Developmental Science 4:1 (2001), pp 61 ± 70
REPORT
Babbling and the lack of auditory speech perception: a matter of
coordination?
Florien J. Koopmans-van Beinum, Chris J. Clement and Ineke van den
Dikkenberg-Pot
Institute of Phonetic Sciences, University of Amsterdam, The Netherlands
Abstract
This paper concentrates on the question whether and where the lack of auditory perception can be traced in the early
sound productions of deaf infants. A sensorimotor description system based on movements in the phonatory and
articulatory speech production systems was developed to classify early infant vocalizations. Canonical babbling is a
strong cue in the normal speech developmental process. Therefore the main question in this work was why deaf infants do
not start to babble in their first year of life like normally hearing children do. Detailed analyses of early vocalizations of
deaf and hearing infants revealed that auditory feedback is needed to lead to coordination of movements of the
phonatory and the articulatory system, and that this coordination capacity is a prerequisite for the development of
normal speech production.
Introduction
During the past three decades the study of early speech
development has widely expanded beyond the fields of
anecdotal stories and diary reports. And although aims
and starting points of studies may differ and the
description tools may vary, the present situation is that
researchers do agree on the occurrence and order of
speech developmental stages from birth onwards (Koopmans-van Beinum & van der Stelt, 1979, 1986; Oller,
1980; Stark, 1980; Roug, Landberg & Lundberg, 1989).
From the studies mentioned above it turns out that
speech production in normally hearing children develops
in a highly organized way. However, it is not totally
clear yet which aspects of early infant vocalizations
represent influences of auditory input and feedback, and
which are anatomically and physiologically governed.
Literature suggests deviant sound production in deaf
infants already in the first year of life, such as no
canonical babbling before 11 months (e.g. Oller & Eilers,
1988). On the other hand it might be surprising that
severe hearing impairment, if it is unexpected in a
family, may often not be detected until the child is over
1 year of age.
In order to get more insight into the influence of (lack
of auditory) speech input on early sound production, we
analysed vocalizations of deaf infants during their first
18 months of life, classifying the sounds by means of a
sensorimotor description system, and we compared the
resulting data with those for hearing children. Special
attention is directed to the occurrence and the characteristics of canonical babbling (as defined by Oller,
1986), since the data of the present study on deaf infants
could nicely be related to the results of our earlier study
on babbling (van der Stelt & Koopmans-van Beinum,
1986).
Sound classification system
As explained in detail in earlier publications (Koopmans-van Beinum & van der Stelt, 1986, 1998; Koopmans-van Beinum, 1990), a basic premise of our work
has been the notion that the human speech production
Address for correspondence: Florien J. Koopmans-van Beinum, Institute of Phonetic Sciences, University of Amsterdam, Herengracht 338, 1016 CG
Amsterdam, The Netherlands; e-mail: f.van.beinum@hum.uva.nl
# Blackwell Publishers Ltd. 2001, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.
62
Florien J. Koopmans-van Beinum, Chris J. Clement and Ineke van den Dikkenberg-Pot
instrument is central to the description of the infant's
early sound productions from birth onwards (e.g. Kent,
1976). Therefore we based our description tools on
universal principles of movements of the human speech
production instrument by applying a sensorimotor
approach, instead of a phonetic ± linguistic description
based on adult phonology. Such a sensorimotor
approach of speech production movements is relatively
free from an interpretation of sound quality in terms of
vowels and consonants and has potentially universal
applicability, as it is based on global aspects of place and
manner characteristics.
