Voice source changes of child and adolescent subjectsundergoing singing training—a preliminary studyC. A. Barlow and D. M. Howard

From the Department of Electronics, University of York, Heslington, York, UK YO10 5DD

Log Phon Vocol 2002; 27: 66–73

Many vocal practitioners have strong beliefs regarding the age at which singing training of a child should begin, and thedifferent ways in which male and female children should be treated. These beliefs are not substantiated by any scienti� cresearch, leading to considerable dispute between vocal coaches and choral directors. The singing voices of over 127 childsingers and non-singers aged 8–18 were analysed using electrolaryngographic measures. Analysis particularly concentratedon the laryngographically derived vocal fold closed quotient (CQ). Results indicated that the voice source characteristics ofsubjects could be divided into groups according to age, gender and the level of vocal training received. Female subjects inparticular exhibited a marked development of voice source production according to the length of training received, whilemale subjects exhibited patterning according to both age (and related pubertal development), and training received. It wasconcluded that the process of training a young voice has a quanti� able effect upon the singing voice production of the child,and in particular on the female voice, while pubertal development also creates measurable effects on the voice sourceproduction of the male child.

Key words: electrolaryngography, singing, closed quotient, adolescent, voice change.

C. A . Barlow, M usic Technology Research Group, Department of Electronics, University of Y ork , Heslington, Y ork , UKTel.:»44 (0)1904 432407. E -mail: cab124@york .ac.uk

ORIGINAL ARTICLE

INTRODUCTION

Over recent years a number of researchers have at-tempted to provide a quanti� able analysis of theeffect upon the adult singing voice of formal vocaltraining. The results of a number of these studiesappear to indicate that formal training in singing hasa measurable effect upon vocal production, particu-larly at the voice source (4, 7, 13, 14). It has beenfurther indicated that such analysis when used in abiofeedback device could provide a training tool,which can increase the rate at which singers improvetheir vocal technique (10, 19).

By far the greatest amount of analysis of singershas been performed upon adult, usually classicallytrained (in what is often known as the ‘bci canto’style) performers, with some work having been un-dertaken with ‘belt’ style singers. Remarkably littlework has studied the effect of formal training uponchildren and younger teenagers, whose voices are stillin the process of developing (2, 6, 17, 21, 22).

BACKGROUND

In the UK, there arc 40 Anglican (Episcopal) cathe-drals, each of which maintains the traditional pattern

of choral worship, in which a considerable part of theservice is set to music and sung by the CathedralChoir. The choir usually sings two to three serviceson a Sunday, as well as at least one service per dayevery weekday. In these choirs the Bass and Tenorlines are sung by adult, professional male singers, andthe alto line is in most cases also made up of adultmale counter-tenors. With little exception, the so-prano (or ‘treble’) line of the choir is sung bychildren.

This tradition has been maintained for hundreds ofyears, for example, the choir school at York Minstercan date its antecedents back to AD 627. For most ofthis period the treble line has been sung by boytrebles, but in recent years a number of cathedralshave admitted girl choirs as well as the traditionalboy choirs. These singers, despite their age, are ‘pro-fessional’ in every sense of the word. They sing 6 daysa week with a professional conductor, performing‘live’ on most of those days, with a rapidly changingrepertoire, in addition to which they perform onradio broadcasts, concert tours and recordings. Inreturn for this, the choristers usually receive scholar-ships towards the cost of fees at the choirschool, mostof which are specialist music schools, providing tu-

© 2002 Taylor & Francis. ISSN 1401-5439 L og Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

itat d

e G

iron

a on

12/

19/1

4Fo

r pe

rson

al u

se o

nly.

Voice source changes of child and adolescent subjects 67

Fig. 1. Laryngeal growthduring puberty (verticalcross section) (derived fromKahane (reference 18)).

ition in a number of musical instruments as well asvoice training.

In addition to cathedral choristers, a large numberof other children also train their voices with profes-sional teachers. While singing in state schools hasdrastically decreased in recent years some schools stillmaintain excellent choirs. Many private schools alsospecialise in music tuition and singing is still consid-ered to be an excellent way to develop musicianshipskills.

