How Do Children Organize Their Speech in the First Years of Life? Insight From Ultrasound Imaging

Article (PDF Available)inJournal of Speech Language and Hearing Research 61(6):1355-1368 · June 2018with 320 Reads
DOI: 10.1044/2018_JSLHR-S-17-0148
Abstract
(in press; excerpt from a non final abstract due to copyright restriction) This study reports on a cross-sectional investigation of lingual coarticulation in 57 typically developing German children (four cohorts from 3.5 to 7 years of age) as compared with 12 adults. It examines whether the organization of lingual gestures for intrasyllabic coarticulation differs as a function of age and consonantal context. Using the technique of ultrasound imaging, we recorded movement of the tongue articulator during the production of pseudo words including various vocalic and consonantal contexts. Results from linear mixed effects models show greater lingual coarticulation in all groups of children as compared to adults with a significant decrease from the kindergarten years (at 3; 4; 5) to the end of the first year into primary school (at 7). Additional differences in coarticulation degree were found across and within age groups as a function of the onset consonant identity (/b/, /d/ and /g/). Results support the view that although coarticulation degree decreases with age, children do not organize consecutive articulatory gestures with a uniform organizational scheme (e.g., segmental or syllabic). Instead results suggest coarticulatory organization is sensitive to the underlying articulatory properties of the segments combined.
Figures - uploaded by Aude Noiray
Author content
All content in this area was uploaded by Aude Noiray
No caption available
… 
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
1!
1
!
2
!
3
!
4
!
How do children organize their speech in the first years of life?
5
!
Insight from ultrasound imaging
6
!
7
!
Aude Noiray12, Dzhuma Abakarova1, Elina Rubertus1, Stella Krüger1, Mark Tiede 2
8
!
9
!
1 Laboratory for Oral Language Acquisition, University of Potsdam, Germany
10
!
2 Haskins Laboratories, New Haven, USA
11
!
12
!
This research was supported by a grant from the Deutsche Forschungsgemeinschaft
13
!
14
!
15
!
16
!
17
!
18
!
19
!
20
!
21
!
22
!
Correspondence concerning this article should be addressed to Aude Noiray, University of
23
!
Potsdam (Germany).
24
!
Contact: anoiray@uni-potsdam.de
25
!
26
!
27
!
28
!
29
!
30
!
31
!
32
!
33
!
34
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
2!
1
!
ABSTRACT
2
!
Purpose: This study reports on a cross-sectional investigation of lingual coarticulation in 57
3
!
typically developing German children (four cohorts from 3.5 to 7 years of age) as compared with
4
!
12 adults. It examines whether the organization of lingual gestures for intrasyllabic coarticulation
5
!
differs as a function of age and consonantal context.
6
!
Method: Using the technique of ultrasound imaging, we recorded movement of the tongue
7
!
articulator during the production of pseudo words including various vocalic and consonantal
8
!
contexts.
9
!
Results: Results from linear mixed effects models show greater lingual coarticulation in all groups
10
!
of children as compared to adults with a significant decrease from the kindergarten years (at 3; 4;
11
!
5) to the end of the first year into primary school (at 7). Additional differences in coarticulation
12
!
degree were found across and within age groups as a function of the onset consonant identity (/b/,
13
!
/d/ and /g/).
14
!
Conclusions: Results support the view that although coarticulation degree decreases with age,
15
!
children do not organize consecutive articulatory gestures with a uniform organizational scheme
16
!
(e.g., segmental or syllabic). Instead results suggest coarticulatory organization is sensitive to the
17
!
underlying articulatory properties of the segments combined.
18
!
19
!
Keywords: Children, Development, Language, Speech motor control, Speech production
20
!
21
!
22
!
23
!
24
!
25
!
26
!
27
!
28
!
29
!
30
!
31
!
32
!
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
3!
How do children organize their speech in the first years of life?
1
!
Insight from ultrasound imaging
2
!
3
!
INTRODUCTION
4
!
5
!
In the domain of spoken language acquisition, great attention has been focused on
6
!
coarticulation, which concerns the overlapping of articulatory gestures for neighboring segments
7
!
(for a review see Hardcastle & Hewlett, 2006). Coarticulation is a fundamental characteristic of
8
!
fluent speech. It is an important mechanism to investigate as it taps into the phonetic instantiations
9
!
of phonological units from various sizes such as phonemes or syllables and therefore offers a
10
!
chance to reveal how unit organization matures over time as children learn to speak their native
11
!
language. In addition, coarticulation engages multiple speech articulators (e.g., the lips, the
12
!
tongue) whose actions must be coordinated in time and in the space of the vocal tract to produce
13
!
intelligible phonetic outputs in the native language. Investigating the development of
14
!
coarticulatory patterns therefore provides a unique opportunity to address both the maturation of
15
!
the speech motor system and its attunement to the phonetic regularities of the language spoken.
16
!
In this study, we were specifically interested in examining how differences in lingual vowel-
17
!
to-consonant coarticulation can shed light on the phonetic organization of speech in young
18
!
German children, from 3 years of age (when they are in kindergarten) to 7 years of age when
19
!
children are in primary school. In addition, we aimed to provide a first quantitative survey of
20
!
anticipatory coarticulation in German learners. Much developmental work on lingual
21
!
coarticulation has focused on English variants. However, studies in languages other than English
22
!
are needed to possibly disentangle universal versus language specific patterns of coarticulation. In
23
!
German at least, most assessments of consonant acquisition have used measures of individual
24
!
production accuracy (e.g., Fox-Boyer, 2006). Using the technique of ultrasound imaging, we
25
!
examined the organization of gestures of the tongue, an organ whose control is essential to
26
!
vowels’ and consonants’ acquisition (e.g., Barbier, Perrier, Ménard, & Payan, 2015; Klein et al.
27
!
2013; Ménard & Noiray, 2011; Noiray, Ménard, & Iskarous, 2013; Rubertus, Abakarova, Ries, &
28
!
Noiray, 2016; Song, Demuth, Shattuck-Hufnagel, & Ménard, 2013; Zharkova, Hewlett,
29
!
Hardcastle, & Lickley, 2014). In the past, articulatory tracking methods such as electromagnetic
30
!
articulography (EMA) and electropalatography (EPG) have been employed in school-aged
31
!
children and adolescents (e.g., EPG: Cheng, Murdoch, Goozée, & Scott, 2007; Gibbon, & Wood,
32
!
2010; EMA: Katz & Bharadwaj 2001; Terband, Maassen, Lieshout, & Nijland, 2011). More
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
4!
recently, the technique of ultrasound imaging has been adapted to the developmental field to make
1
!
articulatory recordings of the tongue possible in young populations (e.g., Barbier, Perrier, Ménard,
2
!
& Payan, 2015; Ménard & Noiray, 2011; Noiray, Ménard, & Iskarous, 2013; Song, et al., 2013;
3
!
Zharkova, 2017). Compared to EMA and EPG, ultrasound is a more suitable technique to use with
4
!
young children because it does neither require long preparation time prior to testing nor invasive
5
!
procedures to track tongue movement during speech (e.g., gluing EMA pellets on young
6
!
children´s tongue or placing an artificial palate). Hence, with the technique of ultrasound imaging,
7
!
it is now possible to revisit questions related to speech organization in young children while
8
!
directly examining the articulatory mechanisms underlying speech production rather than inferring
9
!
those mechanisms from the acoustic outputs.
10
!
11
!
What are the units of speech production in children?
12
!
Finding the units of speech organization in the first years of life has been one of the most
13
!
challenging endeavors for developmental psycholinguists, but the quest is important for advancing
14
!
both theories of language acquisition and clinical assessment of disordered speech.
15
!
Over the past two decades, research examining intra-syllabic coarticulatory patterning in typically
16
!
developing children has provided conflicting results and hypotheses regarding the nature of these
17
!
units. A number of studies have reported less coarticulation in children compared to adults with
18
!
limited influence of the vowel on the preceding consonant. Such findings lead to the hypothesis
19
!
that spoken language organization is initially segmentally driven (e.g., Gibson & Ohde, 2007;
20
!
Green, Moore, & Reilly, 2002; Katz, Kripke, & Tallal, 1991; Kent, 1983). In this view, children
21
!
are supposed to proceed through a sequential maturation process by which articulatory controls for
22
!
individual segments progressively develop into more complex inter-articulator organizations for
23
!
larger units, with increasing intra-syllabic coarticulation as a result. An opposite view holds that
24
!
children initially display greater consonant-vowel (CV) coarticulation than adults, suggesting a
25
!
broader planning unit of their speech than the segmental unit (e.g., Goodell, & Studdert-Kennedy,
26
!
1993; Nijland et al. 2002; Nittrouer & Whalen, 1989; Nittrouer, Studdert-Kennedy, & Neely,
27
!
1996; Rubertus et al., 2015). In this more holistic perspective, maturation of coarticulatory
28
!
patterns would consist in decreasing encroachment between consonantal and vocalic components
29
!
and development of increasingly differentiated controls over individual articulators for a more
30
!
segmental organization of articulatory movements. Other studies have found equivalent patterns
31
!
of coarticulatory degree in children and adults but reported greater variability in children’s
32
!
patterns (e.g., Katz, Kripke, & Tallal, 1991; Munson, 2004; Repp, 1986; Sereno, Baum, Marean,
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
5!
& Lieberman, 1987). To date, organizational units of speech production are still discussed (see
1
!
satellite workshop in Laboratory Phonology 15, July 2016 dedicated to this topic).
2
!
Additional research is evidently needed not only to disentangle the origin(s) of current
3
!
theoretical discrepancies; but also because a detailed understanding of lingual coarticulatory
4
!
development over age in typically developing children (TDs) would provide useful information
5
!
for advancing detection of atypical trajectories (Maas & Mailend, 2017). Indeed, a series of
6
!
experimental studies conducted by Nijland and colleagues revealed inconsistent coarticulatory
7
!
organization in children with developmental apraxia of speech (DAS) compared to TD controls
8
!
(e.g., Nijland, Maassen, & van der Meulen, 2003). While some children with DAS seem to exhibit
9
!
greater coarticulation than TDs, others show the opposite patterns. More recently, Terband (2017)
10
!
examined coarticulatory patterns from 16 children with DAS aged between 5.5 and 7.5 years
11
!
producing /bi, di, bu, du/ and reported greater coarticulation than the 8 age-matched TDs tested for
12
!
comparison. However, deviant coarticulatory patterns in children with DAS were only observed in
13
!
some phonetic contexts but not all. This suggests that the deficit in anticipatory coarticulation
14
!
observed in children with DAS is not uniform as assumed in the ASHA descriptions (ASHA,
15
!
2007) but specific to certain phoneme combinations that may involve more complex articulatory
16
!
coordinations compared to others. Difficulty in coarticulatory organization has also been noticed
17
!
in stuttering children (e.g., Soo-Eun, Ohde, & Conture, 2002) who seem to exhibit smaller
18
!
coarticulatory differences across consonantal contexts than typically developing age-matched
19
!
children. Taken together, these results have important implications as to the links between the
20
!
articulatory properties of the speech material investigated, speech motor control and the breadth of
21
!
coarticulatory organization in atypical development. To provide reference data in German, this
22
!
study focuses on lingual vowel-to-consonant coarticulation in typically developing children.
23
!
!
24
!
Differences in lingual coarticulation degree and resistance across consonantal contexts
25
!
