phonetogram area shows that Fa was most strongly correlated to the lower biological tones
with r = 0.93, less to the middle tones with r = 0.82 and to the higher tones only with r = 0.48, which is
not enough to predict any connection between the values. The correlation of the phonetogram area to
Fo was r = 0.57. Analysis of the relation between intensity values in the phonetogram and the results
shown in table I gave correlation coefficients
88
Voice in Puberty
IV
273
Fig. 1. Typical phonetograms of the age-groups 9.25 years (a), 11.9 years (b), 14.6 years (c) and 18.2
years (d).
all < 0.46, wherefore the results are left out of this study.
All variables demonstrated in table I were clearly related to age with p < 0.001. The coefficient of
variation within groups of age varied between 14 and 100% and showed a fall with increasing age. The
variation was constant in the voice range during' reading and also for the higher and middle biological
tones in the phonetograms, as well as for the adrenal hormone DHEAS. However, the three variables
Fo in speech, lower biological tones in the phonetogram and SHBG showed greatest variation in the
puberty group. In order to elucidate this finding supplementary statistical calculations were made.
Statistical Calculations
Figure 2 shows Fo and two hormonal values related to age. For an individual this relation is not
linear by age, but for the total population the development could be described by a smooth curve or a
linear regression. The analysis of multiple regression with all six hormone values, age and stage of
puberty as independent variables showed that Fo could
89
IV
274
Pedersen/Møller/Krabbe/Munk/Bennett
Table I. Geometric means and ranges of androgen status. testicular function and voice
be described just as well by the two parameters age and log SHBG only. When grouping the values
by stages of puberty (0-1. 2-4, 56) the dependency of Fc, by age disappeared. while the dependency of
log SHBG only remained in the middle group (2-4). which is the interesting one (p < 0.05; table II).
In order to investigate this connection further a regression analysis of log Fo was carried out with
log total serum testosterone, log SHBG and log DHEAS as independent explaining variables for three
types of grouping. namely by stage of puberty, testicular volume or total serum testosterone. With
respect to puberty the same result was found, i.e. only log SHBG was significant and only for the
middle group. As regards testicular volume and total serum testosterone also only log SHBG and.
partly. serum testosterone were significant. but only in the groups with the highest hormone values.
Discussion
We found a clear relation between Fo and androgen status. Multiple regression analysis showed
that SHBG had the highest predictive value ( p < 0.05). The critical point is that there are few children
in the group in whom the changes happen. From a biological
90
Voice in Puberty
IV
275
point of a narrowing of the dependency on pubertal stages 2-4 to hormone development. especially
SHBG and to Fo was observed. Age disappeared as a dependent variable. Studies. including more
children in pubertal groups 2-4. should be carried out to confirm this observation. The results of our
hormonal investigation seem to be in accordance with the findings of other authors [21- 23].
91
IV
276
Pedersen/Møller/Krabbe/Munk/Bennett
Table 11. Coefficients estimated from multiple regression of F„ depending on six hormone values,
age and stage of puberty after reduction of independent parameters
There is no doubt that the well-documented electroglottographic method used by us for Fo
measurement in a reading situation is an improvement in the clinical work. This also applies to the
phonetogram area. the development of which is significantly related to age. Further work has to be
carried out in order to obtain an optimal tool for measurement of intensity [24. 25]. Phonetography as
defined by Seidner and Schutte [6] is a manual analysis of the total sound 30 cm from the mouth. By
means of automatic analysis the measurement. especially of low intensities. car. be refined [26]
resulting in a more exact definition of the total area. The lack of such an exact analysis is probably the
reason why a relation between intensity and androgen status was not found by us. The register phenomena in the phonetograms are in part related to formants and a measuring of the lowest intensity with an
electroglottograph will demonstrate in a better way the relation between biological tones and formants
[27. 28].