Describing early infant vocalizations in this way
requires a number of decisions with respect to the
sound. In a pilot listening experiment, in order to
discriminate between cry and non-cry sounds of young
infants, the respiratory cycle presented itself as the
natural, physiologically based unit in the analysis of the
continuous sound stream. In the next sections the notion
`utterance' has to be interpreted as an infant sound
produced during one single respiratory cycle or breath
unit. For the actual description of infant sounds the two
evolutionary sources of sound production in infancy
were taken into consideration: the larynx making
phonatory movements and the vocal tract making
articulatory movements. These functions are both
secondary: the primary function of the larynx lies in
protecting the lungs, whereas the primary function of
the vocal tract is in sucking, chewing and swallowing. By
basing our description of infant utterances on phonatory
and articulatory movements, we could link up the early
infant sound production system with the generally
accepted `source ±filter' model for adult speech production (Fant, 1960). Within this model the larynx acts as
the basic sound source, whereas the sound is filtered
subsequently by the varying shape of the vocal tract and
by the movements of the articulation organs. Human
speech is characterized perceptually by pitch, loudness,
timbre and perceived duration, and acoustically by
periodicity, intensity, spectral components and physical
duration. Variation, timing and coordination of the
phonatory and articulatory movements in the speech
production mechanism provide segmental as well as
suprasegmental speech characteristics, like vowels, consonants, syllables, intonation patterns, pauses, rate and
rhythm, used in specific languages in specific ways.
Since from birth onwards the infant is equipped with
this source-and-filter system, and since this system in
principle is the same source-and-filter system that adults
use in speech production, this model is pre-eminently
suited for the description of infant speech development,
as it is for adult speech as well. It thus provides us with
the possibility of indicating where in the developmental
# Blackwell Publishers Ltd. 2001
course the universal, anatomically and physiologically
based character of the infant's sound productions
becomes overruled by language-specific features, or
where features can be located that reveal the influence
of the speech input of the surrounding world.
Assuming that it is possible to work with a universally
applicable, clearly defined classification scheme both for
the phonatory and for the articulatory movements of
infants, we can obtain comparable data concerning deaf
and hearing infant speech development, and identify the
role of auditory input and feedback.
Moreover, classification is facilitated when we pay
attention to phonatory and articulatory movements
separately. In our previous research (Koopmans-van
Beinum & van der Stelt, 1979, 1986, 1998; Koopmansvan Beinum, 1990) we provided a description of early
infant vocalizations resulting in six sensorimotor stages
of early speech development for the period from birth
until the occurrence of early words in a clear hierarchical
framework (see Table 1).
The qualitative stages of early vocal development
described above coincide to a great extent with the
Table 1 Overview of speech development stages by means of a
sensorimotor description of movements based on the source ±
filter model of speech production
Stage
Characteristics
Onset
(in weeks)
I
Simple, uninterrupted phonation without any
articulatory movement within one breath unit
0
II
Interrupted phonation within one breath unit,
without any articulatory movement
6
III
One articulatory movement within one breath
unit, combined with (or eventually without)
continuous (I) or interrupted (II) phonation
10
IV
Seeming relapse: a strikingly large decrease of
articulatory movements (III) and an increase
in uninterrupted phonation without
articulation (II); at the same time, however,
all kinds of variations in the phonatory
domain with respect to intonation, duration
and intensity
20
V
Multiple articulatory movements (repetitive
or variegated) during one breath unit,
combined with (or eventually without)
continuous or interrupted phonation (III),
together with all kinds of phonatory
variations (IV), normally indicated as
`babbling'
26
VI
Combining all kinds of utterances of the
preceding stages, using them referentially and
in specific situations, so that meaningful
`words' occur that are recognizable for adults
40
Note: Ages of onset of a new stage are given only as global indications.
Babbling and the lack of auditory speech perception
stages formulated by Oller (1980) and by Stark (1980),
who use more intuitive terms such as `cooing' or
`gooing', `expansion' and `vowel play'. As Roug,
Landberg and Lundberg (1989) have indicated on the
basis of their comparison of a number of studies on early
infant vocalizations, there is general agreement on the
existence of these major stages. However, in general the
stage descriptions offered in the other studies lack an
explicit theoretical framework that can account for the
changes in the developing speech system.
Indicating stages on the basis of age may be crucial
for matching a specific milestone (defined as the onset of
a new stage). It refers to the infant's exhibiting a new
type of behaviour. However, a clear distinction should
be made between, on the one hand, a description of early
speech development on the basis of sensorimotor stages
(this means marking the milestones in the development
of infant sound production) and, on the other hand,
longitudinal descriptions or classifications of infants'
actual sound productions over a certain period of time.
It is clear that a universally applicable classification
system is necessary in order to make true comparisons
possible between qualitative and quantitative results
from various groups of children, such as deaf and
hearing.
