research on language development

ORIGINAL RESEARCH article

Spoken language development and the challenge of skill integration.

• 1 Laboratory for Oral Language Acquisition, Linguistic Department, University of Potsdam, Potsdam, Germany
• 2 Haskins Laboratories, New Haven, CT, United States
• 3 Department of Linguistics, University of Oslo, Oslo, Norway
• 4 Department of Education, Jyväskylä University, Jyväskylä, Finland

The development of phonological awareness, the knowledge of the structural combinatoriality of a language, has been widely investigated in relation to reading (dis)ability across languages. However, the extent to which knowledge of phonemic units may interact with spoken language organization in (transparent) alphabetical languages has hardly been investigated. The present study examined whether phonemic awareness correlates with coarticulation degree, commonly used as a metric for estimating the size of children’s production units. A speech production task was designed to test for developmental differences in intra-syllabic coarticulation degree in 41 German children from 4 to 7 years of age. The technique of ultrasound imaging allowed for comparing the articulatory foundations of children’s coarticulatory patterns. Four behavioral tasks assessing various levels of phonological awareness from large to small units and expressive vocabulary were also administered. Generalized additive modeling revealed strong interactions between children’s vocabulary and phonological awareness with coarticulatory patterns. Greater knowledge of sub-lexical units was associated with lower intra-syllabic coarticulation degree and greater differentiation of articulatory gestures for individual segments. This interaction was mostly nonlinear: an increase in children’s phonological proficiency was not systematically associated with an equivalent change in coarticulation degree. Similar findings were drawn between vocabulary and coarticulatory patterns. Overall, results suggest that the process of developing spoken language fluency involves dynamical interactions between cognitive and speech motor domains. Arguments for an integrated-interactive approach to skill development are discussed.

Introduction

In the first decade of life, most children learn to speak their native language effortlessly, without explicit instruction but with daily exposure and experiencing of their native language as a speech motor activity. With the gradual expansion of children’s expressive repertoire comes the fine tuning of phonological knowledge (e.g., Ferguson and Farwell, 1975 ; Menn and Butterworth, 1983 ; Beckman and Edwards, 2000 ; Munson et al., 2012 ). While relationships between lexical and phonological developments have been well documented over the last decades ( Storkel and Morrisette, 2002 ; Edwards et al., 2004 , 2011 ; Stoel-Gammon, 2011 ; Vihman, 2017 ), research addressing their interaction with spoken language production has often been restricted to production accuracy or duration measures as metrics for assessing spoken language proficiency (e.g., Edwards et al., 2004 ; Munson et al., 2005 ). Likewise, speech motor control studies have provided in-depth analyses of developmental changes in articulatory variability, or movement velocity during word or sentence production ( Smith and Goffman, 1998 ; Smith and Zelaznik, 2004 ; Green et al., 2010 ) without equivalently thorough assessments of children’s phonological or lexical knowledge allowing developmental interactions to be evaluated. Despite a certain imbalance in the focus and analytical approaches of interaction studies, the findings suggest that spoken language proficiency entails dynamical interactions among a set of language-related domains including speech motor skill.

In the present research, we adopted an integrated approach to the study of spoken language development considering parallel developments of the lexical, phonological, and speech motor systems. The study more specifically investigated interactions between domains that have not yet been empirically connected: in particular phonological awareness , the awareness of the particulate nature of the language (e.g., Fowler, 1991 ; Studdert-Kennedy, 1998 , 2005 ) that develops with literacy (reviews in Anthony and Francis, 2005 ; Brady et al., 2011 ; Goswami and Bryant, 2016 ; in German: Fricke et al., 2016 ) and anticipatory coarticulation , a mechanism that is deeply rooted in kinematics (e.g., Parush et al., 1983 ) and motor planning (e.g., Whalen, 1990 ; Levelt and Wheeldon, 1994 ; Grimme et al., 2011 ; Perrier, 2012 ; Davis and Redford, 2019 ) and is fundamental to speech fluency.

While phonological awareness and coarticulatory mechanisms may in principle belong to different realms, we argue that they are developmentally strongly interconnected: phonological awareness relates to the ability to consciously extract functional units of phonological organization from the continuous speech flow (e.g., syllables, segments) and combine those discrete units into new sequences of variable size and meaning (e.g., Metsala, 2011 ). Coarticulation embodies speakers’ structural knowledge of the language, combining and (re)modeling its elementary particles into continuous articulatory movements and acoustic streams, hence contextualizing abstract representations into a decipherable “speech code” ( Liberman et al., 1974 ; Fowler et al., 2016 ). In this perspective, investigating developmental changes in children’s coarticulatory processes may give us an opportunity to track how a combinatorial principle is situated within the representational and production levels and to capture more broadly how motor and cognitive functions come together to develop the skill of spoken language.

While children’s speech organization very early reflects their ability to combine phonetic units, the explicit awareness of the combinatorial nature of their native language forming larger compounds from smaller-sized units follows a more protracted development and seems to climax around the time children acquire literacy (e.g., Gillon, 2007 ). During that period, a gain in phonological awareness allows children to convert the already acquired phonetic units (i.e., sounds they hear and produce by means of distinct speech gestures) into phonological units. However, whether the acquisition of phonological knowledge only relates to attaining literacy or also modifies children’s spoken language organization in fundamental ways remains an empirical question. The alternative direction in which a gain in spoken language practice would stimulate the development of phonological awareness and literacy has also not yet been demonstrated. The present study provides a first step toward addressing this issue by testing whether lexical and phonological skills interact with speech motor control in development. More specifically, we examined whether children with greater knowledge of the segmental makeup of words in their native language exhibited a segmentally specified organization of their speech gestures and reflected in their coarticulatory patterns. We focused on the period encompassing kindergarten to the end of the first primary school year, which is relevant for phonological development as well as for attaining literacy. Our motivations driven from empirical research are further outlined below.

What Are Children’s Units of Spoken Language Organization

In the last decades, a growing number of developmental studies in the area of spoken language ability have focused on coarticulation degree, which characterizes the extent to which the articulatory gestures for neighboring phonemes overlap temporally (e.g., Browman and Goldtstein, 1992 ). Looking specifically at lingual coarticulation, which regards the gestural organization of the tongue, some research has found a developmental decrease in vocalic anticipatory coarticulation over previous segments, within the syllables (e.g., Nittrouer et al., 1996 ; Zharkova et al., 2011 ; Noiray et al., 2018 ) and beyond the syllabic span (e.g., Nijland et al., 2002 ; Rubertus and Noiray, 2018 ). On the basis of these results, Noiray et al. (2019) reasoned that spoken language fluency may entail a gradual narrowing of speech units toward smaller-sized units. In young children, vowels may represent building blocks, which children organize their speech around because of their perceptual salience, long duration, and earlier acquisition compared to consonants (e.g., Polka and Werker, 1994 ; review Nazzi and Cutler, 2019 ). Hence, children’s vocalic and consonantal gestures may be activated more simultaneously than in adults, resulting in an overall larger vocalic influence on previous consonants and a greater degree of vocalic coarticulation than for adults. Instead, adults have been found to organize their speech with more temporally individuated gestures ( Abakarova et al., 2018 ; Rubertus and Noiray, 2018 ). The result of rather large unit size speech organization echoes the multiple findings of whole-word learning ( Vihman and Velleman, 1989 ; Keren-Portnoy et al., 2009 ; Menn and Vihman, 2011 ), transitional probability across syllables (e.g., Jusczyk et al., 1993 ; Saffran et al., 1996 ), or lexically grounded phonological development and production accuracy ( Edwards et al., 2004 ; Velleman and Vihman, 2007 ; Vihman and Keren-Portnoy, 2013 ). The opposite finding of a lesser degree of coarticulation between consonants and vowel gestures in children compared to adults has also been reported (e.g., Katz et al., 1991 ), favoring a more segmental perspective of early spoken units.

Based on our own in-depth examinations of coarticulatory mechanism in both adults ( Abakarova et al., 2018 ) and children ( Noiray et al., 2018 ; Rubertus and Noiray, 2018 ), we have argued that (young) speakers exhibit gradients of coarticulation degree within a continuum from a more syllabic to a more segmental organization. The degree to which segment overlap depends on the gestural demands associated with the combined segments. In adults, contextual differences in coarticulation degree are well attested (e.g., Recasens, 1985 ; Fowler, 1994 ). For instance, syllables recruiting a single organ for the consecutive production of both consonantal and vowel targets (e.g. the tongue in /du/) require from speakers a functional differentiation between the subparts of the tongue (tongue tip, tongue dorsum). This type of syllable further requires greater spatiotemporal coordination in comparison to syllables recruiting two separate primary organs (e.g., the lips and tongue dorsum in /bi/). This phenomenon described within the theory of coarticulatory resistance has been reported in adults across languages over the past decades (review in Recasens, 2018 ). In children, extensive kinematic investigations of coarticulatory processes have been more challenging and hence somewhat restricted in scope compared to adults (e.g., limited variety of stimuli that can be tested in the young age, age range, sample size, scarcity of methodological replications across studies). Yet, once these studies are examined together, they support the view of coarticulatory gradients as observed in adults. While children show overall greater coarticulation degree than adults, they also exhibit contextual effects on coarticulation degree, which result from the particular combination of gestural goals between individual consonants and vowels. Based on those observations, we recently suggested a gestural approach as a “unifying organizational scheme to relate adults’ to children’s patterns. How coarticulatory organization matures over time is then no longer solely a question of direction (toward a greater or lesser coarticulatory degree) or categorical change in phonological organization (e.g., into segments or syllables) but a question of how a primitive gestural scheme shares similar tools (the articulators of speech), constraints, and principles (dynamic interarticulator coordination over time) with adults to instantiate complex phonetic combinations in line with the native language’s phonological grammar” ( Noiray et al., 2019 , p. 3037). In this context, the question of (early) units of speech production may be viewed as a part-whole interaction.

The Development of the Lexical, Phonological, and Motor Domains

While the maturation of the speech motor system is central to spoken language fluency, lexical and phonological developments are equally crucial (e.g., Smith et al., 2010 ; Edwards et al., 2011 ), and research has suggested that they interact dynamically over time (e.g., Beckman et al., 2007 ; Sosa and Stoel-Gammon, 2012 ; Vihman, 2017 ). A main hypothesis motivating the present study is that adults’ coarticulatory patterns do not differ from those of children on the sole basis of greater precision of control from children’s speech production system. Adults also have (1) built an expressive lexicon from which to harness their phonological representations, (2) they have gained an explicit understanding of the structure of their language, and (3) an ability to manipulate this information into a quasi-infinite set of intelligible spoken forms. Hence, considering speech motor development as a goal-directed process – for example, speaking a language fluently – what distinguishes children from adults is that children have not yet built explicit correspondences between phonetic segments and their motor realizations. The rapid growth of the expressive lexicon observed during the kindergarten-to-school years may help children solve this correspondence problem and more generally develop stable relations between representational and executional levels. Vocabulary is indeed often considered the backbone of language acquisition, supporting the development of phonological representations (e.g., Ferguson and Farwell, 1975 ; Metsala, 1999 ) and production accuracy (e.g., Edwards et al., 2004 ; Nicholson et al., 2015 ). Previous research also suggests that children first develop articulatory “routines” for the syllables present in their expressive repertoire (e.g., Menn and Butterworth, 1983 ; Munson et al., 2005 ; Ziegler and Goswami, 2005 ; Vihman, 2017 ). This lexically based process may lay the ground for increased phonetic distinctions along the dimensions of height, fronting and rounding for vowels, place and manner of articulation for consonants, and the maturation of coarticulatory flexibility for a wider range of phonetic environments.

This knowledge is at first experience-based; before entering primary school, children have limited explicit knowledge about the structural organization of their native language, that is, they have limited conscious awareness that the words they hear can be segmented into smaller-sized units (and recombined into new forms; e.g., Liberman et al., 1974 ; Gillon, 2007 ). Note that while the development of phonological awareness differs as a function of orthographic transparency (e.g. Fricke et al., 2016 ) or the age at which children learn how to read (e.g., review in Wimmer et al., 2000 ; Mann and Wimmer, 2002 ; Schaeffer et al., 2014 ; Goswami and Bryant, 2016 ) on average, children in kindergarten show only more or less equivalent proficiency in syllabic units’ awareness to that of school-aged children (in English: e.g., Liberman et al., 1974 ; in German: Ziegler and Goswami, 2005 ; Schaeffer et al., 2014 ) but no advanced phonemic awareness before explicitly learning how to read. Taken together, young listener-speakers would progressively access smaller units allowing them to decipher a wider range of speech forms and manipulate those flexible units to craft increasingly more complex speech flows. Figure 1 provides an illustrative conceptualization of these seemingly parallel developmental trajectories, from more holistic access and production of large units (e.g., lexemes) to more segmentally specified representations and coarticulatory organizations. Developmental overlaps (e.g., from lexeme access to rhyme access) and short-term regressions between learning phases may at times occur (e.g., Anthony et al., 2003 ), as noted in other domains (e.g., “phonological templates” during early word production: Vihman and Vihman, 2011 ; lip-jaw movement variability: Green et al., 2002 ; walking: Thelen and Smith, 1994 ). The developmental pace may also well change over time, as in other domains (e.g., speech motor control: Green et al., 2010 ). Figure 1 highlights the nonlinearity of those developmental processes over time (blue descending and ascending curves). With an advanced knowledge of their native language and a mature control of their speech motor system, adults naturally exhibit more flexible, context-specific organizations with greater or lesser coarticulation degree depending on the gestural properties of the individual segments assembled with one another.

Figure 1 . Theoretical conceptualization of the parallel development of phonological awareness and coarticulatory organization from holistic to more segmental organizations. The horizontal arrow ( x -axis) illustrates developmental time (age in years). The curves indicate the nonlinear change in phonological and coarticulatory organizations over time.

Overall, results from these separate literatures suggest that the development of lexical, phonological, and speech motor abilities are fundamental to the maturation of children’s spoken language. However, to our knowledge, empirical studies examining their interactions with precision have been rare, and this gap has prevented a unifying account of spoken language development. The central hypothesis driving our current research is that the transition from the rather self-paced development of large unit phonological awareness to the more explicit knowledge of the phonemic constituents of the language initiated in primary school should correlate with a significant change in spoken language production from an experience-based holistic organization to a structurally informed, segmentally specified organization of spoken language. Because quantitative longitudinal investigations over a 2- to 3-year span are extremely difficult to conduct, we first opted for a cross-sectional examination of a sample of 41 children in the last 2 years of kindergarten (at 4.5 and 5.5 years of age) and the end of the first grade (at age 7). The latter cohort was chosen to ensure children have been exposed to explicit literacy instruction for a year. With this approach, we first tested for significant interactions between children’s motor, lexical, and phonological skills. Potential implications for causal relations are laid out in the discussion.

Based on our previous research, we expect differences in intra-syllabic coarticulation degree between children and adults but not necessarily between all child cohorts ( Noiray et al., 2019 ). We also anticipated consonantal effects on children’s lingual coarticulatory patterns within each age cohort as found in a preceding study investigating children’s intra-syllabic coarticulation from the age of 3 ( Noiray et al., 2018 ). More specifically, we expected a lower degree of lingual coproduction for consonant-vowel syllables requiring two constriction goals by spatially distinct articulatory organs than from those requiring two constriction goals by a single organ as found in adults (e.g., Iskarous et al., 2013 ; Abakarova et al., 2018 ), albeit to a lesser extent than adults. Importantly, expanding on previous research, we predicted greater phonological awareness and vocabulary would coincide with lower coarticulation degree, i.e., greater segmental differentiation of consonants and vowels in syllables. We further suspected interactions between motor and cognitive domains to be nonlinear and to reflect the complex dynamics in place during the development spoken language fluency. If this were found, it would suggest that the skill of spoken language fluency is not solely tied to production-related considerations but may instead result from and be an integral part of multiple interactions, which are fundamental to the development of each individual skill. If no correlation was to be found, it would on the contrary indicate that representational and production levels may not be tightly coupled in the sense that greater awareness of phonological discreteness does not interact with coarticulatory degree.

Materials and Methods

Participants.

Forty-one monolingual German children all living in the Potsdam region (Brandenburg) were tested: ten 4-year olds (6 females, mean age: 4; 06, called K1 in subsequent analyses), thirteen 5-year-old children (7 females, mean: 5; 06, called K2 hereafter) in kindergarten, and eighteen 7-year-old children at the very end of the first or very beginning of the second grade in primary school (12 females, mean: 7; 02, called P1 hereafter). The discrepancy in sample size was due to greater difficulty in recruiting children in kindergarten. All children were raised in monolingual German families without any known history of hearing, language, or cognitive impairment. They were recruited via the child registry from the BabyLab of the University of Potsdam. Ethics approval was obtained from the Ethic Committee of the University of Potsdam prior to the study. All parents were also fully informed of the study and gave written consent for their child to participate.

