Word-initial voicing in the productions of stops in normal and preterm Italian infants

International Journal of Pediatric Otorhinolaryngology ELSEVIER

31 (1995)191-206

Word-initial voicing in the productions of stops in normal and preterm Italian infants Umberta Bortolini”, Claudio Zmarich b, Renato Fior b, Serena Bonifacio* b “Centro di Studio per le Ricerche di Fonetiea, C.N.R.. Piazza Salvemini 13, 35131 Padova. Italy bServizio di Logopedia. Divisione 0. R.L.. Istituto per I’lnfanzia ‘Burl0 Garofolo’, Via dell’lstria 6511, 34100 Trieste, Italy

Received 30 June 1994; accepted 18 August 1994

Abstract It has been traditionally held that developmental anomalies in language acquisition are more frequent in neonatologically at-risk subjects. There is some suggestion that proficiency in phonology is correlated with motor control development. The purpose of this paper was to compare the patterns of acquisition of the control of the acoustic-phonetic cues for voicing in the speech of premature infants and controls. The measure studied was initial stop consonant voice onset time (VOT), which is known to be the most reliable acoustic cue for the distinction between voiced and voiceless stops. The total population of the study consisted of 7 infants born at less than 37 weeks gestation and a control group of 7 infants born full-term at normal weight; 7 adults, aged 24-26 years, also participated. Each child was recorded under standard recording conditions saying words contrasting labial, dental and velar voiced and voiceless initial stops. Elicited word productions were collected monthly from infants, at different age levels, ranging from 18 to 21 months. The results show that the subjects are more advanced in the acquisition of the appropriate VOT values for the voiceless than for the voiced consonants. This diff%zulty may be related to the increased neuromuscular control and more complex muscle activity necessary for maintaining voicing during the closure, especially for velar stops. It is important to recognize the possibility that increased variability in preterm children may be related to some neuromuscular immaturity. Keywords:

Voice onset time; Speech production;

Preterm children

* Corresponding author. 016%5876/95BO9.50 0 1995Elsevier Science Ireland Ltd. All rights reserved SSDI 0165-5876(94)01091-B

192

U. Bortolini et al. /Int. J. Pediatr. Otorhinolaryngol.

31 (1995) 191-206

1. Introduction This researchis part of a larger investigation concerning the possible effectsof different handicapping conditions present at birth on language learning. It is well known from literature that newborn children at-risk have inferior maturation level in fine motor development [26] as compared to controls, and that the alterations of languagelearning, phonological proficiency [5,8,12,13,22,23,29,32]and later reading abilities [3,4,13,24,31]could be due to these causes. The nature of the relationship (if any) between immaturity of motor control and phonological development is still far from clear, although speech timing and its variability, as a source of information about speechmotor control development, has been the object of much interest recently [6,9,11,21,35]. The present paper reports an acoustical investigation of the development of the voicing contrast in Italian word-initial stops produced by two groups of infants, preterms and normals. The purposes are to compare the pattern of acquisition of the acoustic-phonetic cues for voicing in the speechof at-risk infants and controls, and to discussinter-subject differences in results in relation to phonological proficiency. The measurestudied was the initial stop consonant voice onset time (VOT), which is defined as the time interval between the releaseof the articulators for the stop occlusion and the onset of vocal fold oscillation. VOT is known to be the most reliable acoustic cue for the distinction between voiced and voicelessstops and this temporal characteristic of stop consonant reflects the complex timing of supralaryngeal-laryngeal coordination [ 11. Most languages separate stop consonants into voiced and unvoiced stop consonants and this contrast is achieved through differences in the timing of glottal articulation relative to supraglottal articulation. VOT measuresfall on a continuum, since the timing of glottal articulation relative to supraglottal is free to vary over a continuous range. The standard [16] is to set zero as the time of consonant release; voicing that begins before the releaseof the articulators is assigneda negative value, while voicing that is delayed beyond release is assigned a positive value. The VOT continuum (ms) 0 -100 I--------------------------I--------------------------I voiced

voiceless unaspirated

+lOO voiceless aspirated

In different languages the phonemic contrast between voiced and voiceless stops corresponds to distinct VOT ranges along this VOT continuous range. Extensive cross-language studies by Lisker and Abramson [16-181 (see also [14]) have demonstrated that three categories of stops, having a rough correspondenceacross languages,emerge along the voice onset time continuum: (1)