Conclusion
In a previous study [2] we found that total serum testosterone was never under 10 nmol/I serum
when Fo had reached adult values. In this study we found that SHBG has a possibly significant value (p
< 0.05) as predictive factor of the change in FO from child to adult voice in boys. Cunningham et al.
[23] report that other factors than sex steroids may control the fall in SHBG levels in the second decade
of life usually coincidental with puberty. These factors are still unknown.
We found that F0 during continuous speech is significantly related to the lowest tones in the
phonetogram. In order to encircle the problem of the boys losing high tones in puberty. one way might
be to study more carefully the relation between Fo in speech and the highest tones in the phonetogram.
It is possible that the study of the high tones with supplementary analysis. e.g. of singingformants [27]
and phonation time. may show other relations than those observed for the maximal tones in the
phonetogram.
92
Voice in Puberty
IV
277
Acknowledgement
We are grateful to Docent Med. Dr. P. Kitzing. Department of Phoniatrics and Logopedics,
University ENT Clinic. General Hospital. Malmö, Sweden, for kindly assisting with good and overall
extremely useful advice also for future voice research.
References
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Blatt. I.M.: Training singing children during the phases of voice mutation. Ann. Otol. Rhinol. Lar.
92: 462-468 (1983).
Pedersen. M.F.; Kitzing. P.; Krabbc, S.; Heramb, S.: The change of voice during puberty in 11-16
year old choir singers measured with electroglottographic fundamental frequency and compared to
other phenomena of puberty. Acta oto-lar., Stockh., suppl. 386, pp. 189-192 (1982).
Pedersen, M.F.: Electroglottography compared with synchronized stroboscopy in normal persons.
Folia phoniat. 29: 191-199 (1977).
Pedersen. M.F.: Electroglottography with synchronized stroboscopy in students of music. Study
Sounds 18: 423-434 (1978).
Kitzing. P.: Glottografisk frekvensindikering: thesis University of Lund. Malmö (1979).
Seidner. W.: Schutte. H.K.: Standardisierungsvorschlag Stimmfeldmessung/Phonetographie. Proc.
IXth Congr. Union Eur. Phoniat., Amsterdam 1981. pp. 88-94.
Bloothooft. G.: A computer controlled device for voice-profile registration. Proc. - IXth Congr.
Union Eur. Phoniat., Amsterdam 1981, pp. 8387.
Pedersen. M.F.; Munk, E.: Bennett. P.; Moller, S.: The change of voice during puberty in choir singers measured with phonetograms and compared to androgen status together with other
phenomena of puberty. Proc. Xth Int. Congr. Phonetics, Utrecht 1984. pp. 604-609.
Flatau. T.S.: Gutzmann, H.: Die Singstimme des Schulkindes. Arch. Lar. 20: 327-348 (1907).
Weiss. D.A.: The pubertal change of the human voice. Folia phoniat. 2: 126-169 (1950).
Naidr. J.: Zbohl. M.: Svevcik. K.: Die pubertalen Veränderungen der Stimme bei Jungen im
Verlauf von 5 Jahren. Folia phoniat. 17. 1-18 (1965).
Tosi. 0.; Postan, D., Bianculli, C.: Longitudinal study of children's voices at puberty. Proc. XVth
Int. Congr. Logopedics Phoniatrics, Interlaken 1974, pp. 486-490.
Bastian, H.J.; Unger, E.: Untersuchung des Zusammenhanges von Akzeleration, Mutation und
Dysphonie anhand von Längsschnittuntersuchungen. Arztl. Jugendkd. 71: 205-211 (1980).
Andersen. E.: Skeletal maturation of Danish school children in relation to height. sexual development and social conditions: thesis. Acta paediat. scand., suppl. 185, pp. 97-101 (1968).
Vuorenkoski, V.; Lenko, H.L.: Tjernlund, P., Vuorenkoski, L.; Perheentupa, J.: Fundamental
voice frequency during normal and abnormal growth, and after androgen treatment. Archs Dis.
Childh. 53: 201-209 (1978).