Despite this tradition of child singers, there aremany schools of thought as to how singing tuitionshould be approached for child singers. A child’svocal system undergoes rapid and dramatic changeduring adolescence, during which many singers aremaking demands on their voice as great as that ofany adult. During the pubertal growth spurt, bothmale and female larynxes change dimensions rapidly(F ig. 1), necessitating a constant, reassertion of themuscle control skills needed for speech and singing.There are as many con� icting ideas on how childsingers should be treated during this period as thereare vocal pedagogues who deal with child singers. Inspite of this, there has been very little research under-taken to attempt to determine the effect of regularsinging and vocal training on the voice production ofchildren and adolescents.

This paper explores a pilot study aimed at provid-ing some quanti� able data regarding the effect of

training and regular singing on the voice source ofyoung singers, aged 8–18.

M ethodology

Electrolaryngography uses the measurement of elec-trical impedance to give a direct, noninvasive mea-surement of laryngeal activity. It is widely used for F0

estimation in laboratories and clinics (5, 11).Human tissue is a moderately good conductor of

electricity, and the electrolaryngograph uses a con-stant voltage high frequency electric current passedbetween two electrodes placed externally either sideof the neck at the level of the larynx. The outputwaveform (Lx) (F ig. 2) represents the current � owing

Fig. 2. Idealised electrolaryngograph output wave-form showing closed and open phases (derived from(12: 206).

L og Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

itat d

e G

iron

a on

12/

19/1

4Fo

r pe

rson

al u

se o

nly.

C. A . Barlow and D. M . Howard68

between the electrodes. When the vocal folds areapart, the current must � ow along a longer pathbetween the electrodes, increasing impedance of thesignal. The current will thus be higher when the vocalfolds are in contact than when they are apart. Thecurrent variation can, therefore, be analysed in termsof changes in vocal fold contact area (1: 281).

The opening and closing of the folds can be dividedinto two phases; closed phase (CP) and open phase(OP) (12). In accordance with Howard et al. (12) andRossiter et al. (19), the start point of CP is derived asthe instant of maximum positive peak of differentialof Lx. The end point of CP in each cycle is derived bycalculating when the negative-going Lx waveformcrosses a � xed ratio (3: 7) of the current cycle’samplitude. The length of the cycle (waveform period)is known as Tx. The percentage of the cycle for whichthe vocal folds are in contact is known as the ‘larynxclosed quotient’ or CQ, which is found as:

CQ ¾ [CP:Tx]½100%

As the wave period (Tx) is calculated, the Lx wave-form can also be used to derive real time F0 estima-tion. As the waveform is relatively immune toexternal acoustic factors, it can provide a usefulalternative to acoustic measures of F0 such as peak-picking.

The electroglottograph (EGG) output waveform isessentially similar to the Lx output waveform, al-though it is usually plotted as the inverse. Compari-son of data from EGG and Lx analyses is thereforelegitimate, assuming that the polarity is taken intoaccount. Baken (3), Fourcin and Abberton (8) andHoward (15, 16) give detailed reviews of EGG andelectrolaryngograph operation.

Subjects and data analysis

Seventy-six female and 51 male young singers andnon-singers from 11 schools took part in the study.These included choristers from seven CathedralChoirschools around Britain, in addition to whichfurther subjects were taken from three schools spe-cialising in music, and one non-specialist school. Allexcept four subjects were White European, the excep-tions being three subjects of Asian extraction and onesubject of African –American extraction. Each subjectwas recorded by the researchers in the songschool ormusic department of their school. The youngest sub-jects recorded were at least 8 years old, and the eldestnot more than 18. The 11 schools were visited over aperiod of 1.5 years between February 2000 and July2001.