An important variable to consider when investigating variance in lingual coarticulatory
26
!
patterns within a sample of participants or across populations regards the articulatory properties of
27
!
the sequences produced. In adults, differences in intra-syllabic coarticulation degree within
28
!
individuals reflect differences in consonants’ place of articulation with labial-V syllables showing
29
!
a high coarticulation degree (CD) contrary to alveolar or alveopalatal stop-V syllables, which
30
!
show a lower degree of coarticulation between consonantal and vocalic lingual gestures.
31
!
Interestingly, in adults these patterns have been consistently reported across various languages
32
!
(e.g., in American English: Fowler, 1994; Fowler & Brancazio, 2001; Iskarous et al., 2011;
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
6!
Australian languages: Graetzer, 2006; Canadian French: Noiray et al., 2013; Catalan: Recasens,
1
!
1985; Recasens & Espinosa, 2009; German: Abakarova, Iskarous, & Noiray, 2017; Iskarous,
2
!
Fowler, & Whalen, 2010; Swedish: Lindblom & Sussman, 2012; in Thai, Cairene Arabic, and
3
!
Urdu: Sussman, Hoemeke, & Ahmed, 1993).
4
!
A main hypothesis is that differences in CD are related to the degree of coarticulatory
5
!
resistance of the consonant (e.g., Bladon & Al-Barmeni, 1976; Fowler, 1994; Fowler &
6
!
Brancazio, 2001; Recasens, 1985; Recasens & Espinosa, 2006). In this view, resistance varies
7
!
across consonants’ place and manner of articulation as a result of differences in articulatory
8
!
demands on target articulators, which affect their degree of temporal and spatial overlap with
9
!
adjacent segments (e.g., Fowler & Saltzman, 1993). The more constraints on an articulator are
10
!
involved in the production of a consonant, the more resistant the consonant may be to large
11
!
coarticulatory overlap with contiguous vowels. For example, in adult speakers, alveolar stops /t, d/
12
!
resist lingual coarticulation with adjacent vowels more than labial stops /p, b/ do (e.g., Iskarous et
13
!
al., 2010; Recasens, 1985; Sussman, Hoemeke, & Ahmed, 1993). Various studies across
14
!
languages have outlined that palatal consonants such as [ɲ] exert more coarticulatory resistance
15
!
than alveolars such as [n] (in Catalan: Rodriguez & Recasens, 2016; in English: Fowler &
16
!
Brancazio, 2001; in German: Hoole, Gfroerer, & Tilmann, 1990). These findings corroborate the
17
!
predictions from the Degree of Articulatory Constraints model of coarticulation (DAC, Recasens,
18
!
1999), which directly relates the degree and direction of coarticulation to the demands imposed on
19
!
the tongue body for consecutive articulatory gestures. Taken together, results suggest that CD
20
!
varies along a continuum depending on consonant identity and its degree of resistance to
21
!
coarticulation with adjacent segments (for a detailed discussion, see Iskarous, Mooshammer,
22
!
Hoole, Recasens, Shadle, Saltzman, & Whalen, 2013). Given that adults’ coarticulatory
23
!
organization depends on the interaction of articulatory gestures, any study looking at the
24
!
ontogenetic development of coarticulation would gain in explanatory power by considering
25
!
articulatory gestures’ intrinsic properties from which coarticulatory overlap versus resistance
26
!
originates.
27
!
28
!
There is converging evidence from acoustic studies (e.g., Nittrouer, 1993; Nittrouer, 1995;
29
!
Sussman, Duder, Dalston, & Cacciatore, 1999; Reidy, 2015) and articulatory studies (e.g., in 7
30
!
and 5-year old American English: Katz & Bharadwaj, 2001; in 5 and 13-year old Scottish children
31
!
in Zharkova et al., 2015; in 4-5 year old Canadian French children in Noiray, Ménard & Iskarous,
32
!
2013) that children’s coarticulatory patterns differ across phonetic contexts. Greater lingual
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
7!
anticipatory coarticulation is observed for heterorganic sequences (e.g., labial C-V syllables)
1
!
compared to homorganic sequences for which the same target organ is recruited for both the
2
!
consonant and vowel (e.g., alveolar C-V). Studies employing the Locus Equation approach in
3
!
child speech (mostly in English) have reported a decreasing degree of coarticulation from labial to
4
!
velar to alveolar stops (e.g., Goodell & Studdert-Kennedy, 1993; Sussman, Duder, Dalston, &
5
!
Cacciatore, 1999), with labial and velar stops sometimes yielding similar CD depending on
6
!
speakers (e.g., Noiray, Ménard, & Iskarous, 2013; review in Gibson & Ohde, 2007). Interestingly,
7
!
alveolar stops considered as more resistant to coarticulation than labials, show a decrease in CD
8
!
with age. One hypothesis is that children progressively develop synergistic relationships among
9
!
muscles and functional subparts of the tongue (e.g., tongue body and tongue tip) to achieve lingual
10
!
constrictions (e.g., the tongue body moving front to support the tongue tip in achieving the
11
!
alveolar constriction) (Noiray, Ménard, & Iskarous, 2013). This point will be further addressed in
12
!
the Discussion.
13
!
14
!
With respect to fricatives, results diverge across studies. Some report age-related differences
15
!
in coarticulation (e.g., Maas & Mailend, 2017; Nittrouer et al., 1989) with greater CD in children
16
!
than adults (e.g., in English: Nittrouer, et al., 1996) while others do not (e.g., Katz, Kripke, &
17
!
Tallal, 1991). In German, a recent acoustic study investigating coarticulation between fricatives /s/
18
!
or /ʃ/ and vowels in children aged 4 to 6 reported greater vocalic influence over fricatives in
19
!
preschoolers than adults (Kleber, 2015). In general, fricatives are complex consonants, whose
20
!
productions stabilize later than stops’ (Nittrouer, 1995), especially labials that are present in the
21
!
early babbling repertoire (in German: Fox-Boyer, 2006; 2009; in English: Prather, 1975; Stoel-
22
!
Gammon & Dunn, 1985).
23
!
Overall, available evidence suggests that children’s coarticulatory patterns exhibit sensitivity
24
!
to contextual effects early in age. Hence, the question of early coarticulatory organization may be
25
!
framed like in adults, as gradient distinctions along a continuum rather than supporting a binary
26
!
organization as often suggested in the developmental literature (segmental versus syllabic).
27
!
28
!
RESEARCH QUESTIONS
29
!
Given the theoretical findings outlined above, our study asked the following questions: First,
30
!
does lingual CD overall differ in German children as compared to German adults? Given
31
!
preschoolers’ immature phonological and speech motor systems (e.g., Smith, 2010), we expected
32
!
children’s coarticulatory patterns to differ significantly from adults. The discrepancies found as to
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
8!
whether children generally coarticulate equally, more, or less than adults made it difficult to
1
!
formulate specific expectation regarding the direction of the difference. Second, we tested whether
2
!
CD systematically differs between children and adults regardless of the onset consonant identity.
3
!
Given most developmental studies have investigated age-related differences in lingual
4
!
coarticulation in one or two phonetic contrasts, e.g., /ti, ta/, Zharkova, 2017; /si, su, ʃi, ʃu/,
5
!
Nittrouer et al., 1996), we expanded this research to consonants varying in places and manners of
6
!
articulation.
7
!
Third, we examined whether coarticulation degree in children varies as a function of the
8
!
consonant’s identity as it does in adults across languages. Taking preliminary results in Canadian
9
!
French preschoolers (Noiray et al., 2013), we expected German children to show modulations in
10
!
CD according to consonants’ place and manner of articulation with a lower CD in consonantal
11
!
contexts that have been shown in adults to resist coarticulatory overlap, and greater CD in velar
12
!
and labial contexts supporting large coarticulatory overlap.
13
!
14
!
METHODS
15
!
Participants
16
!
The production task was administered to four cohorts of children (total: 57) and one adult
17
!
cohort (total: 12). The groups consisted of 17 3-year old children (age range: 3;05 3;07, mean:
18
!
3;06), 14 4-year old children (age range: 4;05 4;07, mean: 4;06), 13 5-year old children (age
19
!
range: 5;05 – 5;07, mean: 5;06), and 13 children at the end of the first school year (last month and
20
!
a half) or beginning of second year (first month and a half) (age range: 7;00 – 7;05, mean: 7;02).
21
!
The latter will be referred to as grade 1 children. All five cohorts were monolingual German
22
!
speakers. Parental questionnaires insured that none of the participants had any language-related,
23
!
hearing, or visual impairment. The twelve German adults (age range: 19 – 34 years, mean: 25;08)
24
!
also presented no history in language or hearing impairments.
25
!
26
!
Stimulus material
27
!
A German female model speaker recorded production material. The stimuli were presented
28
!
auditorily in a repetition task to all participants. They consisted of disyllabic trochaic C1VC2ǝ
29
!
pseudo words embedded in a carrier phrase with the German female article /aɪnə/ (e.g., “eine
30
!
bide”). There were three stop consonants: /b/, /d/, /g/. For the 4-year-old group, the school-aged
31
!
children, and adults, the additional consonant /z/ was included. We also used the tense and long
32
!
vowels /i:/, /y:/, /e:/, /a:/, /u:/, and /o:/. C1Vs were designed as a fully crossed set of Cs and Vs
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
9!
while the second C2ǝ syllable was added in a way that C1 was not the same consonant as C2.
1
!
Intra-syllabic coarticulation was measured in the first CV syllable, between C1 and V. We aimed
2
!
for all children to repeat the CV syllables 6 times while adults produced 9 repetitions. Pseudo
3
!
words were presented in randomized blocks to prevent habituation effects. To make the recording
4
!
playful and meaningful for our young participants, stimuli were presented as a new language that
5
!
they would use with aliens during a space journey. This scenario fitted the experimental procedure
6
!
developed in our lab very well.
7
!
8
!
Experimental procedure
9
!
Recordings took place at the Laboratory for Oral Language Acquisition (LOLA) at
10
!
University of Potsdam (Germany) in an experimental room that is well suited for child studies and
11
!
decorated to match our space journey storyline. All participants were recorded within the
12
!
SOLLAR platform (Sonographic and Optical Linguo-Labial Articulation Recording system,
13
!
Noiray et al., 2015). SOLLAR is a multi-data recording platform embedded into a spaceship to
14
!
stimulate children’s interest. This child-friendly platform allows for the recording of the audio
15
!
speech signal (microphone Shure, sr.: 48kHz), tongue movement via ultrasound imaging (Sonosite
16
!
scanner, sr.: 48Hz), and labial-shape tracking via video recording (camera SONY, sr.: 50Hz). The
17
!
same technical set-up was used for adults to ensure similar experimental conditions for all
18
!
participants. The audio signal was recorded in relation to two devices: 1) synchronous with the
19
!
ultrasound device and 2) synchronous with the video camera. This information was used to
20
!
generate a universal time code for all data. Video and acoustic signals were then synchronized
21
!
using cross-correlation function within Matlab. This method has been reliably used in previous
22
!
speech production studies (e.g., in adults: Noiray, Ménard, Cathiard, & Abry, 2011; Noiray,
23
!
Iskarous, & Whalen, 2014; in children: Noiray, Ménard, Cathiard, Savariaux, & Abry, 2008;
24
!
Rubertus et al., 2015).
25
!
26
!
In SOLLAR, the ultrasound probe is positioned below participants’ chin to record the
27
!
tongue surface contour on the midsagittal plane. It is placed in a custom-made probe holder that is
28
!
constructed with a system of light springs and ball bearings to allow the probe to move smoothly
29
!
down with the jaw while the participant speaks. It is mounted in an adjustable custom-made
30
!
pedestal and on an electrical table to be adjustable for the participant´s height. The child sits
31
!
perpendicular to the probe holder with the small probe positioned below his chin between the
32
!
maxillary bones. In this study, we did not use a fixed headset to maximize the naturalness of the
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
10!
speech recorded and avoid blocking jaw movements, which would require participants to modify
1
!
their natural articulatory strategies. As we were interested in the maturation of coarticulatory
2
!
patterns over age and laboratory settings may already affect natural speech style, maximizing
3
!
naturalness was a crucial criterion for us to decide against a fixed headset. A discussion of our set
4
!
up´s advantages and limitations can be found in the Limitation and Perspective section.