Production Task

The speech production task consisted in the repetition of trochaic pseudowords (i.e., conforming to German phonotactics) of the form consonant 1 -vowel-consonant 2 -schwa ( C 1 V C 2 ǝ). Target phrases used as stimuli were pre-recorded by a native German female adult speaker. Three consonants varying in place of articulation: /b/, /d/, and /g/ and six tense, long vowels /i/, /y/, /u/, /a/, /e/, and /o/ were used. Pseudowords were chosen instead of real words to combine consonants and vowels varying in lingual gestures and coarticulatory resistance. Target pseudowords were embedded in a carrier phrase with the article /aɪnə/ resulting in utterances such as /aɪnə ba:də/. Utterances were repeated six times in semi-randomized blocks. To measure lingual coarticulation, we employed the technique of ultrasound imaging (Sonosite edge, fps: 48 Hz) that permits recording movement from participants’ tongue over time while producing various speech materials ( Noiray et al., 2013 ). In this study, tongue imaging was integrated in a space journey narrative to stimulate children’s motivation to complete the task. Children were seated in a pilot seat with seatbelts, facing the operating console from a space rocket replica. The ultrasound probe on which children positioned their chin was integrated into a customized probe-holder as part of the rocket console (for a full description of the method, see Noiray et al., 2018 ). The acoustic speech signal was recorded synchronously with the ultrasound tongue video via a microphone (Shure, fps: 48KHz).

Assessment of Phonological Awareness and Vocabulary

Assessments of various levels of phonological awareness (rhyme, onset segment, and individual phonemes) were conducted with the Test für Phonologische Bewusstheitsfähigkeiten (TPB; Fricke and Schäfer, 2008 ). Prior to testing, children were familiarized with all images used as test items. The procedure for each of the TPB test is briefly summarized below; a complete description can be found by Fricke and Schäfer (2008) . The tests were scored according to the test instructions, and raw scores were considered for subsequent analyses.

Rhyme Production

Children are shown a picture and are instructed to produce (non)words that rhyme with the word corresponding to the target picture (e.g., Puppe: Muppe, Kuppe, Wuppe ). Children are instructed to provide as many rhymes as they can. However, to make the task comparable for every child, we scored children’s proficiency differently from the test instructions: for each of the 12 target words, children scored 1 point if they succeeded in giving at least one correct rhyme; if not, they scored zero. This way, we could assess the stability and generalization of the rhyming skill rather than relying on raw number of rhymes produced (e.g. if a child produced six rhymes for two target words but then failed for all other target words).

Onset Segment Deletion

Children are shown a picture and are instructed to delete the onset segment from the word represented by the picture and utter the resulting nonword (e.g. Mond: ond; Zahn: ahn). Note children were precisely instructed what to delete (e.g. “delete “m” from Mond”). A total of 12 words is tested in each age cohort.

Phoneme Synthesis

Children are instructed to produce a word after hearing a pre-recorded female voice uttering its phonemes one by one (e.g. fee: [f-e:], dose: [d-o:-z-Ə], salat: [z-a-l-a:-t]). For the onset segment deletion task, the TPB assessment uses a total of 12 words for each age cohort.

Expressive Vocabulary

Expressive vocabulary was tested with Patholinguistische Diagnostik bei Sprachentwicklungsstörungen (PDSS; Siegmüller and Kauschke, 2010 ) and widely used to assess German children’s lexical repertoire. The test consists of a 20-word picture naming task assessing nouns for the target ages (see Table 1 for an overview). In subsequent analyses, we used a composite score for phonemic awareness (PA hereafter that includes the two tasks tapping phoneme-size awareness: onset deletion and phoneme synthesis).

Table 1 . Summary of the results from the assessments tapping phonological awareness (Rhyme, Composite PA) and expressive vocabulary (VOC) conducted in 4-year-old (K1), 5-year-old (K2), and 7-year-old children at the end of first grade (P1).

We focused on output phonological tasks as well as expressive vocabulary because we were interested in their direct relationship with children’s speech production. Given that young children have a limited attention span, we could also assess children’s actual proficiency with better confidence than when conducting long series of cognitively demanding assessments. All assessments were conducted in our laboratories by experimenters trained by a speech language pathologist.

Statistical Analyses

Consistent with previous research, intra-syllabic coarticulation degree was estimated in terms of whether the lingual gesture for a target vowel was anticipated in the previous consonant (see review on vowels’ degrees of aggressiveness in the context of different consonants: Iskarous et al., 2010 ). We focused on the antero-posterior tongue dorsum position that is highly relevant in terms of articulatory and acoustical contrasts between vowels (e.g., Delattre, 1951 ). We calculated differences in tongue dorsum position between the production of consonants and following vowels. A tongue dorsum position for a consonant (e.g., /g/) that varies in the context of various vowels (e.g., /a/, /i/) indicates vocalic anticipation onto the previous consonant and hence a high coarticulation degree. On the contrary, low coarticulation degree is reflected by an absence of change in tongue dorsum position during the consonant in the context of various vowels (review in Iskarous et al., 2010 ).

Differences in coarticulation degree were estimated for each consonantal context from the midpoint of the consonant (C 1 ) compared to the vowel midpoint (V). A few preliminary processing steps were necessary. First, the corresponding midsagittal tongue contours for both C 1 and V were extracted from the ultrasound video based on the acoustic speech signal labeling. The tongue contours were then analyzed using SOLLAR (Noiray et al., submitted), a platform created in our laboratory for the analysis of kinematic data (Matlab environment). For each target tongue contour, a 100-point spline was automatically generated, and the x - and y -coordinates for each point were extracted. In subsequent analyses, we used the horizontal x -coordinate for the highest y -coordinate point of the tongue dorsum to reflect its variation in the anterior-posterior dimension (e.g., anterior position for /i/, posterior position for /u/, e.g., Abakarova et al., 2018 ). Data were normalized for each participant by setting the most anterior tongue dorsum position during the target vowel midpoints to 0 and the most posterior tongue dorsum position to 1. Tongue dorsum positions for consonant midpoints were then scaled within this range.

To test for developmental differences in coarticulation degree, we employed linear mixed effects models (LMER), using the “lme4” package in R (version 1.1–19; Bates et al., 2015 ). Coarticulation degree was calculated by regressing the horizontal position of the tongue dorsum at consonant midpoint (PEAKC 1 _X) on the horizontal position of the tongue dorsum at vowel midpoint (PEAKV_X) for each age group (K1, K2, and P1). Two interaction terms were used: Coarticulation and Consonant (C 1 ) and Coarticulation and Age. By-subject C 1 and by-word random slopes for PEAKV_X were included as random effects.

To test for an effect of phonological awareness and vocabulary on children’s coarticulation degree, we then employed Generalized Additive Modeling (GAM), a statistical approach allowing us to test for linear and nonlinear relationships ( Winter and Wieling, 2016 ; Wood, 2017 ; for a comprehensive tutorial, see Wieling, 2018 ). To date, this approach has only been used in psycholinguistic research with adults (e.g., Strycharczuk and Scobbie, 2017 ; Wieling et al., 2017 ) and only recently in the developmental domain ( Noiray et al., 2019 ). In this study, we were interested in the effect of three variables on the degree of coarticulation: RHYME, COMPOSITE_PA (a composite computed from the sum of the scores obtained for both phonemic awareness tasks: onset segment deletion and phoneme synthesis, see section “Descriptive Statistics for Phonological Awareness and Vocabulary”), and VOC. We used the function bam of the mgcv R package (version 1.8–26) and itsadug (version 2.3). Our dependent variable was again PEAKC 1 _X with respect to PEAKV_X. We predicted this value on the basis of a nonlinear interaction, which is modeled by a tensor product smooth (te). A tensor product smooth can model both linear and nonlinear effects across a set of predictors and their interaction (see Wieling, 2018 ) here between: RHYME, COMPOSITE_PA or VOC, and PEAKV_X. The resulting estimated degrees of freedom (edf) indicate whether the relation is linear (value close to 1) or nonlinear (values above 1).

Testing for Developmental Differences in Coarticulation Organization

Table 2 shows the results from the LMER testing for age-related differences in coarticulation degree across all consonants and vowels. No significant difference was noted across the three target age groups. However, differences in coarticulation degree were found across consonantal contexts, with a lower coarticulation degree in alveolar /d/ context as compared to labial /b/ context (estimate: −0.11793, p < 0.05). Coarticulation degree did not differ across other consonantal contexts.

Table 2 . Results from the linear mixed effects model testing for age comparisons in coarticulation degree between the 4-year-old group (K1), 5-year-old group (K2), and 7-year-old group (P1).

Descriptive Statistics for Phonological Awareness and Vocabulary

Pearson product-moment correlations were computed to assess relationships between all developmental assessments. For the rhyming task, we conducted the task in 40 of the 41 children because one P1 child did not want to conduct the rhyming task. A strong positive 0.94 correlation ( p < 0.001) was found between scores for onset deletion and phoneme synthesis. In subsequent analyses, testing the effect of phonological awareness on coarticulatory organization, we therefore computed a composite score as the sum of the scores obtained in the two tasks. This score was taken to reflect children’s phonemic awareness (COMPOSITE_PA), that is, of phonemic units in comparison to the awareness of larger phonological units (rhymes).

Figure 2 provides an overview of the score distribution for each of the four developmental assessments conducted across child cohorts. Dot plots were used to highlight variations in the number of children obtaining a target score. Table 1 provides a summary of the descriptive statistics reflecting children’s phonological awareness and expressive vocabulary. Mean score and range reflect the number of correct items (raw scores). While mean scores increased with age for all language-related skills, results (1) revealed stark individual differences within the same age-group and (2) overlap in scores across age groups for rhyme and expressive vocabulary. For the phonological tasks targeting the awareness of phonemic units (onset segments and individual phonemes), children in kindergarten had overall great difficulty completing the tasks (despite being familiarized with pre-test items), while children in the first grade could complete the tasks with various levels of proficiency.

Figure 2 . Score distribution for each of the four developmental assessments conducted across age groups (K1, K2, and P1). From left to right: rhyme production, onset deletion, phoneme synthesis, and vocabulary. The filled colored circles from different sizes represent different numbers of participants sharing a similar score.

The Welch t test was conducted to test for developmental differences in phonological awareness and vocabulary. Performance on rhyme production for the scoring procedure we employed did not yield any significant differences among age groups (K1–K2: t = −0.58, df = 17.47, p < 0.6; K1–P1: t = −0.58238, df = 17.47, p < 0.6; K2–P1: t = −1.9085, df = 12.524, p < 0.08). With regard to the composite score computed to target the awareness of phonemic units, 5-year-old children (K2) did not differ in performance from 4-year olds (K1) ( t = −1, df = 12, p < 0.4). Only 7-year-old children (P1) showed greater proficiency than K2 ( t = −15.572, df = 21.128, p < 0.0001 4.693e-13) and K1 ( t = −30.006, df = 14, p < 0.0001). Hence, a developmental increase in awareness of segmental units was found between children in kindergarten altogether and those in the first year of primary school, which yielded an overall high correlation between age and PA composite of 0.9 ( p < 0.0001). Regarding vocabulary, similar directions were found. K1 children did not exhibit lower proficiency than K2 ( t = −0.95914, df = 19.728, p < 0.4), only when compared to P1 children ( t = −7.0665, df = 16.375, p < 0.0001). K2 children also had lower vocabulary scores than P1 children ( t = −4.0338, df = 16.257, p < 0.001). However, unlike for phonemic awareness, the correlation between age and vocabulary was not significant (0.12, p < 0.3).

Interaction Between Phonological Awareness and Coarticulation Degree

Given the results from the developmental assessments, we adopted the following statistical approach: we first tested the interaction between rhyme proficiency as an index of intermediate unit-size awareness and coarticulation degree for all children. We then further tested for a separate interaction between phonemic awareness (COMPOSITE_PA, named PA for short hereafter) or vocabulary (VOC) and coarticulation degree. We conducted GAM analyses to illuminate potentially nonlinear interactions.

First and foremost, an interaction between rhyme awareness and coarticulation degree was found across all three consonantal contexts ( p < 0.0001). More specifically, greater rhyming skills were associated with lower coarticulation degree. Furthermore, the estimated degrees of freedom (edf) were all above 1, which indicates that rhyme proficiency was non-linearly related to an increase in children’s coarticulation scores. Nonlinear interactions between rhyme and coarticulation degree were found in each consonantal context ( Table 3 ). The nonlinearity was the highest in the alveolar context (edf: 10.778), followed by the velar and labial contexts. This means that the pattern of interaction between rhyme and coarticulation degree was specific to the gestural organization of the consonant-vowel combinations.

Table 3 . Tensor smooth terms of the generalized additive model testing for an interaction between rhyme and coarticulation degree for all children per consonantal context /b/, /d/, /g/. edf: estimated degrees of freedom.

Table 4 presents an overview of the GAM model testing for an interaction between phonemic awareness (PA) and coarticulation degree. A negative correlation was found, that is, greater phonemic proficiency coincided with lower coarticulation degree. This interaction differed significantly across consonant contexts ( p < 0.0001). The nonlinearity of the interaction was again the most prominent in the alveolar context and lowest in the labial context. Figure 3 presents three-dimensional visualizations of the nonlinear interaction patterns obtained for each consonantal context, called terrain maps. These visualizations (also called contour plots) provide further insights into the direction of the observed interaction between PA and coarticulation degree. More specifically, they depict differences in the tongue dorsum position during the production of each stop consonant (/b, d, g/ from left to right plot) with respect to the tongue dorsum position during the production of the subsequent target vowel ( y -axis) as a function of children’s PA score ( x -axis). In the plot, changes are expressed by means of a color scaling. The color scheme in the small upper right rectangle provides a referential color coding for various tongue dorsum positions scaled from 0 to 1. While blue shades characterize more anterior tongue dorsum positions (as expected for anterior vowels such as /i/), orange shades correspond to more posterior tongue positions (e.g., for /u/). The full-size plots themselves display the tongue position during the consonant as a function of its subsequent vowel position ( y -axis) and PA scores obtained (value on the x -axis). If the tongue dorsum position of the consonant is highly influenced by the upcoming vowel (i.e., if coarticulation degree is high), the color distribution within the plots is expected to resemble the referential color scaling provided for the vowel tongue dorsum positions (i.e., yellow color for more posterior and blue color for more anterior tongue dorsum positions). The red contour lines are used similarly to isolines in topographic maps (e.g. for hiking) to indicate locations sharing the same (predicted, based on all trials) value. Here, the values are not altitude landmarks, but tongue dorsum positions. Hence, red contour lines characterize locations of identical consonant tongue dorsum positions across a set of PA scores (from 0 to 24) as a function of their vocalic environment. The direction and shape of the contour line provide information whether changes in tongue dorsum position are linear (straight line) or not (curved line).

Table 4 . Tensor smooth terms of the generalized additive model testing for an interaction between phonemic awareness (composite_PA) and coarticulation degree for all children per consonantal context /b/, /d/, /g/.

Figure 3 . Terrain maps illustrating changes in the tongue dorsum gesture across three consonantal contexts (/b/: left column, /d/: middle column, /g/: right column) as a function of tongue dorsum position for target vowels ( y -axis) and composite phonological awareness scores from 0 (the minimal score obtained) to the maximal score of 25 ( x -axis).

Let us now take a concrete example. In the labial context /b/, we can see that for a target vocalic tongue dorsum position of 0.3 (value on the y-axis), the corresponding position at the consonant midpoint is about 0.4 (value on the red contour line) for children who have obtained a PA score close to 0. From a score of 10 upward, the tongue dorsum position during the consonant becomes slightly more posterior (i.e., above the 0.4 red contour line, hence further away from the target 0.3 value for its subsequent vowel).