Subjects indicated whether or not they were a‘singer’ or a ‘non-singer’, and if a singer the length of

time they had been training their voice. For thepurposes of this study, ‘singers’ were considered to bethose who undertake regular individual formalsinging training with a professional coach:singingteacher, or who sang at least twice per week in agroup environment with a professional singing coachor choral director (this does not include schoolchoirs). This information was con� rmed by the choir-master or music teacher of the children. Subjects werealso asked their date of birth as a developmentalindicator.

The subjects voices were recorded (a) reading alouda passage of spoken text approximately 90 s in dura-tion and (b) singing a 2 octave scale to the vowel of:a: covering the pitch range of G major. The spokentext was a section of ‘Arthur the Rat’, which has beenused in a number of similar studies. In the case offemales or boy trebles the scale was sung from G3(196 Hz) to G5 (784 Hz). In the case of changed malevoices, the scale was sung from G2 (98 Hz) to G4(392 Hz), or as near to G4 as the subject could reach.Subjects were asked to project the sound at as consis-tent a level as possible, ‘as if they were singing asolo’. They were not required to sing the entire scalein one breath, but asked to breathe when necessary tomaintain as regular a volume as possible. In the eventof a subject running out of breath whilst singing, thesubject was asked to repeat the scale.

The output waveform from an electrolaryngograph(Lx) and the speech pressure waveform (Sp) from anAKG CK77 omnidirectional condensor microphonewere recorded onto the two channels of a FostexDA-5 DAT (Digital Audio Tape) recorder. The mi-crophone was powered by a Mackie 1402 VLZ-PROlow noise mixer. The Lx signal was viewed on anoscilloscope during the recording to maintain correctelectrode positioning.

Data analysis was undertaken using the SpeechStudio™ and Quantitative Analysis™ software run-ning on a PC with PIII 450 Hz processor � tted withthe Laryngograph™ (23) processor card. The record-ings were digitally transferred to the PC for storageand analysis.

The Speech Studio software was used to output atext � le of sample number, derived F0, and CQ foreach subject, which was then analysed in a spread-sheet. Quantitative Analysis™ was used to plot scat-tergrams of CQ:log F0 for visual analysis.

Results

For the purposes of this study, the spoken text wasused to return only the mean speaking pitch of eachsubject as a further developmental indicator in addi-tion to the recorded age.

L og Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

itat d

e G

iron

a on

12/

19/1

4Fo

r pe

rson

al u

se o

nly.

Voice source changes of child and adolescent subjects 69

Fig. 3. Examples of Qxplots. 18-year-old male(L) and 10-year-old fe-male (R).

The data for the sung scale was plotted for eachsubject as a CQ:log F0 (Qx) scattergram. This plotallows a visual analysis of the general trend of CQagainst changing pitch (Fig. 3).

CQ :log F0 trend analysis. Each of the Qx scatter-grams was analysed and the pattern of the slopenoted (positive, negative or neutral gradient, and thepoint of any gradient change). These results wererecorded in a database containing age, training andgender information. This was then analysed fortrends amongst particular demographic groups.

SUBJECTS 12 YEARS OF AGE AND UNDER

The development of the larynx in both male andfemale children occurs about 6 months after pubertalonset, about age 11.5 years in female children, and 12years in male children (20). The cartilage growth,which triggers the larynx development, is part of thegeneral ‘growth spurt’, which occurs in adolescentsduring puberty. Children aged 12 or under can, there-fore, normally be assumed to be vocally either pre-pubescent or in the early stages of adolescence (SexMaturity Rating 1–2), as indicated by Tanner (20).The voice classi� cation index developed by Gackle(9) indicates that female subjects in this age groupwill be in either Stage I (Prepubertal) or Stage IIA(Pre-Menarcheal). Larynx development will begin tooccur in some subjects, but will be in its early stages.This means that laryngeal dimensions will vary byonly a small amount for this age group. For thisreason, subjects aged 12 years and under wereanalysed as a single group. A further indicator thatthe male subjects of this age range are prepubescent isthat all subjects are singing in the ‘treble’ (soprano)register, rather than tenor or baritone registers. Ofthe children analysed, none of the male subjects aged12 or under had undergone voice change. Onanalysing the Qx plots, it was found that they fellinto � ve general categories according to the gradientof the best � t line.

