Dept. for Speech, Music and Hearing

Quarterly Progress andStatus Report

Auditory perception of voicequalities and speaking

fundamental frequency inMulibrey-nanism and in some

other children with growthfailure

Vuorenkoski, V. and Tjernlund, P. andPerheentupa, J.

journal: STL-QPSRvolume: 13number: 2-3year: 1972pages: 064-074

http://www.speech.kth.se/qpsr

STL-QPSR

IV. MEDICAL APPLICATIONS

A. AUDITORY PERCEPTION OF VOICE QUALITIES AND SPEAKING FUNDAMENTAI, FREQUENCY IN MULIBREY -NANISM AND IN SOMZ OTHER CHILDREN WITH GROWTH FAILURE

V. ~uorenkosk i " , P. ~jernlund"", and J. ~ e r h e e n t u ~ a * ' ~

Summa rv

Auditory and instrumental analysis of the voice of 20 children with different types of growth failure revealed various kinds of abnormalities. The fundamental frequency was found to be higher than normal , with no correlation to age. In the auditory judging of voice qualities, severa l voices were regarded abnormal, the sex of the subjects was not co r rec t - ly identified, and the age was estimated younger than the chronological age. Various articulatory and phonatory d i so rde r s were observed in the mater ial . The clinical significance of voice analysis in children with growth failure i s discussed.

There i s a multiplicity of causes of growth retardation in children.

Some of them affect postnatal growth only. Of these, the normal variants

of heredi tary small stature and heredi tary slow growth and maturation

a r e the most common; especially the coincidence of these two features

appears a s marked stunting of the child. The well known deficiency

s tates of thyroid and growth hormone must be recognized and effectively

t r e a t e d a s ear ly a s possible. Children with growth failure which i s a l -

ready apparent a t birth a r e a s common but l e s s well classified. Various

aberrat ions of genes or chromosomes cause such prenatal dwarfism. In

other instances a known o r unknown external harmful influence of a tem-

porary nature gives r i s e to permanent damage to the growth of t i ssues in

utero. New methods a r e needed to character ize the various aspec ts of

growth failure for classification purposes.

Mulibrey-nanism i s a recently described prenatal dwarfism(4) which

i s inherited by an autosomal recessive gene. One of the character is t ics

of this condition was reported to be a pcculiar type of high-pitched voice;

this observation initiated a systematic study of a possible correlation

" Dept. of Speech Communication, KTH and Dept. of Pediatr ics , Ke skipohjanmaa Central Ho spital, Kokkola, Finland.

;Y 46 Employed a t the Dept. of Speech Communication, KTH, 1964- 1970. ;'%(4b Children' s Hospital, University of Helsinki, Finland.

STL-QPSR 2 -3/1972 65.

between ea r ly voice deviations and growth retardation. A collection of

tape recordings of different types of nanism was s tar ted in 197 0 a t the

Children' s Hospital, University of Helsinki.

P r i m a r y observations had suggcstcd the occurrence of different types

of abnormalit ies and pecularit ies in the voices of these patients:

(1) Many patients with growth failure seem to have some kind of voice deviation already a t an ear ly age;

(2) A relatively high fundamental frequency (F0) i s a general feature;

(3) Some specific voice qualities may make it possible to differentiate certain types of nanism (e. g. hfulibrey) f rom others;

(4) Some patients may have individual voice qualities on the basis of which we can make a personal identification during infancy.

Method and mater ia l

Voice recordings from 20 patients representing different types of 1

I

growth failure were included in this study. Fifteen had prenatal nanism I I I I

(cases 1-15); five had postnatal nanism (cases 16-20), two of these fam- 1 ilia1 short stature with delayed growth and maturation ( cases 19 -20). I

i I

Their chronological age varied between 3.9 and 13.2 yea r s a t the t ime 1 the voice recording was made. There were 1 I boys and 9 gir ls . F o r

the clinical data, see Table IV-A-I.

The voice sample used was the same for each subject, a short sentence

of command in Finnish "Tulkaa hiihtamaan ! ' ' [ tulkn :hi:ht m r :nl (in

English: "Come to ski! ' I ) . Figs. IV-A- 1 and IV-A-2 show sound spectro-

g rams of this sentence, recorded f rom cases 18, 19, I , 10, 8 , and 6.