Actually, if we want to differentiate between various
types of phonation and articulation that sometimes
occur in parallel, we need a matrix representation of the
classification results, as displayed in Table 2. Along one
axis a number of phonatory distinctions with respect to
the utterance in question are indicated:
*
*
*
*
*
NoPhon
UnIntPhon
IntPhon
VarUnIntPhon
VarIntPhon
no phonation
uninterrupted phonation
interrupted phonation
variegated uninterrupted phonation
(variation in duration, intonation
and loudness)
variegated interrupted phonation
Table 2 Matrix display used for the classification of phonatory
and articulatory aspects of infant utterances
Articulation type
Phonation type
NoArt
OneArt
TwoArt
NoPhon
UnIntPhon
IntPhon
VarUnIntPhon
VarIntPhon
±
Stage I
Stage II
Stage IV
Stage IV
Stage III
Stage III
Stage III
Stage III ‡ IV
Stage II ‡ III ‡ IV
Stage V
Stage V
Stage V
Stage IV ‡ V
Stage IV ‡ V
Note: Cells contain possible sound productions as described in the stage model
provided in Table 1.
# Blackwell Publishers Ltd. 2001
63
Along the other axis articulatory aspects are indicated:
*
*
*
NoArt
OneArt
TwoArt
no articulatory movement
one articulatory movement
two or more articulatory movements
during two- or more-syllabic utterances
By means of these tools each utterance can be
classified uniquely and each cell of the matrix represents
a clearly defined class of utterances. Moreover vocal,
consonantal and syllabic structures, as well as intonational, temporal and rhythmic patterns, are thus
represented in their most basic elements. For specific
purposes some of the codings were further elaborated
with respect to place and manner of articulation
(Clement, den Os & Koopmans-van Beinum, 1995;
Clement, forthcoming).
Within this basic coding system it is always possible to
vary the number of distinctions to be used. In our
project on differences between vocalizations of deaf and
hearing infants (see below) these codings were sufficient
to compare the phonatory and articulatory development
of the children's early sound productions.
Babbling in hearing infants
In a previous study concerning 54 normally hearing
children (van der Stelt & Koopmans-van Beinum, 1986)
we studied the relationship between infants' early sound
productions as a result of their physiological development and their other developing motor functions. For
that study we concentrated on the babbling milestone
because of its easy and clear identifiability. The
following questions were to be answered:
*
*
*
At what age do infants start babbling?
Is the onset of babbling different for the two sexes?
Is babbling related to other motor functions with an
onset in the first year of life?
All 54 infants in our sample sooner or later started to
babble. The mean age was 30.8 weeks with a standard
deviation of 6.3 weeks and a range of 30 weeks. In the
eighteenth week the first infant was reported to babble.
In the thirty-fifth week three-quarters of the infants were
babbling (mean age of 28 weeks and a standard
deviation of 3.6 weeks) and in the thirty-eighth week
the vast majority (50 out of 54) had started to babble
(see Figure 1).
Since the last infant in our sample started to babble
only in the forty-eighth week, it will be clear that we were
highly interested in this case. The neurological development of the child concerned, a healthy boy, seemed not to
be delayed. He was a very active child with regard to his
64
Florien J. Koopmans-van Beinum, Chris J. Clement and Ineke van den Dikkenberg-Pot
Figure 1
Distribution of the onset of babbling of 54 infants in weeks. Mean age (30.8 weeks) is indicated by the broken line.
motor development and was able to walk at the age of 10
months. At the age of 17 years, he was reported to be a
normal, healthy schoolboy, finishing his high school
perfectly on schedule. This case clearly demonstrates the
variability in early speech development, which obviously
complicates the diagnosis of a possible delay.
No significant differences (Student's t distribution)
were found between boys and girls with respect to the
onset of babbling and other aspects of motor development. As for the question whether the onset of babbling is
directly related to the onset of one or more of the other
motor functions in the first year of life, we have seen
already in the case of the last babbler, who mastered
all other motor functions, that a relationship is not
necessarily present. Moreover, the girl who babbled in the
eighteenth week was a relatively late smiler, but was
amongst the middle group for the other functions.