The plots were grouped according to the variationin slope patterns. These divided into discrete groupsdepending on whether the gradient was positive ornegative, and how many changes in gradient occurred(Table 1). Two male subjects returned such poorquality signals from the electrolaryngograph thattheir data were not included in the analysis (noted inTable 9).

Non -singers

The results of this analysis demonstrated that bothmale and female non-singers trend towards the group1 pattern (constant negative gradient), with the fe-male sample group showing considerably less varia-tion than the male (Table 2). There is a near 73%instance of group 1 patterns for the female group and55% for the male, with the next highest instance ineither group being 18%.

Table 1. Vocal groupings according to slope of Qxplot

Classi� cation Pattern Idealised plot

Negative gradientGroup 1slope throughou tplot

Horizontal (zeroGroup 2gradient) plot

Group 3 Varying gradients(positive-negativepositive or negative-positive-negative)

Group 4 Initially negative thenswitching to positivegradient (‘V’ shape)Positive gradientGroup 5throughout plot

L og Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

itat d

e G

iron

a on

12/

19/1

4Fo

r pe

rson

al u

se o

nly.

C. A . Barlow and D. M . Howard70

Table 2. Distribution of Qx patterning for male andfemale non -singers age 12 and under

Number of subjectsPlot % of group

Male 12 and under —non-singersSample size 11

Group 1 6 54.50 0Group 22Group3 18.22 18.2Group 41 9.1Group 5

Female 12 and under —non-singersSample size 11

8 72.7Group 1Group 2 2 18.2Group 3 1 9.1

0Group 4 0Group 5 0 0

The female group in particular has a very strongcorrelation between the mean length of training andthe particular Qx pattern being exhibited by a groupof singers (Table 3). The male group, while thecorrelation is not as strong, still indicates some signthat increased training changes vocal production ac-cording to a reasonably regular pattern. Groups 2–4in the male group demonstrate a change in patternwith increased training that corresponds closely tothe female singers. Only one subject demonstrated thegroup 5 plot, so could possibly be anomalistic, as hehad only been training for 1.5 years, considerably lessthan would otherwise be expected for this plot type(4, 15).

SUBJECTS 13 » YEARS OF AGE

Subjects of 13 or more years of age will vary inpubertal development according to both age andgenetic traits. While the pubertal onset is usuallyabout 11.5 years of age, it can occur earlier or laterthan that. This being said, most children will havereached SMR stage 3 (mid-puberty) by the age of 13,and SMR 4 (late puberty) by age 15. By the age of18, most subjects will be at SMR 5, having full adultgrowth and characteristics.

Over this period from 13 to 18 years of age, bothmale and female children undergo dramatic changesin their vocal tract, which could have implications forthe production of the voice. The Qx plots of subjectswere analysed in terms of age and mean speaking F0

for indicators of pubertal development, as well asduration of voice training.

Female singers

The female singers in the sample demonstrated verysimilar patterns to the younger group of femalesingers (Table 4). While the subjects exhibiting group1 patterns has a mean duration of training slightlyhigher than that of group 2, groups 2–5 show aprogression through the plots that is again a closecorrelation between pattern and mean duration oftraining. Furthermore, the training levels of the

Table 3. Distribution of Qx patterning for male andfemale singers 12 and under

Mean length ofNumber of subjectsPlot training (years)

Male 12 and under —singersSample size 20

Group 1 2 2.8Group 2 6 1.8

2.986Group 3Group 4 2 3.2Group 5 1 (1.5)

Female 12 and under —singersSample size 21

5Group 1 2.7Group 2 5 3.1Group 3 5 3.3Group 4 5 4.3Group 5 1 (4.8)

Two subjects no result.

Singers

Analysis of the results of the trained singers in the 12and under age group demonstrates that some pattern-ing occurs with increased levels of singing training.