5
!
Upon arrival at the laboratory, children were familiarized with the experimenters and the
6
!
SOLLAR platform. They were comfortably seated in a car seat that included seat belts as part of
7
!
the SOLLAR spaceship. Two experimenters were involved in the recording to maximize the
8
!
quality of the data collected. While one experimenter monitored the recording equipment and
9
!
controlled for the quality of the data collection (e.g., quality of tongue images, position of the
10
!
child via the video camera), the other experimenter maintained a face-to-face connection with the
11
!
child, controlled for head movement, and executed the stimulus presentation. The two
12
!
experimenters were well trained with the devices and with recording children. Prior to testing, we
13
!
organized pilot recordings with adults and children to optimize the experimental procedure, timing
14
!
and general approach to be employed with children. Despite our efforts, articulatory data
15
!
collection with young children remains challenging; hence, we could not guarantee experimental
16
!
conditions as optimal as when testing adults. As in other developmental studies (e.g., Zharkova et
17
!
al., 2017), we therefore conducted qualitative examinations of the video data post recording to
18
!
select tokens for analyses (cf. Appendix A).
19
!
The production task consisted in repeating the auditorily presented stimuli. Adults were
20
!
recorded with the same experimental set-up except that we excluded the space journey storyline.
21
!
All participants were compensated for their participation in the study, and children received a
22
!
present.
23
!
24
!
Data processing
25
!
The acoustic speech signal was used as a reference to detect relevant time points in the
26
!
articulatory signal recorded with ultrasound imaging. For adults, the segmentation was done semi-
27
!
automatically using WebMAUSBasic (Kisler, Schiel, & Sloetjes, 2012) and subsequent manual
28
!
adjustments. For children, two to three trained students at the linguistic department of University
29
!
of Potsdam labeled segments from correct target CV syllables. Manual adjustments and labeling
30
!
were done within Praat (Boersma & Weenink, 1996). Formant patterns (especially F2 and F3) as
31
!
well as changes in periodic cycle were used to determine phoneme onsets and offsets (e.g., for
32
!
vowel, the first pulse with the visible formant structure was used as reference for onsets as well as
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
11!
the end of the formant structure for offsets). Boundaries were systematically adjusted with the
1
!
automatic function “move to nearest zero crossing” provided in Praat to guarantee consistency in
2
!
boundary settings throughout labeling. Cases for which transcription was problematic were
3
!
discussed with the labeling team. To measure CD differences, two time points were extracted
4
!
from the acoustic speech signal: the temporal midpoint of the acoustically defined first consonant
5
!
(hereafter referred to as C50), and the temporal midpoint of the acoustically-defined vowel (V50).
6
!
Corresponding ultrasound images of the tongue were then extracted based on the
7
!
synchronized acoustic speech signal. More specifically, SOLLAR script selects the video frame
8
!
with the time code most closely matching the each of the two target time points. As the ultrasound
9
!
data are recorded in 60Hz, the interval between each ultrasound frame was 16.6667ms. This
10
!
means that in the worst case, if the requested time falls exactly in between frames, the selected
11
!
ultrasound frame would be 8.33ms off with respect to the acoustical landmark. We judged this
12
!
potential issue minimal (e.g., accuracy of ±40 ms in Zharkova & Hewlett, 2009). For each relevant
13
!
frame (C50, V50), tongue contours were detected with SOLLAR custom-made scripts for
14
!
MATLAB (2016a; Fig 1). For each tongue contour, an estimate of the tongue body position along
15
!
the front-back dimension was obtained by extracting the x and y coordinates of the highest point
16
!
of the tongue body. This point was taken as reference for vocalic gestures. We discarded /da/
17
!
sequences because /a/ being a low vowel, the highest point on the tongue body would occur more
18
!
in the region of the tongue blade which would not be representative of the vocalic gesture.
19
!
20
!
FIGURE'1
21
!
!
22
!
!
23
!
Statistical analysis
24
!
A table providing an overview of the number of CV repetitions across and within age groups
25
!
used for the statistical analyses can be found in Appendix A.
26
!
27
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
12!
Testing for overall developmental differences in CD
1
!
First, we examined whether overall CD differences could be observed across the five
2
!
cohorts investigated in this study. To achieve this, we fitted linear mixed-effects models using the
3
!
lme4 package in R (Bates, Maechler, Bolker, & Walker, 2014). We regressed the horizontal
4
!
position of the highest point on tongue body at consonant midpoint (PEAKX_C1_050) on the
5
!
horizontal position of the highest point on tongue body at vowel midpoint (PEAKX_V50), age
6
!
cohort (COHORT), consonant (CONSONANT1) and the interaction of PEAKX_V50 and age
7
!
cohort (PEAKX_V50: COHORT). Age cohort and consonant were treatment coded with C3 and
8
!
/b/ as baselines respectively. The structure of random effects for this model and for all the
9
!
following linear mixed models presented in the paper was determined following the approach
10
!
suggested by Bates and colleagues (Bates et al., 2015). The approach combines principal
11
!
components analysis (PCA) to determine the maximal number of dimensions for a model that is
12
!
supported by the data (“RePsychLing” package, Bates et al., 2015) with likelihood ratio tests to
13
!
assess goodness of fit. We began by testing the full random effects structure for subject and word.
14
!
If the maximal model converged, we used PCA to check whether this number of dimensions was
15
!
supported by the data. If the PCA showed that the number of dimensions was not supported we
16
!
proceeded with dropping smallest variance components. If the maximal model failed, we dropped
17
!
variance components until the identification was achieved. As a result, random intercepts and
18
!
random by-consonant slopes for subjects were included as random effects. The models'
19
!
assumptions were checked by visual inspection of the residual plots. Outliers were checked
20
!
individually and either removed (in case of experimental errors) or corrected (in case of
21
!
processing errors). Removing outliers did not result in any changes in outcome pattern.
22
!
The p-values were corrected to account for multiple comparisons following the truncated
23
!
closed test procedure from Westfall (1997) as implemented in the glht function of "multcomp"
24
!
package (Hothorn, Bretz, Westfall, & Heiberger, 2008). All pairwise comparisons for the
25
!
PEAKX_V50: COHORT were obtained by manually setting the contrast matrix.
26
!
Testing for consonant-specific effects on CD across age groups
27
!
In the second step, to test whether CD for a specific consonant differed between age groups,
28
!
we fitted a linear mixed model for each consonant with PEAKX_C1_050 as response variable and
29
!
PEAKX_V50 and COHORT as well as interaction thereof as fixed effects. PEAKX_C1_050 was
30
!
power transformed to better approximate normality using “BoxCox” function from R package
31
!
“forecast” (Hyndman, 2017). Cohort was a five-level factor for all stops’ models and a three-level
32
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
13!
factor for the alveolar fricative model, with treatment coding, and C3 as baseline. For all the five
1
!
models, the random effects explored consisted in the full random effects structure for subject and
2
!
word. The resulting random effects structure in all cases included by-subject random intercepts
3
!
and slopes for PEAKX_V50.
4
!
Testing consonant specific effects on CD within age groups
5
!
Third, we fitted linear mixed effects models to statistically compare the effect of onset
6
!
consonant identity on CD within each age group. All five models included PEAKX_C1_050 as
7
!
response variable and PEAKX_V50, CONSONANT1 and their interaction (PEAKX_V50:
8
!
CONSONANT1) as fixed effects. PEAKX_C1_050 was again power transformed with the
9
!
“BoxCox” function from the R package “forecast” (Hyndman, 2017). For the 3- and 5-year-olds,
10
!
consonant was a three level categorical predictor (b, d, g). For 4- and 7-year-olds, as well as adults
11
!
the consonant was a four level predictor (b, d, g, z). We used dummy coding with /b/ as baseline.
12
!
The structure of random effects for each model was determined following the strategy described
13
!
above. For each of the five models we began by testing the full random effects structure for
14
!
subject and word. We followed the same approach as above (cf. Testing for differences in CD).
15
!
All pairwise comparisons for the PEAKX_V50: CONSONANT1 were obtained by manually
16
!
setting the contrasts matrix. As separate models were fitted for each cohort, results are only
17
!
comparable within cohort.
18
!
All statistical analyses were carried out in R (Version 3.4.0, R Core Team, 2017). In the next
19
!
section, we present the results of each statistical analysis as output of pairwise comparisons with a
20
!
short descriptor of the comparisons, the effect estimates with associated standard errors (SE), the
21
!
test statistics (t-value) and multiplicity adjusted p-values. The +/- signs of the estimates determine
22
!
the direction of the effect. To take an example, the negative estimate comparing coarticulation
23
!
degree between 4 and 3-year-olds (noted as 4-3 in Table 1) means the 4-year olds show a lower
24
!
slope as compared to the 3-year olds.
25
!
26
!
RESULTS
27
!
Developmental differences in CD
28
!
Before addressing the question of consonant-specific effects on CD, we first evaluated
29
!
whether overall CD differences could be observed across the five cohorts investigated in this
30
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
14!
study. Table 1 reports the results from the linear mixed effects model comparing the effect of the
1
!
horizontal tongue body position at vowel midpoint (V50) on the horizontal tongue body position
2
!
at consonant midpoint (C50) between different age groups (adjusted p-values). Results are pooled
3
!
across consonants and do not include the alveolar fricative /z/ as this consonant was not collected
4
!
in all age groups.
5
!
TABLE 1
6
!
7
!
Group comparisons for coarticulation degree
estimated between C50 and V50
Age
Estimates
p-value
4-3
-0.0182195
.6483
5-3
-0.0187760
.6483
G1-3
-0.0832647
<.001 ***
Adult-3
-0.2209096
<.001 ***
5-4
-0.0005565
.97733
G1-4
-0.0650452
.00139 **
Adult-4
-0.2026901
<.001 ***
G1-5
-0.0644888
.00139 **
Adult-5
-0.2021336
<.001 ***
Adult-G1
-0.1376449
<.001 ***
8
!
9
!
10
!
As can be noted in Table 1, significant differences in CD were found between all children
11
!
cohorts regardless of age and adults with overall greater coarticulation in children relative to
12
!
adults (p< .001). In addition, CD differed significantly between the oldest group of children at
13
!
grade 1 and the three younger groups of children at age 3 (p< .001), 4, or 5 years of age (p< .01).
14
!
However, CD differences between the three groups of younger children did not yield significance.
15
!
16
!
17
!
Consonant specific effects on CD across groups
18
!
Second, we tested for age-related differences in CD as a function of onset consonant
19
!
identity. Table 2 reports results from the linear mixed-effects models comparing CD differences
20
!
across the five age cohorts investigated. Here again, we observed substantial differences between
21
!
adults and children.
22
!
23
!
TABLE 2
24
!
25
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
15!
1
!
2
!
3
!