Cumulative distribution curves of the onset of
babbling and the onset of other motor functions in this
group of 54 children were highly comparable, suggesting
a strong physiological underlying base. However, a pure
physiological underlying base could not account for the
severely delayed onset of babbling in deaf children as
reported by Oller and Eilers (1988).
Vocalizations of deaf infants
The general question about the role of audition and the
influence of the surrounding language input on the
# Blackwell Publishers Ltd. 2001
speech production of (normally hearing) infants during
their first year of life has been addressed in several
studies (for an overview see, for example, Jusczyk,
1997). Results suggest that there is a general influence of
the surrounding speech input on early infant sound
productions demonstrable already in the first year.
Therefore we may expect that a deficiency in auditory
perception will reveal its influence on the vocalizations
of deaf infants when they are carefully analysed and
compared with those of hearing infants, from a very
young age onwards.
As far as we know it has not yet been determined
systematically whether and where deviations can be
traced during the pre-lexical stage in severely hearingimpaired infants, who lack auditory speech input as
well as the internal auditory feedback of their own
speech productions. In a double project using monthly
audio-recorded longitudinal vocalization data of six
deaf and six hearing infants from 5.5 until 18 months
of age, and partly from 2.5 months onwards, we
investigated the influence of auditory perception
(project Clement from 2.5 to 12 months; project van
den Dikkenberg from 12 to 18 months, and partly
beyond). We analysed the sound productions by
applying the sensorimotor coding system for perceptual classification of phonatory and articulatory
aspects, and by measuring acoustical characteristics.
In the present paper we concentrate on the classification of the infants' utterances in types of phonation
and articulation.
Babbling and the lack of auditory speech perception
Participants
Twelve mother ± infant pairs participated in the longitudinal study. Six infants were profoundly hearing
impaired (HI), and the six other infants had normal
hearing (NH). All infants had hearing parents. No clear
health problems such as cognitive or motor delays were
found in a health screening immediately after birth
(Apgar score), nor later on as tested by the Denver
Developmental Screening Test and the Bayley Developmental Scales applied at 12 and 18 months (Bayley,
1969). The NH infants were matched with the HI infants
along the following criteria: sex, birth order, duration of
pregnancy, age of the mother, socioeconomic status of
the parents (defined in terms of education), dialect of the
parents (defined in terms of residence and regional
origin) and degree of dialect use.
The HI infants (five boys and one girl), who were the
second or third child in the family, had an average
hearing loss of over 90 dB in the best ear, which was
established by Auditory Brainstem Response Audiometry in the first months of life. The loss was confirmed
by pure-tone audiometric tests at a later age. All HI
infants participated in early intervention programs,
including ear training. They all used hearing aids,
although not all of the children used them regularly
during the first year. In all cases the cause of deafness
was genetically based. Two HI children were raised
with total communication (TC), two with Dutch sign
language=TC, and two mainly by the oral method.
Table 3 gives an overview of the auditory characteristics
of the six HI children.
Although the HI children are diverse with regard to
the degree of `loss with hearing aids', a preliminary
attempt to split the group into one subgroup with better
hearing (HI-1, HI-2, HI-3) and one subgroup with worse
hearing (HI-4, HI-5, HI-6) did not provide any
significant differences in the results of the analyses.
Table 3 Overview of the auditory characteristics of the HI
children
Participant
HI-1
HI-2
HI-3
HI-4
HI-5
HI-6
Hearing
loss (dB),
best ear
Loss with
hearing
aids (dB)
Hearing
aids from
age
(months)
Language
method
Start of
recording
(months)
97
93
110
>120
>120
>100
55
55
65
Not tested
Not tested
>100
2.0
3.5
4.5
No
6.5
7.5
Oral
TC
Oral
NGT=TC
NGT=TC
TC
2.5
5.5
5.5
2.5
3.5
5.5
Note: The various language methods concern the oral method, total communication (TC) and Dutch sign language (NGT).
# Blackwell Publishers Ltd. 2001
65
Neither did a threefold subdivision with respect to
language method. For both types of division the
resulting groups are actually too small to make
statistically sound comparisons. Although it is clear
that the lack of homogeneity of the children might have
its consequences for eventual generalization of the
results, neither the degree of `loss with hearing aids'
nor the language method chosen by the parents could be
a selection criterion in our project. The difficulty of
finding very young HI infants, especially in a country
where most of the infants are born at home and undergo
their first auditory tests only at the age of 9 months,
makes it almost impossible to apply too strong a
selection criterion.