‘Voicing lead’: negative VOT values, ranging from about -125 to -75 ms, with a median value of -100 ms. Italian voiced stops are of this type.

U. Bortolini et al. /Int. J. Pediatr. Otorhinolaryngol. 31 (1995) 191-206

(2) (3)

193

‘Short voicing lag’: positive VOT values, ranging from 0 to +25 ms, with a median value of +lO ms. Italian voiceless stops and English voiced stops are of this type. ‘Long voicing lag’: highly positive VOT values, ranging from +60 to +lOO ms, with a median value of +75 ms. English voiceless stops are of this type.

The articulatory gestures underlying short voicing lag stops in initial utterance position are in specific ways less complicated than for the other two types of VOT. Voicing lead requires muscle gesturesin addition to those needed for short voicing lag stops. In fact, in the production of voiced initial stop consonants some mechanism, external to the larynx, must operate to sustain an adequate transglottal pressure drop during stop closure. The long voicing lag stop requires more carefully controlled tuning between stop and laryngeal closure and the adduction of the vocal folds requires more complex muscle activity [7,28,30,33,34]. In order to produce the voicing contrast the young child must learn to coordinate the timing of velopharyngeal closure, closure of supraglottal articulators, vocal fold oscillation and releaseof supraglottal articulators. Children’s earlier productions of stop consonants fall in the short lag voicing range for both voiced and voicelesstargets [15,27]. It is apparent that the acquisition of voicing contrast will be circumscribed by both physiologically and acoustic parameters. Differences in the age at which similar distinctions are made across languages may actually be the results of the different phonetic parameters employed for voicing contrast by languages. Italian is a language where the voice categories are significantly different from those found in English (see above). The hypothesis tested in this paper was that differences in duration of VOT between the two subject groups can be seenas a measure of line motor skills development. In fact, a study of Piper et al. [26] suggest that infants born at very early gestational agesexhibit significant differences in their fine motor development compared with their older counterparts. This delay in line motor development may be an indicator of later developmental disparities such as learning disorders. 2. Procedure The population of this study consisted of 7 low-risk preterm infants, born at less than 37 weeks gestation (mean gestational age: 34 weeks), and a control group of 7 children, born full term at normal weight (seeTable 1). Routine clinical tests on the preterm subjects at 12 months excluded any loss of hearing or cognitive delay. All the children were developing normally. A group of 7 adults, aged 24-26, also participated. The test was administered to preterm and normal infants at 18, 19, 20 and 21 months of chronological age. Each child was recorded, under standard recording conditions (using Uher model 4200 portable tape recorder with Electrovoice model 635A microphone), saying each of the 12 test words at least three times. The test items were the following minimal pair pseudo-words, contrasting labial, dental and velar voiced and voicelessstops: ‘papa’, ‘baba’, ‘pipi’, ‘bibi’, ‘tata’, ‘dada’, ‘titi’, ‘didi’, ‘kaka’, ‘gaga’, ‘kiki’, ‘gigi’.

(I. Etortolini et al. /Int. J. Pediatr. Otorhinolaryngol.

31 (1995) 191-206

Table 1 Gestational age(weeks) weightat birth (g) and Apgar scoresat 1 and 5 min for each subject of the preterm and normal groups Subjects’ characterisics Subjects

Gestational age (weeks)

Weight (is)

36 36 34 32 32 35 34

40 40 39 39 38 40 40

Awar 1 min

5min

2630 2530 2310 1970 2120 1940 2520

I 2 9 4 4 7 8

10 6 10 8 7 10 10

3300 3300 3150 2960 2930 3000 3810

8 IO 8 8 9 2 10

10 10 IO 10 10 4 10

Pretenns

1. 2. 3. 4. 5. 6. 7.