Consonant effects on the degree of coarticulation
between C50 and V50 across age group
cons
Age
Estimates
p-value
b
4-3
-0.0103
.404
5-3
-0.0101
.404
G1-3
-0.0247
.011 *
Adult-3
-0.0736
<.001 ***
5-4
0.0002
.979
G1-4
-0.0144
.130
Adult-4
-0.0633
<.001 ***
G1-5
-0.0146
.130
Adult-5
-0.0635
<.001 ***
Adult- G1
-0.0489
<.001 ***
d
4-3
0.0240
.96175
5-3
-0.0166
.96175
G1-3
-0.1120
.79641
Adult-3
-0.4997
.00433 **
5-4
-0.0406
.96175
G1-4
-0.1360
.79641
Adult-4
-0.5237
.00320 **
G1-5
-0.0954
.79641
Adult-5
-0.4831
.00433 **
Adult- G1
-0.3877
.02631 *
g
4-3
-0.2019
.0632 .
5-3
-0.0802
.3853
G1-3
-0.3345
<.001 ***
Adult-3
-0.5393
<.001 ***
5-4
0.1217
.2336
G1-4
-0.1326
.2336
Adult-4
-0.3374
<.001 ***
G1-5
-0.2543
.0121 *
Adult-5
-0.4591
<.001 ***
Adult- G1
-0.2048
.0486 *
z
G1-4
-1.6770
.486
Adult-4
-2.9100
.430
Adult- G1
-1.2340
.597
4
!
In the case of labial-V coarticulation, all child cohorts differed significantly from adults (p<
5
!
.001). In addition, 3-year-olds differed from grade 1 children (p< .05). Velar-V coarticulatory
6
!
patterns also showed greater CD in each child cohort compared to adults (p< .001 with the 3; 4; 5
7
!
year-olds but p< .05 with grade 1 children). Further, for velar-V sequences, both the 3 and 5-year-
8
!
olds showed greater CD than children at grade 1 (p< .001 and p< .05 respectively). For the
9
!
alveolar stop /d/, CD was significantly greater in children relative to adults (p< .01 for the 3; 4; 5
10
!
year-olds and p< .05 for grade 1). However, this time no difference was found between the 3-year-
11
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
16!
olds and children at grade 1. Finally, as regards the alveolar fricative /z/, no difference in CD
1
!
between the 4-year-olds, the grade 1 children and adults was found.
2
!
Consonant specific effects on CD within groups
3
!
Table 3 presents results from linear mixed effects models assessing the effect of onset consonant
4
!
identity on CD pattern within each age group.
5
!
6
!
TABLE 3
7
!
8
!
9
!
10
!
11
!
12
!
13
!
14
!
15
!
16
!
17
!
18
!
19
!
20
!
21
!
22
!
23
!
24
!
25
!
Results show a similar decreasing degree of coarticulation for all cohorts from labial, to velar,
26
!
followed by the alveolar stop consonant. With respect to the alveolar fricative and stop, the 4-
27
!
years-olds as well as the grade 1 children showed a similar direction of effects with greater CD in
28
!
the fricative compared to the stop context. However, this difference was not significant in the 4-
29
!
years-old children contrary to older children at the end of grade 1 (p< .05). Finally, neither the 4-
30
!
Consonant-specific effects on the degree of coarticulation
between C50 and V50 within age group
Cons
Estimates
SE
t-value
p-value
Age 3
d-b
g-b
g-d
-0.0274 0.0029 -9.498 <.001 ***
-0.0149 0.0026 -5.786 <.001 ***
0.0125 0.0030 4.201 <.001 ***
Age 4
d-b
g-b
z-b
g-d
z-d
z-g
-0.1310 0.0152 -8.633 <.0001 ***
-0.1197 0.0136 -8.779 <.0001 ***
-0.1224 0.0156 -7.842 <.001 ***
0.0112 0.0166 0.678 .776
0.0085 0.0177 0.481 .776
-0.0027 0.0167 -0.163 .870
Age 5
d-b
g-b
g-d
-0.2168 0.0208 -10.422 <.0001 ***
-0.1384 0.0203 -6.810 <.0001 ***
0.0784 0.0216 3.639 .000273 ***
Grade 1
d-b
g-b
z-b
g-d
z-d
z-g
-8.6322 0.7235 -11.931 <.001 ***
-6.7567 0.6846 -9.870 <.001 ***
-6.4479 0.7885 -8.177 <.001 ***
1.8755 0.7643 2.454 .0272 *
2.1843 0.8489 2.573 .0272 *
0.3088 0.8171 0.378 .7055
Adults
d-b
g-b
z-b
g-d
z-d
z-g
-1.3134 0.1153 -11.390 <.001 ***
-0.6636 0.1086 -6.113 <.001 ***
-0.9059 0.1127 -8.041 <.001 ***
0.6497 0.1249 5.202 <.001 ***
0.4074 0.1299 3.135 .00172 **
-0.2423 0.1227 -1.974 .04842 *
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
17!
years-olds nor the grade 1 children showed a difference between the velar /g/ and fricative /z/
1
!
while adults did.
2
!
3
!
DISCUSSION
4
!
The present study investigated the maturation of intra-syllabic coarticulatory organization in
5
!
typically developing German children from 3 to 7 years of age. Investigation of coarticulation in
6
!
this period is not only relevant for understanding the maturation of spoken language fluency or
7
!
speech motor control, it also provides normative data necessary for disentangling atypical
8
!
trajectories from the typical variability observed in the young age (e.g., in developmental apraxia
9
!
of speech: Nijland et al., 2002; Nijland et al., 2003; stuttering children: Soo-Eun, et al., 2002;
10
!
speech sound disorders: Cleland, Scobbie, & Wrench, 2015; phonological disorders: Gibbon,
11
!
1999; hearing impairment: Bernhardt, Gick, & Bacsfalvi, 2005). Indeed, if we understand well
12
!
how coarticulatory mechanisms mature in typically developing children, we should be better
13
!
equipped to detect deviations from typical trajectories early in the child’s language development.
14
!
In light of previous research, we were specifically interested in the following questions.
15
!
Does intra-syllabic lingual coarticulation in German children differ between children and adults?
16
!
If so, is this difference observed across consonants? Finally, does CD vary based on the onset
17
!
consonants’ identity within children, as it does in German adults (Iskarous et al., 2013; Abakarova
18
!
et al., 2017)? In the next two sections, results are discussed with respect to the two overarching
19
!
themes they address, namely: the general ontogeny of coarticulation and its relations to speech
20
!
motor control development, its implication for ongoing discussions about the units of speech
21
!
organization. Finally, we discuss the limitations and perspectives of the study.
22
!
Age-related changes in coarticulation and speech motor control development
23
!
The challenging question of whether children organize their speech in segments versus
24
!
syllables versus phonological words or lexical items is two-fold: it requires finding the
25
!
phonological units guiding children’s speech production as well as the motor units embedding
26
!
those higher-level units. From a motor perspective, a main goal for children is to learn to
27
!
discretize continuous articulatory movements into distinct articulatory gestures conveying the
28
!
specific phonetic properties of their native language. In that perspective, coarticulation is an
29
!
interesting mechanism to investigate as it offers a window into understanding how speech motor
30
!
control develops with age and experience with the native language. In this study, we could not
31
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
18!
assess the direct role of individual experience; however, by employing a cross-sectional design
1
!
with four tightly grouped cohorts of German children spreading the kindergarten period (at an
2
!
average of 3; 4; 5 years of age) and the beginning of primary school (at 7 years of age) we could
3
!
estimate the apparent changes in CD patterns across childhood.
4
!
5
!
First, as predicted, results indicate that lingual patterns in all children groups differ from
6
!
adults with greater CD in children relative to adults. In addition, we found that the three youngest
7
!
groups of children in kindergarten differed from older children at the end of grade 1. This suggests
8
!
that the period between the end of kindergarten and the end of the first year into primary school
9
!
corresponds to an important transition with respect to children´s phonological and speech motor
10
!
control development, that is children have decreased the spatio-temporal overlap for consonant-
11
!
vowel articulatory gestures. However, by the end of the first school year, German children’s CD
12
!
patterns are still greater than adults’. The maturation of lingual coarticulation is still ongoing.
13
!
Second, we found age-related differences in CD across the three consonantal contexts (/b/,
14
!
/d/ and /g/) in the direction of greater CD in children than adults. This suggests that vocalic
15
!
gestures invade the temporal domain of the syllable more than in adults. Interestingly, in labial
16
!
context, 3-year-olds differed from children at grade 1. Given that jaw patterns are controlled the
17
!
earliest in age (e.g., review in Green et al., 2002), that labials are dominant in the babbling
18
!
repertoire (e.g., in German: Fox-Boyer, 2006; 2009), one may expect CD in syllables involving
19
!
labials to approximate adults’ CD patterns early in age. Yet, our results do not support this
20
!
hypothesis. In fact, they show that at the age of 7, children have certainly more mature lingual CD
21
!
patterns than at 3 but they still differ significantly from adults as to the phasing between two
22
!
articulatory gestures: a well-practiced motor routine involving the jaw (with support of the lips)
23
!
for the labial consonant, another gesture involving the tongue for the achievement of the vowel.
24
!
This result somehow contradicts previous reports that by age 6, children’s coupling of the jaw and
25
!
lips approximates those of adults (e.g., Green et al., 2002). This may be due to the fact that in their
26
!
study, the authors only examined spontaneous productions of /baba/ sequences that are
27
!
prototypical in the babbling and first words of children while we examined /b/-V syllables
28
!
involving various vocalic contexts in a repetition task. In our study, age-related differences were
29
!
also found for the velar stop /g/ with greater vowel-related influence upon CD in all groups of
30
!
children compared to adults and in the 3 and 5-year-olds compared to older children at 7. Hence,
31
!
in homorganic syllables involving a single organ for the production of both consonantal and
32
!
vocalic gestures such as in /g/-V, results suggest that at 7, children are on their way to achieve
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
19!
adult-like CD patterns but they are not quite yet like adults. For the alveolar stop /d/, age-related
1
!
differences between the groups of children disappear; the only difference in CD pattern was noted
2
!
between children and adults. Finally, results for the fricative alveolar /z/ did not reveal any
3
!
significant difference within the groups of children or as compared to adults. This result does not
4
!
corroborate previous reports (e.g., in German based on measures of spectral center of gravity in 4
5
!
to 6-year-old children: Kleber, 2015). Discrepancy in results may be due to methodological
6
!
differences between acoustical and articulatory studies along with the age and particular choice of
7
!
stimuli investigated.
8
!
However, our final observation, that children displayed consonantal effects on CD patterns
9
!
within age groups provides further details on children´s speech motor control and most
10
!
specifically on the absence of age-related difference for the alveolar fricative. We indeed found
11
!
greater CD in the context of the labial stop /b/, followed by the velar /g/ then the alveolar stop /d/.
12
!
As briefly outlined in the introduction, a main explanation for these differences in CD stems from
13
!
consonants’ degree of flexibility versus resistance to overlap with neighboring vowels. While
14
!
some consonants allow sizeable coarticulation with neighbors (e.g., labials), others make strong
15
!
articulatory demands on the tongue to preserve intelligibility (e.g., alveolars) and therefore resist
16
!
blending with adjacent vowels (e.g., Fowler, 1994; Fowler & Brancazio, 2001; Recasens, 1985;
17
!
Recasens & Espinosa, 2009). Previous work has shown that during the alveolar stop /d/, the
18
!
tongue body and tongue tip function as a collaborative network (or functional synergy) so that the
19
!
base of the tongue moves the front to support the anterior part of the tongue into achieving the
20
!
alveolar occlusion (e.g., Iskarous et al., 2010; Iskarous et al., 2011). Results from the present study
21
!
seem to support the hypothesis that children as young as three years of age display synergistic
22
!
relationships among the functional subparts of the tongue to achieve the alveolar constriction (e.g.,
23
!
for /d/). However, the significant differences in CD noted between children and adults indicate
24
!
that these synergies are not yet mature by the end of grade 1. Such discrepancy is also exemplified
25
!
by the finding that both 4-year-olds and children at the end of grade 1 did not show any difference
26
!
in CD between the velar /g/ and alveolar fricative /z/ while adults did. The horizontal position of
27
!
the tongue body in syllables including /z/ was affected by the subsequent vocalic gesture as much
28
!
as in velar context. This suggests that the articulation of alveolar fricatives remains challenging
29
!
throughout childhood as exemplified in other studies (e.g., in German: Fox-Boyer, 2009; in
30
!