Table 4. Distribution of Qx plots for female singers age 13–18

Mean length of training (years)Plot Number of subjects Mean age (years) Mean speaking F0 (Hz)

Female 13 and over —singersSample size 28Group 1 3 13.6 242 3.2Group 2 3 15 240 2.8

16 3.9227Group 3 14.68Group 4 14 240 4.4

Group 5 3 16 239 7.5

L og Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

itat d

e G

iron

a on

12/

19/1

4Fo

r pe

rson

al u

se o

nly.

Voice source changes of child and adolescent subjects 71

Table 5. Qx patterning of female non -singers aged 13–18

Number of subjectsPlot Mean age (years) Mean speaking F0 (Hz) % of group

Female 13 and over —non-singersSample size 16

15 244Group 1 56.25914 2355 31.25Group 2n:a n:a 0Group 3 016 1901 6.25Group 414 220Group 5 6.251

groups are remarkably close to those of the samegroups in the younger age group.

There is no indication that age or mean speakingfrequency are signi� cant in this patterning, as there isno clear trend shown by either of these measures.

Female non -singers

This group again demonstrates a strong inclinationtowards the group 1 pattern (55%), with group 2 at31% and the others insigni� cant (Table 5). The indi-cation from this would appear to be that untrainedvoices generally exhibit a decreasing CQ with in-creased F0.

M ale subjects

The male subjects in the over 13 age group demon-strated a very different patterning from the othergroups. Whereas the other groups appear to show theshape of the Qx plot, to be primarily a function ofthe level of training received by the subject, theindications for male adolescents appear to demon-strate a function of age (and, therefore, presumablyof physical development).

While there were three apparent patterns of Qx inthe male adolescent category, the small sample num-ber means that these patterns are not conclusive.However, from the available data, patterns for ado-lescent males are suggested as follows (Table 6).

Pubertal and post-pubertal (changed voice) malesubjects did not demonstrate any correlation of thepattern of their Qx plot and the level of singingtraining that they had received (Table 7). There issome indication, however, that the Qx plot graduallyceases to vary with changed frequency, but � attensout so that there is eventually little or no change inCQ with changed F0 as boys complete puberty andreach adult maturity.

The subjects analysed as group 1 exhibited a down-ward sloping plot. The mean speaking frequency ofthese subjects is 171 Hz, and therefore still consider-ably higher than mean adult speaking F0. The pitchrange of the sung scale of these subjects was found tovary between D3 and E5. None of the subjects could

attain the highest notes of the treble scale (F5, G5),or the lower notes of the bass:tenor scale (G2–C3).This suggests that while vocal change has started, thesubjects are still vocally only in early pubescence, andtherefore are likely to still follow the patterns ofprepubescent subjects. As these subjects have notreached adult speaking pitch, and exhibit a plottypical of untrained prepubescent subjects, it is sug-gested that if a larger sample of this age group wereobtained, the Qx plots of subjects in will vary accord-ing to training levels according to Table 4 above.These suggestions will, however, need extensive test-ing before they can be presented as conclusive.

The other three groups all exhibit values close toadult speaking pitch. G roup 4 contains by far thehighest proportion of the subjects in the samplegroup, and further analysis of the subjects indicatesthat this pattern is evenly divided between the trainedsingers—the most experienced of whom had beentraining for nearly 8 years—and non-singers. Giventhat the mean age of the subjects is the highest of allthe group, this suggests that the ‘horizontal’ or ‘� at’Qx plot in which CQ remains constant with changingF0 is achieved as male subjects reach adult maturity.This correlates with results demonstrated in a previ-ous study (12), in which adult male singers all exhib-ited Qx plots with little variation in CQ with varyingF0, although mean CQ varied with increased training.

Table 6. Vocal groupings of adolescent males accord -ing to slope of Qx plot

Idealised plotClassi� cation Pattern

Group 1 See Table 1Subjects can exhibitany of the plots I–Vshown in Table 1above, depending ontrainingPositive:negativeGroup 2(inverted ‘V’)

Group 3 Negative:horizontalslopeHorizontal patternGroup 4

L og Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

itat d

e G

iron

a on

12/

19/1

4Fo

r pe

rson

al u

se o

nly.