Kevin Fabrizio Francesco Giulia Lorenza Athina Debora

Gestational age mean = 34.0 S.D. = 1.6 Controls I.

2. 3. 4. 5. 6. 7.

Davide Nina Sara Andrea Jessica Allegra Ludovica

Gestational age mean = 39.4 S.D. = 0.7

The infant and adult productions were collected in randomized order. Measurement of VOT values were made for the initial stop consonant of each target word from both acoustic waveform and wide-band spectrograms,obtained with a Macintosh 2ci computer using Signalyze 2.2 (InfoSignalTM Inc.), with AudioMedia 2.0 0 Digidesign. The sampling of the signal was set at 44 kHz/l6 bits. As measuredfrom a wide-band spectrogram, VOT is defined as the time elapsed from the onset of the releaseburst to the first of the regularly spacedvertical striations which indicate glottal pulsing [ 171.When measuredfrom an expanded waveform display, VOT is defined as the time interval from the onset of the consonant releaseburst, as reflected by the first sudden change to the wave-shape, to the point at which a periodic wave (voicing) was first clearly detected (Fig. 1). All the tokens selected for analysis are produced with oral and velopharyngeal complete closure. The criteria for rejection of an utterance from VOT measurement were (percentage to the overall productions, i.e. 1354 tokens for normals and 705 tokens for preterms); phonological processes (normals 7.0%, preterms 14.7%);

V. Bortolini et al. / Ini. J. Pediatr. Otorhinolaryngol.

4--

25,9

msec

31 (1995) 191-206

195

+

Fig. 1. Acoustic representation of VOT interval: wide-band spectrogram, waveform and expanded waveform for /kaka/.

spirantization (normals 8.9%, preterms 3.7%); voice airflow leaking (normals 3. I%, preterms 2.8%). The term ‘airflow leaking’ (Fig. 2) refers to the partial lowering of the soft palate and opening of the velopharyngeal port, in order to sustain an adequate transglottal pressure drop during stops. Although hypothesized by some authors as a mechanism possibly used to sustain phonation 125,281,others questioned its importance in the normal adult production of voiced consonants [19,33]. The productions clearly characterized by the presenceof the nasality acoustic markers on the sonogram were not accepted, Other exclusions refer to tokens characterized by: ‘multiple bursts’, i.e. the presenceof one or more vertical striations of the waveform a few milliseconds after the first burst of consonant release (normals lS%, preterms I .I%); whisper or breathy voice (normals 1.2%, preterms I. 1%); ‘stress shift’, i.e. the change of the word stressfrom the first syllable to a following one (normals 0.3%, preterms 1,3%); and, finally, ‘voice breaking’ (normals I,OO/,preterms 0.7%). This term refers to a

196

lJ. Bortolini et al. / Int. J. Pediatr. Otorhinolaryngol.

I

-321

msec

31 (1995) 191-206

_____)

Fig. 2. Acoustic representation of a ‘voice airflow leaking’ instance: narrow-band spectrogram, wide-band spectrogram, waveform and expanded waveform for /bita/.

clear break in voicing during the closure interval of a voiced consonant (Fig. 3). Flege and Brown [lo] comment on this phenomenon: ‘it is possible that given the small physical size of speakers,the voicing observed before the interruptions in their productions of ibl was sustained by only passive cavity enlargement’. After the break, if the vocal folds are not strictly pressedtogether, voicing may restart only by using a supplementary strategy to enlarge the vocal cavity, such as the lowering of the larynx. Furthermore, mechanical or environmental causes,as an extremely low or high vocal intensity (normals 2.5%, preterms 2.3%), or the presenceof noises and voice overlay (normals 4.7%, preterms 3.7%) brought a rejection of 23.0% of the total set of utterances from normal children and 25.5% from preterm children.