English: review in Maas & Mailend, 2017). Beyond the positioning of the tongue tip and tongue
31
!
blade in the alveolar region, fricatives require fine glottal control and therefore precise
32
!
coordination between the supralaryngeal and laryngeal areas which is not yet mastered at 10 (e.g.,
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
20!
Koenig et al., 2008). The observations made for the two alveolars in the present study are
1
!
interesting but evidently preliminary, and call for more detailed examinations of lingual synergies
2
!
for individual articulatory gestures. In future studies, other developmental factors such as
3
!
anatomical growth of the vocal tract should also be carefully considered.
4
!
5
!
Taken together, our findings indicate that by the end of their first year in primary school,
6
!
children have substantially changed the organization of their lingual coarticulatory patterns in
7
!
comparison to the kindergarten period but they do no yet approximate adults’ patterns. This
8
!
suggests that the process of learning to coordinate the spatio-temporal properties of articulatory
9
!
gestures to produce adult-like patterns of intra-syllabic coarticulation degree is protracted and not
10
!
uniform across childhood (e.g., Smith, 2010) as found for other motor behaviors (e.g., Thelen &
11
!
Smith, 1994). The present study provided a first account in German. More research is obviously
12
!
needed to assess the age at which children transition towards adult-like patterns of lingual
13
!
coarticulation, and to fully understand the complexities underlying the maturation of intra and
14
!
inter-organ coordination.
15
!
16
!
Units of intra-syllabic organization: Segment? Syllable? Articulatory gesture?
17
!
An overarching motivation to conduct this research stemmed from the conflicting reports
18
!
found in the literature as to whether intra-syllabic coarticulation is organized in children as a
19
!
single chunk with a large vocalic influence observed throughout the syllable (e.g., Goodell, &
20
!
Studdert-Kennedy, 1993; Nijland et al. 2002; Nittrouer & Whalen, 1989; Rubertus et al., 2015) or
21
!
in a more sequential manner with minimal overlap between both segments (Gibson & Ohde, 2007;
22
!
Green, et al., 2002; Kent, 1983; Zharkova et al., 2012). Examinations of consonant-related effects
23
!
on CD across and within age groups provide new insight on this question.
24
!
In our adult reference group, the domain of coarticulatory organization varied as a function
25
!
of the onset consonant identity, which suggests that intra-syllabic organization is not uniform
26
!
across CV syllables but instead depends on the articulatory demands associated with consecutive
27
!
phonemes. In German children, results overall corroborate previous research arguing for a
28
!
broader-size organization!in which the vowel exerts more influence on the previous consonant
29
!
relative to adults. This general finding supports a holistic perspective by which children start with
30
!
a word-based or syllable-based unit of speech organization and progressively reduce their domain
31
!
of coarticulatory organization as they learn to abstract phonemes from the words and syllables
32
!
they are embedded as well as their associated articulatory gestures (e.g., Nittrouer et al., 1996).
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
21!
However, the fact that we observe consistent consonantal effects on CD within age groups is a
1
!
strong indication that at the age of 3, children have departed from a purely holistic organization to
2
!
integrate articulatory gestures associated with individual segments.
3
!
Results suggest that children do not coarticulate consonants and vowels with an unvarying
4
!
degree of overlap irrespective of the segments combined (e.g., Gibson & Ohde, 2007; Katz &
5
!
Bharadwaj, 2001; Noiray et al., 2013; Zharkova et al., 2015). In fact, results show that the degree
6
!
with which lingual vocalic gestures invade the temporal domain of previous consonants depends
7
!
on whether the production of both the consonant and vowel involves the tongue. Taken together,
8
!
previous research as well as present findings suggest a crucial step in becoming mature speakers
9
!
may not be for children to globally increase or decrease coarticulation but to achieve flexible
10
!
patterns of coarticulation depending on the combination of segments. This hypothesis is in line
11
!
with the theory of Articulatory Phonology (Browman & Goldstein, 1992 and sub.), which argues
12
!
that articulatory (or constriction) goals represent dynamic units of action guiding the activity of
13
!
speech articulators. From a developmental standpoint, Articulatory Phonology provides an
14
!
interesting framework to explain differences in intra-syllabic coarticulatory organization, which
15
!
departs from traditional phonological descriptions. In the first years of language acquisition,
16
!
children have minimal structural knowledge about their native language, that is they have limited
17
!
awareness that words can be decomposed into segments and recombined into new meaningful
18
!
forms (e.g., Gillon, 2007; Liberman et al., 1974). Yet, they are able to concatenate sounds into an
19
!
intelligible flow to communicate with the world. Hence, one reason for the unresolved controversy
20
!
over children’s units of speech production may stem from the premise that children’s productions
21
!
can be described in terms of adults’ categories (e.g., segments, syllables). However, up to date,
22
!
there hasn’t been any strong evidence showing that children organize their speech in categories
23
!
similar to those used to describe adults’ productions. Instead, children may organize their speech
24
!
in very different units - that are neither segments nor syllables - that researchers fail to capture
25
!
because of the inadequacy of the methodologies employed focusing on finding adult-based units
26
!
or because of the limitations in the linguistic material tested (often one or two contrasts). Results
27
!
from adult studies suggest that even adults do not organize their speech along traditional
28
!
phonological lines but rather in terms of articulatory goals that are dynamically organized to
29
!
integrate contextual variability (e.g., Browman & Goldstein, 1992; Iskarous et al., 2011; Recasens
30
!
& Espinosa, 2009).
31
!
Results from this cross-sectional investigation support the hypothesis of an articulatory basis
32
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
22!
for speech organization in preschool and school-aged children. They further support the view that
1
!
CD varies within a continuum with gradient distinctions resulting from the coproduction of
2
!
articulatory gestures. When a consonant imposes strong demands on an articulator that is shared
3
!
with its neighboring vowel, coarticulatory resistance leads towards a more phone-sized
4
!
organization. When no such constraint exists between consecutive phonemes, speakers may
5
!
employ a broader domain of articulation. More in-depth investigations are evidently needed to
6
!
uncover the impact of articulatory goals in spoken language organization. Fortuitously, in the last
7
!
decade, optimization of ultrasound imaging for child studies has allowed for a direct and user-
8
!
friendly access to children’s articulation. More recently, a platform combining a talking head with
9
!
ultrasound images has been developed (e.g., Fabre, Hueber, Girin, Alameda-Pineda, & Badin,
10
!
2017), which provides new opportunities for advancing our understanding of early language
11
!
organization in typical and atypical populations beyond acoustic analyses or standard assessment
12
!
of articulatory capabilities. In a new project, we have also started examining the maturation of
13
!
coarticulatory organization in relation to speech motor control development in greater details as
14
!
well as the impact of phonemic awareness on coarticulatory (re)organization in late childhood.
15
!
16
!
LIMITATION AND PERSPECTIVE OF THE STUDY
17
!
18
!
In this study, we were interested in consonant-related effects on coarticulatory organization
19
!
in children and in the phenomenon of consonants resistance across a range of vowels. We
20
!
therefore conducted analyses across vocalic contexts. Other studies have investigated the specific
21
!
role of individual vowels upon lingual organization (e.g., in Catalan adults: Rodriguez &
22
!
Recasens, 2016). Future studies should expand on such examinations to disentangle consonant-
23
!
related as well as vowel-related effects on children’s coarticulatory patterns as compared to adults.
24
!
25
!
Another important aspect to discuss regards the experimental approach endorsed in this
26
!
study. Controlling for head movement has been a longstanding challenge for psycholinguists
27
!
aiming to collect data from children´s speech articulators. Each approach (fixed headset, hand-
28
!
held, microphone-stand, probe holder) presents certain advantages and drawbacks we briefly
29
!
outline below. Using a fixed headset prevents the head from moving with respect to the jaw and
30
!
therefore makes comparisons across tongue curves more reliable than when the ultrasound probe
31
!
is free to move. However, this method has also issues, which were outlined in the method section.
32
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
23!
Importantly, fixed headsets block the jaw and therefore require children to employ articulatory
1
!
strategies that may not reflect their actual speech. Given previous research showing the
2
!
importance of the jaw in early childhood (e.g., Green et al., 2002; Grigos, 2009; Smith &
3
!
Zelaznik, 2004) and its use in adults for vowel contrast (e.g., Noiray, Iskarous, & Whalen, 2014),
4
!
it appeared fundamental for us to not prevent speakers from moving their jaw freely, especially in
5
!
the young age. Some studies have used microphone stands to attach ultrasound probes (e.g.,
6
!
Noiray et al., 2013). While this device does not block jaw motion, it encourages the participant to
7
!
move his head up to compromise for the impossibility to move the jaw down. Finally, other
8
!
studies have used a hand-held probe approach (e.g., Zharkova et al., 2015) and reported on
9
!
noticeable differences with the headset approach. Such difference may result from the hand sliding
10
!
in the lateral and horizontal dimensions as well as from inconsistent contact with the chin floor
11
!
affecting the probe’s vertical dimension and more generally the image quality (e.g., shorter tongue
12
!
curve, blurry tongue curve line).
13
!
Our customized probe-holder also presents advantages and limitations. It was designed to
14
!
maximize naturalness of speech while controlling for probe motion (see Method section for
15
!
details). While this set-up certainly succeeds at not impeding natural motor movements from the
16
!
face, it may affect the reliability of tongue curvature comparisons because of a physically induced
17
!
misalignment of tongue shape rather than linguistically induced distinctions. We used various
18
!
strategies to address this limitation (e.g., seatbelts, two experimenters, an experimenter and a star
19
!
as a visual fixation point in front of the child) and conducted qualitative examinations of the
20
!
tongue contour using video data post-recording to discard data for which children moved as in
21
!
other studies (e.g., Zharkova et al., 2017). Of course, qualitative examinations remain somewhat
22
!
subjective but we remain confident that with the quantitative dataset we collected, our results
23
!
bring insightful knowledge to the field.
24
!
25
!
For our analyses, we used the highest point on the tongue body as used in previous research
26
!
(e.g., Noiray et al., 2013). This approach allowed us to depart from the end curves of the tongue
27
!
that can be inaccurate regardless of the recording (e.g., headset, microphone stand, probe holder)
28
!
or analysis approach (e.g., automatic tongue contour detection, manual detection). Other
29
!
approaches exist such as examining the whole tongue shape to account for coarticulatory
30
!
differences (e.g., Zharkova et al., 2015). The reliability of this method may be affected by the
31
!
quality of the tongue imaging at the two ends of the tongue curve across tokens and speakers but it
32
!
provides insightful results as to the tongue bunching. Hence, both approaches provide
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
24!
complementary information on the maturation of articulatory organization in the young age.
1
!
Various parameters may affect the quality of tongue imaging irrespective of the method
2
!
employed. The experimenter must decide on the method that best suits the population and
3
!
theoretical questions addressed. Considering the limitations outlined above, future studies should
4
!
prioritize large-scale descriptions across childhood employing the same experimental and
5
!
analytical approaches for all subjects. Including comparative data with adults is also essential to
6
!
have a reference regarding the maturity of the production patterns investigated. Most quantitative
7
!
analyses of lingual coarticulation in children (including ours) do not provide absolute values easily
8
!
comparable to the adult literature but rather compare numerical values for one age cohort relative
9
!
to the others within-study.