The different vowels and consonants a r e roughly segmented but these

spectrograms a r e shown only a s a guicle for prel iminar orientation. A

detailed study of the physical analysis of the articulatory features will be

presented in a forthcoming study. The children were asked to say the

sentence in a loud voice to imagined friends in the room close by.

The recorded voice samples werc studied both by instrumental and by

auditory analysis. The resu l t s of the Fo analysis a r e presented in this

paper f rom the i n s t r ~ r n e n t ~ ~ l analysis. The Fo and the subsequent calcula- i I

t ions were analyzed f rom each samplc using an automatic computerized 1 method. This method has been described in detail e l s e ~ h e r e ' ~ ) . In short ,

the fundamental i s isolated using a lo\-/ pass fi l ter. The zerocrossings

a r e detected and convcrted to a pulse train. These pulses a r e the input

F i g . I V - A - l a - c. "tulkaa h i ih t amaan" r e c o r d e d f r o m cases 18, 19, a n d 1.

S e g m e n t s : 1 = t 4 = k 7 = i: 10 = 13 = n 2 = u 5 = a: 8 = h 11 = m 3 = 1 6 = h 9 = t 12 = z :

Fig . IV-A- l a .

Fig. IV-A- 1 b.

9'1 S'L 'I 'L E'L Z'L CL OL 6 ' 8' &' 9' S * 2' E' 2 ' 1' 0

Fig. IV-A-2b.

Fig. IV-A-2c.

As for the auditory judging of voice i tems a s growth retarded, the

total mean frequency was . 52. The voices of7 boys and one gir l ( s ee

Table IV-A-I) were identified a t a significant level a s abnormal while

the voices of 3 boys and 5 g i r l s were regarded a s normal , the judging

being contradictory in the case of one boy and 3 giris. We can examine

the influence of the age of thc child on the e stirnation level in Fig.

IV-A-3, which shows a correlation scat ter diagram of the mean es t ima-

tion a s growth retarded by 56 judges and thc: chronological age of the

20 chi ldren; no correlation can be traced.

The judges estimated f rom the different voices .58 for boys and

.42 for g i r l s , which corresponds to thc actual sex rat io (1 1 boys, 9

girls). The total mean for cor rec t identifications was .68 . Among the

different voice i tems, the sex of 9 boys and 5 g i r l s were well identified.

One boy (case 8) was identified a s a gir l , and two g i r l s (cases 10 and

14) were identified a s boys. Judging of the sex was contradictory for

one boy (case 3) and for two g i r l s (cascs 7 and 13). The sex ideni ifica-

tion frequency varied between . 14 (case 10) and 1.00 ( cases 18 and 20).

The total mean of a l l age e stimations was 1.9 yea r s le s s than the

mean chronological age; this difference i s highly significant. Of the

different voice i tems, 5 boys and 4 g i r l s were estimated to be younger

than their chronological age a t a significant level and no one was signi-

ficantly e stimated to be older than the chronological age. The deviation

in age estimation varied between -8.0 y e a r s (case 6 ) and t i . 3 y e a r s

(case 2).

Taking into consideration a l l the three judgings together (diagnosis,

sex, and age) , cases 11, 15, 18, and 20 can be regarded a s most normal.

The most abnormal oncs wcre cases 3, 8 , 10, and 17.

If we examinc the 6 Mulibrey cases separately we can see that only

two of them, cases 4 and 6, have been judged to be normal. The sex of

one Mulibrey patient (case 3) was not well identified. The age deviation

i s significant in only one of the cases (case 6) but i s a l so the most r e - markable one: 8 years .

It was of intercst to compare the F and the age. Fig. IV-A-4 shows 0 a correlation scat ter diagram of chronological age and F - no co r re l a - 0 tion can be traced. Similar scatter diagram of bone age and F did not

0

Fig . IV-A-3. Cor re la t ion s c a t t e r d i a g r a m of chronologica l age and m e a n es t ima t ions a s growth r e t a r d e d by 56 judges. p = > 0 . 0 5 .