Hagg. U.: Taranger, J.: Maturation indicators and the pubertal growth spurt. Am. J. Orthod.
82:299309(1982).
Kahane, J.C.: Growth of the human prepubertal and pubertal larynx. J. Speech Hear. Res. 25: 4464550982).
Hirano, M.; Kurita. S.: Nakashima, T.: The structure of the vocal folds: in Stevens, Hirano, Vocal
fold physiology, pp. 33-41 (Tokyo 1981).
Komiyama. S.; Watanabe. H.; Ryu, S.: Phonographic relationship between pitch and intensity of
the human voice. Folia phoniat. 36: 1-7 (1984).
Tanner. J.M.: Growth of adolescence: 2nd ed. (Blackwell, Oxford 1962).
Sizonenko. P.C.; Paunier, L: Hormonal changes in puberty: correlation of plasma
dihydroepiandrosterone. testosterone, FSH and LH with stags of puberty and bone age in normal
boys and girls and in patients with Addison's disease or hypogonadism or with premature or later
adrenarche. J. clin. Endocr. Metab. 41: 894-904 (1975).
Tanner, J.M.: Endocrinology of puberty; in Brook. Clinical paediatric endocrinology, pp. 207-223
(Blackwell. Oxford 1981).
Cunningham. S.K.: Loughlin. T.; Culliton, M.: McKenna. T.J.: Plasma sex hormone-binding
globulin levels decrease during the second decade of life irrespective of pubertal status. J. clin.
Endocr. Metab. 58: 915-918 (1984).
Klingholz. F.: Martin. F.: Die quantitative Auswertunc der Stimmfeldmessung. Sprache Stimme
Gehor, ;- 106-1 10 (1983).
93
IV
278
Pedersen/Møller/Krabbe/Munk/Bennett
25 Gramming. P.; Pedersen, M.F., Kitzing. P.: Datoriserade fonetogram for objektivcring av
röstfunktionsstörningar. Läkarstamman Stockholm, Otolaryngol. Sect.
26 Pedersen, M.F.; Hansen, T.L.; Hansen, H.L.; Munk, E.: Phonetograph apparature for use in clinical
praxis. 22nd Scand. Congr. Otolaryngol.
27 Frank, F.; Donner, F.: Vom Knabenalt zum Startenor. Eine sonographische Analyse. Sprache
Stimme Gehor 5: 21-23 (1981).
28 Stürzebecher. E.: Wagner, H.: Becker, R.: Rauhut, A.: Seidner, W.: Einrichtung zur simultanen
Registrierung von Stimmfeld und hohem Sangerformant. HNO Praxis, Leipzig 7: 223-226 (1982).
M.F. Pedersen, MD,
Copenhagen Boys' Choir,
Sjaelør Boulevard 135,
DK-2500 Copenhagen (Valby) (Denmark)
94
V
Clin. Otolaryngol. 1993, 18, 488-491
A longitudinal pilot study on phonetograms/voice
profiles in pre-pubertal choir boys
M.F.PEDERSEN
The Voice Clinic, Medical Center, Østergade 18, DK-1100 Copenhagen, Denmark
Accepted for publication 30 April 1993
PEDERSEN M.F.
(1993) Clin. Otolaryngol. 18, 488-491
A longitudinal pilot study on phonetograms/voice profiles in pre-pubertal choir boys
The phonetogram or voice profile, measuring frequency combined with intensity of the
voice, has replaced simple measurement of tone ranges for analysis of professional boys
choirs. Knowledge of the relationship between pubertal, sex hormones, and phonetogram
development has been established in stratified studies. The connection between the singing
voice in puberty and other pubertal phenomena needs documentation in longitudinal studies
also.
Therefore, a pilot study was carried out, where three boys aged 13-15 years were analysed
at intervals of 2 months.
Measurement of phonetograms, free and total serum testosterone and sex hormone
binding globulin (SHBG) showed a significant relationship between SHBG and lowering of
the lowest tones in the phonetograms (P < 0.01) confirming earlier stratified studies.