10
!
11
!
CONCLUSION
12
!
This study addressed the maturation of lingual coarticulatory patterns in typically
13
!
developing German children in comparison to adults. Results show that already at age 3,
14
!
differences in the degree of intra-syllabic coarticulation correlate with differences in phonetic
15
!
properties of the onset consonant. Overall, results provide evidence that German children’s speech
16
!
is not uniformly organized along abstract phonological lines, in either segments or syllables as
17
!
often argued in the literature (at least for English learning children). Instead, coarticulatory
18
!
organization is sensitive to the articulatory properties associated with individual segments and
19
!
their compatibility once combined into a continuous speech flow. This result replicates previous
20
!
findings in adults and older children and expands it to German children from 3 to 7. This is an
21
!
important result because it not only concurs to clarify typical maturation of coarticulatory
22
!
mechanism; it could potentially supplement assessment of atypical production patterns. Future
23
!
research should further test the role of lingual gestures for coarticulatory organization in typically
24
!
developing children and examine whether articulatory demands observed for consecutive
25
!
articulatory gestures may impede acquisition of spoken language fluency.
26
!
27
!
ACKNOWLEDGMENTS
28
!
This research has been supported by the DFG (1098 and 255676067). We are grateful to
29
!
Carol Fowler for stimulating discussions at various stages of this research and to Martijn Wieling
30
!
for his statistical insights. We also thank the three reviewers for their helpful feedback. Our
31
!
gratitude to Jan Ries (University of Potsdam) for his assistance in developing Matlab scripts, to
32
!
the Baby-Lab at University of Potsdam (in particular to Barbara Hoehle and Tom Fritzsche) for
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
25!
helping us with participants’ recruitment, and the team at Laboratory for Oral Language
1
!
Acquisition (LOLA) involved in data recording and processing. Importantly, we thank all the
2
!
participants, adults and children without whom this research would have never been possible.
3
!
4
!
5
!
REFERENCES
6
!
Abakarova, D., Iskarous, & Noiray, A. (2017). Quantification of coarticulation resistance in
7
!
German with ultrasound. In Speech Motor Control, Groningen.
8
!
9
!
ASHA (2007). Childhood Apraxia of Speech [technical report]. American Speech-Language-
10
!
Hearing Association. Available from www.asha.org/policy.
11
!
Barbier, G., Perrier, P., Ménard, L., Payan, Y., Tiede, M., & Perkell, J. (2015). Speech planning in
12
!
4-year-old children versus adults: Acoustic and articulatory analyses. In 16th Annual
13
!
Conference of the International Speech Communication Association, Dresden.
14
!
15
!
Bates, D., Kliegl, R., Vasishth, S., & Baayen, H. (2015). Parsimonious mixed models. arXiv
16
!
preprint arXiv:1506.04967.
17
!
18
!
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2014). Fitting linear mixed-effects models using
19
!
lme4. arXiv preprint arXiv:1406.5823.
20
!
21
!
Bernhardt, B., Gick, B., Bacsfalvi, P., & AdlerBock, M. (2005). Ultrasound in speech therapy
22
!
with adolescents and adults. Clinical Linguistics & Phonetics, 19(6-7), 605-617.
23
!
24
!
Bladon, R. A., & Al-Bamerni, A. (1976). Coarticulation resistance in English /l/. Journal of
25
!
Phonetics, 4, 135–150.
26
!
27
!
Boersma, P., & Weenink, D. (1996). Praat, a system for doing phonetics by computer, version 3.4,
28
!
Report 132, Institute of Phonetic Sciences of the University of Amsterdam, 1-182.
29
!
30
!
Browman, C. P., & Goldstein, L. (1992). Articulatory phonology: An overview. Phonetica, 49(3-
31
!
4), 155-180.
32
!
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
26!
Cheng, H. Y., Murdoch, B. E., Goozée, J. V., & Scott, D. (2007). Physiologic development of
1
!
tongue–jaw coordination from childhood to adulthood. Journal of Speech, Language, and
2
!
Hearing Research, 50(2), 352-360.
3
!
4
!
Cleland, J., Scobbie, J. M., & Wrench, A. A. (2015). Using ultrasound visual biofeedback to treat
5
!
persistent primary speech sound disorders. Clinical linguistics & phonetics, 29(8-10), 575-
6
!
597.
7
!
8
!
Fabre, D., Hueber, T., Girin, L., Alameda-Pineda, X., & Badin, P. (2017). Automatic Animation
9
!
of an Articulatory Tongue Model from Ultrasound Images of the Vocal Tract. Speech
10
!
Communication, 93, 63-75.
11
!
12
!
Fowler, C. A. (1994). Invariants, specifiers, cues: An investigation of locus equations as
13
!
information for place of articulation. Attention, Perception, & Psychophysics, 55(6), 597-
14
!
610.
15
!
16
!
Fowler, C. A., & Brancazio, L. (2000). Coarticulation resistance of American English consonants
17
!
and its effects on transconsonantal vowel-to-vowel coarticulation. Language and Speech,
18
!
43(1), 1-41.
19
!
Fox-Boyer, A. (2009). Kindliche Aussprachestörungen: phonologischer Erwerb,
20
!
Differenzialdiagnostik, Therapie. Idstein: Schulz-Kirchner.
21
!
Fox-Boyer, A. (2006). Evidence from German-speaking children. Phonological Development and
22
!
Disorders in Children. In: Hua, Z./Dodd, B.(Hgg.) A Multilingual Perspective. Clevedon,
23
!
Buffalo, Toronto: Multilingual Matters LTD, 56-80.
24
!
25
!
Gibbon, F. E. (1999). Undifferentiated lingual gestures in children with articulation/phonological
26
!
disorders. Journal of Speech, Language, and Hearing Research, 42(2), 382-397.
27
!
28
!
Gibson, T., & Ohde, R. N. (2007). F2 locus equations: Phonetic descriptors of coarticulation in
29
!
17-to 22-month-old children. Journal of Speech, Language, and Hearing Research, 50(1),
30
!
97-108.
31
!
32
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
27!
Gillon, G. T. (2007). Phonological awareness: From research to practice. Guilford Press.
1
!
2
!
Graetzer, S. (2006). Consonantal coarticulation resistance in vowel-consonant-vowel sequences in
3
!
two Australian languages. In Proceedings of the 11th Australasian International Conference
4
!
on Speech Science and Technology,1(1), 270-275.
5
!
6
!
Green, J. R., Moore, C. A., & Reilly, K. J. (2002). The sequential development of jaw and lip
7
!
control for speech. Journal of Speech, Language, and Hearing Research, 45(1), 66-79.
8
!
9
!
Grigos, M. I. (2009). Changes in articulator movement variability during phonemic development:
10
!
a longitudinal study. Journal of Speech, Language, and Hearing Research, 52(1), 164-177.
11
!
12
!
Hardcastle, W. J., & Hewlett, N. (2006). Coarticulation: Theory, data and techniques. Cambridge
13
!
University Press.
14
!
15
!
Hothorn, T., Bretz, F., Westfall, P., & Heiberger, R. M. (2008). Multcomp: simultaneous
16
!
inference in general parametric models. R package version 1.0-0. R Foundation for
17
!
Statistical Computing. Vienna.
18
!
19
!
Hyndman R. J. (2017). Forecast: Forecasting functions for time series and linear models. R
20
!
package version 8.1, http://github.com/robjhyndman/forecast.
21
!
22
!
Iskarous, K., Mooshammer, C., Hoole, P., Recasens, D., Shadle, C. H., Saltzman, E., & Whalen,
23
!
D. H. (2013). The coarticulation/invariance scale: Mutual information as a measure of
24
!
coarticulation resistance, motor synergy, and articulatory invariance. The Journal of the
25
!
Acoustical Society of America, 134(2), 1271-1282.
26
!
27
!
Iskarous, K., Shadle, C. H., & Proctor, M. I. (2011). Articulatory–acoustic kinematics: The
28
!
production of American English/s. The Journal of the Acoustical Society of America, 129(2),
29
!
944-954.
30
!
31
!
Iskarous, K., Fowler, C. A., & Whalen, D. H. (2010). Locus equations are an acoustic expression
32
!
of articulator synergy. The Journal of the Acoustical Society of America, 128(4), 2021-2032.
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
28!
1
!
Katz, W. F., & Bharadwaj, S. (2001). Coarticulation in fricative-vowel syllables produced by
2
!
children and adults: A preliminary report. Clinical linguistics & phonetics, 15(1-2), 139-
3
!
143.
4
!
5
!
Kent, R. D. (1983). The segmental organization of speech. In The production of speech (57-89).
6
!
Springer New York.
7
!
8
!
Kisler, T., Schiel, F., & Sloetjes, H. (2012, July). Signal processing via web services: the use case
9
!
WebMAUS. In Digital Humanities Conference 2012.
10
!
11
!
Kleber, F. (2015). Direction of coarticulation in vowel-fricative sequences in L1-German children.
12
!
In Proceedings of the International Child Phonology Conference 2015, St. John's.
13
!
14
!
Klein, H. B., Byun, T. M., Davidson, L., & Grigos, M. I. (2013). A multidimensional investigation
15
!
of children's/r/productions: Perceptual, ultrasound, and acoustic measures. American
16
!
Journal of Speech-Language Pathology, 22(3), 540-553.
17
!
18
!
Koenig, L. L., Lucero, J. C., & Perlman, E. (2008). Speech production variability in fricatives of
19
!
children and adults: Results of functional data analysis. The Journal of the Acoustical
20
!
Society of America, 124(5), 3158-3170.
21
!
22
!
Liberman, I. Y., Shankweiler, D., Fischer, F. W., & Carter, B. (1974). Explicit syllable and
23
!
phoneme segmentation in the young child. Journal of experimental child psychology, 18(2),
24
!
201-212.
25
!
26
!
Lindblom, B., & Sussman, H. M. (2012). Dissecting coarticulation: How locus equations happen.
27
!
Journal of phonetics, 40(1), 1-19.
28
!
29
!
Maas, E., & Mailend, M. L. (2017). Fricative contrast and coarticulation in children with and
30
!
without speech sound disorders. American journal of speech-language
31
!
pathology, 26(2S), 649-663.
32
!
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
29!
Ménard, L., & Noiray, A. (2011). The development of lingual gestures in speech: Comparing
1
!
synthesized vocal tracts with natural vowels. Faits de langue, 37, 189-202.
2
!
3
!
Munson, B. (2004). Variability in/s/Production in Children and AdultsEvidence From Dynamic
4
!
Measures of Spectral Mean. Journal of Speech, Language, and Hearing Research, 47(1),
5
!
58-69.
6
!
7
!
Nittrouer, S., Studdert-Kennedy, M., & Neely, S. T. (1996). How children learn to organize their
8
!
speech gestures: Further evidence from fricative-vowel syllables. Journal of Speech,
9
!
Language, and Hearing Research, 39(2), 379-389.
10
!
11
!
Nittrouer, S. (1995). Children learn separate aspects of speech production at different rates:
12
!
Evidence from spectral moments. The Journal of the Acoustical Society of America, 97(1),
13
!
520-530.
14
!
15
!
Nittrouer, S. (1993). The emergence of mature gestural patterns is not uniform: Evidence from an
16
!
acoustic study. Journal of Speech, Language, and Hearing Research, 36(5), 959-972.
17
!