Method
Audio-recordings lasting about half an hour each were
made every 2 weeks in the first year and every month in
the second year. Recordings were started as soon as
possible after diagnosis (i.e. at the ages of 2, 3 and 5
months for two, one and three HI infants, respectively)
and at the age of 2 months for the NH children. The
parents themselves made the recordings at their homes
in order to keep the situation during the audiorecordings as natural as possible. The mothers were
asked to talk with their children in a face-to-face
situation while the children were sitting in an upright
position and to start recordings as soon as the children
were vocalizing in their usual way.
All non-vegetative comfort utterances of the infants
produced during the first 10 min of each monthly
recording were transcribed and counted. No laughing
sounds were included by two trained phoneticians (interjudge agreement > 90%). No specific selection was made
in `speech-like' and `non-speech-like' vocalizations, since
it is not known beforehand which aspects of sound
production are essential in the developmental process of
speech production. Next, 50 infant utterances per
recording were selected evenly (i.e. each second or third
utterance, depending on the total number) out of the ten
transcribed minutes, and were stored in digital form for
further acoustic analysis. Subsequently each utterance
was coded as belonging to one of three possible types of
articulation (inter-judge agreement 91%) and one of five
possible types of phonation (inter-judge agreement
85%), based on the classification of movements in
infant speech development as described above (see
Table 2). For all selected utterances syllabic structures
were determined and for all articulatory movements
occurring in the selected utterances place and manner
were analysed as well (neither are presented in this
66
Florien J. Koopmans-van Beinum, Chris J. Clement and Ineke van den Dikkenberg-Pot
paper). Moreover acoustic measurements on utterance
duration and intonation were performed (only slightly
referred to in this paper).
Results
Overall results, calculated by averaging the occurrences
of the various phonation and articulation types per
month and per child over the whole research period for
the two groups, reveal significant differences (Mann ±
Whitney U test, p < 0:05) for a number of sound types
(see Table 4).
Since the data of one deaf and one hearing child (both
girls) after the first year are incomplete, we removed
them for further analysis, leaving us with material on ten
children with respect to the second year of life.
Sound productions of the following types are produced significantly more by the hearing children than by
the deaf:
*
*
*
uninterrupted phonation combined with one articulatory movement;
uninterrupted phonation combined with two or more
articulatory movements;
interrupted phonation combined with two or more
articulatory movements.
Sound productions of the following types are produced significantly more by the deaf children than by the
hearing:
*
*
variable uninterrupted phonation without any articulatory movement;
variable interrupted phonation without any articulatory movement.
These results seem to indicate that combining phonation with articulation, actually the coordination of
Table 4 Mean occurrence of each of the combined
articulation=phonation types for deaf and hearing children,
averaged over all months and all children per group
Articulation type
NoArt
OneArt
TwoArt
Total
Phonation type
HI
NH
HI
NH
HI
NH
HI
NH
NoPhon
UnIntPhon
IntPhon
VarUnIntPhon
VarIntPhon
Total
±
18.8
1.9
11.4
4.6
36.6
±
17.4
1.7
6.5
1.8
27.5
2.0
6.9
0.7
1.7
0.7
12.2
0.5
13.4
1.2
2.5
0.6
18.8
0.0
0.8
0.0
0.3
0.1
1.2
0.0
2.5
0.5
0.5
0.2
3.8
2.0
26.5
2.6
13.4
5.4
50.0
0.5
34.0
3.6
9.5
2.6
50.0
Note: Bold numbers indicate significant differences between the two groups
(Mann ± Whitney U test, p < 0:05).
# Blackwell Publishers Ltd. 2001
movements of the two different sound production
systems, is a capacity that causes problems for the deaf
children. However, averaging results from all monthly
measurements may obscure developmental trends in the
data. We therefore analysed the monthly data for the
five types of phonation=articulation where the two
groups of children differed significantly, by means of
running averages over 3 month periods. Figure 2
displays the data on uninterrupted phonation combined
with one articulatory movement and the data on
uninterrupted phonation and interrupted phonation
both combined with two or more articulatory movements.