U. Bortolini et al. /Int. J. Pediatr. Otorhinolaryngol. 31 (1995) 191-206

197

Fig. 3. Acoustic representation of a ‘voice breaking’ instance: narrow-band spectrogram, wide-band spectrogram, waveform and expanded waveform for /dadi/.

The resulting number of analyzable tokens of two subject groups were very different. Normal subjects as a whole generally produced more than preterm subjects (1043 tokens versus 526). In order to make the two groups comparable, the study randomly selecteda number of tokens close to that of preterm subjects.For the VOT analysis, the study considered the subject groups only if at least four subjectsin each group said the target words at least twice. At 18 months most of the preterm children repeated the tafget words less than twice. 3. Results

First the distributions of VOT mean values of 7 adults producing the sametarget words will be examined. Adult VOT mean values, as shown in Table 2, reveal differ-

198

U. Bortolini et al. /Int. J. Pediatr. Oiorhinolaryngoi.

31 (1995) 191-206

Table 2 Number of tokens, range, mean and standard deviation of VOT values for stop consonants (before /a/ and /ii for the adults group stops

Count

Range (ms)

Mean W

SD. (ms)

pa Pi ba bi

68 62 61 66

0.0-I 5.9 7.5-15.9 -147.8-16.0 -185.1--33

10.0 12.7 -61.8 -84.7

3.2 3.4 38.7 38.6

ta ti da di

65 61 65 61

10.8-30.2 10.8-34.7 -165.9-19.2 -190.7--24.8

16.8 21.9 -71.9 -87.8

3.9 5.5 38.9 37.3

ka ki ga gi

69 71 65 63

8.5-53.4 16.3-71.5 -161.6-26.8 -210.6-42.6

21.3 40.7 -64.3 -69.5

13.0 39.4 41.6

7.7

encesin mean VOT as a function of place of articulation and of coarticulatory effect of the following vowel. The study’s data, in fact, provide some evidence for the well-recognized difliculty in maintaining voicing during a velar stop, and for the longer VOT durations in stops coarticulated with high vowels than in those coarticulated with low vowels. Fig. 4 clearly illustrates the greater variability or standard deviation for voiced than for voiceless stops. ADULTS

-50

-100

-150

I

I

I

I

I

I

pa

pi

ba

bi

ta

I

ti

I

da

I

di

I

ka

I

ki

I

ga

I

gi

SYLIABLEs

Fig. 4. Means, bracketed by standard deviation, of VOT values for adults.

V. Bortolini et al. /Int. J. Pediatr. Otorhinolaryngol.

31 (1995) 191-206

199

Number of subjects, number of tokens, range, mean and standard deviations of VOT values for each stop consonant (before /a/ and /i/) for each subject group aged 19, 20 and 21 months were calculated (Table 3). Table 3 Number of subjects, number of tokens, range, mean and standard deviation of VOT values (mediated values in the -a and -i context) for stop consonants for the adults’ group and the two groups of children aged 19, 20 and 21 months

P b t d k is

Mean (ms)

SD. (ms)

10.8-34.7 -190.7-19.2 8.5-71.5 -210.0-42.6

11.3 -13.7 19.3 -79.9 34.1 -66.9

3.5 40 5.4 38.8 12.6 43.5

21 25 25 12 14

-113.4-45.5 -141.8-29.8 -31.1-55.6 -201.2-124.8 11.2-78.1

4.2 -55.9 13.8 -19.1 33.3

33.6 59.3 15.5 83.6 19.0

23 26 24

2.7-50.6 -211.5-58.7 5.4-56.6

17.4 -41.9 18.2

12.9 81.4 13.5

13

1.7-66.1

32.2

17.4

No. of subjects

Count

I I 7 I I I

130 127 126 132 140 128

Range b-4 0.0-23.1

-185.1-16.0

19 months Normals ::