18
!
Nittrouer, S., & Whalen, D. H. (1989). The perceptual effects of child–adult differences in
19
!
fricativevowel coarticulation. The Journal of the Acoustical Society of America, 86(4),
20
!
1266-1276.
21
!
22
!
Noiray, A., Ries, J., & Tiede, M. (2015). Sonographic & Optical Linguo-Labial Articulation
23
!
Recording system (SOLLAR). Oral presentation at Ultrafest VII, Hong Kong.
24
!
25
!
Noiray A., Ménard L., & Iskarous K. (2013). The Development of Motor Synergies in Children:
26
!
Ultrasound and Acoustic Measurements. The Journal of the Acoustical Society of America,
27
!
133(1), 444-452.
28
!
29
!
Noiray, A., Iskarous, K., & Whalen, D. H. (2014). Variability in English vowels is comparable in
30
!
articulation and acoustics. Laboratory phonology, 5(2), 271-288.
31
!
32
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
30!
Noiray, A., Cathiard, M. A., Ménard, L., & Abry, C. (2011). Test of the movement expansion
1
!
model: Anticipatory vowel lip protrusion and constriction in French and English speakers.
2
!
The Journal of the Acoustical Society of America, 129(1), 340-349.
3
!
4
!
Noiray A., Cathiard M.A., Abry C. & Ménard L. & Savariaux, C. (2008). Emergence of a vowel
5
!
gesture control Attunement of the anticipatory rounding temporal pattern in French children.
6
!
In S. Kern, F. Gayraud, & E. Marsico (eds), Emergence of Linguistic Abilities. Cambridge
7
!
Scholars Publishing: New Castle, 100-116.
8
!
9
!
Prather, E., Hendrick, D., & Kern, C. (1975). Articulation development in children aged two to
10
!
four years. Journal of Speech and Hearing Disorders, 40, 179- 191.
11
!
12
!
Recasens, D. (1985). Coarticulatory patterns and degrees of coarticulatory resistance in Catalan
13
!
CV sequences. Language and Speech, 28(2), 97-114.
14
!
15
!
Recasens, D., & Espinosa, A. (2009). An articulatory investigation of lingual coarticulatory
16
!
resistance and aggressiveness for consonants and vowels in Catalan. The Journal of the
17
!
Acoustical Society of America, 125(4), 2288-2298.
18
!
19
!
Recasens, D., Fontdevila, J., & Pallarès, M. D. (1995). Velarization degree and coarticulatory
20
!
resistance for/I/in Catalan and German. Journal of Phonetics, 23(1-2), 37-52.
21
!
22
!
Reidy, P. F. (2015). The spectral dynamics of voiceless sibilant fricatives in English and
23
!
Japanese. The Ohio State University.
24
!
25
!
Repp, B. H. (1986). Some observations on the development of anticipatory coarticulation. The
26
!
Journal of the Acoustical Society of America, 79(5), 1616-1619.
27
!
28
!
Rubertus, E., Abakarova, D., Ries, J., & Noiray, A. Anticipatory V-to-V Coarticulation in German
29
!
Preschoolers. In C. Draxler & F. Kleber (eds.) Tagungsband der 12. Tagung Phonetik und
30
!
Phonologie im deutschsprachigen Raum.
31
!
32
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
31!
Sereno, J. A., Baum, S. R., Marean, G. C., & Lieberman, P. (1987). Acoustic analyses and
1
!
perceptual data on anticipatory labial coarticulation in adults and children. The Journal of
2
!
the Acoustical Society of America, 81(2), 512-519.
3
!
4
!
Smith, A. (2010). Development of neural control of orofacial movements for speech. In The
5
!
Handbook of Phonetic Sciences, Second Edition, 251-296.
6
!
7
!
Smith, A., & Zelaznik, H. N. (2004). Development of functional synergies for speech motor
8
!
coordination in childhood and adolescence. Developmental psychobiology, 45(1), 22-33.
9
!
10
!
Song, J. Y., Demuth, K., Shattuck-Hufnagel, S., & Ménard, L. (2013). The effects of
11
!
coarticulation and morphological complexity on the production of English coda clusters:
12
!
Acoustic and articulatory evidence from 2-year-olds and adults using ultrasound. Journal of
13
!
Phonetics, 41(3), 281-295.
14
!
15
!
Soo-Eun, C., Ohde, R. N., & Conture, E. G. (2002). Coarticulation and formant transition rate in
16
!
young children who stutter. Journal of Speech, Language, and Hearing Research, 45(4),
17
!
676.
18
!
19
!
Stoel-Gammon, C., & Dunn, C. (1985). Normal and disordered phonology in children. Austin,
20
!
Texas: pro.ed.
21
!
22
!
Sussman, H. M., Duder, C., Dalston, E., & Cacciatore, A. (1999). An Acoustic Analysis of the
23
!
Development of CV Coarticulation. A Case Study. Journal of Speech, Language, and
24
!
Hearing Research, 42(5), 1080-1096.
25
!
26
!
Sussman, H. M., Hoemeke, K. A., & Ahmed, F. S. (1993). A crosslinguistic investigation of
27
!
locus equations as a phonetic descriptor for place of articulation. The Journal of the
28
!
Acoustical Society of America, 94(3), 1256-1268.
29
!
30
!
Team, R. C. (2000). R language definition. Vienna, Austria: R foundation for statistical
31
!
computing.
32
!
33
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
32!
Terband, H. (2017). Deviant coarticulation in childhood apraxia of speech (CAS) does not include
1
!
hyperartion. In Speech Motor Control, Groningen, Netherlands.
2
!
3
!
Terband, H., Maassen, B., Van Lieshout, P. H. H. M., & Nijland, L. (2011). Stability and
4
!
composition of functional synergies for speech movements in children with developmental
5
!
speech disorders. Journal of Communication Disorders, 44(1), 59-74.
6
!
7
!
Thelen, E., & Smith, L. B. (1994). A dynamic systems approach to the development of perception
8
!
and action. Cambridge, MA: MIT Press.
9
!
10
!
Westfall, P. H. (1997). Multiple testing of general contrasts using logical constraints and
11
!
correlations. Journal of the American Statistical Association, 92(437), 299-306.
12
!
13
!
Zharkova, N. (2017). Voiceless alveolar stop coarticulation in typically developing 5-year-olds
14
!
and 13-year-olds. Clinical linguistics & phonetics, 1-11.
15
!
16
!
Zharkova, N., Hardcastle, W. J., Gibbon, F., & Lickley, R. (2015). Development of lingual motor
17
!
control in children and adolescents. Proceedings of the 18th ICPhS, Glasgow.
18
!
19
!
Zharkova, N., Hewlett, N., Hardcastle, W. J., & Lickley, R. J. (2014). Spatial and temporal lingual
20
!
coarticulation and motor control in preadolescents. Journal of Speech, Language, and
21
!
Hearing Research, 57(2), 374-388.
22
!
23
!
Zharkova, N., Hewlett, N., & Hardcastle, W. J. (2012). An ultrasound study of lingual
24
!
coarticulation in/sV/syllables produced by adults and typically developing children. Journal
25
!
of the International Phonetic Association, 42(2), 193-208.
26
!
27
!
Zharkova, N., & Hewlett, N. (2009). Measuring lingual coarticulation from midsagittal tongue
28
!
contours: Description and example calculations using English/t/and/ɑ. Journal of
29
!
Phonetics, 37(2), 248-256.
30
!
31
!
32
!
FIGURE LEGENDS:
33
!
Figure 1: Example of ultrasound image recorded within SOLLAR platform. Left panel presents
34
!
This manuscript is not for public dissemination. It is published in its final form in Journal of Speech, Language and
Hearing Research, June, 2018. https://jslhr.pubs.asha.org/article.aspx?articleid=2681858
!
!
33!
the initial tongue image as recorded on the ultrasound scanner; the right panel shows the
1
!
highlighted tongue contours. In each image, the left portion corresponds to the anterior part of the
2
!
tongue.!
3
!
Table 1: Results from Linear Mixed Effects models across age groups for the horizontal position
4
!
of the tongue at C50 with respect to V50.
5
!
6
!
Table 2: Results from Linear Mixed Effects models across age groups for the horizontal position
7
!
of the tongue at C50 with respect to V50. Results (adjusted p-values) are presented per consonant.
8
!
9
!
Table 3: Results from Linear Mixed Effects models within age group for the horizontal position of
10
!
the tongue at C50 with respect to V50. Results (adjusted p-values) are presented per consonant.
11
!
12
!
Appendix A: Syllable count per age group
13
!
14
!
15
!
16
!

Supplementary resources

  • Article
    Full-text available
    This study examines the temporal organization of vocalic anticipation in German children from 3 to 7 years of age and adults. The main objective was to test for non-linear processes in vocalic anticipation, which may result from the interaction between lingual gestural goals for individual vowels, and those for their neighbors over time. The technique of ultrasound imaging was employed to record tongue movement at five time points throughout short utterances of the form V1#CV2. Vocalic anticipation was examined with Generalized Additive Modeling, an analytical approach allowing for the estimation of both linear and non-linear influences on anticipatory processes. Both adults and children exhibit non-linear patterns of vocalic anticipation over time with the degree and extent of vocalic anticipation varying as a function of the individual consonants and vowels assembled. However, noticeable developmental discrepancies were found with vocalic anticipation being present earlier in children´s utterances at 3-4-5 years of age in comparison to adults and to some extent 7-year-old children. A narrowing of speech production organization from large chunks in kindergarten to more segmentally differentiated and contextually-specified organizations from kindergarten to primary school to adulthood, although variation in the temporal overlap of lingual gestures for consecutive segments is already present in the youngest cohorts. In adults, non-linear anticipatory patterns over time suggest a strong differentiation between the gestural goals for consecutive segments. In children, this differentiation is not yet mature: vowels show greater prominence over time and seem activated more in-phase with those of previous segments relative to adults.
  • Article
    Full-text available
    This study aims at evaluating speech motor skills in the fluent speech of a cohort of stuttering Italian children. Stuttering is a neurodevelopmental disorder that may arise from an innate limitation of the speech motor control system, which fails to prepare and organize the movements required for fluent speech (Van Lieshout, Hulstijn, & Peters, 2004). Anticipatory coarticulation in CV sequences and stability of speech movements have been used as measures of the maturity of articulatory processes for fluent speech production. This study aims to assess if direct measures of speech dynamics can identify impaired mechanisms in stuttering speech during a phrase-repetition task. The Ultrasound Tongue Imaging data of eight school-aged children, half of whom stutter while the other half don’t, show different articulatory patterns between the two groups, for both motor aspects under investigation. Articulatory data show that the stuttering group presents a higher degree of intra-syllabic coarticulation compared to the control group and decreased stability (i.e. more variability) through multiple repetitions of the same alveolar and velar item. Outcomes of this study suggest that the speech motor control system of children who stutter is less mature in preparing and executing the speech gestures required for fluent speech. This study contributes to shedding light on the impaired articulatory patterns involved in stuttering speech and to identifying the diagnostic markers of the disorder by evaluating the speech of children close to the onset of stuttering.
  • Article
    Full-text available
    In the first years of life, children differ greatly from adults in the temporal organization of their speech gestures in fluent language production. However, dissent remains as to the maturational direction of such organization. The present study sheds new light on this process by tracking the development of anticipatory vowel-to-vowel coarticulation in a cross-sectional investigation of 62 German children (from 3.5 to 7 years of age) and 13 adults. It focuses on gestures of the tongue, a complex organ whose spatiotemporal control is indispensable for speech production. The goal of the study was threefold: 1) investigate whether children as well as adults initiate the articulation for a target vowel in advance of its acoustic onset, 2) test if the identity of the intervocalic consonant matters and finally, 3) describe age-related developments of these lingual coarticulatory patterns.