Keywords
phonetograms/voice profiles puberty androgens
The voices of pubertal boys have been studied for centuries by music teachers and doctors.1-3 Since
measurement of tone ranges has been replaced by phonetography4-6 with a supplementary analysis of
intensity variation for each semitone in the tone range of a voice, new knowledge of the dramatic loss
of high singing notes in choir boys and its relation to pubertal development has been established in
stratified studies.7-10 Possibly the lowering of the boys voices can be predicted by the fall in the sex
hormone binding globulin (SHBG) between 13 and 15 years of age but only at a significance level of P
< 0.05.9
Longitudinal studies to confirm the relationship between the lowering of the voice and hormonal
changes, especially of SHBG. have not yet been carried out. Therefore, a pilot study seemed
appropriate to analyse the changes of shapes of the phonetograms with lowering register shifts.
Table 1. Mean values for each choir boy are presented for the singing and the pubertal period. A
statistically significant relationship between sex hormone binding globulin (SHBG) and the change of
the lowest tones in the pubertal group was found (R 0.68, P < 0.01)
95
Phonetograms/voice profiles in pre-pubertal choir boys
489
Material and methods
Three musically gifted and voice trained boys from a music
school in Copenhagen were chosen at random at the end of
the 7th grade, where the boys begin to leave the boys choir. A
consent from the parents was obtained. Indirect laryngoscopy
and stroboscopy showed normal vocal cords. Development
was normal, evaluated by a pediatrician by staging of height,
weight, pubic hair and orchidometry. 11.12
Measurement of phonetograms/voice profiles was carried
out at intervals of 2 months for I year in a voice laboratory.
At every clinic visit venous blood samples were taken for
analysis of total serum testosterone, free testosterone, and sex
hormone binding globulin (SHBG) and measured at the
hormone department at Statens Serum Institut, Copenhagen.
9.13-15
The analysis of phonetograms was based on the standards
set by the Union of European Phoniatricians and included
computed calculation of area in dB x semitones with
averaging of phonetograms in each boy and calculation of
95% intervals. 4-6 The phonetograms were analysed during the
period where the boys were singing in the choir and the
pubertal voice.
Register shifts by listeners test were compared with calculation of the middle of the most narrow part of the
phonetogram16 and evaluation of the stroboscopy results. 17
Tone ranges, and the lowest and highest semi-tones were also
determined.
Results
Figure 1. Six phonetograms with 2
months interval of one prepubertal boy
from age 13 7/12 to 14 6/12 years. The
third phonetogram (in December) was
best. In January the boy was excluded
from the choir (cl. 262 Hz).
Table 1 shows the results of measurements for each boy
between the ages of 13 and 15 years. Statistically a relationship was found between SHBG and the change of the lowest
tone (R2 0.68 P < 0.01).
Figure I shows six phonetograms of one of the boys with 2
months intervals for 1 year. The lowering of the lowest tone
and the falling of register levels was noted.
Figure 2 (boys I, II & III) presents the computed average
phonetograms and 95% confidence intervals for the three
boys in their last singing period compared with the period
after leaving the choir due to pubertal voice changes. In boy
1, only one phonetogram was recorded after exclusion, in boy
II only one was recorded before exclusion, but in boy III.
three phenotograms were obtained in the singing period and
three after the voice change.
A general finding at the time where the music teacher
stopped the boys singing in the choir was, apart from
lowering of the lowest tones also a reduction of the intensity
variation capability especially at the register shifts.
Discussion and conclusion
A longitudinal pubertal pilot study for 1 year in three boys at
the time when the boys have to stop their choir career,
96
490 M.F.Pedersen
V
Figure 2. Computed average phonetograms with 95% confidence intervals of the three choir boys (IIII) in their last singing period compared with the pubertal voice period. Boys I and 11 had one
phonetogram in the pubertal and singer period respectively. Boy III had three phonetograms in each
period. (cl. 262 Hz).
confirms an earlier stratified study of 8-19-year-old choir boys in the following way:
(I) The phonetogram changes shape when the boy stops singing in area, intensity range, register shifts
and lowest tones.