Moving on to the alveolar context, it can be noted that the position of the tongue dorsum during the alveolar /d/ stop remains overall in a central (green shade) to anterior position (blue shade) regardless of the upcoming vowel. This shows that the tongue dorsum position during the alveolar stop resists vocalic influences due to more immediate gestural constraints requiring a more anterior to central tongue dorsum position. However, scores starting from 10 (about half the maximal score) onward are associated with a change toward a more central tongue dorsum position as compared to children with poorer PA scores. In labial and velar contexts, the color scaling characterizes more faithfully the range of vocalic targets in the antero-posterior dimension: from blue for anterior vowels to orange for more posterior vowels. This is very clear for children with a poor PA score: the tongue dorsum position for all vowels is well anticipated in the consonant. The color patterning differs in children with higher PA scores reflecting a more central tongue dorsum position (larger green portion) and hence a lower coarticulation degree. Furthermore, in velar context, the contour lines are flatter with central vowels (e.g., on y -axis: 0.5–0.6 values) and more non-linear in the context of posterior vowels (0.8 and above). In the labial context, the interaction between phonemic awareness and coarticulation degree is slightly nonlinear (edf value: 3). In Figure 3 , the red contour lines look overall flat, except with anterior vowels (e.g., 0.3 value and below). Overall, Figure 3 shows that the interaction of PA and coarticulation degree: (1) approximates linearity in labial and velar contexts contrary to the alveolar context and (2) varies as a function of the various combination of individual consonants and vowels. The implications of these nonlinear relationships between phonological and motor domains are discussed in section “Nonlinear Interactions Between Vocabulary, Phonological Awareness, and Coarticulatory Organization.”

These visual outputs differ markedly from standard numerical reports. They are quite valuable for speech production research in general and more so for the developmental field (e.g., Figure 3 ) because the continuous color scaling used in these plots can reveal gradients in target effects or interactions between parameters and hence potentially identifying nonlinear patterning. In the case of spoken language acquisition, these permit departing from categorization of children’s articulations in terms of abstract phonological targets (which they are in the process of acquiring) and instead obtain more faithful descriptions of the variety of articulatory expressions for a given target. This type of description is particularly relevant in the developmental field because like adults – and even to a greater extent than adults – children do not produce words or segments uniformly across repetitions. Acoustic and articulatory variability are indeed ubiquitous in child speech (e.g., Heisler et al., 2010 ). The color scaling in the GAM contour plots hence provides a fair depiction of the variations in tongue dorsum positions within regions associated with a specific target (e.g., individual vowels) or in interaction with a phonetic environment (e.g., a specific vowel in the context of a specific consonant).

Interaction Between Expressive Vocabulary and Coarticulation Degree

Last, we tested for an interaction between children’s expressive vocabulary and their pattern of coarticulation degree. A significant effect was found in all three consonantal contexts ( Table 5 , p < 0.0001). Overall, nonlinear patterns of interactions between domains were noted. However, those were not uniform across consonant and vowel combinations ( Figure 4 ). In the labial context, an increase in vocabulary score coincides with lower coarticulation degree. For example, in anterior vowels that have a 0.2 tongue dorsum position value ( y -axis), the corresponding tongue dorsum position during the labial stop production has a value of 0.3 in children with low vocabulary while close to 0.4 in children with advanced vocabulary. Similar trends are observed in syllables including an alveolar onset, but the interaction between vocabulary and coarticulation degree is this time more nonlinear (more pronounced curved lines) and complex than in the labial context. For children with more proficient vocabulary (e.g., score 16 upward), the tongue dorsum position is slightly more central in the case of anterior vowels (e.g., 0.2). Consonantal tongue positions in the context of central vowels (e.g., 0.6) are characterized by a slightly oscillatory behavior from more to less to more central. Last, tongue position for the alveolar stop flanked by posterior vowels (e.g., 0.8) also shows a nonlinear pattern with an overall central tongue dorsum position. Last, in the velar context, the relation between vocabulary and coarticulation degree also translates into slightly more central tongue dorsum positions in children with higher vocabulary score. To summarize, greater expressive vocabulary is associated with a more central tongue dorsum during the consonant and hence lesser influence from individual vowels.

Table 5 . Tensor function terms of the generalized additive model testing for an interaction between expressive vocabulary and coarticulation degree for all children per consonantal context /b/, /d/, /g/.

Figure 4 . Terrain maps illustrating changes in the tongue dorsum gesture across three consonantal contexts (/b/: left column, /d/: middle column, /g/: right column) as a function of tongue dorsum position for target vowels ( y -axis) and vocabulary scores from 13 (the minimal score obtained) to the maximal score of 25 ( x -axis).

In this study, we asked whether children’s phonological awareness and expressive vocabulary have an impact on anticipatory coarticulation . Our general motivation for this research stemmed from independent findings made in speech motor control and developmental phonology suggesting an increasing access to and use of phonemic units during the kindergarten-to-primary school period. Results drawn from a cross-sectional investigation of 41 children provide the first empirical evidence that vocabulary and phonological awareness interact dynamically with coarticulation degree during the period from kindergarten to primary school. In general, greater phonemic awareness and vocabulary were associated with greater segmental differentiation of tongue gestures in children’s coarticulatory organization. We expand below on the implications of those findings for the development of spoken language fluency.

Age-Related Versus Skill-Based Descriptions of Spoken Language Development

In the past decade, a fair amount of empirical research has reported greater vocabulary and phonological awareness in school-aged children than children in kindergarten (in German: Kauschke, 2000 ; Wimmer and Mayringer, 2002 ; Schäfer et al., 2014 ; in English: Carroll et al., 2003 ; Ziegler and Goswami, 2005 ). However, results from the present study suggest that age-driven categorizations are not always the only suitable ways to characterize skill development or at least they may underestimate its complexity. Several findings uphold this argument.

First of all, the language-related assessments conducted in this study provide a mixed validation of prior findings regarding a developmental increase in expressive vocabulary and phonological awareness. Indeed, our sample of kindergarten children was seemingly as proficient as first-grade children in expressive vocabulary as attested by the absence of significant age differences. Likewise, they were as proficient as first-grade children in their rhyming skills, which suggest that by the age of 4.5, they have gained awareness of intermediate size phonological components. This may be due to rhyming practices being initiated early in age, via singing, counting rhyming games at home or in kindergarten. With respect to tasks probing phonemic units, the two youngest cohorts did not differ from each other but showed significantly lower awareness than school-aged children at age 7. Interestingly in our study, the only 5-year old who could actually perform the phonemic task was able to read a few words and had knowledge about some letters. Hence, success in these tasks may emerge only once children have been explicitly trained in phonemic decoding/encoding, either in primary school in the context of reading acquisition (e.g., Ziegler and Goswami, 2005 ; Schaeffer et al., 2014 ) or with parents at home. We discuss this point further in section “An Integrated-Interactive Approach to Skill Development.”

Second, children within the same age group did not behave all in the same way but instead exhibited substantial individual variability ( Figure 2 ), a phenomenon also previously noted (e.g., review in Sosa and Stoel-Gammon, 2012 ; see also Wimmer and Mayringer, 2002 ; Schäfer et al., 2014 ). In the present study, this was the case in all three age groups and for all assessments, except for tasks probing phonemic awareness in kindergarteners (onset segments, phoneme synthesis) for which we noted a floor effect. Regarding first-grade children, it seems that while they have gained substantial awareness of sub-lexical units in comparison to children in kindergarten, it takes longer to be fully proficient in manipulating phonemic units (cf. the scores distribution, Figure 2 ). Regarding vocabulary, wide disparities across children from the same age are well-established (e.g., CDI reports within and across languages). Similar conclusions have been drawn regarding children’s coarticulatory patterns (e.g., at 4 years of age in Nittrouer and Burton, 2005 ; Barbier et al., 2015 ; at 5 years of age in Zharkova, 2017 ; overlap between 3–4-year and 5-year olds in Noiray et al., 2019 ) and here again with no systematic age-related difference in coarticulatory degree across consonantal contexts.

It is not uncommon for developmental researchers to point to between-age overlaps and/or substantial within age-group differences in various abilities. The question is then why those differences are observed. A simple answer may be that children are at different individual stages in their developmental trajectory. For instance, well-attested vocabulary spurts seem to depend on pre-existing achievements (e.g. reaching the 50 words milestone) rather than be the result of biological age progression (see review of lexical development in Nazzi and Bertoncini, 2003 ). Other studies have underlined stronger developmental dependencies based on proficiency rather than age (e.g., between phonological development and motor ability, e.g., Smith, 2006 ; Goffman, 2010 ; between vocabulary and production accuracy, e.g., Edwards et al., 2004 ; Vihman and Croft, 2007 ). When that is the case, age-related interpretations are problematic because they may attribute evidence (e.g., a decrease in coarticulation degree) to the wrong source or hide complex relationships between factors that are individual-specific rather than age-dependent. This is not to argue that age does not matter: the development of speech motor skill along with lexical and phonological knowledge can actually be described within a maturational perspective because all skills develop in the time domain. It is hence not surprising that correlations between age and phonological awareness were found in our study – albeit not with all PA tasks and not with vocabulary. However, while age-based descriptions of language acquisition may be interpreted in the perspective of biologically-driven developments, it may instead be the effect of experience upon the learning mechanism (i.e., the exposure to and practice speaking the language) that gives maturation its transformational power (e.g., in perception: Kuhl et al., 1992 ; Hay, 2018 ). Uncovering how experience shapes (spoken) language acquisition independent of age has been not only thrilling but also enduring challenge for psycholinguists because experience unfolds within an extended time scale and results from multiple interactions in a continuously variable environment that remains difficult to replicate in lab environments.

To summarize, the results reported in this study provide good incentives for future research to draw skill-based comparisons of children’s linguistic ability. With this approach, we will not only account for the complex developmental relationships across domains taking place in the first decade of life, we will also better capture the complexity of (spoken) language acquisition arising from both experience-based and biologically driven processes than if our analyses are restricted to age comparisons. This leads us to the discussion of the role of skill interactions for (spoken) language development.

Nonlinear Interactions Between Vocabulary, Phonological Awareness, and Coarticulatory Organization

As reported in previous sections, no uniformly strong differences in coarticulation degree emerged between 4-, 5- and 7-year-old children ( Table 2 ). However, children showing poor phonological awareness indicated overall greater coarticulation degree than children with higher scores. This suggests that for children with poorer phonemic representations, lingual gestures for consecutive consonants and vowels may be activated together with substantial vocalic anticipation. Further, we noted no uniform relation between coarticulation and phonemic awareness across children’s scores, by which each unit change in one domain would result in an equivalent (linear) unit change in the other domain of interest. In our sample of children, the relationship between domains was non-linear and therefore more complex: an increase in children’s phonemic awareness score was at times not associated with any equivalent change in coarticulatory pattern until reaching a certain stage. Last, those non-linear interactions varied across phonetic contexts (cf. edf values). The shape of the skill interactions indeed differed as a function of the identity of the coarticulated consonants and vowels and the compatibility of their gestural goals (cf. colored terrain maps). For instance, in the case of a syllable involving two gestures from two anatomically distinct organs (the lips for the labial /b/ and the tongue for any vowel), vocalic influences remained high regardless of children’s phonemic proficiency (rather flat isolines and all colors well represented; Figure 3 ). However, in the context of the alveolar /d/ stop that involves two consecutive lingual gestures within a short-temporal span (tongue dorsum for both /d/ and subsequent vowels), non-linear interactions were more noticeable. Children with advanced awareness of the smallest phonemic units (e.g., higher scores) exhibited slightly more central tongue dorsum positions than children with poorer ability (larger blue portion characterizing an anterior tongue position). This suggests a gradual functional decoupling between the anterior (tip-blade) and the posterior subparts of the tongue (dorsum-back). While the tongue remains in a rather anterior position during the alveolar stop production, the tongue dorsum seems a little more central as if to anticipate the production upcoming vocalic gesture. Non-linear interactions were also visible in syllables including a velar onset. Variation in phonemic awareness coincided with variation in the palatal-to-velar constriction location as a function of the vowel (see Recasens, 2014 ). While lower phonemic awareness was associated with greater vocalic influences (full color scale represented, Figure 3 ), greater awareness correlated with more central tongue positions during the consonant articulation. This finding corroborates previous research reporting a lack of speech motor independence in the early age (e.g., Nittrouer et al., 1996 ) and provides additional evidence for an important interaction with phonemic awareness, which seems particularly relevant for the coarticulation of complex gestural goals involving a single organ.

Nonlinearities were also observed in the interaction between vocabulary and coarticulatory patterns. First, results indicated that children with greater expressive vocabulary showed lower intra-syllabic coarticulation degree independently of age (cf. 0.12 correlation) and hence greater sensitivity to the gestural demands underlying various consonant-vowel combinations, while children with poorer vocabulary showed larger coarticulatory units with greater vocalic influence over previous consonants. Given numerous findings supporting a lexically grounded development of phonological representations and its impact on production accuracy (e.g., Ferguson and Farwell, 1975 ; Metsala, 1999 ; Beckman and Edwards, 2000 ; Edwards et al., 2004 , 2011 ; Munson et al., 2005 ; Vihman and Keren-Portnoy, 2013 ), our results supplement existing evidence that a rich lexical repertoire leads to greater phonological differentiation, by showing it may also support greater motor differentiation and flexibility in coarticulatory patterns depending on the gestural demands associated with consecutive segments. In the present study, the interaction between vocabulary and coarticulation degree in the alveolar context provides a compelling example that children with more proficient vocabulary show greater differentiation between the tongue dorsum and tongue tip for coarticulating consecutive consonantal and vocalic gestures recruiting the same organ. Second, the nonlinear nature of the interaction between vocabulary and coarticulation also suggests that the coupling between domains does not develop incrementally but rather that it may be when individual children reach a certain size of expressive vocabulary that the interaction with production weighs in children’s coarticulatory organization.

Taken together, results support the view of a by-stage approach to skill development. Milestones and developmental stages have long been identified in various developmental domains (e.g., walking: Thelen and Smith, 1994 ; perception: e.g. Best, 1994 ; Maye et al., 2002 ; Werker, 2018 ; spoken language: e.g., Kuhl, 2011 ; language processing: e.g., Vilain et al., 2019 ) and provide researchers with referential landmarks for a better understanding of typical trajectories, as well as useful tools for the diagnosis and prediction of potential deviations from typical pathways. In the domain of spoken language development, canonical babbling stands as an undisputed milestone allowing children to move toward a more complex quality of the speech production skill (e.g., production of the first meaningful words). This study points to a similar mechanism for skill interaction. In the same way children continuously develop individual skills (e.g., spoken language, expressive vocabulary), there may be milestones and developmental stages characterizing periods for which an interaction is (more significantly) activated. The outcome of this interaction would lead children to progress toward a new developmental stage. Taking again the relation between phonemic awareness and coarticulation, an average score reaching above 10 may characterize a developmental stage by which phonemic differentiation is maturing both at the representational and speech motor levels.

An Integrated-Interactive Approach to Skill Development

In a preceding study, we had argued that the question “whether children organize their speech in segments versus syllables versus phonological words or lexical items is twofold: It requires finding the phonological units guiding children’s speech production and the motor units embedding those higher-level units” ( Noiray et al., 2018 , p. 8). The research conducted since then motivates us to endorse an integrated-interactive approach to (spoken) language acquisition. By integrated, we mean that the gradually acquired knowledge about different unit types and sizes does not constrain children to move from one organizational scheme to another (e.g., from holistic to segmental representation of speech or vice versa). Instead, this knowledge would integrate into an increasingly more complex and flexible language system allowing children to gradually manipulate a greater variety of phonetic compounds and structural organizations ( Noiray et al., 2019 ). At the production level, this integrative process is exemplified in preschool-age children using gradients of coarticulation degree to accommodate the varying gestural demands of consecutive consonants and vowels ( Noiray et al., 2019 ). At the representational level, the way phonological awareness has been traditionally assessed directly reflects an integrative approach to phonological development: children’s structural knowledge of their native language is usually tested incrementally with tasks tapping different levels of unit complexity (e.g., words, syllables, rhymes, and segments). Phonological awareness may therefore be envisioned as an integrative learning process: it is only once children have fully integrated all organizational levels and can manipulate them into various ways that they have reached adult-like phonological representations.