The data with respect to the utterance type `uninterrupted phonation combined with one articulatory
movement', illustrated as dots in Figure 2, differ
significantly for the deaf and the hearing children from
7.5 ± 9.5 months onwards until the end of the whole
period studied in this project (Mann ± Whitney U test,
p < 0:05). Mastering the coordination of the two sound
production systems in its most elementary form is a
prerequisite for later word production, as this rather
simple phonation=articulation combination is the basic
and earliest acquired word form. Although the deaf
children start to make these utterances in the period
from 4.5 to 8.5 months (in the back of the vocal tract),
they do not continue producing this type of utterance at
other articulation places (in the front and the central
part of the vocal tract) as the hearing children
increasingly do (see for more details Clement, forthcoming).
The data with respect to the utterance type `uninterrupted phonation combined with two or more
articulatory movements' (squares in Figure 2) differ
significantly for the deaf and the hearing children
from 6.5 ± 8.5 months onwards until the end of the
whole period studied in this project (Mann ± Whitney
U test, p < 0:05). The data with respect to the utterance
type `interrupted phonation combined with two or
more articulatory movements' (triangles in Figure 2)
differ significantly for the deaf and the hearing
children over the period from 6.5 ± 8.5 to 12.5 ± 14.5
months. Both types actually belong to the group
of utterances characterized as babbling (see Tables 1
and 2).
All six hearing children started to produce canonical
babbling (phonation combined with two or more
articulatory movements) at the expected age: in our
recordings two children at 5.5 months, one at 6.5
months and three at 7.5 months of age. Unexpectedly
one of the six deaf infants (HI-2) started to babble at
7.5 months of age as well. Since the average hearing
loss of this child is 93 dB for his best ear, there is no
Babbling and the lack of auditory speech perception
67
Figure 2 Mean number of occurrences for three phonation=articulation types with articulatory movements, in which deaf and hearing
infants differ significantly. Data points are presented as running averages over 3 month periods; n indicates the total number of monthly
data sets involved in the 3 month periods, for the deaf and the hearing infants separately (C, A, uninterrupted phonation combined with
one articulatory movement; E, D, uninterrupted phonation combined with two or more articulatory movements; G, F, interrupted
phonation combined with two or more articulatory movements).
reason to be suspicious about the degree of hearing
impairment: in spite of his severe hearing impairment
he started to babble at the same age as normally
hearing children do on average. One explanation might
be that the child has a usable hearing residue,
particularly in the lower frequency range. None of
the other HI children in this project started to babble
before 18 months of age.
For the hearing children as well as for the deaf child
that started to babble the mean number of occurrences
of the utterance type `uninterrupted phonation combined
with two or more articulatory movements' (squares in
Figure 2) is higher than that of the utterance type
`interrupted phonation combined with two or more
articulatory movements' (triangles in Figure 2), which
utterances can be found only scarcely for the deaf boy.
This is not surprising since the uninterrupted phonation
type is the simplest and earliest acquired one in
babbling. Producing the combination of multiple
articulatory movements with repetitive starting and
stopping movements of the phonatory system, as in
the utterance type `interrupted phonation combined with
# Blackwell Publishers Ltd. 2001
two or more articulatory movements', is clearly much
more complicated.
Figure 3 shows the results concerning `variable
uninterrupted phonation without any articulatory
movement' and those concerning `variable interrupted
phonation without any articulatory movement'. These
data concern utterance types for which the deaf and the
hearing children differ significantly as well (Mann ±
Whitney U test, p < 0:05), but now the deaf children
outnumber the hearing children. With respect to the
utterance type with variable intonation patterns with
uninterrupted phonation but without any articulatory
movement, presented as dots in Figure 3, the two
groups of children differ significantly from 9.5 ± 11.5
months onwards until the end of the whole period
studied in this project. It is evident that this type of
sound does not require any coordination of the two
sound production systems and therefore might be
favorite for the deaf children, probably since it does
provide some tactile sensation.