Fig. IV-A-4. Correlation scat ter diagram of chronological age and Fo in 20 children. p = > 0.05.

-50.0 < = d -uaxpl!q= oz zoj O J pue (saspnr 95) a8e paleurgsa ueaur 30 luexOerp xaJle=s uogeTaxzo3 .S-V-AI - 8 : ~

MEAN ESTIMATED AGE,YEARS

TOTAL 293 356 400 .52 NS .68 x -1.9 NS 4C prenatal nanism with dysmorphic features (minor Significance levels for co r rec t and incorrect

anomalies) i s h e r e called pr imordial diagnosis, sex, and for age deviation ** famil ia l shor t s ta turetdelayed growth and maturation NS p = > . 05 , x p = '( .05 , xx p = < .O1

'

*** brain anomaly -- Table IV-A-I. Diagnosis, sex, age, bone age, deviation of height f rom the population mean for the bone

age, relative weight, fundamental frequency (FO), analysis and voice identification of CF

different voice i tems (the 56 judges condensed in one group). 00 .

Sex

boy

boy

boy

g i r l

boy g i r l

g i r l

boy

boy g i r l

g i r l

g i r l

g i r l

g i r l

boy g i r l

boy

boy

boy

boy

f

Voice Type of no. nanism

1. mulibrey

2 , mulibrey

3. mulibrey

4. mulibrey

5. mulibrey

6. mulibrey '

7. primordial*

8 , primordial

9. primordial

10. primordial

1 I , prenataltfamilial

12. prenataltfamilial

13. prenataltfamilial

14. prenatali-familial

15. prenatal+familial

16. hyposomatotropic

17. hyposomatotropic 9 1 ~ ~

18, hyposomatotropic

19. famil ia l

- 20. familial*"

Age years

3.9

4.9

6.7

6.9

8 .4

13.8

6.5

8.8

9.0

12.3

5.0

5 .1

5 .4

9.4

10.0

10.0

12.1

13.2

7.5

9.2

Bone Age years

3.7

5.0

6.7

4.8

8 .5

13.0

4.0

7.5

6 .5

15.5

4.0

2.5

3.3

7 .5

8 .4

6 .8

5.2

8 .5

3.0

5.8

Height devia- tion S.D.units

-3.6

-2.0

-4.5

-2.9

-4.3

-5.6

. f - 0 -1.5

-2.7

-8.4

-4.3

-1.5

-0 .5

-0.9

-1.2

-1.6

-0.7

-0.5

t 3 . 5

2 0

Weight dev/ height dev

0.97

1.20

0.90

0.98

0.9:

1.06

1.05

1.35

0.94

0.25

0.54

0.71

0.68

1.23

1.21

0.78

0.56

0.70

0.56

0.82

F O , Hz P25 MD P 7 5

277 317 361

397 431 504

359 418 442

296 318 384

303 322 347

322 387 423

296 389 444

315 380 408

291 358 396

241 300 337

246 372 429

322 383 447

216 314 352

275 380 403

274 320 339

340 418 444

275 340 389

282 342 374

264 301 420

273 329 357

F r e - quency for sex

.80 xx

. 7 5 x x

. 4 1 N S

, 6 8 x

. 8 2 x x

.96 xx

-37 NS

.16 xx

.75 xx

. 14 xx

.93 xx

.77 xx

.55 NS

.20 xx

. 9 3 xx

. 9 i x x

.75 xx

1.00 xx

.79 xx

1 . 0 0 ~ ~

Identi- fication diag- nosis

.95 x x

. 7 8 x x

. 7 3 x x

. 3 7 x x

. 7 5 x x

.29 xx

.45 NS

.60 NS

. 61 NS

.36 x

.36 x

.29 xx

.55 NS

.26 xx

.20 xx

. 7 5 x x

.68 x

.09 xx

1.00 xx

. 3 2 x

Age estimate deviation

years -

+O.8 NS

t 1 . 3 N S

-1.ONS

-1.0 NS

-1 .1NS

-8.0 xx

-1.1 NS

-2.3 x

-3.7 xx

-3.5 x

-0.6 PTS

+0.3 NS

-0.8 NS

-2.7 x

-2.2 x

- 4 . i x

-4.4 x

-3.9 x

-0 .4 NS

t 0 . 6 N S

STL-QPSR 2 -3/1972 69.

show a correlation either. On the other hand, the correlation between

the estimated age and Fo approaches but does not reach a significant

level, see Fig. IV-A-5. Examining the different cases , i t i s noticeable

that the four most significant age deviations ( cases 6, 9, 16, 17) c o r -

respond to very high Fo values (387, 357, 418, 340 HZ), despite the

ages of these children (13.8, 9 .0 , 10.0, 12. 1 years ) .