The process of combinatoriality is not unique to language. In their discussion of language discreteness, Studdert-Kennedy and Goldstein (2003) had remarked on striking structural similarities between the way languages pattern and the way other processes in nature pattern (e.g., in biology, physics, chemistry). They argue for a “particulate principle” ( Abler, 1989 ) under which “units that combine into a larger unit do not disappear or lose their integrity: they can re-emerge or be recovered through mechanisms of physical, chemical, or genetic interaction, or, for language, through the mechanisms of human speech perception and language understanding” ( Studdert-Kennedy and Goldstein, 2003 , pp. 52–53). Congruent with this theoretical position, we consider a view of (spoken) language in which various structural types of combinations – gestures, segments, syllables, and words – are not mutually exclusive but reflect complementary levels of linguistic organizations that all contribute to the richness and complexity of language systems (e.g., Goffman et al., 2008 ; Noiray et al., 2019 ). From very early in development, the process of coarticulation itself binds gestures, sounds, phonetic units together to create compounds that ultimately lend meaning to speech streams. This imparts to coarticulation a special role for (spoken) language development beyond its usual circumscription to low-level motor processes. By tracking the maturation of coarticulatory organization, we can indeed capture the gradual binding of representational and executional levels. Expanding on that view, the present findings provide evidence for subtle differences in the implementation of this relationship due to the very nature of the phonemes represented in children’s mind and their motor expressions. From our preceding studies ( Noiray et al., 2013 , 2018 , 2019 ; Rubertus and Noiray, 2018 ) and research conducted in the domains of lexical and phonological development, it seems that holistic and segmental organizations (both in representation and production) develop together, albeit probably at different paces at different times. For instance, lexically based organizations may prevail at an early stage because they support object-word correspondences and referencing which are particularly relevant for children at an early stage of their life, while segmental representations may develop more slowly because they are more abstract and not bound to real-world objects. While variability in individual trajectories is evidently to be expected (e.g., Smith et al., 2010 ), overall there is converging evidence in typically developing children that these types of organization integrate with one another in the course of developing spoken language fluency (e.g., Vihman, 2015 ).

Furthermore, we argue for an interactive approach to (spoken) language development in which various skills develop together and are equally important to the uniqueness of human communication. While the literature abounds with studies highlighting developmental interactions between phonological awareness and various cognitive domains (e.g. literacy: Ziegler and Goswami, 2005 ; or with vocabulary: Charles-Luce and Luce, 1995 ; Muter et al., 2004 ; Hilden, 2016 ), the present study sheds light on the interaction between cognitive and speech motor skills. Results suggest that motor, lexical, and phonological developments collaborate dynamically over time by contact with the language (i.e., via increasingly richer exposure and practice speaking the language). This is a fairly significant finding that has various implications.

First, it may challenge models of adult speech production that have suggested a modular approach with lexical, phonological, and motor processes considered as separate components sequentially orchestrated (e.g., Levelt and Wheeldon, 1994 , Figure 1; Levelt, 1999 , Figure 1). It may also promote a revision of speech production models that have considered interactions across domains but with a top-down approach, whereby motor execution depends on the output of preceding cognitive or neural processes (e.g., in Levelt and Wheeldon’s model: motor execution is comprised within phonological encoding but implemented as the final component, p. 245; in Guenther and Vladusich, 2012 ’s DIVA model: between the motor, auditory, and somatosensory domains, Figure 1, review in Tourville and Guenther, 2011 ). If interactions between the lexical, phonological, and motor domains exist in the developing speech system of children, those should prevail in adults’ speech organization or at least residuals from such relationships may remain. Assuming a developmental continuity from children to adults’ speech production, models of speech production would benefit in taking the ontogenetic findings into account and perhaps adopt a more integrated-interactive perspective. By doing so, it may be possible to move forward in the longstanding quest for determining the nature of the units of speech production (see, for example, discussion in Pierrehumbert, 2003 ; Hickok, 2014 ).

Second, the finding of interactions across domains is relevant for the clinical field. Indeed, while predictive studies have usually tested how skill X at a time T1 predicts the stage of another skill Y at time T2 (e.g., Walley et al., 2003 ; Edwards et al., 2004 ), no study has to our knowledge ventured to examine how interactions between specific skills change over developmental time or predict the stage of another interaction at a later time. Although the present study was not designed to demonstrate a specific causal direction in the relationships observed, it is highly likely that speech motor, lexical, and phonological skills mutually influence each other over time. There is enough evidence in infant and child research supporting both directions (e.g., motor, lexical and phonological developments: Menn and Butterworth, 1983 ; DePaolis et al., 2013 ; articulatory filter hypothesis: Vihman, 1996 ; DePaolis et al., 2011 ; Majorano et al., 2014 ; phonological templates: Vihman and Croft, 2007 ; Vihman and Wauquier, 2018 ; role of articulatory skills for later phonemic awareness). Given that coarticulated speech is initiated years before children gain adult-like knowledge about the structural combinatoriality of their native language, an effect of coarticulatory practice on the development of phonological awareness is not an implausible scenario. In the first 4-to-5 years of life, children acquire a basic awareness of the structural combinatoriality of sounds (phonetic awareness) because they can form new words (real words or imaginary creations) and converse comfortably with others. This raises the question whether phonological awareness is indispensable to adult-like fluent speech or only to fluent reading. To elucidate whether it is only a by-product of literacy acquisition that happens to create collateral changes to children’s speech organization, it will be crucial to examine whether the maturational trajectories of illiterate adults or children’s coarticulatory patterns are similar to those of literate children. If they do, it may suggest that developing adult-like coarticulatory patterns does not entail any advanced awareness of the structural combinatoriality of their native language. Instead, maturation of coarticulatory patterns may relate more to children tuning their speech motor system to the phonetic regularities of their native language and therefore interact more significantly with perceptual rather than phonological development. Expanding on this hypothesis, the process of language acquisition may encompass two types of interactions: one serving oral communication and primarily involving perceptual, motor, and lexical skills; another one developing in a more protracted fashion for the purpose of literacy acquisition and involving primary interactions between motor, lexical, and phonological skills. Comparisons with preschool-aged children with advanced phonemic awareness would also provide a compelling experimental framework for assessing the role of phonological awareness with respect to speech motor control skill for developing adult-like patterns of coarticulation. In a recently funded project, we have initiated a first step in this direction, testing for interactions between various levels of phonological awareness, reading proficiency, and production fluency in typically developing school-aged children ( Popescu and Noiray, 2019 ) in comparison to children at risk or diagnosed with reading disorders.

Limitations and Perspectives for Future Research

Overall, results from the present study provide strong evidence that the process of developing spoken language fluency encompasses dynamic interactions between vocabulary, phonological awareness, and speech motor control in German children. While this represents a promising first step, further empirical work is obviously needed to understand these multidimensional interactions in greater detail. Generalized additive modeling (GAM) represents an innovative and powerful method because it can unveil nonlinear relationships between cognitive and motor domains and estimate their interrelated change over time. In the present study, it was possible to use GAM models to illuminate nonlinear patterns of interactions, which would have remained hidden if we had used linear mixed models. Note, however, our dataset presents some weaknesses. For instance, the examination of vocabulary being limited to nouns in this study, our assessment of children’s expressive lexicon was limited, and hence, correlation should be considered with caution. As mentioned earlier, it was not possible to reliably test for the combined effect of vocabulary together with phonological awareness on coarticulatory coarticulation due to dataset requirements (e.g., recording many more children and obtaining many more scores per participant). For generalized additive modeling to provide reliable results, large sample-sized investigations are also necessary, which remain challenging in the developmental field due to various methodological constraints and time-consuming data processing. However, given the growing statistical expertise among developmental psycholinguists combined with greater effort to conduct synergistic data collection across laboratories, there is no doubt that future quantitative studies will succeed in teasing apart their (in)dependent effect on the development of spoken language fluency.

The present study is part of a longer-term project aiming to elucidate whether the expansion of vocabulary and phonological awareness contributes to increasingly more segmentally specified coarticulatory organizations from kindergarten to primary school. This question is not only important for theories of language acquisition but also for clinical practice. Assessments of deviant coarticulatory patterns have primarily tested their motor origins (e.g., apraxia of speech: Nijland et al., 2002 ; speech sound disorder: Maas and Mailend, 2012 ; phonological disorders: Gibbon, 1999 ; stuttering: Lenoci and Ricci, 2018 ). Evidence of an intricate relationship with other linguistic components of the language system would certainly affect the way diagnosis and treatment are envisioned. The opposite question whether increased practice coarticulating a wide range of phonetic combinations supports greater phonemic differentiation and the stabilization of motor correspondences would be equally exciting in terms of its implications for language-related cognitive development. In this study, we have first demonstrated that important interactions between cognitive and motor domains occur in the course of developing spoken language fluency. We believe our findings now warrant longitudinal investigations to further test whether the interactions observed are bi-directional and hence fundamental to the growth of each individual skill or unilateral.

Last, if phonological awareness is the knowledge of the discrete and coarticulation represents its continuous articulatory-acoustic make-up, it will be important in future studies to design analytical approaches that can adequately account for the development of this intricate relationship over time. Dynamical systems seem a promising avenue in that respect. In a recent discussion of speech dynamics, Iskarous emphasizes that dynamical systems “do not assume separate sets of principles to describe discrete and continuous aspects of a system. Rather, the discrete description is shown to predict the continuous one, using the concept of a differential equation” ( Iskarous, 2017 , p. 8). The present study provides an ontogenetic perspective illustrating how access to various levels of phonological discreteness (words, syllables, segments) interacts with the organization of the continuous: from the production of syllabic entities to the fine integration of segmentally specified gestures. In future research on this topic, we aim to combine dynamical systems theory with longitudinal data to address how this dynamical relationship precisely unfold in the developing language system of children.

The present study tested whether developmental differences in coarticulation degree widely reported in the literature over the past decades were strictly related to maturational differences in speech motor abilities or also interacted with children’s language-related abilities. An examination of children’s coarticulatory patterns in relation to their lexical and phonological proficiency allowed us to uncover developmental differences that would remain unexplained if each skill was considered separately. Other domains, which have not been examined in the present study, are likely to play a role and should be thoroughly considered in future studies (e.g., assessment of literacy, phonological memory). The question of what skill interactions allow children to become fluent language users and how those evolve dynamically over time have become pressing issues for developmental researchers. However, for those to be uncovered interdisciplinary collaborations will be necessary, between developmental biology, psychology, and linguistics. While all domains have separately argued that multiple developments are intricately connected over time, only actual collaborations across disciplines will generate a unified account of language development.

Data Availability Statement

The datasets generated for this study are available on request to the corresponding author.

Ethics Statement

The study reported in the manuscript has been approved by the Ethic Committee of the University of Potsdam in Germany. The goals of the research, the children population recorded, the method, and recruitment procedure have been described and reviewed by the Committee prior to giving a positive review.

Author Contributions

AN provided the theoretical framework of the study, obtained the funding, and designed the empirical questions resulting in the manuscript. AN and AP conceptualized and designed the statistical analyses. AN, AP, and LH organized the dataset for subsequent statistical analyses. AP performed all statistical analyses. AN, AP, HK, ER, SK, and LH contributed to ultrasound data collection and processing and/or administration and scoring of the behavioural assessments. HK trained the team in administration and scoring the developmental assessments. AN wrote the manuscript. AN and AP provided all visualizations and edited the first draft. HK, ER, and SK provided feedback on the pre-final draft. All authors read the manuscript and agreed on its submission.

This research was generously supported by the Deutsche Forschungsgemeinschaft (DFG) grant N° 255676067 and 1098 and PredictAble (Marie Skłodowska-Curie Actions, H2020-MSCA-ITN-2014, N° 641858).

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

Many colleagues have contributed to the success of this study to whom we are indebted: Martijn Wieling for his careful guidance in the statistical analyses of the present dataset and Bodo Winter for useful related advice, Jan Ries and Mark Tiede for co-developing the SOLLAR platform used in this research, the BabyLab at University of Potsdam recruitment assistance (in particular Barbara Höhle and Tom Fritzsche), the team at Laboratory for Oral Language Acquisition (LOLA) involved in data recording and processing, and all participants enrolled in the study. We thank two reviewers for their thorough and insightful input. We are also grateful to Carol Fowler for stimulating discussions and for reviewing an earlier draft of this manuscript. Last, we shall thank the various scholars cited in this manuscript whose referential work has been a great source of inspiration. In that respect, a special thought for Michael Studdert-Kennedy who first sparked enthusiasm for this research. The publishing of this manuscript was supported by the Deutsche Forschungsgemeinschaft (DFG) and the Publishing fund of the University of Potsdam.

Abakarova, D., Iskarous, K., and Noiray, A. (2018). Quantifying lingual coarticulation in German using mutual information: an ultrasound study. J. Acoust. Soc. Am. 144, 897–907. doi: 10.1121/1.5047669

PubMed Abstract | CrossRef Full Text | Google Scholar

Abler, W. L. (1989). On the particulate principle of self-diversifying systems. J. Soc. Biol. Struct. 12, 1–13.

Google Scholar

Anthony, J. L., and Francis, D. J. (2005). Development of phonological awareness. Curr. Dir. Psychol. Sci. 14, 255–259. doi: 10.1111/j.0963-7214.2005.00376.x

CrossRef Full Text | Google Scholar

Anthony, J. L., Lonigan, C. J., Driscoll, K., Phillips, B. M., and Burgess, S. R. (2003). Phonological sensitivity: a quasi-parallel progression of word structure units and cognitive operations. Read. Res. Q. 38, 470–487. doi: 10.1598/RRQ.38.4.3

Barbier, G., Perrier, P., Ménard, L., Payan, Y., Tiede, M., and Perkell, J. (2015). “Speech planning in 4-year-old children versus adults: acoustic and articulatory analyses” in Proceedings 16th Annual Conference of the International Speech Communication Association (Germany: Dresden).

Bates, D., Maechler, M., Bolker, B., Walker, S., Christensen, R. H. B., Singmann, H., et al. (2015). Package ‘lme4’. Convergence 12.

Beckman, M. E., and Edwards, J. (2000). The ontogeny of phonological categories and the primacy of lexical learning in linguistic development. Child Dev. 71, 240–249. doi: 10.1111/1467-8624.00139

Beckman, M. E., Munson, B., and Edwards, J. (2007). Vocabulary growth and the developmental expansion of types of phonological knowledge. Lab. Phonol. 9, 241–264.

Best, C. T. (1994). “The emergence of language-specific phonemic influences in infant speech perception” in The development of speech perception: The transition from speech sounds to spoken word . 167–224.

Brady, S. A., Braze, D., and Fowler, C. A. (2011). Explaining individual differences in reading: Theory and evidence . Psychology Press.

Browman, C. P., and Goldstein, L. (1992). Articulatory phonology: an overview. Phonetica 49, 155–180. doi: 10.1159/000261913

Carroll, J. M., Snowling, M. J., Stevenson, J., and Hulme, C. (2003). The development of phonological awareness in preschool children. Dev. Psychol. 39:913. doi: 10.1037/0012-1649.39.5.913

Charles-Luce, J., and Luce, P. A. (1995). An examination of similarity neighbourhoods in young children’s receptive vocabularies. J. Child Lang. 22, 727–735. doi: 10.1017/S0305000900010023

Davis, M., and Redford, M. A. (2019). The emergence of perceptual-motor units in a production model that assumes holistic speech plans. Front. Psychol. 10:2121. doi: 10.3389/fpsyg.2019.02121

Delattre, P. (1951). The physiological interpretation of sound spectrograms . Publications of the Modern Language Association of America. 864–875.

DePaolis, R. A., Vihman, M. M., and Keren-Portnoy, T. (2011). Do production patterns influence the processing of speech in prelinguistic infants? Infant Behav. Dev. 34, 590–601. doi: 10.1016/j.infbeh.2011.06.005

DePaolis, R. A., Vihman, M. M., and Nakai, S. (2013). The influence of babbling patterns on the processing of speech. Infant Behav. Dev. 36, 642–649. doi: 10.1016/j.infbeh.2013.06.007

Edwards, J., Beckman, M. E., and Munson, B. (2004). The interaction between vocabulary size and phonotactic probability effects on children’s production accuracy and fluency in nonword repetition. J. Speech Lang. Hear. Res. 47, 421–436. doi: 10.1044/1092-4388(2004/034)

Edwards, J., Munson, B., and Beckman, M. E. (2011). Lexicon–phonology relationships and dynamics of early language development–a commentary on Stoel-Gammon’s ‘relationships between lexical and phonological development in young children’. J. Child Lang. 38, 35–40. doi: 10.1017/S0305000910000450

Ferguson, C. A., and Farwell, C. B. (1975). Words and sounds in early language acquisition: English initial consonants in the first fifty words. Lg 51, 419–439.

Fowler, A. E. (1991). “How early phonological development might set the stage for phoneme awareness” in Phonological processes in literacy: A tribute to Isabelle Y. Liberman . Vol. 106 . eds. Brady, S. A., and Shankweiler, D. P., 97–117.

Fowler, C. A. (1994). Invariants, specifiers, cues: an investigation of locus equations as information for place of articulation. Percept. Psychophys. 55, 597–610. doi: 10.3758/BF03211675

Fowler, C. A., Shankweiler, D., and Studdert-Kennedy, M. (2016). “Perception of the speech code” revisited: speech is alphabetic after all. Psychol. Rev. 123, 125–150. doi: 10.1037/rev0000013

Fricke, S., and Schäfer, B. (2008). Test für phonologische Bewusstheitsfähigkeiten (TPB) . Idstein: Schulz-Kirchner.