C. A . Barlow and D. M . Howard72

Table 7. Qx patterning of male subjects aged 13»

Number of subjectsPlot Mean age (years) Mean speaking F0 (Hz) % of group

Male subjects 13 and overSample size 20

3 13.6 171Group 1 2 154 15.5 140.5Group 2 2 202 15 122 5.5Group 3 10

11 15.8 138Group 4 3.8 55

Indications from groups 2 and 3 are less evident.While subjects in group 2 have a higher mean agethan those in group 3, subjects in group 3 have alower speaking pitch, indicating more advanced pu-bertal development. Their numerical ordering is,therefore, arbitrary, and should not suggest that thesepatterns necessarily occur in that order. Again, due tothe sample size, further research needs to be con-ducted before this grouping can be presented asconclusive.

CONCLUSION

Results from this study indicate that voice sourceproduction of children changes in a predictable andmeasurable pattern with increased vocal training forprepubescent male and female children, and also foradolescent females. The vocal production of adoles-cent males appears to change in a regular pattern asa function of pubertal development. The rapid natureof this change means that it is dif� cult to tell theeffect, if any, that training levels have upon thisprocess.

On comparison to the Qx plots obtained fromadult female singers (15) (F ig. 4), it can be suggestedthat there are similarities between the plots forgroups 3–5 for female:prepubertal subjects and theplots obtained from adult singers. In each case thereis a slope change at about the register break G4, andalso the gradual assumption of a constant upwardslope with increased training. It can be suggestedfrom this that training a child’s voice is likely toresult in improved technique of at least voice produc-tion at source. There was no indication in the studythat the subjects who trained their voices profession-ally from an early age had suffered as a result. Thesmall change in the female larynx during the pubertalgrowth period was insuf� cient to cause any dramaticchange in vocal production, and so there appears tobe no reason why female singers should not continuetraining from childhood to adulthood.

The results of this study cannot infer whether ornot training a male voice through adolescence isbene� cial, as the data is inconclusive.

F igures 5 and 6 show idealised Qx plots for maleand female child and adolescent singers, which havebeen derived from this study. It is important to notethat, while no instances have occurred in this study,the male group 1 could theoretically contain a zerogradient Qx plot. This could result in its confusionwith the plot analysed as group 3 (Fig. 6), which isalso of zero gradient. Classi� cation of individualswould, therefore, have to consider factors such asvocal range, mean spoken F0 and physical develop-ment to assist in determining group.

Fig. 4. Ox plot patterning for adult female singers(16: 171).

Fig. 5. Idealised Qx plots for prepubertal male and allchild:adolescent female singers.

Fig. 6. Idealised Qx plots for male singers frompre-pubescent to post-pubescent.

L og Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

itat d

e G

iron

a on

12/

19/1

4Fo

r pe

rson

al u

se o

nly.

Voice source changes of child and adolescent subjects 73

FUTURE WORK

While this study has demonstrated that there is ameasurable change in voice source production as afactor of training and pubertal development, in orderfor this data to be useful it will be necessary toundertake further study.

Areas identi� ed for further study are as follows:

(i) Longitudinal study of a number of subjectsover a period of time in order to characterisethe effect of training on a single voice overtime, and with relation to speci� c developmen-tal characteristics.

(ii) Investigate the effect of training on adolescentmales using a greater sample number and moredevelopmental indicators (height, weight, shoul-der width) to determine whether or not train-ing is effective on the male changing voice.

(iii) Analyse numerical data for CQ values againstfrequency for more accurate quantisation ofvoice production characteristics.