(2) The lowest tones in the phonetograms are significantly related to SHBG (P < 0.01) and seem to be
the most stable change at the end of a boy's singing career.
(3) The combined discription of phonetograms and androgens (testosterone and SHBG) serves as a
tool for further studies of voice regulation.18-20
References
1
2
WEISS D.A. (1950) The pubertal change of the human voice. Folia Phoniatr. 2, 126-159
KLINGHOLZ F., JOLK A. & MARTIN F. (1989) Stimmfeld Untersuchungen bei Knabenstimmen
(Tolzer Knabenchor. SpracheStimme Gehör 13, 107-111
3 KONZELMANN K., MOSER M. & KITTEL G. (1989) Stimmfeld Untersuchungen im Kindesund Pubertätsalter. HNO 14, 145-152
4 PEDERSEN M.F. & LINDSKOV HANSEN H. (1986) Computerized phonetograms for clinical
use. Folia Phoniatr. Proceedings XXth Congr. of Int. Assoc. Logoped. Phoniatr., 170-171
5 PEDERSEN M.F., LINDSKOV HANSEN T., LINDSKOV HANSEN H. & MUNK. E. (1984) A
phonetograph for use in clinical practice. Acta Otol. Suppl. 412, 138
6 PEDERSEN M.F. (1991) Pilotstudie der Stimmfunktion vor und nach Behandlung von
Hirngeschädigten. In Die Krankheit der Stimme. pp. 162-171. Ed. H. Gundermann Fischer. Verlag,
Stuttgart
7 PEDERSEN M.F., MØLLER S., KRABBE S., BENNETT P. & MUNK E.
(1986) Phonetograms in choir boys compared with voice categories, somatic puberty and androgen
development. J. Res. Sing. 9, 39-49
8 PEDERSEN M.F., MØLLER S., KRABBE S. & BENNETT P. (1986) Fundamental voice
frequency measured with electroglottography during continuous speech. A new extract secondary
sex characteristic in boys in puberty. Int- J. Paed. Otorhinolaryngol. 11,21-27
9 PEDERSEN MF., MOLLER, S., BENNETT P. & MUNK E. (1985) A multi-variate statistical
analysis of electroglottographically measured fundamental frequency in choir boys compared with
androgens and other phenomena of puberty. Folia Phoniatr. 37: 271-278
10 PEDERSEN M.F. (1991) Die biologische Entwicklung der Stimme in der Pubertat. In
Dokumentation pp. 28-37. Ed. Bundesverband Deutscher Gesangpadagogen, Detmold, Germany
11 TANNER J.M. (1962) Growth and adolescence. Blackwell & Mott Ltd, Oxford
97
V
Phonetograms/voice profiles in pre-pubertal choir boys
491
12 BROOK C.G.D (ED.) (1989) Clinical Paediatric Endocrinology. 2nd edn. Blackwell Scientific
Publications, Oxford
13 GUPTA D., ATTANASIO A. & RAAF S. (1975) Plasma oestrogen and androgen concentrations
in children during adolescence. J. Clin. Endocrinol. Metabol. 40, 636-643
14 PARKER L.N. (1989) Adrenal androgens in clinical medicine. Academic Press. San Diego,
California
15 YOUNG M.C., ROBINSON 1.A., READ G.F., RIAD-FAHMy D. & HUGHES I.A. (1988) 170 Hprogesterone rhythms in congenital adrenal hyperplasia. Arch. Dis. Child. 63, 617-623
16 PEDERSEN M.F. & MUNK E. (1983) Register examination during mutation in the Copenhagen
boys choir. Proceedings of the XIXth int. Congr. of Logoped. and Phoniatr. II 686-687
17 PEDERSEN M.F., HACKI T. & LOEBELL E. (1988) Videostroboscopy of register shifts in choir
boys (videofilm). Ed. Abt. Phoniatrie und Pedoaudiologie. Medizinische Hochschule Hannover
(for sale) presented at Voice Symposium New York
18 WENNINK J., WAAL H.A., VAN DE D., SHOEMAKER, R., SHOEMAKER H. &
SHOEMAKER J. (1989) Luteinizlng hormone and follicle stimulating hormone secretion patterns
in boys puberty measured with highly sensitive immuneradiometric assays. Clin. Endocrinol. 31,