Fricke, S., Szczerbinski, M., Fox-Boyer, A., and Stackhouse, J. (2016). Preschool predictors of early literacy acquisition in German-speaking children. Read. Res. Q. 51, 29–53. doi: 10.1002/rrq.116

Gibbon, F. E. (1999). Undifferentiated lingual gestures in children with articulation/phonological disorders. J. Speech Lang. Hear. Res. 42, 382–397. doi: 10.1044/jslhr.4202.382

Gillon, G. T. (2007). Effective practice in phonological awareness intervention for children with speech sound disorder. Persp. Lang. Learn. Educ. 14, 18–23. doi: 10.1044/lle14.3.18

Goffman, L. (2010). “Dynamic interaction of motor and language factors in normal and disordered development” in Speech motor control . eds. Maassen, B., and van Lieshout, P. (Oxford University Press), 137–152.

Goffman, L., Smith, A., Heisler, L., and Ho, M. (2008). Speech production units in children and adults: evidence from coarticulation. J. Speech Lang. Hear. Res. 51, 1423–1437. doi: 10.1044/1092-4388(2008/07-0020)

Goswami, U., and Bryant, P. (2016). Phonological skills and learning to read : Routledge.

Green, J. R., Moore, C. A., and Reilly, K. J. (2002). The sequential development of jaw and lip control for speech. J. Speech Lang. Hear. Res. 45, 66–79. doi: 10.1044/1092-4388(2002/005)

Green, J. R., Nip, I. S., and Maassen, B. (2010). “Some organization principles in early speech development” in Speech motor control . eds. Maassen, B., and van Lieshout, P. (Oxford University Press), 171–188.

Grimme, B., Fuchs, S., Perrier, P., and Schöner, G. (2011). Limb versus speech motor control: a conceptual review. Mot. Control. 15, 5–33. doi: 10.1123/mcj.15.1.5

Guenther, F. H., and Vladusich, T. (2012). A neural theory of speech acquisition and production. J. Neurolinguistics 25, 408–422. doi: 10.1016/j.jneuroling.2009.08.006

Hay, J. (2018). Sociophonetics: the role of words, the role of context, and the role of words in context. Top. Cognit. Sci. 10, 696–706. doi: 10.1111/tops.12326

Heisler, L., Goffman, L., and Younger, B. (2010). Lexical and articulatory interactions in children’s language production. Dev. Sci. 13, 722–730. doi: 10.1111/j.1467-7687.2009.00930.x

Hickok, G. (2014). The architecture of speech production and the role of the phoneme in speech processing. Lang. Cogn. Neurosci. 29, 2–20. doi: 10.1080/01690965.2013.834370

Hilden, R. (2016). Empirische Studie zum Zusammenhang von Lexikon und Phonologischen Bewusstheitsfähigkeiten bei monolingual deutschen Kindern im Alter von 5;0 bis 6;6 Jahren. Bachelor’s thesis.

Iskarous, K. (2017). The relation between the continuous and the discrete: a note on the first principles of speech dynamics. J. Phon. 64, 8–20. doi: 10.1016/j.wocn.2017.05.003

Iskarous, K., Fowler, C. A., and Whalen, D. H. (2010). Locus equations are an acoustic expression of articulator synergy. J. Acoust. Soc. Am. 128, 2021–2032. doi: 10.1121/1.3479538

Iskarous, K., Mooshammer, C., Hoole, P., Recasens, D., Shadle, C. H., Saltzman, E., et al. (2013). The coarticulation/invariance scale: mutual information as a measure of coarticulation resistance, motor synergy, and articulatory invariance. J. Acoust. Soc. Am. 134, 1271–1282. doi: 10.1121/1.4812855

Jusczyk, P. W., Cutler, A., and Redanz, N. J. (1993). Infants’ preference for the predominant stress patterns of English words. Child Dev. 64, 675–687. doi: 10.1111/j.1467-8624.1993.tb02935.x

Katz, W. F., Kripke, C., and Tallal, P. (1991). Anticipatory coarticulation in the speech of adults and young children: acoustic, perceptual, and video data. J. Speech Hear. Res. 34, 1222–1232. doi: 10.1044/jshr.3406.1222

Kauschke, C. (2000). Der Erwerb des frühkindlichen Lexikons: eine empirische Studie zur Entwicklung des Wortschatzes im Deutschen. Vol. 27 . Gunter Narr Verlag.

Keren-Portnoy, T., Majorano, M., and Vihman, M. M. (2009). From phonetics to phonology: the emergence of first words in Italian. J. Child Lang. 36, 235–267. doi: 10.1017/s0305000908008933

Kuhl, P. K. (2011). Early language learning and literacy: neuroscience implications for education. Mind Brain Educ. 5, 128–142. doi: 10.1111/j.1751-228X.2011.01121.x

Kuhl, P. K., Williams, K. A., Lacerda, F., Stevens, K. N., and Lindblom, B. (1992). Linguistic experience alters phonetic perception in infants by 6 months of age. Science 255, 606–608. doi: 10.1126/science.1736364

Lenoci, G., and Ricci, I. (2018). An ultrasound investigation of the speech motor skills of stuttering Italian children. Clin. Linguist. Phon. 32, 1126–1144. doi: 10.1080/02699206.2018.1510983

Levelt, W. (1999). “Producing spoken language” in The neurocognition of language . (Cambridge University Press), 83–122.

Levelt, W. J., and Wheeldon, L. (1994). Do speakers have access to a mental syllabary? Cognition 50, 239–269. doi: 10.1016/0010-0277(94)90030-2

Liberman, I. Y., Shankweiler, D., Fischer, F. W., and Carter, B. (1974). Explicit syllable and phoneme segmentation in the young child. J. Exp. Child Psychol. 18, 201–212. doi: 10.1016/0022-0965(74)90101-5

Maas, E., and Mailend, M. L. (2012). Speech planning happens before speech execution: online reaction time methods in the study of apraxia of speech. J. Speech Lang. Hear. Res. 55:S1523-34. doi: 10.1044/1092-4388(2012/11-0311)

Majorano, M., Vihman, M. M., and DePaolis, R. A. (2014). The relationship between infants’ production experience and their processing of speech. Lang. Learn. Develop. 10, 179–204. doi: 10.1080/15475441.2013.829740

Mann, V., and Wimmer, H. (2002). Phoneme awareness and pathways into literacy: a comparison of German and American children. Reading Writing 15, 653–682. doi: 10.1023/a:1020984704781

Maye, J., Werker, J. F., and Gerken, L. (2002). Infant sensitivity to distributional information can affect phonetic discrimination. Cognition 82, B101–B111. doi: 10.1016/S0010-0277(01)00157-3

Menn, L., and Butterworth, B. (1983). Development of articulatory, phonetic, and phonological capabilities. Lang. Prod. 2, 3–50.

Menn, L., and Vihman, M. (2011). “Features in child phonology” in Where do phonological features come from . 261–301.

Metsala, J. L. (1999). Young children’s phonological awareness and nonword repetition as a function of vocabulary development. J. Educ. Psychol. 91:3. doi: 10.1037/0022-0663.91.1.3

Metsala, J. L. (2011). “Lexical reorganization and the emergence of phonological awareness” in Handbook of early literacy research . Vol. 3 . eds. Neuman, S. B., and Dickinson, D. K. 66–84.

Munson, B., Beckman, M., and Edwards, J. (2012). “Abstraction and specificity in early lexical representations: climbing the ladder of abstraction” in The Oxford handbook of laboratory phonology . American Speech-Language-Hearing Association (Oxford: Oxford University Press), 288–309.

Munson, B., Kurtz, B. A., and Windsor, J. (2005). The influence of vocabulary size, phonotactic probability, and wordlikeness on nonword repetitions of children with and without specific language impairment. J. Speech Lang. Hear. Res. 48, 1033–1047. doi: 10.1044/1092-4388(2005/072)

Munson, B., Swenson, C. L., and Manthei, S. C. (2005). Lexical and phonological organization in children. J. Speech Lang. Hear. Res. 48, 108–124. doi: 10.1044/1092-4388(2005/009)

Muter, V., Hulme, C., Snowling, M. J., and Stevenson, J. (2004). Phonemes, rimes, vocabulary, and grammatical skills as foundations of early reading development: evidence from a longitudinal study. Dev. Psychol. 40, 665–681. doi: 10.1037/0012-1649.40.5.665

Nazzi, T., and Bertoncini, J. (2003). Before and after the vocabulary spurt: two modes of word acquisition? Dev. Sci. 6, 136–142. doi: 10.1111/1467-7687.00263

Nazzi, T., and Cutler, A. (2019). How consonants and vowels shape spoken-language recognition. Annu. Rev. Linguist. 5, 25–47. doi: 10.1146/annurev-linguistics-011718-011919

Nicholson, H., Munson, B., Reidy, P., and Edwards, J. (2015). “Effects of age and vocabulary size on production accuracy and acoustic differentiation of young children’s sibilant fricatives” in ICPhS .

Nijland, L., Maassen, B., Meulen, S. V. D., Gabreëls, F., Kraaimaat, F. W., and Schreuder, R. (2002). Coarticulation patterns in children with developmental apraxia of speech. Clin. Linguist. Phon. 16, 461–483. doi: 10.1080/02699200210159103

Nittrouer, S., and Burton, L. T. (2005). The role of early language experience in the development of speech perception and phonological processing abilities: evidence from 5-year-olds with histories of otitis media with effusion and low socioeconomic status. J. Commun. Disord. 38, 29–63. doi: 10.1016/j.jcomdis.2004.03.006

Nittrouer, S., Studdert-Kennedy, M., and Neely, S. T. (1996). How children learn to organize their speech gestures: further evidence from fricative-vowel syllables. J. Speech Lang. Hear. Res. 39, 379–389. doi: 10.1044/jshr.3902.379

Noiray, A., Abakarova, D., Rubertus, E., Krüger, S., and Tiede, M. (2018). How do children organize their speech in the first years of life? Insight from ultrasound imaging. J. Speech Lang. Hear. Res. 61, 1355–1368. doi: 10.1044/2018_JSLHR-S-17-0148

Noiray, A., Ménard, L., and Iskarous, K. (2013). The development of motor synergies in children: ultrasound and acoustic measurements. J. Acoust. Soc. Am. 133, 444–452. doi: 10.1121/1.4763983

Noiray, A., Wieling, M., Abakarova, D., Rubertus, E., and Tiede, M. (2019). Back from the future: nonlinear anticipation in adults’ and children’s speech. J. Speech Lang. Hear. Res. 62, 3033–3054. doi: 10.1044/2019_JSLHR-S-CSMC7-18-0208

Parush, A., Ostry, D. J., and Munhall, K. G. (1983). A kinematic study of lingual coarticulation in VCV sequences. J. Acoust. Soc. Am. 74, 1115–1125. doi: 10.1121/1.390035

Perrier, P. (2012). “Gesture planning integrating knowledge of the motor plant’s dynamics: a literature review from motor control and speech motor control” in Speech planning and dynamics . eds. Fuchs, S., Weirich, M., Pape, D., and Perrier, P. (Peter Lang Publishers), 191–238.

Pierrehumbert, J. B. (2003). Phonetic diversity, statistical learning, and acquisition of phonology. Lang. Speech 46, 115–154. doi: 10.1177/00238309030460020501

Polka, L., and Werker, J. F. (1994). Developmental changes in perception of nonnative vowel contrasts. J. Exp. Psychol. Hum. Percept. Perform. 20, 421–435. doi: 10.1037//0096-1523.20.2.421

Popescu, A., and Noiray, A. (2019). “Reading proficiency and phonemic awareness as predictors for coarticulatory gradients in children” in Proceeding of BUCLD 44 , Boston.

Recasens, D. (1985). Coarticulatory patterns and degrees of coarticulatory resistance in Catalan CV sequences. Lang. Speech 28, 97–114.

Recasens, D. (2014). Coarticulation and sound change in romance . Vol. 329, John Benjamins Publishing Company.

Recasens, D. (2018). “Coarticulation” in The Oxford research encyclopedia of linguistics .

Rubertus, E., and Noiray, A. (2018). On the development of gestural organization: a cross-sectional study of vowel-to-vowel anticipatory coarticulation. PLoS One 13:e0203562. doi: 10.1371/journal.pone.0203562

Saffran, J. R., Aslin, R. N., and Newport, E. L. (1996). Statistical learning by 8-month-old infants. Science 274, 1926–1928. doi: 10.1126/science.274.5294.1926

Schaeffer, D. J., Krafft, C. E., Schwarz, N. F., Chi, L., Rodrigue, A. L., Pierce, J. E., et al. (2014). An 8‐month exercise intervention alters frontotemporal white matter integrity in overweight children. Psychophysiology 51, 728–733. doi: 10.1111/psyp.12227

Schäfer, B., Bremer, M., and Herrmann, F. (2014). Onset and phoneme awareness and its relationship to letter knowledge in German-speaking preschool children. Folia Phoniatr. Logop. 66, 126–131. doi: 10.1159/000368228

Siegmüller, J., Kauschke, C., van Minnen, B., and Bittner, D. (2010). Test zum Satzverstehen von Kindern: Eine profilorientierte Diagnostik der Syntax.

Smith, A. (2006). Speech motor development: integrating muscles, movements, and linguistic units. J. Commun. Disord. 39, 331–349. doi: 10.1016/j.jcomdis.2006.06.017

Smith, A., and Goffman, L. (1998). Stability and patterning of speech movement sequences in children and adults. J. Speech Lang. Hear. Res. 41, 18–30. doi: 10.1044/jslhr.4101.18

Smith, A., Sadagopan, N., Walsh, B., and Weber-Fox, C. (2010). Increasing phonological complexity reveals heightened instability in inter-articulatory coordination in adults who stutter. J. Fluen. Disord. 35, 1–18. doi: 10.1016/j.jfludis.2009.12.001

Smith, A., and Zelaznik, H. N. (2004). Development of functional synergies for speech motor coordination in childhood and adolescence. Dev. Psychobiol. 45, 22–33. doi: 10.1002/dev.20009

Sosa, A. V., and Stoel-Gammon, C. (2012). Lexical and phonological effects in early word production. J. Speech Lang. Hear. Res. 55, 596–608. doi: 10.1044/1092-4388(2011/10-0113)

Stoel-Gammon, C. (2011). Relationships between lexical and phonological development in young children. J. Child Lang. 38, 1–34. doi: 10.1017/S0305000910000425

Storkel, H. L., and Morrisette, M. L. (2002). The lexicon and phonology. Lang. Speech Hear. Serv. Sch. 33, 24–37. doi: 10.1044/0161-1461(2002/003)

Strycharczuk, P., and Scobbie, J. M. (2017). Fronting of southern British English high-back vowels in articulation and acoustics. J. Acoust. Soc. Am. 142, 322–331. doi: 10.1121/1.4991010

Studdert-Kennedy, M. (1998). “The particulate origins of language generativity: from syllable to gesture” in Approaches to the evolution of language . eds. Hurford, J., Studdert-Kennedy, M., and Knight, C. (Cambridge), 202–221.

Studdert-Kennedy, M. (2005). “How did language go discrete” in Evolutionary prerequisites for language . ed. Tallerman, M. (Oxford University Press).

Studdert-Kennedy, M., and Goldstein, L. (2003). Launching language: the gestural origin of discrete infinity. Stud. Evol. Lang. 3, 235–254. doi: 10.1093/acprof:oso/9780199244843.003.0013

Thelen, E., and Smith, L. B. (1994). A dynamic systems approach to the development of perception and action . Cambridge, MA: MIT Press.

Tourville, J. A., and Guenther, F. H. (2011). The DIVA model: a neural theory of speech acquisition and production. Lang. Cognit. Proc. 26, 952–981. doi: 10.1080/01690960903498424

Velleman, S. L., and Vihman, M. M. (2007). “Phonology in infancy and early childhood: implications for theories of language learning” in Phonology in context . (London: Palgrave Macmillan), 25–50.

Vihman, M. M. (1996). Phonological development: The origins of language in the child : Blackwell Publishing.

Vihman, M. (2015). “20 perception and production in phonological development” in The handbook of language emergence . Vol. 87 . (Wiley Online Library), 437.

Vihman, M. M. (2017). Learning words and learning sounds: advances in language development. Br. J. Psychol. 108, 1–27. doi: 10.1111/bjop.12207

Vihman, M., and Croft, W. (2007). Phonological development: toward a “radical” templatic phonology. Linguistics 45, 683–725. doi: 10.1515/LING.2007.021

Vihman, M. M., and Keren-Portnoy, T. (eds.) (2013). The emergence of phonology: Whole-word approaches and cross-linguistic evidence . Cambridge University Press.