  • Preprint
    Full-text available
    A conceptual framework for modeling articulatory control is presented in a tutorial fashion. The framework incorporates three mechanisms: selection, coordination, and intention. Selection is mechanism for governing the choice and ordering of articulatory movements, and operates through an activation code. Coordination is a mechanism for governing the control of movement timing in precise manner, and operates through a phase code regulated by coupled oscillators. Intention is a mechanism for determining the target state of the vocal tract, and operates through a spatial code derived from integrating over parameter fields. All three mechanisms are inherently dynamic, and their interactions provide a basis for understanding a wide variety of phonetic and phonological patterns in development and across languages. An integrated model of the mechanisms is described and applied to a range of empirical phenomena.
  • Article
    Full-text available
    (In press. Pre-final version, due to copyright) In previous research, Mutual Information (MI) was employed to quantify the physical information shared between consecutive phonological segments, based on electromagnetic articulography data (EMA). In this study, MI is extended to quantifying coarticulatory resistance (CR) versus overlap in German adult speakers using ultrasound imaging. Two measurements are tested as input to MI: 1) the highest point on the tongue body and, 2) the first coefficient of the discrete Fourier transform (DFT) of the whole tongue contour.
  • Book
    This monograph arose from a conference on the Production of Speech held at the University of Texas at Austin on April 28-30, 1981. It was sponsored by the Center for Cognitive Science, the College of Liberal Arts, and the Linguistics and Psychology Departments. The conference was the second in a series of conferences on human experimental psychology: the first, held to commemorate the 50th anniversary of the founding of the Psychology Department, resulted in publication of the monograph Neural Mechanisms in Behavior, D. McFadden (Ed.), Springer-Verlag, 1980. The choice of the particular topic of the second conference was motivated by the belief that the state of knowledge of speech production had recently reached a critical mass, and that a good deal was to be gained from bringing together the foremost researchers in this field. The benefits were the opportunity for the participants to compare notes on their common problems, the publication of a monograph giving a comprehensive state-of-the-art picture of this research area, and the provision of enormous intellectual stimulus for local students of this topic.
  • Article
    Visual biofeedback is the process of gaining awareness of physiological functions through the display of visual information. As speech is concerned, visual biofeedback usually consists in showing a speaker his/her own articulatory movements, which has proven useful in applications such as speech therapy or second language learning. This article presents a novel method for automatically animating an articulatory tongue model from ultrasound images. Integrating this model into a virtual talking head enables to overcome the limitations of displaying raw ultrasound images, and provides a more complete and user-friendly feedback by showing not only the tongue, but also the palate, teeth, pharynx, etc. Altogether, these cues are expected to lead to an easier understanding of the tongue movements. Our approach is based on a probabilistic model which converts raw ultrasound images of the vocal tract into control parameters of the articulatory tongue model. We investigated several mapping techniques such as the Gaussian Mixture Regression (GMR), and in particular the Cascaded Gaussian Mixture Regression (C-GMR) techniques, recently proposed in the context of acoustic-articulatory inversion. Both techniques are evaluated on a multispeaker database. The C-GMR consists in the adaptation of a GMR reference model, trained with a large dataset of multimodal articulatory data from a reference speaker, to a new source speaker using a small set of adaptation data recorded during a preliminary enrollment session (system calibration). By using prior information from the reference model, the C-GMR approach is able (i) to maintain good mapping performance while minimizing the amount of adaptation data (and thus limiting the duration of the enrollment session), and (ii) to generalize to articulatory configurations not seen during enrollment better than the GMR approach. As a result, the C-GMR appears to be a good mapping technique for a practical system of visual biofeedback.
  • Article
    Purpose: The purpose of this study was, first, to expand our understanding of typical speech development regarding segmental contrast and anticipatory coarticulation, and second, to explore the potential diagnostic utility of acoustic measures of fricative contrast and anticipatory coarticulation in children with speech sound disorders (SSD). Method: In a cross-sectional design, 10 adults, 17 typically developing children, and 11 children with SSD repeated carrier phrases with novel words with fricatives (/s/, /ʃ/). Dependent measures were 2 ratios derived from spectral mean, obtained from perceptually accurate tokens. Group analyses compared adults and typically developing children; individual children with SSD were compared to their respective typically developing peers. Results: Typically developing children demonstrated smaller fricative acoustic contrast than adults but similar coarticulatory patterns. Three children with SSD showed smaller fricative acoustic contrast than their typically developing peers, and 2 children showed abnormal coarticulation. The 2 children with abnormal coarticulation both had a clinical diagnosis of childhood apraxia of speech; no clear pattern was evident regarding SSD subtype for smaller fricative contrast. Conclusions: Children have not reached adult-like speech motor control for fricative production by age 10 even when fricatives are perceptually accurate. Present findings also suggest that abnormal coarticulation but not reduced fricative contrast is SSD-subtype-specific. Supplemental materials: S1: https://doi.org/10.23641/asha.5103070. S2 and S3: https://doi.org/10.23641/asha.5106508.
  • Conference Paper
    Full-text available
    A successful characterization of vocal tract control during speech needs to account for regular variability in the degree of coarticulatory overlap allowed by different speech segments. While some segments allow for large degree of articulatory overlap, others show high coarticulation resistance (CR) i.g. ability to resist influence from neighbors and retain control over articulators across contexts. Despite the importance of the CR phenomenon for theories and models of speech production, a unified quantitative measure of coarticulation resistance has not been established yet. The most prominent description of CR, DAC scale (Recasens & Espinosa, 2009), has some limitations: first, it subjectively divides segments into several categories based exclusively on the degree of lingual coarticulation they exhibit across contexts. Second, this categorization is based on the measures of coarticulation that quantify only linear dependencies. Generally, a wide variety of experimental techniques and corresponding quantification methods make it difficult to directly compare CR estimates across studies. Recently, Iskarous et al. (2013) suggested measuring CR with Mutual Information (MI), or the amount of information shared by a given segment with other segments across contexts. The MI is non-parametric method that does not make assumptions about distribution but rather estimates it from data. The MI values have been shown by Iskarous et al (2013) to capture the CR effects of place and manner, as well as time differences in CR. In our study we extend the application of MI to quantifying coarticulation from ultrasound images. We investigate the effect of consonantal context on vowel-to-consonant coarticulation in German adults by looking at the position of the highest point of the tongue. Our results show that CR of German consonants exhibits the pattern /z > d > g > b/ in horizontal dimension and /z > g > d > b/ in the vertical dimension. These finding corroborates those made by Iskarous et al (2013) for German voiceless consonants for tongue body articulator using EMA data. This suggests that MI measure allows for cross-methodological comparisons and generalizations of quantitative findings. Temporal aspects of coarticulation resistance for different segments are currently being investigated by comparing MI values at different time points during consonant production. We are also applying MI to quantify whole tongue contours to compare aspects of coarticulation captured by different methods of tongue shape quantification.
  • Article
    Full-text available
    In this study, vowel-on-consonant lingual coarticulation at [t] closure offset was compared in 5-year-old children and 13-year-old adolescents. The study aimed to establish whether, by the end of the closure, children from the younger age group adjust the tongue shape to the following vowels to the same extent as adolescents. Ten 5-year-olds and ten 13-year-olds, all speakers of Scottish Standard English, produced [t]-vowel syllables with the vowels [i] and [a], in a carrier phrase. Measures of tongue shape based on midsagittal ultrasound imaging data were used to compare anticipatory coarticulation and within-speaker variability across groups. Both age groups changed the extent of tongue dorsum bunching in order to coarticulate the consonant with the following vowels. The 5-year-old children, unlike the adolescents, did not consistently modify the bunching location within the tongue curve to accommodate the tongue shape to that of the upcoming vowel. Token-to-token variability was significantly greater in the younger age group. The results suggest that vowel-on-[t] coarticulatory patterns produced by typically developing children are affected by the development of motor control, with articulatory constraints on the tongue limiting the extent of lingual coarticulation in 5-year-old children. The findings on typical coarticulation development are relevant for clinical practice, and they highlight the need for more detailed descriptions of how phonetic characteristics of speech sounds affect coarticulation throughout childhood.
  • Article
    A new method for quantifying contextual variability at different regions of the tongue using ultrasound spline data reveals that tongue body coarticulatory resistance for Catalan consonants and vowels in VCV sequences decreases in the progression [ʎ, ɲ, ʃ] > [s, r] > [l, ɾ, t, n] > [ð] and [i, e] > [a] > [o] > [u]. These consonant and vowel hierarchies support the degree of articulatory constraint model of coarticulation according to which coarticulatory resistance depends on whether a given lingual region is involved in the formation of a closure or constriction and on the severity of the manner of articulation requirements. Data show that this coarticulatory scenario holds not only at the palatal zone, as revealed by previous coarticulation studies, but at the velar and pharyngeal zones as well. Partial exceptions are [s] and [i], which may allow for some more contextual variability than expected at the back of the vocal tract. Another major finding is that tongue body coarticulatory resistance and aggressiveness are highly positively correlated. The implications of these experimental results for speech production organization and sound change are discussed.
  • Conference Paper
    Full-text available
    We present here a customized method developed jointly by scientists at LOLA (Potsdam University) and Haskins Laboratories (New Haven) for the recording of both tongue and lip motion during speech tasks in young children. The method is currently being used to investigate the development of 1) coarticulation (resistance and anticipatory coarticulation, cf. two other abstracts submitted); and 2) articulatory coordination in preschoolers compared with adults who have mature control of their speech production system. Children are recorded with a portable ultrasound system (Sonosite Edge, 48Hz) with a small probe fixed on a custom-made probe holder and ultrasound stand. The probe holder was specifically designed to allow for natural vertical motion of the jaw but prevent motion in the lateral and horizontal translations. The set up is integrated into a child-friendly booth that facilitates integrating the production tasks into games. Ultrasound video data are collected concurrently with synchronized audio recorded via a microphone (Shure, 48kHz,), pre-amplified before being recorded onto a desktop computer. In addition to tongue motion, a frontal video recording of the face is obtained with a camcorder (Sony HDR-CX740VE, fps: 50Hz). This video is used to track lip motion for subsequent labial measurements, and to track head and probe motion for transforming contours extracted from the ultrasound images to a head-based coordinate system. The speech signal is also recorded via the built-in camcorder microphone, and synchronization of both video signals (from the ultrasound and the camcorder) is performed through audio cross-correlation in post-processing. Lip motion is characterized with a video shape tracking system (Lallouache 1991) previously used for examining anticipatory coarticulation in adults (Noiray et al., 2011) and children (Noiray et al., 2004; 2008). During production tasks, the lips of our young participants are painted in blue as this color maximized contrast with the skin. In post-processing these blue shapes are then tracked for calculation of lip aperture, interolabial area and upper lip protrusion. Tongue contours derived from ultrasound are relative to the orientation of the probe with respect to the tongue surface. To correct for jaw displacement and (pitch) rotation of the head we compute two correction signals similar to the HOCUS method described in Whalen et al. (2005), but in this case derived from tracking the positions of blue reference dots in the video signal using custom Matlab procedures. The displacement of the probe relative to the centroid of dots placed on each speaker's forehead provides a vertical correction signal. The orientation of dots placed on the cheek observed within the video image through a mirror oriented at 45° giving a profile view provides a pitch rotation correction signal around the lateral axis. Application of these two signals to the extracted contours allows for their consistent comparison in a head-centric coordinate system.