Table IV-A-I1 shows the resu l t s f rom the voice identification tes t in

the different sub-groups. Comparing the total means , group I (pediatri-

cians and nur ses experienced in the care of some of the patients studied)

had better values in a l l the three estimations than group I1 (pediatricians

and nurses not familiar with the patients, and group I11 (acousticians not

familiar with the patients). Group I was significantly better than groups

I1 and 111 in estimating the voices a s growth retarded but the difference

between groups I1 and I11 was not significant. A s for the different voice

i tems, both groups I and 11 estimated six voices a s growth retarded but

group I considered only two to be normal, while group 11 estimated seven

voices a s normal. Group I11 was cautious in this estimation: only two

voices were regarded a s growth retarded and six a s normal.

In sex estimation al l three groups mere grosso mod0 equal. Group I:

ten i tems right and two wrong a t a significant level; group 11: 12 i tems

right and two wrong, and group 111: nine i tems right and three wrong.

In age e stirnation group I was significantly better than groups I1 and 111,

which had an equal mean deviation f rom the calendar age, 2.4 years . If

we look a t the different i tems, group I estimated nine cases younger and

four case s older than the chronological age; group II 16 case s younger

and one case older, and group I11 14 causes younger and one case older

than the chronologicnl age.

A s for the identification of Mulibrey, group I recognized cases 1, 2,

and 3 a s Mulibrey but not a t a significant level due to a relatively large

number of incorrect identifications.

Discussion

Auditory and instrumental analysis made f rom a short voice sample

recorded f r o ~ n 20 children with different types of growth retardation r e -

vealed some significant deviations from normality.

-- --- __l_r__--- -. ,

ZDENTIFICA TION FREQUENCY for

SEX

I: 11 I11

AGE ESTIMATE DEVIATION

I I1 I11

TOTAL . 6 2 . 5 2 . 4 3 . 7 2 . 7 1 . 64 -0.9 - 2 . 4 - 2 . 4

Significance levels f o r c o r r e c t and incor rec t diagnosis, sex, and for age deviation: NS p = > .05, x p = ( .05 , xx p = < 0.01

Table IV-A-11. Voice identification: frequency of c o r r e c t diagnosis and sex, and deviation of es t imated age f rom chronological age in the different voice samples by the t h r e e groups of judges: I =

4 pediatricians and n u r s e s with some knowledge of these groups of patients, N = 16, 11 = o pediatricians and n u r s e s with no knowledge of these groups of patients, N = 20, I11 = acousticians with no knowledge of patients, N = 20 .

STL-QPSR 2 -3/1972

cases regarded a s abnormal. A s to the subject ive observation of the

phonatory qualities, about half of the children in this s e r i e s seems to

have some kind of general phonatory subnormality, mainly variable

degrees of hoarseness o r nasality. The nature and significance of these

subnormalities will be discussed in a la te r paper.

The sex of the 20 cases was relatively well identified; in only three

cases , one boy and two g i r l s , the identification was significantly wrong.

On the average, boys were incorrectly estimated a s g i r l s more often

than g i r l s a s boys. One of the g i r l s identified a s a boy had the lowest

Fo value in the whole mater ia l (300 112). She was the only patient in our

se r i e s who had been t reated with anabolic s teroids for he r growth r e -

tardation. It i s most probable that the androgenic effect of that medica-

tion was responsible for the lowering of he r voice, a s well a s for the

undue advancement of he r bone age (Table IV-A -I).

A s to the auditory judging of age, there was a general tendency to

estimate the subjects to be younger than their chronological age. In

spite of an obvious trend, the correlation between the F and the e s t i - 0 mated age was not significant. This i s certainly explained in p a r t by the

masking effects of phonatory dysfunctions.