Vihman, M. M., and Velleman, S. L. (1989). Phonological reorganization: a case study. Lang. Speech 32, 149–170.

Vihman, M. M., and Vihman, V. A. (2011). “From first words to segments” in Experience, variation and generalization: Learning a first language. Vol. 7 . eds. Arnon, I., and Clark, E. V., 109–133.

Vihman, M., and Wauquier, S. (2018). “Templates in child language” in Sources of variation in first language acquisition: Languages, contexts, and learners . eds. Hickmann, M., Veneziano, E., and Jisa, H. 27–44.

Vilain, A., Dole, M., Lœvenbruck, H., Pascalis, O., and Schwartz, J. L. (2019). The role of production abilities in the perception of consonant category in infants. Develop. Sci. :e12830. doi: 10.1111/desc.12830

Walley, A. C., Metsala, J. L., and Garlock, V. M. (2003). Spoken vocabulary growth: its role in the development of phoneme awareness and early reading ability. Read. Writ. 16, 5–20. doi: 10.1023/A:1021789804977

Werker, J. F. (2018). Perceptual beginnings to language acquisition. Appl. Psycholinguist. 39, 703–728. doi: 10.1017/S0142716418000152

Whalen, D. H. (1990). Coarticulation is largely planned 7/3. J. Phon. 18, 3–35. doi: 10.1016/S0095-4470(19)30356-0

Wieling, M. (2018). Analyzing dynamic phonetic data using generalized additive mixed modeling: a tutorial focusing on articulatory differences between L1 and L2 speakers of English. J. Phon. 70, 86–116. doi: 10.1016/j.wocn.2018.03.002

Wieling, M., Veenstra, P., Adank, P., and Tiede, M. (2017). Articulatory differences between L1 and L2 speakers of English. Proceedings paper Proceedings of the 11th International Seminar on Speech Production , Melbourne, August, 2019.

Wimmer, H., and Mayringer, H. (2002). Dysfluent reading in the absence of spelling difficulties: a specific disability in regular orthographies. J. Educ. Psychol. 94, 272–277. doi: 10.1037/0022-0663.94.2.272

Wimmer, H., Mayringer, H., and Landerl, K. (2000). The double-deficit hypothesis and difficulties in learning to read a regular orthography. J. Educ. Psychol. 92, 668. doi: 10.1037/0022-0663.92.4.668

Winter, B., and Wieling, M. (2016). How to analyze linguistic change using mixed models, growth curve analysis and generalized additive Modeling. J. Lang. Evol. 1, 7–18. doi: 10.1093/jole/lzv003

Wood, S. N. (2017). Generalized additive models: An introduction with R . Chapman and Hall/CRC.

Zharkova, N. (2017). Voiceless alveolar stop coarticulation in typically developing 5-year-olds and 13-year-olds. Clin. Linguist. Phon. 31, 503–513. doi: 10.1080/02699206.2016.1268209

Zharkova, N., Hewlett, N., and Hardcastle, W. J. (2011). Coarticulation as an indicator of speech motor control development in children: an ultrasound study. Mot. Control. 15, 118–140. doi: 10.1123/mcj.15.1.118

Ziegler, J. C., and Goswami, U. (2005). Reading acquisition, developmental dyslexia, and skilled reading across languages: a psycholinguistic grain size theory. Psychol. Bull. 131, 3–20. doi: 10.1037/0033-2909.131.1.3

Keywords: language acquisition, coarticulation, speech motor control, phonological awareness, vocabulary, speech production

Citation: Noiray A, Popescu A, Killmer H, Rubertus E, Krüger S and Hintermeier L (2019) Spoken Language Development and the Challenge of Skill Integration. Front. Psychol . 10:2777. doi: 10.3389/fpsyg.2019.02777

Received: 07 May 2019; Accepted: 25 November 2019; Published: 17 December 2019.

Reviewed by:

Copyright © 2019 Noiray, Popescu, Killmer, Rubertus, Krüger and Hintermeier. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Aude Noiray, [email protected]

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Review Article
• Open access
• Published: 19 June 2020

The social brain of language: grounding second language learning in social interaction

• Ping Li   ORCID: orcid.org/0000-0002-3314-943X 1 &
• Hyeonjeong Jeong   ORCID: orcid.org/0000-0002-5094-5390 2  

npj Science of Learning volume  5 , Article number:  8 ( 2020 ) Cite this article

38k Accesses

40 Citations

39 Altmetric

Metrics details

• Human behaviour

For centuries, adults may have relied on pedagogies that promote rote memory for the learning of foreign languages through word associations and grammar rules. This contrasts sharply with child language learning which unfolds in socially interactive contexts. In this paper, we advocate an approach to study the social brain of language by grounding second language learning in social interaction. Evidence has accumulated from research in child language, education, and cognitive science pointing to the efficacy and significance of social learning. Work from several recent L2 studies also suggests positive brain changes along with enhanced behavioral outcomes as a result of social learning. Here we provide a blueprint for the brain network underlying social L2 learning, enabling the integration of neurocognitive bases with social cognition of second language while combining theories of language and memory with practical implications for the learning and teaching of a new language in adulthood.

Similar content being viewed by others

A methodological perspective on learning in the developing brain

Anna C. K. van Duijvenvoorde, Lucy B. Whitmore, … Kathryn L. Mills

The language network as a natural kind within the broader landscape of the human brain

Evelina Fedorenko, Anna A. Ivanova & Tamar I. Regev

Skills and motivations underlying children’s cumulative cultural learning: case not closed

E. Reindl, A. L. Gwilliams, … C. Tennie

The study of the neuroscience of cognition has made great strides in the last two decades, thanks to the rapid developments in non-invasive neuroimaging techniques and the corresponding data analytics. At the same time, the study of language acquisition, including second language (L2) learning by children and adults, has also progressed significantly from behavioral research toward neurocognitive understanding, thanks also to new methods including neuroimaging. These two domains of study (i.e., cognitive neuroscience and language learning) have seen increasingly happy marriages of approaches, theories, and methodologies in the last two decades, driven largely by the New Science of Learning 1 , a framework for studying learning at the intersection of psychology, neuroscience, education, and machine learning. Specifically, this framework argues that learning should be studied along three important dimensions: a computational process, a social process, and a process supported by brain circuits linking perception and action. Meltzoff and colleagues 1 further suggested that human language acquisition provides a bona fide example for connecting computational learning, social learning, and brain circuits for perception and action. Despite the call from this multi-disciplinary perspective, researchers in cognitive neuroscience and language acquisition have remained to focus on the individual learner, especially in the study of adult L2 learning. This tradition has seriously limited our understanding of a key aspect of what it means to learn: learning in the social context, interactively.

The study of language learning focused on the individual might have had its origin in the tradition of generative linguistics 2 , 3 , according to which linguistics as a science should study the language competence of the idealized speaker and the corresponding innate mechanisms that enable humans to learn language. Although the neuroscience of language has largely avoided accepting the generative tradition, the focus on the individual, and consequently, the brain structure and function of the individual (i.e., the “single-brain” approach 4 ), has not changed as a field (see Fig. 1 for illustration ) . This is unfortunate, since language serves a social communicative purpose and is fundamentally a social behavior. Note that there are some exceptions to this focus, especially in (a) the study of child language learning (see discussion next), and (b) social neuroscience, which has begun to focus on how brains respond to social interactions using methodologies such as hyper-scanning 4 , 5 . In addition, although leading models of the neurobiology of language do not incorporate a social component 6 , there have been recent efforts to extend the landscape to include pragmatic reasoning 7 , theory of mind 8 , and social interaction 9 .

Left: Traditional approaches for “single-brain” study of language learning; Right: “Social-interactive brain” research and emerging methods.

In this paper, we advocate an approach focused on grounding L2 learning in social interaction; we call this approach “Social L2 Learning” (SL2). Specifically, we define “social interaction” here as “learning through real-life or simulated real-life environments where learners can interact with objects and people, perform actions, receive, use, and integrate perceptual, visuospatial, and other sensorimotor information, which enables learning and communication to become embodied.” Notwithstanding generative linguistics and individual-brain study approaches, the field of first language (L1) acquisition has clearly demonstrated that children, from the earliest stages, depend on social interactions to learn 1 . This dependence may be initially coordinated through “joint attention” and shared intentionality between the infant and the parent/caregiver 10 . Computational models that incorporate social-interactive cues from mother–child interactions perform significantly better than models with no such cues included 11 , 12 . Kuhl et al. 13 further indicated that social learning is crucial even when children learn an L2: American babies exposed to Mandarin Chinese through a “DVD condition” (pre-recorded audiovisual or audio-only material) did not demonstrate learning of Mandarin phonetic categories as did babies who were exposed to the same material through a “live condition” (experimenter interacting with the infant during learning). For adults, however, folk wisdom suggests that they can learn an L2 rapidly without social cues (e.g., through intensive training in a classroom) and may be less dependent on the presence of peer learners. Limited evidence, however, suggests that social cues such as joint attention may also enhance L2 learning success through orienting the learner’s attention to the correct meaning among competing alternatives 14 .

Theoretical frameworks for understanding the social brain of language Learning

The proposed SL2 model is focused on grounding L2 learning in social interaction based on both behavioral and brain data. A number of important theoretical framworks have already paved the way for the SL2 model, some of which are separately known in the domains of psycholinguistics, memory, and cognition, respectively.

First, while the classic Critical Period Hypothesis 15 suggests a biology-based account of effects of age of acquisition (AoA) on learning, the Competition Model, in its various formulations 16 , 17 , 18 , 19 , 20 , provides a social-based and interactive-emergentist account of the differences between L1 and L2 learning. Upon this account, the principles of learning are not fundamentally different between the child learning an L1 and the adult learning an L2 (e.g., contra the “less is more” hypothesis 21 ), but the processes and contexts within which learning takes place may be significantly different. For children, language learning is a natural event that unfolds in the environment where they grow up. They can naturally integrate the rich perceptual and sensorimotor experiences from this environment, interacting with the objects and people and performing actions in it. Picking up and using a spoon while hearing the sound “spoon” is part of the learning process, which differs from the process where adults sitting in the classroom look at a picture of spoon and associate it to an existing label in their native language. According to MacWhinney 17 , adult L2 learning is susceptible to several major “risk factors”, factors that prevent adults from acquiring a foreign language to native competence. These include thinking in L1 only (which implies the need to translate from L2 to L1 rather than directly using L2 as a medium), social isolation (learning as an individual or through in-group communities only), and lack of perception-action resonance (lack of direct contact with the target objects or actions in the environment while learning L2). These risk factors, particularly social isolation and lack of perception-action-based contexts, may explain why adult learners display the strong parasitic L2-on-L1 representations 22 : on the one hand, adults typically start to learn L2 when they have already established a solid L1 (“entrenchment” in L1), which lends easily to L2-to-L1 translation and association; on the other hand, they lack a dynamic and variable environment to build direct relations between L2 words and the objects/concepts to which the words refer 23 . With regard to the risk factors of thinking in L1 and social isolation, empirical evidence has shown that study-abroad experience may provide some environmental support, particularly in attenuating L1 to L2 interference for late adult learners 24 .

These theoretical perspectives are consistent with a larger trend in psycholinguistics to examine language learning and bilingualism not as an individualized but a general communicative experience. Adults show significant differences in how they learn two (or more) languages, the frequency and contexts with which they use the languages, and the communicative purposes for which each language is needed, therefore showing that bilingualism is a highly dynamic developmental process 19 , 25 , 26 , 27 . The SL2 approach advocated here also echoes a movement in the broader language science, from sociocultural theory 28 to usage-based language learning 29 and conversational analysis 30 , all of which view language learning as a socially grounded process. Ellis 31 summarizes this movement with regard to its focus on “how language is learned from the participatory experience of processing language during embodied interaction in social and cultural contexts where individually desired outcomes are goals to be achieved by communicating intentions, concepts, and meaning with others.”

Second and independently, human memory research suggests that item-based learning (encoding) and use (retrieval) are highly interdependent. This is due to the associative nature of memory, in which the cognitive operations used for encoding stimulus items directly impact their subsequent retrieval. A well-established hypothesis in this regard is the “encoding-specificity” principle 32 , according to which semantic memories are more successfully retrieved if they are recalled in the same context as when they were originally encoded (e.g., if word lists were encoded underwater they would be recalled better underwater than on dry land 33 ). Related to this hypothesis is the “levels of processing” theory 34 that suggests deeper, more elaborative, or richer semantic processing during encoding would lead to more successful retrieval than shallow or surface-level processing of the same material. If encoding involves more elaborative semantic processing, e.g., using multimodal information, it will have a positive impact on memory retention and retrieval. Both the “dual encoding” theory 35 and the multimedia learning theories 36 suggest that elaborative processing with multimodal sensory information could enhance the quality of semantic memory, hence leading to better recall. One of the predictions here is a “multimodal advantage” such that, for example, people learn better with words and pictures together than with words alone 37 .

There have been several studies that build on the encoding-specificity principle to account for bilingual language processing. Marian and Kaushanskaya 38 proposed a language-dependent memory hypothesis to explain bilingual semantic/conceptual representation, according to which language is encoded in the episodic memory of an event and therefore forms part of one’s autobiographical memory. It is this episodic encoding that influences the accessibility of semantic memories. They observed that memories were more accessible when retrieved in the same language in which they were originally encoded or learned. Furthermore, this language specificity in bilingual memory is influenced by variables such as AoA, proficiency in the L2, and history of usage in the two languages 39 , 40 ; for example, richer memories were associated with an earlier age of L2 learning.

The richness of memory with regard to AoA may be explained by the rich episodic experiences/events associated with specific perceptual-sensory features in the environments, perhaps because early L1 learning includes these experiences but late L2 learning typically does not. This leads us to the embodied cognition theory 41 , 42 , according to which body-specific (e.g., head, hand, foot) and modality-specific (e.g., auditory, visual, tactile) experiences form an integral part of the learner’s mental representation of concepts, objects, and actions. This contrasts with classic cognitive theories of symbolic representation that argue that cognition and cognitive operations are modular, and that language is unrelated to the rest of cognition including perception and action 43 , 44 . The embodied cognition theory highlights the whole-body interaction with the context, that is, “interaction between perception, action, the body and the environment” 45 , and when engaged, will also activate the brain’s perceptual and sensorimotor cortex 46 , 47 . Although the embodied cognition hypotheses have been examined in many studies of brain and behavior, so far, the focus has been on native L1 speakers; whether and how body-specific and modality-specific experiences play the same role in L2 learning has not received much attention 48 , 49 . Our SL2 model argues for the important role of social interaction for L2 learning and draws on the link between learning and perception and action, as suggested by the New Science of Learning framework 1 .

Social interaction for second language learning: neuroimaging evidence

How do the theoretical frameworks above shed light on our SL2 approach in understanding the social brain of L2 learning? Although many recent neuroimaging studies have examined brain changes resulting from L2 learning 50 , most of this literature has focused on traditional L2 learning methods such as rote memorization or translation-based learning, in either classroom settings or lab-based intensive training 51 , 52 , 53 , 54 , 55 . Their findings suggest largely the engagement of language-related neural networks (e.g., the classic frontal-parietal network) and memory-related brain regions (e.g., the medial temporal region for the learning and consolidation of linguistic information; see Fig. 1 for illustration). So far, only a handful of studies have provided initial evidence on the neural networks implicated in social-based L2 learning, pointing to the following key patterns.

First, the supramarginal gyrus (SMG) and the angular gyrus (AG) could play a significant role. In one of the first studies in this domain, Jeong et al. 56 trained Japanese speakers to learn Korean words under two conditions, either through L1 translation or simulated social interaction in which the participants watched videos that showed joint activities in real-life situations (e.g., the L2 target word “Dowajo”, meaning help me in English, is shown in the video with an actor trying to move a heavy bag and asking another actor for help). The authors then asked participants to retrieve the target L2 words in a functional magnetic resonance imaging (fMRI) session. The results indicated that the words learned through videos with social interactions produced more activation in the right SMG whereas the words learned from translation produced more activity in the left middle frontal gyrus (MFG). Interestingly, retrieval of L1 words (acquired by these participants in childhood through daily life) also produced greater activation in the right SMG. These findings can be interpreted to suggest that L2 words learned via social interaction (as simulated in videos through short-term training) are processed in a similar fashion as L1 words.