With respect to the utterance type with variable
intonation patterns without any articulatory move-
68
Florien J. Koopmans-van Beinum, Chris J. Clement and Ineke van den Dikkenberg-Pot
Figure 3 Mean number of occurrences for two phonation=articulation types without articulatory movements, in which deaf and
hearing infants differ significantly. Data points are presented as running averages over 3 month periods; n indicates the total number of
monthly data sets involved in the 3 month periods, for the deaf and the hearing infants separately (C, A, variable uninterrupted
phonation without any articulatory movement; G, F, variable interrupted phonation without any articulatory movement).
ment with interrupted phonation, presented as triangles
in Figure 3, deaf and hearing children differ significantly from 7.5 ± 9.5 months onwards until the end of
the whole period studied in this project. This type of
utterance is particularly interesting since it contains the
repetitive movement that is characteristic for babbling
at least in one of the two sound production systems
(phonation). The deaf children start to outnumber the
hearing children in this type of sound at the same time
that hearing children start babbling, i.e. making
repetitive movements in both sound production systems (phonation and articulation) which requires a
good coordination of the movements of the two
systems.
Another type of utterance in which the deaf children
differ from the hearing children, although not significantly in the overall mean values (not bold in Table 4
and not presented in the figures), is one articulation
without any phonation (NoPhon OneArt). From the age
of 9.5 ±11.5 months up to 11.5 ±13.5 months the deaf
children again outnumber the hearing children significantly (Mann ±Whitney U test, p < 0:05), a period where
the hearing children are quite active with respect to
# Blackwell Publishers Ltd. 2001
utterances in which OneArt and TwoArt are combined
with phonation. The tactile sensation of the articulatory
movement may favour this type of utterance for the deaf
children, but coordination with the phonatory system is
lacking.
Conclusion and discussion
Frequently severely HI children may remain undiscovered as such until their second year of life. Their
frequent vocalizations and their active multimodal way
of communication disguise the lack of auditory perception. Nevertheless it turns out that the vocalizations of
deaf and hearing children show clear differences already
in the first year of life, continuing in the second year,
such as in the number of utterances (higher for deaf
infants), utterance duration, and the place and manner
of articulation (van den Dikkenberg-Pot & Koopmansvan Beinum, 1997; Clement, forthcoming; van den
Dikkenberg-Pot, forthcoming).
The use of the sensorimotor approach for the
classification of early infant vocalizations provides us
Babbling and the lack of auditory speech perception
with the possibility of investigating in detail which
aspects of the vocalizations actually differ. Our study
showed that deaf children display the same developmental order as hearing children with respect to the use
of each of the two sound production systems involved
in speech (the phonatory system and the articulatory
system). This development of the separate systems
seems to be physiologically and neurologically governed. Also series of repetitive (rhythmic) movements
in one of the two sound production systems (phonation only) are present in the deaf infants at the same
time as babbling (i.e. repetitive, rhythmic movements
in the phonatory and the articulatory system simultaneously) starts in hearing infants, just like the
repetitive movements in hammering, chewing, bumping etc. (see also the contributions of Ejiri and
Masataka and of Takei in this issue). However, the
main differences between deaf and hearing children
arise where the two systems have to be coupled:
uninterrupted phonation combined with one articulatory movement (stage III in Table 2), and interrupted phonation combined with two or more
articulatory movements (stage V in Table 2). The deaf
child that nevertheless started to babble at the same
age as normally hearing children do on average could
obviously make use of a hearing residue that provided
him with enough auditory feedback for an adequate
coordination of the two sound production systems.
Although only part of the data of the whole project
has been presented here, it seems legitimate to state that
infants' auditory perception and feedback are a prerequisite for the coordination of the movements in the two
systems together, necessary in canonical babbling. Since
canonical babbling contains all basic elements of (adult)
speech, it is a strong cue in the normal speech
developmental process.
So far the results of our study are an indication that
the capacity to produce canonical babbling requires two
factors. One is a biologically governed developmental
skill present in all healthy children to produce repetitive,
rhythmic movements. The other is the ability to
correctly coordinate the movements of the two elementary sources of sound production ± the phonatory
instrument and the articulatory instrument ± a skill that
requires a well functioning auditory perception system
or at least some form of auditory feedback.
Acknowledgement
These projects are financially supported by the Institute
for the Deaf in Sint-Michielsgestel, The Netherlands,
and by the joint Dutch Institutes for the Deaf.