The p r e sent ser ie s of patients with growth failure was clinically

heterogenous. The la rges t single subgroup consisted of six children

with Mulibrey-nanism. The four children with what we have called

pr imordial nanism (prenatal growth failure with dysmorphic features) ,

and the five other children with a prenatal growth failure do not r ep re -

sent homogeneous groups comparable to the Mulibrey ser ies . Three

children with hyposomatotropic nanislx and two children diagnosed a s

representing the normal variant of delayed growth and maturation were

a l so included. They may not be fully representative of these etiological

groups. Our se r i e s , a s a whole, i s large enough to provide a general

view on the voice abnorrrlalitie s associated with growth fai lure , though

defining the possible specific voice character is t ics of the different sub-

groups awaits further work.

The one fcature common to the whole se r i e s i r respect ive of etiolog-

ical and clinical differences was an abnormally high fundamental f r e -

quency. Whether this i s a distinctive feature of an abnormally slow

STL-QPSR 2 -3/1972 7 3 .

growth in contrast to the normal slow variant, remains unclear. P r e -

sumably, the ra te of maturation of voice conforms to the general ra te

of maturation. The normal slow grolT/ers a r e expected to end with a

normal adult voicc, whereas the patl~ological case s may remain different.

Effective t reatment i s available to relatively few children with growth

failure. The h'$rposomatotropic patients will grow a t a higher than normal

rate and catch up their agemates when substitution therapy with human

growth hormone i s instituted. Anabolic s te ro ids have been used in many

different types of growth failure. They do accelerate the growth but may 1

advance the bone maturation even more a d m a y thus even decrease the ;

final height. It will be interesting to follow the effect of these t reatments ,

on the voice qualities. Especially, the anabolic s teroids , being in fact

derivatives of male hormone, may lower the fundamental frequency a s a I

I

specific "side-effect", a s had presumably happened in one of our patients.

Following of F may provide a sensitive mean for detecting an undesirably 0 strong androgenic effect before this i s evident by other means of observa-

tion. This may help to adjust the dose of the drug to the level appropriate

to the individual patient. On the basis of these observations we have added

the use of a simple Fo indicator (3)s in the follow-up routine of our patients

receiving medication for the advancement of growth.

Acknowledgment

The authors express their gratitude to Mrs. P i r jo Fahls t rom, Institute

of Occupational Health, Helsinki, for the statist ical work in this study.

References:

Eguchi, S. and Hirsh, I. J. : Development of Speech Sounds in Children, Acta oto-laryng., Suppl. - 257 (1969).

Hollien, H. and Paul , P. : "A sccond evaluation of the speaking funda- mental frequency cha rac te r i s t i c s of post-adolescent girls", Language and Speech - 12 (1969), pp. 119-124.

~ h r t o n ~ , J. : "Visual a ids for speech correction", Int. Symp. on Speech Communication Ability and Profound Deafness, Stockholm 1970, A. G. Bell Association for the Deaf, Washington, D. C. 1972(ed. G. Fant).

Perheentupa, J. , Autio, S. , Leist i , S. , and Raitta, C. : I1Mulibrey- nanism: dwarfism with muscle , l iver , brain and eye involvementf1, P roc . of the XVI Northern Pediatr ic Congress, Turku 1970, Acta paediat. Scand. Suppl. 206 (1970), pp. 74-75.

made by AB Specialinstrument, Stockholm, Sweden.

STL-QPSR

(5) Sundberg, J. and Tjernlund, P. : "Real time notation of performed melodies by means of a computer", paper 21 S 6 in the Congress Volume of the 7th ICA, Budapest, 197 1.

(6) Vuorenkoski, V., Wasz-Hockert, O . , Lind, J . , Koivisto, M . , and Partanen, T. J. : "Training thc auditory perception of some specific types of the abnormzl pain c r y in newborn and young infants", STL-QPSR 4/197 1, pp. 37 -48.

(7) Weinberg, B. and Zlatin, M. : "Speaking fundamental frequency character is t ics of five - and six-year old children with mongolism", J. Speech and Hearing Research - 13 (1970), pp. 418 -425.