Second, the right inferior parietal cortex (IPL, including both SMG and AG) has been implicated more strongly in virtual reality-based (VR) interactive learning as compared with non-virtual, word-to-picture association, learning 57 . Legault and colleagues found that cortical thickness, a structural brain measure of gray-matter thickness from the surface of the cortex to the white matter, is associated with different contexts of learning: after 2–3 weeks of intensive L2 vocabulary training across seven sessions, the VR learners showed a positive correlation in the right IPL with performance across all training sessions, while the non-VR learners showed a positive correlation at the final stages only in the right inferior frontal gyrus (IFG), a region associated with effective explicit language training 58 (though there is counter evidence 59 ). Furthermore, cortical thickness in the right SMG was correlated with higher accuracy scores of the delayed retention test, but only for the VR learning group. The VR group was engaged in 3D virtual environments in which the learners could dynamically view or play with the objects in an interactive manner.

Third, the right SMG is shown to be more activated in simulated partner-based learning than individual-based learning of word meanings, indicating that the mere presence of a social partner would facilitate L2 word learning 59 , like in child language learning. Verga and Kotz 59 further found that participants with higher learning outcomes showed higher activity in the right IFG during an interactive learning condition but not during an individualized non-interactive learning condition. Levels of activity in the right lingual gyrus (LG) and right caudate nucleus (CN), previously implicated in visual search process and visuospatial learning, were also found to correlate with temporal coordination between a learner and a partner during simulated interactive learning.

These brain imaging data suggest that social-based L2 learning versus classroom-based individual learning conditions can lead to distinct neural correlates; for example, social learning of L2 may engage more strongly the brain regions for visual and spatial processing 57 , 59 , which may have consequences on both encoding (learning) and retrieval of information (memory). In contrast to the idea that only the child brain may respond to social learning, these findings suggest that the adult brain displays significant neuroplasticity in response to social interaction. Jeong et al. 56 showed that if an L2 word was initially encoded in a more socially interactive condition (through video simulations), it engaged the relevant brain areas as in L1, areas that would not become activated if learning had occurred through word association or translation as in a typical L2 classroom.

Figure 2 illustrates the proposed neural correlates of social interaction in the frontal, parietal, and subcortical regions for L2 learning. The strong engagement of the SMG, AG, IFG, along with the visual (LG) and subcortical regions (CN), may form an important neural network for understanding how SL2 is instantiated in the human brain. Importantly, this network highlights the stronger role of the right-hemisphere brain regions as compared with the typical left-lateralized language networks. The IFG has long been implicated in lexical-semantic processing and its integration with memory 60 , which is shown bilaterally in both hemispheres in Fig. 2 . The other regions, the SMG, AG, LG, CN, are illustrated in Fig. 2 on the right hemisphere. The role of this “right-heavy” network is evidence of the significant neurocognitive impacts of social L2 learning as opposed to traditional methods (Fig. 1 ).

The left hemisphere regions (blue) handle lexical-semantic processing, while the right hemisphere cortical plus the subcortical regions (green) participate in social learning. IFG inferior frontal gyrus, SMG supramarginal gyrus, AG angular gyrus, LG lingual gyrus, CN caudate nucleus, MTG/ITG middle temporal gyrus/inferior temporal gyrus (Note: the right hemisphere is depicted on the left side and left hemisphere on the right side).

There are a number of important issues for further consideration with regard to the SL2 network charted in Fig. 2 . First, it is important to understand how the various areas collaborate and communicate with each other during learning and memory. A true brain network is one that involves modules, communities, and pathways that are dynamically connected and organized. An important research direction in neuroscience today is the network science approach towards the analysis of functional/structural brain patterns underlying cognition, and significant advances have been made in applying this approach to the understanding of neural circuits of learning and memory, including L2 learning 61 , 62 , 63 . It remains to be understood how the left frontal IFG and right parietal IPL areas (including SMG and AG) form a dynamic network in support of SL2 learning, alongside the visual and subcortical regions (LG and CN). It is possible that the LG and CN regions play an important early role in visuospatial analysis and learning in social settings, which feeds into action-based lexico-semantic and conceptual integration that heavily involves the SMG and AG regions, as evidenced in studies by Verga and Kotz 59 , Jeong et al. 56 , and Legault et al. 48 . The IFG then coordinates this network with significant participation of semantic memory and cognitive control as well as lexical retrieval 64 . In this regard, the IFG also plays a significant role in modulating competition between L1 and L2 in a language control network 19 , 65 .

Second, a related issue for further study is how such neural networks evolve during development, which would allow us to understand the degree to which time of learning (e.g., AoA), extent of learning, and increased proficiency may impact the dynamic changes in the neural network 50 . Elsewhere significant progress has been made in this domain 20 , 66 , 67 , but the focus there has been on the relationship between cognitive control and bilingualism and the related debate on bilingual cognitive advantage (see a recent discussion 25 ). Methodologically, to study the developmental process we will also need to pursue longitudinal neuroimaging work 51 as well as short-term intensive training paradigms. Finally, much work is needed for understanding how the SL2 network may overlap with neural networks implicated in other types of social interaction 68 , 69 . Hagoort and Indefrey 7 , 70 suggested that pragmatic inference in language processing involves the “theory of mind” (ToM) or the mentalizing network 71 , 72 , in which the medial prefrontal (mPFC), along with the temporoparietal junction (TPJ) regions, play an important role in social reasoning such as thinking about other people’s beliefs, emotions, and intentions. Not surprisingly, the extended language network (ELN) hypothesis for narrative text comprehension 73 significantly overlaps with the ToM network, involving mPFC and the TPJ in building story coherence, drawing inference, and interpreting pragmatic meaning in the narrative story being read. The ELN network allows the reader to follow the plots, empathize with the characters, and take the protagonist’s perspectives 74 , 75 . We hypothesize that the SL2 network in Fig. 2 dynamically connects to mPFC and TPJ implicated in ToM and social reasoning, although this hypothesis needs to be examined carefully by comparing learning with social interaction versus without.

New approaches toward SL2 as a theoretical hypothesis and a practical model

Embodied semantic representation in l1 and l2.

In a typical adult L2 learning setting, students rely on translation/association of two languages and rote memory, unlike the child who acquires the L1 with sensorimotor experiences in an enriched perceptual environment. For example, in an L2 classroom, the teacher introduces a new L2 word (e.g., Japanese “inu”) by its translation equivalent in the L1 (e.g., English “dog”) and the learner’s task is to form paired associations between L1 and L2 when learning the L2 vocabulary. Although this method is efficient early on, it leads to what is called a parasitic lexical representation: the L2 word is conveniently linked to a conceptual system already established through the L1 19 , 22 . Because the task of word association or translation does not encourage direct L2-to-concept relations, the link from the L2 word to the concept is weak, and has to be indirectly mediated via the L1-to-concept link 23 . More significantly from the SL2 perspective is the “collateral damage” of this parasitism: the new L2 representation lacks the relevant perceptual-spatial-sensorimotor features (e.g., shape, size, motion and location of “inu” or dog), features that are an integral part of the lexical-semantic representation in the L1.

Why can’t the adult L2 learner take the newly acquired L2 representation and map it to the rich embodied features in the L1 representational system, given that would be the most efficient way? Several computational models 22 , 76 , 77 have systematically manipulated the timing of adding new L2 items to L1 lexical structure during simultaneous or sequential learning of the two languages and showed that the L2 lexical organization is sensitive to AoA: the later L2 is learned, the less well organized and more fragmented the L2 representations are. Thus, parasitism is characteristic of L2 semantic learning in late adulthood. Hernandez et al. 19 and Li 78 attributed this to the mechanism of “entrenchment”, in which the lexical structure established by the L1 early on is entrenched to resist radical changes during later L2 learning. The entrenchment may have led to late adult L2 learner’s inability to map L2 forms directly to the rich L1 lexico-semantic representations. In a recent neuroimaging study comparing L1 vs. L2 embodied semantic representations, Zhang et al. 79 showed that L1 speakers engage a more integrated brain network connecting key areas for language and sensorimotor integration during lexico-semantic processing, whereas L2 speakers fail to activate the necessary sensorimotor information, recruiting a less integrated embodied brain system for the same task.

The persistent parasitism could also be attributed to the different contexts in which the two languages have been learned. Recent evidence from affective processing indicates that affective-specific experiences are more strongly evoked in L1 than in L2 words due to the different contexts of social learning (e.g., family vs. workplace interactions) and the co-evolution of emotional regulation systems with early language systems 80 , 81 , 82 . Consistent with embodied semantic differences 79 , such emotionality differences between L1 and L2 have been found most reliable when the L2 is a later-learned or less proficient language 80 , showing evidence that the L2 representation, if acquired late, cannot easily incorporate the rich social and affective features of the L1 representation.

How can the L2 learner break away from this parasitism so as to establish the L2 representation on a par with the L1 representation? SL2 provides a theoretical framework for addressing this question from an embodied cognition perspective. Recent work suggests that embodied actions, even when no direct social interaction is involved, can impact learning outcomes simply by engaging the body, for example, through gestures. Mayer et al. 49 showed neurocognitive differences between (a) L2 vocabulary learning with gestures that activated the superior temporal sulcus, STS, and the premotor areas, versus (b) learning without gestures that activated the right lateral occipital cortex only. Critically, learners in the gesture condition showed significantly better memory for L2 words, hence more sustained retention, than the non-gesture learners, even after 2–6 months. Such findings point to the significance of embodied “body-specific” (hands in this case) activities for learning, and are consistent with the sensorimotor-based neural accounts of semantic representation 20 . According to the “hub-and-spoke model” 83 , 84 , “modality-specific” versus “modality-independent” (or “amodal”) representations are realized in different neural circuits, in visual/auditory/motor areas versus anterior temporal lobe, respectively. However, the outcome conceptual system must encode knowledge through integrating higher-order relationships among sensory, motor, affect, and language experiences. In this regard, one of the outstanding questions raised by Pulvermüller 84 was whether semantic learning from embodied experience and context could lead to different semantic representations in the mind and the brain. This question becomes particularly relevant when we examine the contexts of SL2 learning.

Simulated social interaction, technology, and the brain

In addition to the cognitive and neuroscience models that support SL2 theoretically, recent advances in technology have enabled us to study SL2 as a practical model toward building embodied representations in the L2 through technology-based learning. Because of the L1 vs. L2 embodied representation differences 79 , the L2 learner should aim at integrating modality-specific information with the newly acquired L2 amodal representations, in order to fully approach native-like conceptual-semantic representations. Technology-based learning could aid in this process from the earliest stages of learning, given the ample evidence from (a) technology-enhanced child language learning 85 , 86 , (b) prevalence of technology-based multimedia learning for both children and adults 36 , and (c) evidence of multimedia learning effects on the brain 37 . For example, in child language, despite a clear advantage of live learning compared to screen-based DVD learning 13 , it is now shown that direct face-to-face human interaction is not a necessary condition for infant foreign language learning. Children can benefit from technology such as Skype and other screen media platforms, provided that these technologies can deliver simulated social interactions, for example, through video chats 85 . Lytle et al. 86 showed that when the same learning materials from Kuhl et al. 13 were delivered to children through play sessions with an interactive touchscreen video, children can indeed learn from the videos. This study clearly points to both the role of interactive social play (simulated through touchscreen videos) and the impact of technology, breaking the simple dichotomy between live human learning (as effective) vs. screen-based learning (as ineffective).

In real-life learning situations, students observe and integrate multiple sources of information including actions and intentions of the speaker for using specific words in specific contexts. In a follow-up study of Jeong et al. 56 , Jeong et al. 87 examined fMRI evidence during learning (i.e., encoding), under both traditional translation and simulated social interaction conditions. The authors controlled for the amount of visual information in the two conditions by using L1 text and L1 videos as baseline comparisons. In the simulated video condition, participants had to infer the meaning of L2 target words by observing social interactions of others. Learning of L2 words in this condition resulted in additional activation in the bilateral posterior STS and right IPL. Compared with learning through L1 translation, this condition also resulted in significant positive correlations between performance scores at delayed post-test and neural activities in the right TPJ, hippocampus, and motor areas.

Jeong et al.’s new findings showed that simulated social interaction methods, compared with traditional translation/association methods, may result in stronger neural activities in key brain regions implicated for memory, perception and action, which can boost both recall and sustained long-term retention. These results are consistent with the semantic memory encoding and retrieval theories reviewed earlier. They are also consistent with recent multimedia learning effects on the brain, reflected in the bimodal encoding advantage that materials learned in multimodal conditions (e.g., learned from videos that engage both auditory and visual channels 37 ) may lead to sustained neural activities in AG, mPFC, hippocampus, posterior cingulate, and subcortical areas. These brain areas, including mPFC, TPJ, and hippocampus, significantly overlapped with the SL2 brain network and the ToM network that relies on social learning and reasoning (Fig. 2 ).

Videos or other multimedia platforms, although very effective as discussed, nevertheless have their limits with regard to social interaction and “whole-body” embodiment/engagement as in real life. Recent technological advances in immersive technologies (e.g., virtual reality, VR and augmented reality, AR) enable social interaction to a greater extent, by simulating real-world contexts and promoting student learning through active and self-exploratory discovery processes 88 . VR also provides a new platform to connect cognition, language learning, and social interaction, as it allows researchers to simulate the process of learning in its natural ecology without sacrificing experimental rigor 89 , 90 . In the current consideration, and in light of Competition Model and Embodied Cognition theories discussed, VR provides a tool for students to learn L2 in a new way. Specifically, it enables the adult learner, like the child L1 learner, to directly map (“perceptually ground”) the L2 material during learning onto objects, actions, and episodic memory to form embodied semantic representations in the L2.

Although VR has been applied to L2 teaching and learning, systematic and experimental research is still scarce in understanding the effects of VR as a function of both features of the technology and characteristics of the learner 90 . Lan et al. 91 and Hsiao et al. 92 provided early evidence in this regard. The authors trained American students to learn Mandarin Chinese vocabulary through Second Life , a popular desktop virtual platform of gaming and social networking, and demonstrated that (a) the virtual learners needed only about half of the number of exposures to gain the same level of performance as learners through traditional associative learning, and (b) virtual learners showed faster acceleration of later-stage learning. More importantly, clear individual differences in learning were observed: the low-achieving learners tended to follow a fixed route in the virtual space (using the “nearest neighbor” strategy to learn), whereas the high-achieving learners were more exploratory, grouping together similarly sounding words or similarly looking objects for learning. Interestingly, such individual difference patterns could be captured by statistical methods such as “roaming entropy” to quantify the degree or variability of movement trajectories in self-directed exploration of space, a measure previously shown to correlate with neural development during spatial navigation 93 : better learners showed higher roaming entropy, indicating more exploratory analyses of the virtual environment. Thus, navigation patterns in the VR may reflect how learners conceptually organize the environment and their abilities to explore it interactively.

Most L2 virtual learning studies, like Lan et al. 91 , have relied on desktop virtual platforms like Second Life rather than more interactive and immersive VR (iVR). Limited evidence suggests that iVR, with its more realistic simulation of the visuospatial environment and more bodily activity and interaction, leads to higher accuracy in memory recall tasks 94 . It is likely that iVR, compared to desktop VR, more strongly engages the perceptual-motor systems and maximizes the integration of modality-specific experience, and therefore generates better embodied representation 90 . In Legault et al. 48 , participants wore head-mounted displays to view and interact with objects/animals in an iVR kitchen or zoo, and showed significantly better performance of L2 vocabulary attainment than learning through the L2-to-L1 word-to-word association method. Further, the kitchen words were learned better than the animal words, presumably because the learner could more directly manipulate the virtual objects in the kitchen (e.g., squatting and picking up a broom and moving it around; see Fig. 3 for an illustration) than they could with the virtual animals in the zoo. The iVR kitchen environment thus conferred more “whole-body” interactive experience to the learner, especially with respect to the engagement of the sensorimotor system 95 .

a In the iVR kitchen, the learner used her handset to point to any item and hear the corresponding word (e.g., “dao”, Chinese knife in the example). b The learner could pick up and move objects (broom in the example) by pressing a trigger button with index finger; ( c ) position of the learner picking up the item (broom)—the learner consented to the use of her photo here. d Left panel: Effect of learning context (iVR vs. word-word association); Right panel: effect of category of learning (iVR kitchen vs. iVR zoo). Error bars indicate 95% confidence intervals (CIs). * indicates significant effect (from Legault et al. 48 ; copyright permission from MDPI).

In terms of the SL2 framework, VR has the promise of providing a context of learning for children and adults on equal footing, and in particular, it simulates “situated learning”, a condition whereby learning takes place through real-world experiences and visuospatial analyses of the learning environment, experiences and analyses that are often absent in a typical classroom 88 . Therefore, the positive benefits of SL2 learning based on either real or simulated social interactions are clear, including at least the aforementioned aspects of (a) embodied, native-like, neural representation 56 , (b) more sustained long-term memory 49 , and (c) less susceptibility to L1 interference 24 . These benefits not only apply to foreign language learning, but also other educational contents such as spatial learning and memory 90 and learning of subjects in STEM (i.e., science, technology, engineering, and mathematics) 88 .