t d k g

67 5 4

I

-

-

Preterms t:

t

4 4 6

d k

-

8

-

4

-

20 months

E t d k 8

I 6 6

5 5

21 33 31 25 41

-33.8-103.2 -176.2-41.8 6.1-140.5 -149.6-46.9 6.2-14.9

18.5 -60.9 28.3 -49.5 28.7

24.2 58.5 25.8 60.0 16.3 -

23 31 34 21 38

3.9-113.0 -182.7-59.1 -200.4-38.5 -159.7-33.5 -208.8-158.4

18.1 -39.5 4.5 -35.4 18.4

23.8 64.5 42.5 65.3 54.0

Preterms f: t

d k k?

5 6 I 6

5

200

U. Bortolini et al. /ht. J. Pediatr. Otorhholaryngol. 31 (1995) 191-206

Table 3 (continued) No. of subjects 21 Illen& Normals P b t d k B

Count

Range cm

Mean (ms)

7 7 I 6 7 4

46 43 33 21 34 17

5.4,-16.1 -260.6-38.1 -56.2-48.2 -194.6-67.7 9.5-96.3 -216.9-34.2

26.9 -49.1 19.7 -58.3 39.6 -51.6

18.5 70.2 18.2 78.0 23.1 96.3

Preterms I 67

46 45

-176.3-84.8 -94.6-50.6

-39.1 11.5

60.9 24.0

:

47

22 32

-120.5-29.9 -72.1-109.4

-26.0 20.1

52.6 30.9

k B

6 -

31

-194.8-74.4 -

-

18.4

52.5 -

-

No significant difference was found in normal and preterm infants for the coarticulatory effect of the following vowel (two tailed f-test, P < 0.05). Mean VOT values, bracketed by standard deviation for each subject group at different age levels are presented in Figs. 5-7. Comparing the pattern of VOT as a cue for voicing it appears that: At 19 months(Fig. 5): the present data reveal: (1) normal and preterm subjectsproduce no significant difference between the means for any voiced and voiceless stop NORMALS

PRETERMS

50 _. _ ._ __ ._ ,_ .._. _ .__ _ d

0 - .._..

.-

5

.. -

t .- .-.

I .,.,- 1 .._.

-

50 -" -.' -

..

_ _

_.

_

-

_

_ _

_ -

I -50

-150

_.__ __ _

I p

,,,_.

I bt

I

.. .- .-

_.

i d

0 9

STOPS

I k

._

-150

1

'I

p pbtdkg

'I btd

'I

-

'I

'I k

'I g

STOPS

Fig. 5. Means, bracketed by standard deviation, of VOT values for the subjects’ groups at 19 months.

U. Bortolini et al. /Int. J. Pediatr. Otorhinolaryngol. 31 (1995) 191-206 NORMALS

PRETERMS

100 -

100 -

50 _..,.__, ,_ ._.. _ ._ , ._ _. -.__ 1_,,,,_

0

201

_ , _.. ._. _

~

5o-

.. .._

_..._._.._.__....

-- _.....-..._

0 -. _. _... _ _ ,_ _ ._ _ ._ _ .._ -

o-

_

l -50

- .._...-

-. _ -

.

-'

_ .-....-...._..._

B

5

-50

-..-...

'

-

-

-

! _. -...._ s _..._ - ._..._

_

--

-

$ -100

_

._.... -

-150

-

_. _

I

I

p

b

_

_ .._.

I

t

_..._..,_,

I

I

I

d

k

g

_

-100

-.

..' --.

-150

-. _-

I

I

p

b

--

.. -..