551-564
19 BLAUSTEIN J.D. (1986) Steroid receptors and hormone action in the brain. Ann. New York Acad.
Sci. Repro. 474, 400-414 20 PEDERSEN M.F. & BEHREND W.
20 PEDERSEN M.F. & BEHREND W. (1989) A comparative pubertal study of thomaner choir
Leipzig and Copenhagen boys choir Copenhagen, presented at Voice Symposium, Philadelphia,
USA
98
99
ACTA UNIVERSITATIS OULUENSIS
SERIES D MEDICA 401
385. Perälä, Jukka (1996) Varicocele: diagnosis and post treatment follow-up color duplex US and
comparison of two embolization methods, 83 p.
386. Veijola, Riitta (1996) Genetic susceptibility and the natural history of insulindependent diabetes
mellitus, 97 + 37 p.
387. Kemppainen, Ritva (1996) Human lysyl oxidase and hereditary diseases of copper metabolism,
56 + 20 p.
388. Heikkinen, Jari (1996) Human lysyl hydroxylase: Structure of the gene, and characterization of
gene defects associated with the type VIA variant of the Ehlers-Danlos syndrome, 64 + 23 p.
389. Uitto, Lahja (1996)DNA polymerase s. Characterization of proofreading exonuclease activity,
cell proliferation dependent expression of its catalytic subunit, and dispensability in SV40 DNA
replication, 72 + 23 p.
390. Tomas, Eija (1996) The effects of antiestrogen treatmet on the uterus and insulin-like growth
factor-system in postmenopausal breast cancer patients, 82 +28 p
391. Qiao, Qing (1996) Early detection of non-insulin-dependent diabetes mellitus in adult Finns, 71
p. + appendix + 36 p.
392. Vuori, Juhani (1996) Development of immunoassays for carbonic anhydrase III and myoglobin
and clinical usefulness of the assays in evaluation of cardiac and skeletal muscle unjuries, 70 +
39 p.
393. Leskinen, Hanna (1996) Vasoactive factors as regulators of atrial natriuretic peptide release in
response to changes in volume load, 88 + 55 p.
394. Peng, Zhiqi (1996) The mechanical properties of rat bone in experimental osteoporosis. The
effect of runnung exercise on rat bone after ovariectomy, 73 + 41 p.
395. Hakkola, Jukka (1996) Expression of xenobiotic-metabolizing cytochrome P450 enzymes in the
human feto-placental unit, 73 + 38 p.
396. Pienimäki, Päivi (1996) Pharmacokinetics of carbamazepine and oxcarbazepine with special
reference to plancental transfer and metabolism, 78 + 43 p.
397. Lauri, Timo (1996) Cardiovascular responses of the dog to surface-induced hypothermia and
rewarming, 68 + 50 p.
398. Väisänen, Leena (1996) Family grief and recovery process when a baby dies. A qualitative study
of family grief and healing processes after fetal or baby loss, 150 p. + appendix.
399. Pokela, Risto (1996) Surgery for aneurysms and dissections of the thoracic and
thoracoabdominal aorta, 61 + 33 p.
400. Satta, Jari (1997) Expansion and rupture of abdominal aortic aneurysms, 49 + 31 p.
Book orders:
OULU UNIVERSITY PRESS
400, FIN-90571 Oulu, Finland
Distributed by
OULU UNIVERSITY LIBRARY P.O.B.
P.O.B. 450, FIN-90571 Oulu, Finland
100