ACKNOWLEDGEMENTS

This study would have not been possible withoutthe cooperation and assistance of many people.Particular thanks go to: Neil Taylor, Master of theMusic, Shef� eld Cathedral; Tony Jones, Head ofMusic, High Storrs School, Shef� eld; Richard Shep-herd, Headmaster, York Minster School; PhilipMoore, Master of the Music, York Minster; SimonLole, Master of the Music, Salisbury Cathedral;Richard Moore, Headmaster, Ripon CathedralChoirschool; Stewart Smith, Head of Music, AbbeyGate College, Chester; Phyl Barlow, Abbey GateSchool, Chester; Matthew Owens, Master of theMusic, Edinburgh Episcopal Cathedral; MalcomArcher, Muster of the Music, Wells Cathedral;Richard Tanner, Master of the Music, BlackburnCathedral; Alasdair Jamieson, Head of Music,Bootham School, York.

REFERENCES

1. Abberton ER, Howard DM, Fourcin AJ. Laryngo-graphic assessment of normal voice: a tutorial. ClinLinguistics Phon 1989; 3: 281–96.

2. Andrews ML. Intervention with young voice users: aclinical perspective. J Voice 1993; 7: 160–4.

3. Baken RJ. Clinical measurement of speech and voice.Boston: College Hill, 1987.

4. Barlow CA. The developing voice from child to adult —a laryngographic analysis. Unpublished M.Sc. disserta-tion, University of York, UK, 1999.

5. Childers DG, Hicks DM, Moore GP, Alsaka YA. Amodel for vocal fold vibratory motion, contact area, andthe electrolaryngogram. J Acoust Soc Am 1986; 80:1309–20.

6. Cooksey JM. Do adolescent voices ‘break’ or do they‘transform’? Voice 1993; 2: 15–39.

7. Evans M, Howard DM. Larynx closed quotient infemale belt and opera qualities: a case study. Voice 1993;2: 7–14.

8. Fourcin AJ, Abberton ERM. First applications of a newlaryngograph. Med Biol Rev 1971; 21: 172–82.

9. Gackle ML. The adolescent female voice: characteristicsof change and stages of development. Choral J 1991; 31(8): 17–25.

10. Garner PE, Howard DM. Real-time display of voicesource characteristics. Log Phon Vocol 1999; 24: 19–25.

11. Gilbert HR, Potter CR, Hoodin R. The laryngograph asa measure of vocal fold contact area. J Speech HearingRes 1984; 27: 178–82.

12. Howard DM, Lindsey GA, Allen B. Towards thequanti� cation of vocal ef� ciency. J Voice 1990; 4:205 –12.

13. Howard DM. Quanti� able aspects of different singingstyles. Voice 1992; 1: 47–62.

14. Howard DM, Barlow CA, Welch GF. Vocal productionand listener perception of trained girls and boys in theEnglish Cathedral Choir. Proceedings of the 18th Inter-national Seminar of the International Society for MusicEducation, 2000.

15. Howard DM. Variation of electrolaryngographicallyderived closed quotient for trained and untrained adultfemale singers. J Voice 1995; 9: 163–72.

16. Howard DM. Instrumental voice measurement: usesand limitations. In: The voice clinic handbook . London:Whurr Publishers, 1998. pp. 323–79.

17. Huff-Gackle L. The young adolescent female voice (ages11–15): classi� cation, placement and development oftone. Choral J 1985; 25 (8): 15–8.

18. Kahane JC. A morphological study of the humanprepubertal and pubertal larynx. Am J Anat 1978; 151:11–20.

19. Rossiter DP, Howard DM, DeCosta M. Voice develop-ment under training with and without the in� uence ofreal time visually presented biofeedback. J Acoust SocAm 1996; 99: 3253–6.

20. Tanner JM. Foetus into man-physical growth fromconception to maturity. London: Open Books, 1978.

21. Vaughan VC, Litt IF (Eds.) Child and Adolescentdevelopment. Philadelphia: Saunders, 1990.

22. Weiss DA. The pubertal change of the human voice.Folia Phoniatrica 1950; 2: 125–59.

23. Internet www page at URL \ http:::www.laryngograph.com.

L og Phon Vocol 27

Log

oped

Pho

niat

r V

ocol

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

itat d

e G

iron

a on

12/

19/1

4Fo

r pe

rson

al u

se o

nly.