I

t

STOPS

-

.-

-

I

I

I

d

k

g

-

STOPS

Fig. 6. Means, bracketed by standard deviation, of VOT values for the subjects’ groups at 20 months.

pair at any place of articulation; (2) a concentration of stops in the short voicing lag category with only bilabial stops that show short lead mean values. Larger standard deviation values for preterms than normals in the production of/b/ suggestthat these children a;e just beginning to produce p/b distinction into their phonetic repertoire. Sometokens of voiced dental stops were produced in the short voicing lead only for normal children, since preterms produced too few tokens to be represented. At 20 months (Fig. 3): all children VOT distributions for voiced and voiceless labial and dental stops began to assumethe bimodal form. In fact, most of the tokens had negative values for voiced stops and positive for voiceless. A closer inspection PFUZTERMS

NORMALS 100

100 l-

I f ’ 0 3 50

c

0

$

__ _ 1

_. _

_

_. -.

- ._ -

-

_

_

-

_

1

-

.,,__ .- .._ * _. ._ .-

-50

-

*

- .-

_

_ _

-

_

1

. _ _

i

-. - - __-r- _ _

50

_ - _ _

0

_. _

3

..-. .-. ..-. - ..-

-50 $

L

_

-100

-150

1

__ _ __ -

'

p

'

btd

. _

'

STOPS

_ -

'

'

k

_

_

'

9

-

-100

-150

i

I

I

p

b

I

t

I

1

I

d

k

g

STOPS

Fig. 7. Means, bracketed by standard deviation, of VOT values for the subjects’ groups at 21 months.

202

U. Bortolini et al. /Int. J. Pediatr. Otorhinolaryngol.

31 (1995) 191-206

of individual data of preterm subjectsreveals a considerable overlap in VOT distribution between voiced and voiceless stops. In these subjects standard deviation is generally larger than that for controls. At 21 months (Fig. 4) VOT mean values of all three stops fall into separateregions along the VOT continuum. The VOT distributions are generally bimodal with very little overlap in normal subject productions. In contrast, a considerable overlap in VOT distribution is observed for preterm bilabial and dental consonants, whereas for velars these subjects did not produce the voiced consonant. One possible explanation can be given in terms of aerodynamics. In order for vocal cord vibration to take place it is necessaryfor air to flow past the vocal cords. In order for airflow to occur, quite simply, there must be somewherefor the air to go. In the caseof labial and dental stops there is a general mouth cavity, and fully voiced stops are in fact facilitated by a slight opening of the jaw, expansion of the cheeks and elevation of the velopharyngeal port. For the velar stops with the cavity limited to the pharynx, there is little room for expansion. Consequently, velar stops require the most effort to produce in the first place. The differencesof the mean VOT values for subject groups as a function of voicing category for hornorganic stop consonants are all statistically significant (P < 0.001 for f-test). Furthermore, at 21 months the non-normalized data can be condidered on an individual base,due to the increased number of productions. A closer inspection into these data shows variations in the number of subjects exhibiting significant differences(at the level of P < 0.05 for a Mann-Whitney U-test) in VOT values associated with the production of voiced and voiceless word initial stop consonants. In fact, voicing distinction is statistically significant for labial stops in 4 normal subjects,and in 3 preterm subjects, for dentals in 3 normals and only 1 preterm subject, and for velars in 4 normals and only 1 preterm subject (seeTables 4, 5). The difference also of VOT total values for non-normalized data of the two subject groups as a function of voicing category are all statistically significant for the Mann-Whitney U-test, as shown in the last column of Tables 4 and 5. 4. Discussion Basedon the data from these VOT analysesthe following conclusion was drawn - that the pretetm children did exhibit VOT patterns that differed from those produced by normal subjects in the following aspects: (1) The distinction between voiced and voiceless stops in word initial position emergesrelatively late as measured by differences in mean VOT within each age group: at 21 months, only 2 normal children and 1 preterm child exhibit significant differencesin mean VOT, for voiced-voicelesslabial, dental and velar stops. The results show that subjects are more advanced in the acquisition of the appropriate VOT values for the voiceless than for the voiced consonants. Results coming from developmental studies on the acquisition of voicing in languageshaving voiced stops with negative values of VOT, like Spanish [20] and French [2], demonstrated that 2-year-old children do not show evidenceof having acquired the VOT values for the initial voiced stops.