# Blackwell Publishers Ltd. 2001
69
References
Bayley, N. (1969). Manual for the Bayley scales of infant
development. New York: The Psychological Corporation.
Clement, C.J. (forthcoming). PhD thesis, University of
Amsterdam.
Clement, C.J., den Os, E.A., & Koopmans-van Beinum, F.J.
(1995). The development of early vocalizations of deaf and
normally hearing infants in the first eight months of life. In K.
Elenius & P. Branderud (Eds), Proceedings ICPhS 95 (Vol. 1,
pp. 138±141). Amsterdam: University of Amsterdam.
van den Dikkenberg-Pot, I. (forthcoming). PhD thesis,
University of Amsterdam.
van den Dikkenberg-Pot, I., & Koopmans-van Beinum, F.J.
(1997). Number and duration of spoken utterances of deaf
and normally hearing children between twelve and eighteen
months. Some preliminary results. Proceedings of the
Institute of Phonetic Sciences Amsterdam, (Vol. 21, pp. 91 ±
100). Amsterdam: University of Amsterdam.
Fant, G. (1960). Acoustic theory of speech production. The
Hague: Mouton.
Jusczyk, P.W. (1997). The discovery of spoken language.
Cambridge, MA: MIT Press.
Kent, R.D. (1976). Anatomical and neuromuscular maturation
of the speech mechanism: evidence from acoustic studies.
Journal of Speech and Hearing Research, 19, 421± 445.
Koopmans-van Beinum, F.J. (1990). The source ± filter model
of speech production applied to early speech development.
Proceedings of the International Conference of Spoken
Language Processing 1990, Kobe, Japan (Vol. 2, pp. 1321±
1324).
Koopmans-van Beinum, F.J., & van der Stelt, J.M. (1979).
Early stages in infant speech development. Proceedings
of the Institute of Phonetic Sciences Amsterdam, (Vol. 5,
pp. 30 ± 43). Amsterdam: University of Amsterdam.
Koopmans-van Beinum, F.J., & van der Stelt, J.M. (1986).
Early stages in the development of speech movements. In B.
Lindblom & R. ZetterstroÈm (Eds), Precursors of early speech
(pp. 37 ±50), Wenner Gren. Int. Symp. Series 44. New York:
Stockton Press.
Koopmans-van Beinum, F.J., & van der Stelt, J.M. (1998).
Early speech development in children acquiring Dutch.
Mastering general basic elements. In S. Gillis & A. de
Houwer (Eds), The acquisition of Dutch (pp. 101± 162).
Amsterdam=Philadelphia: John Benjamins.
Oller, D.K. (1980). The emergence of the sounds of speech in
infancy. In G.H. Yeni-Komshian, J. Kavanagh & C.A.
Ferguson (Eds), Child phonology, Vol. I: Production
(pp. 93 ±112). New York: Academic Press.
Oller, D.K. (1986). Metaphonology and infant vocalizations.
In B. Lindblom & R. ZetterstroÈm (Eds), Precursors of early
speech (pp. 21 ± 35), Wenner-Gren Int. Symp. Series 44. New
York: Stockton Press.
Oller, D.K., & Eilers, R.E. (1988). The role of audition in
infant babbling. Child Development, 59, 441±449.
Roug, L., Landberg, I., & Lundberg, L.-J. (1989). Phonetic
development in early infancy: a study of four Swedish
70
Florien J. Koopmans-van Beinum, Chris J. Clement and Ineke van den Dikkenberg-Pot
children during the first eighteen months of life. Journal of
Child Language, 16, 19 ± 40.
Stark, R.E. (1980). Stages of speech development in the first
year of life. In G.H. Yeni-Komshian, J. Kavanagh & C.A.
Ferguson (Eds), Child phonology, Vol. I: Production
(pp. 73 ± 92). New York: Academic Press.
# Blackwell Publishers Ltd. 2001
van der Stelt, J.M., & Koopmans-van Beinum, F.J. (1986). The
onset of babbling related to gross motor development. In B.
Lindblom & R. ZetterstroÈm (Eds), Precursors of early speech
(pp. 163± 173), Wenner Gren. Int. Symp. Series 44. New
York: Stockton Press.