U. Bortolini et al. /InI. J. Pediatr. Otorhinolaryngol.

31 (1995) 191-204

203

204

U. Bortolini et al. /Int. J. Pediatr. Otorhinolaryngol. 31 (1995) 191-206

U. Bortolini et al. /Int. J. Pediatr. Otorhinolaryngol.

31 (1995) 191-206

205

(2) The VOT patterns between the two groups of subjectsindicate that the voicing contrast is acquired earlier in labial and dental stops than in velars. It appears, therefore, that control of voicing for initial velar stops matures later, in both normal and at-risk children. The results provide someevidence of later acquisition of negative VOT values in velar consonants by at-risk children than by Controls. (3) The standard deviations are generally greater in preterm subjectsthan in Controls, and in voiced stops than in voicelessin both normal and preterm children. This variability can be considered another indication of the difficulty in producing voiced stops. This difficulty may be related to the increased neuromuscular control and more complex muscle activity previously reported as necessaryfor maintaining voicing during the closure, especially for velar stops. It is important to recognize the possibility that increased variability in preterm children may be related to some neuromuscular immaturity.

111Abramson, AS. (1977) Laryngeal timing in consonant distinctions. Phonetica 34, 295-303. 121Allen, G.D. (1985) How the young French avoids the pre-voicing problem for word-initial voiced stops. J. Child Lang. 12, 37-46. 131 Aram, D.M., Hack, M., Hawkins, S., Weissman,B.M. and Borawski-Clark, E. (1991) Very-LowBirthweight children and speechand language development. J. Speech Hear. Res. 34, 1169-I 179. [41 Bishop, D.V.M. and Adams, C. (1990)A prospective study of the relationship between specific language impairment, phonological disorders and reading retardation. J. Child Psychol. Psychiatry 3I, 1027-1050. 151Byers Brown, B., Bendersky, M. and Chapman, T. (1986)The early utterances of preterm infants. Br. J. Disorders Commun. 21, 307-319. 161Cooper, W.E. (1977) The development of speech timing. In: Segalowitz, S.J. and Gruber, F.A. (Eds.), Academic Press, New York, pp. 357-373. I71 Dixit, R.P. (1989) Glottal gestures in Hindi plosives. J. Phonet. 17, 213-237. 181D’Souza, S.W., McCartney, E., Nolan, M. and Taylor, LB. (1981) Hearing, speechand language in survivors of severe perinatal asphyxia. Arch. Dis. Child. 60, 20-24. I91 Fametani, E. (1992)Articolazione e coarticolazione nello sviluppo fonologico, in AAVV, Fonetica e Fonologia nello sviluppo normale patologico de1linguaggio, Assicopy, Bologna, pp. 39-63. 1101Flege, J.E. and Brown, W.S., Jr. (1982)The voicing contrast between English /p/ and ib/ as a function of stress and position-in-utterance. J. Phonet. 10, 335-345. 1111Hawkins, S. (1984) Motor control in speech:evidence from studies of temporal coordination. In: Lass, N. (Ed.), Speech and Language, Vol. II. Academic Press, New York, pp. 317-374. 1121 Hubatch, L.M., Johnson, C.J., Kistler, D.J., Bums, W.J. and Moneka, W. (1985) Early language abilities of high-risk infants. J. Speech Hear. Disord. 50, 195-207. 1131 Jensen,T.S., Boggild-Andersen, B., Schmidt, J., Ankerhus, J. and Hansen, E. (1988) Perinatal risk factors and first-year vocalizations: influence on preschool language and motor performance. Develop. Med. Child Neurol. 30, 153-161. I141 Keating, P., Linker, W. and Huffman, M. (1983) Patterns in allophone distribution for voiced and voiceless stops. J. Phonet. 1I, 277-290. 1151 Kewley-Port, D. and Preston, M. (1974) Early apical stop production: a voice onset times analysis. J. Phonet. 2, 195-210. 1161Lisker, L. and Abramson, AS. (1964)A cross-languagestudy of voicing in initial’stops: acoustical measurements.Word 20, 384-422. [I71 Lisker, L. and Abramson, A.S. (1967) The voicing dimension: some experiments in comparative

206

U. Bortolini et al. / Int. J. Pediotr. Otorhinolaryngol. 31 (1995) 191-206

phonetics. Proceedingsof 6th International Congress of Phonetic Sciences,Prague, pp. 563-567, (Academia, Prague 1970). (181 Lisker, L. and Abramson, AS. (1971) Distinctive features and laryngeal control. Language 47, 767-785. 1191 Lubker, J.F. (1973) Transglottal airflow during stop consonant production. J. Acoust. Sot. Am. 53, 212-215. [20] Macken, M.A. and Barton, D. (1980) The acquisition of the voicing contrasts in Spanish: a phonetic and phonological study of word-initial stop consonants. J. Child Lang. 7, 433-458. [2l] Menn, L. (1983) Development of articulatory, phonetic, and phonological capabilities. In: Butterworth, B. (Ed.), Language Production, Vol. 2. Academic Press, London, pp. 3-50. [22] Menyuk, P., Liebergott, J. and Schultz, M. (1986) Predicting phonological development. In: Lindblom, B. and Zetterstrom, R. (Eds.), Precursor of Early Speech. The MacMillan Press Ltd, pp. 79-93. [23] Menyuk, P., Liebergott, J., Schultz, M., Chesnick, M. and Ferrier, L. (1991) Patterns of early lexical and cognitive development in premature and full-term infants. JSpeech Hear. Res. 34, 88-94. [24] Menyuk, P., Chesnick, M., Weis Liebergott, J., Korngold, B., D’Agostino, R. and Belanger, A. (1991) Predicting reading problems in at-risk children. J. Speech Hear. Res. 34, 893-903. 1251 Nihalani, P. (1975)Velopharyngeal opening in the formation of voiced stops in Sindhi. Phonetica 32, 89- 102. [26] Piper, M.C., Byrne, P.J., Darrah, J. and Watt, J.M. (1989) Gross and line motor development of preterm infants at eight and I2 months of age. Develop. Med. Child Neural. 31, 591-597. [27] Preston, MS., Yeni-Komshian, Cl. and Stark, R. (1967) A study of voicing in initial stops found in the pre-linguistic vocalization of infants from different language environments. Haskins Laboratory Status Report on Speech Research IO, 109-120. WI Rothenberg, M. (1968) The breath-stream dynamics of simple released-plosiveproduction. Bibl. Phonetica 6, l-l 17. 1291 Stewart, A.L. (1984) Severe perinatal hazards. In: Rutter, M. (Ed.), Developmental Neurop sychiatry. Churchill Livingston, Edinburgh, pp. 15-31. [301 Tyler, A.A. and Watterson, T.L. (1991) VOT as an indirect measureof laryngeal function. Semin. Speech Lang. 12, 131-141. [311 Vohr, B.R. and Garcia-Coil, C.T. (1985)Neurodevelopmental and school performance of very-lowbirth-weight infants: a seven year longitudinal study. Pediatrics 76, 345-350. 1321 Vohr, B.R., Garcia-Coil, C.T. and Oh, W. (1988)Language development of low birthweight infants at two years. Develop. Med. Child Neurol. 30, 608-615. [331 Westbury, J.R. (1983) Enlargement of the supraglottal cavity and its relation to stop consonant voicing. J. Acoust. Sot. Am. 73, 1322-1336. [341 Westbury, J.R. and Keating, P.A. (1986)On the naturalnessof stop consonant voicing. J. Linguist. 22, 145-166. [351 Wilson, W.R. and Morton, K. (1990) Reconsideration of the action-theory perspective on speech motor control. Clin. Linguist. Phonet. 4, 341-362.