VR is an excellent example of the power of today’s technology-based learning, and it urges us to study how students can take advantage of rapidly developing technologies for better learning outcomes. We need to pay attention to the specific key features that support VR learning (e.g., immersive experience, spatial navigation, and user interactivity), the individual differences therein (e.g., cognitive characteristics of the learner including memory and motivation), and the underlying neurocognitive mechanisms (e.g., sensorimotor integration) that enable VR as an effective tool 90 . In this regard, the SL2 approach we advocate here will have the potential of not only benefitting students in terms of reaching native-like linguistic representation and communicative competence, but also providing specific recommendations to teachers in the classroom, especially for those struggling students who may need help in integrating multiple sources of information through contextualized learning. For example, as indicated by Legault et al. 48 , it is the struggling students (“the less successful learners”) who benefitted more from VR learning than from non-VR learning, whereas for the successful learners, VR versus non-VR learning did not make a significant difference. Consistent with the larger trend in education to promote personalized learning and active learning in STEM 96 , there is a movement for today’s classroom instructions to be structured differently from the traditional “teacher-centered” instructional methods, to encourage more “student-centered” interactions and in-depth discussions (e.g., the “flipped classroom” model). E-learning technologies including VR play a significant role in this movement.

Future directions

New exciting research in the neurocognitive mechanisms of SL2 has just begun. To understand different aspects of L2 learning from a multi-level language systems and multiple networks perspective 7 , neuroimaging studies should extend their focus from the lexico-semantic level to phonological, morphological, syntactic, and discourse levels with the SL2 approach. For example, if, as in infant L1 learning, L2 phonology can be learned through socially enriched linguistic exposure (e.g., multi-talker variability, visible articulation), then even late L2 adult learners may advance to native competence 97 . It is also important to examine how social interaction impacts the acquisition of different types of syntactic rules (e.g., cross-linguistically different syntactic features), as demonstrated in a recent fMRI study of the acquisition of possessive constructions in Japanese Sign Language 98 . The relationship between lexical versus syntactic acquisition is also a topic of significant research interest. While lexical learning typically elicits stronger involvement of the declarative system, morphosyntactic learning likely involves to a greater extent the procedural memory system 99 , 100 . How L2 lexical learning may also engage the procedural memory system in light of the SL2 brain network (Fig. 2 ) needs to be seriously considered and carefully examined in future studies.

Despite the significant effects of social L2 learning, individual differences have been observed as discussed 48 , 92 . It is therefore important to examine in greater detail both the contexts of learning and the characteristics of the learner 90 . Specifically, the magnitude of the effects might depend on the interaction between features of social learning and the learner’s cognitive and linguistic abilities. It is possible that highly interactive, embodied experiences are more helpful to some than to others 48 : learners who are poor at abstract associative learning may benefit more from social-interactive learning. A challenge to future research will be to identify the nature of the interaction between the individual learner’s inherent abilities and the richness of the social learning context.

Finally, a number of new directions present further research opportunities. For example, systematic investigation is needed for understanding the role of various types of non-verbal information that may contribute to positive L2 learning outcomes. Previous cognitive neuroscience studies have provided empirical evidence that non-verbal information (e.g. gesture, communicative intention) facilitates speech comprehension and production, as well as language learning in children and adults 49 , 101 , 102 . Furthermore, it is important to study how SL2 facilitates affective processing such as emotion and motivation 81 , 103 and consequently how it engages the brain’s limbic and subcortical reward systems. As discussed earlier, there is evidence that emotional responses are more strongly associated with L1 than L2 and social contexts may be a significant contributor to this association 80 , 81 . Indeed, social interaction has been studied as one of the most crucial contributors to the development of learning motivation in L2 acquisition 104 , 105 . The SL2 approach provides a framework for integrating previous findings and hypotheses with new insights from affective and cognitive neuroscience to fully understand the social brain of language learning.

Meltzoff, A., Kuhl, P., Movellan, J. & Sejnowski, T. Foundations for a new science of learning. Science 325 , 284–288 (2009).

CAS   PubMed   PubMed Central   Google Scholar  

Chomsky, N. Syntactic structures . Janua Linguarum 4 (The Hague, Mouton, 1957).

Chomsky, N. Aspects of the Theory of Syntax (MIT, Cambridge, MA, 1965).

Redcay, E. & Schilbach, L. Using second-person neuroscience to elucidate the mechanisms of social interaction. Nat. Rev. Neurosci. 20 , 495–505 (2019).

Babiloni, F. & Astolfi, L. Social neuroscience and hyperscanning techniques: past, present and future. Neurosci. Biobehav. Rev. 44 , 76–93 (2014).

PubMed   Google Scholar  

Hickok, G. & Poeppel, D. The cortical organization of speech processing. Nat. Rev. Neurosci. 8 , 393–402 (2007).

CAS   PubMed   Google Scholar  

Hagoort, P. The neurobiology of language beyond single-word processing. Science 366 , 55–58 (2019).

Ferstl, E. C. Neuroimaging of text comprehension: Where are we now? Ital. j. linguist. 22 , 61–88 (2010).

Google Scholar  

Verga, L. & Kotz, S. A. How relevant is social interaction in second language learning? Front. Hum. Neurosci. 7 , 550 (2013).

PubMed   PubMed Central   Google Scholar  

Tomasello, M. The social-pragmatic theory of word learning. Pragmatics 10 (4), 401–413 (2000).

Yu, C. & Ballard, D. A unified model of early word learning: Integrating statistical and social cues. Neurocomputing 70 (13-15), 2149–2165 (2007).

Li, P., & Zhao, X. In Research Methods in Psycholinguistics and the Neurobiology of Language: A Practical Guide (eds. de Groot, A. & Hagoort, P.) 208–229 (Malden, MA: John Wiley & Sons, Inc., 2017).

Kuhl, P., Tsao, F. M. & Liu, H. M. Foreign-language experience in infancy: effects of short-term exposure and social interaction on phonetic learning. PNAS 100 (15), 9096–9101 (2003).

Verga, L. & Kotz, S. A. Help me if I can’t: Social interaction effects in adult contextual word learning. Cognition 168 , 76–90 (2017).

Lenneberg, E. H. Biological Foundations of Language (John Wiley & Sons, New York, 1967).

Bates, E., & MacWhinney, B. In Mechanisms of Language Acquisition (ed. MacWhinney, B.) 157–194 (Lawrence Erlbaum Associates, 1987).

MacWhinney, B. The Routledge Handbook of Second Language Acquisition (eds. Gass, S. & Mackey, A.) 211–227 (New York: Routledge, 2012).

Li, P., & MacWhinney, B. In The Encyclopedia of Applied Linguistics (ed. Chapelle, C. A.) 1–5 (John Wiley & Sons, Inc., Malden, MA, 2013).

Hernandez, A., Li, P. & MacWhinney, B. The emergence of competing modules in bilingualism. Trends Cogn. Sci. 9 , 220–5 (2005).

Hernandez, A. & Li, P. Age of acquisition: its neural and computational mechanisms. Psychol. Bull. 133 (4), 638–650 (2007).

Johnson, J. S. & Newport, E. L. Critical period effects in second language learning: the influence of maturational state on the acquisition of English as a second language. Cogn. Psychol. 21 , 60–99 (1989).

Zhao, X. & Li, P. Bilingual lexical interactions in an unsupervised neural network model. IJB 13 , 505–524 (2010).

Kroll, J. F. & Stewart, E. Category interference in translation and picture naming: evidence for asymmetric connections between bilingual memory representations. J. Mem. Lang. 33 , 149–174 (1994).

Linck, J., Kroll, J. & Sunderman, G. Losing access to the native language while immersed in a second language: Evidence for the role of inhibition in second-language learning. Psychol. Sci. 20 (12), 1507–1515 (2009).

DeLuca, V., Rothman, J., Bialystok, E. & Pliatsikas, C. Redefining bilingualism as a spectrum of experiences that differentially affects brain structure and function. PNAS 116 (15), 7565–7574 (2019).

Grosjean, F. In The Psycholinguistics of Bilingualism (eds. Grosjean, F. & Li, P.) 5–25 (Wiley & Sons, Inc., Malden, MA, 2013).

Li, P. In The Handbook of Language Emergence (eds. MacWhinney, B., & O’Grady, W.) 511–536 (John Wiley & Sons, Inc., Maiden, MA, 2015).

Lantolf, J. Sociocultural theory and L2: State of the art. Stud. Second Lang. Acquis. 28 , 67–109 (2006).

Tomasello, M. Constructing a Language: A Usage-based Theory of Language Acquisition (Harvard University Press, Cambridge, 2003).

Hall, J. K. The contributions of conversation analysis and interactional linguistics to a usagebased understanding of language: expanding the transdisciplinary framework. Mod. Lang. J. 103 , 80–94 (2019).

Ellis, N. C. Essentials of a theory of language cognition. Mod. Lang. J. 103 , 39–60 (2019).

Tulving, E. & Thomson, D. M. Encoding specificity and retrieval processes in episodic memory. Psychol. Rev. 80 (5), 352–373 (1973).

Godden, G. & Baddeley, A. Context-dependent memory in two natural environments: on land and underwater. Br. J. Psychol. 6 , 355–369 (1975).

Craik, F. I. & Lockhart, R. S. Levels of processing: a framework for memory research. J. Verbal Learning Verbal Behav. 11 (6), 671–684 (1972).

Paivio, A. Mental Representations: A Dual Coding Approach . (Oxford University Press, Oxford, 1990).

Mayer, R. E., Moreno, R., Boire, M. & Vagge, S. Maximizing constructivist learning from multimedia communications by minimizing cognitive load. J. Educ. Psychol. 91 (4), 638–643 (1999).

Liu, C., Wang, R., Li, L., Ding, G., Yang, J. & Li, P. Effects of encoding modes on memory of naturalistic events. J. Neurolinguist. 53 , 100863 (2020).

Marian, V. & Kaushanskaya, M. in Relations Between Language and Memory : Sabest Saarbrucker Beitrage zur Sprach- und Transl (ed. Zelinsky-Wibbelt, C.) 95–120 (Frankfrut, Peter Lang, 2011).

Marian, V. & Neisser, U. Language-dependent recall of autobiographical memories. J. Exp. Psychol. Gen. 129 (3), 361–368 (2000).

Marian, V. & Kaushanskaya, M. Language context guides memory content. Psychon. Bull. Rev. 14 (5), 925–933 (2007).

Barsalou, L. W., Niedenthal, P. M., Barbey, A. K. & Ruppert, J. A. In The Psychology of Learning and Motivation: Advances in Research and Theory (ed. Ross, B. H.) 43–92 (Elsevier Science, 2003).

Glenberg, A. M., Sato, M. & Cattaneo, L. Use-induced motor plasticity affects the processing of abstract and concrete language. Curr. Biol. 18 , R290–R291 (2008).

Chomsky, N. Lectures on Government and Binding (Foris Publications, Dordrecht, Holland, & Cinnaminson, NJ, 1981).

Fodor, J. A. The Modularity of Mind (MIT Press, Cambridge, MA, 1983).

Barsalou, L. W. In Embodied Grounding: Social, Cognitive, Affective, and Neuroscientific Approaches (eds. Semin, G. R. & Smith, E. R.) 9–42 (Cambridge University Press, 2008).

Aziz-Zadeh, L. & Damasio, A. Embodied semantics for actions: findings from functional brain imaging. J. Physiol. Paris 102 (1-3), 35–9 (2008).

Willems, R. M. & Casasanto, D. Flexibility in embodied language understanding. Front. Psychol. 2 , 116 (2011).

Legault, J. et al. Immersive virtual reality as an effective tool for second language vocabulary learning. Languages 4 (1), 13 (2019).

Mayer, K. M., Yildiz, I. B., Macedonia, M. & von Kriegstein, K. Visual and motor cortices differentially support the translation of foreign language words. Curr. Biol. 25 (4), 530–535 (2015).

Li, P., Legault, J. & Litcofsky, K. A. Neuroplasticity as a function of second language learning: anatomical changes in the human brain. Cortex 58 , 301–24 (2014).

Grant, A. M., Fang, S.-Y. Y. & Li, P. Second language lexical development and cognitive control: a longitudinal fMRI study. Brain Lang. 144 , 35–47 (2015).

Qi, Z., Han, M., Garel, K., Chen, E. & Gabrieli, J. White-matter structure in the right hemisphere predicts Mandarin Chinese learning success. J. Neurolinguist. 33 , 14–28 (2015).

Yang, J., Gates, K., Molenaar, P. & Li, P. Neural changes underlying successful second language word learning: An fMRI study. J. Neurolinguist. 33 , 29–49 (2015).

Breitenstein, C. et al. Hippocampus activity differentiates good from poor learners of a novel lexicon. NeuroImage 25 (3), 958–968 (2005).

Tagarelli, K., Shattuck, K., Turkeltaub, P. & Ullman, M. Language learning in the adult brain: a neuroanatomical meta-analysis of lexical and grammatical learning. NeuroImage 193 , 178–200 (2019).

Jeong, H. et al. Learning second language vocabulary: neural dissociation of situation-based learning and text-based learning. NeuroImage 50 (2), 802–9 (2010).

Legault, J., Fang, S., Lan, Y. & Li, P. Structural brain changes as a function of second language vocabulary training: Effects of learning context. Brain Cogn. 134 , 90–102 (2019).

Stein, M., Winkler, C., Kaiser, A. & Dierks, T. Structural brain changes related to bilingualism: does immersion make a difference? Front. Psychol. 5 , 1116 (2014).

Verga, L. & Kotz, S. A. Spatial attention underpins social word learning in the right fronto-parietal network. NeuroImage 195 , 165–173 (2019).

Thompson-Schill, S. Neuroimaging studies of semantic memory: inferring “how” from “where”. Neuropsychologia 41 (3), 280–292 (2003).

Bressler, S. & Menon, V. Large-scale brain networks in cognition: emerging methods and principles. Trends Cogn. Sci. 14 (6), 277–290 (2010).

Bassett, D. & Sporns, O. Network neuroscience. Nat. Neurosci. 20 (3), 353–364 (2017).

Li, P. & Grant, A. Second language learning success revealed by brain networks. Biling. Lang. Cogn. 19 , 657–664 (2016).

Hagoort, P. On Broca, brain, and binding: a new framework. Trends Cogn. Sci. 9 (9), 416–423 (2005).

Abutalebi, J. & Green, D. Bilingual language production: the neurocognition of language representation and control. J. Neurolinguist. 20 (3), 242–275 (2007).

Nichols, E. S. & Joanisse, M. F. Functional activity and white matter microstructure reveal the independent effects of age of acquisition and proficiency on second-language learning. NeuroImage 143 , 15–25 (2016).

Sun, X., Li, L., Ding, G., Wang, R. & Li, P. Effects of language proficiency on cognitive control: Evidence from resting-state functional connectivity. Neuropsychologia 129 , 263–275 (2019).

Noordzij, M. L. et al. Brain mechanisms underlying human communication. Front. Hum. Neurosci. 3 , 14 (2009).

Redcay, E. et al. Live face-to-face interaction during fMRI: a new tool for social cognitive neuroscience. NeuroImage 50 (4), 1639–47 (2010).

Hagoort, P. & Indefrey, P. The neurobiology of language beyond single words. Annu. Rev. Neurosci. 37 (1), 347–362 (2014).

Frith, C. & Frith, U. Social cognition in humans. Curr. Biol. 17 (16), R724–R732 (2007).

Adolphs, R. The social brain: neural basis of social knowledge. Annu. Rev. Psychol. 60 , 693–716 (2009).

Ferstl, E. C., Neumann, J., Bogler, C. & von Cramon, D. Y. The extended language network: a meta-analysis of neuroimaging studies on text comprehension. Hum. Brain. Mapp. 29 (5), 581–93 (2008).

Mason, R. & Just, M. The role of the theory-of-mind cortical network in the comprehension of narratives. Lang. Linguist. Compass 3 , 157–174 (2009).

Li, P. & Clariana, R. B. Reading comprehension in L1 and L2: An integrative approach. J. Neurolinguist. 50 , 94–105 (2019).

Peñaloza, C., Grasemann, U., Dekhtyar, M., Miikkulainen, R. & Kiran, S. BiLex: a computational approach to the effects of age of acquisition and language exposure on bilingual lexical access. Brain Lang ., in press, https://doi.org/10.1016/j.bandl.2019.104643 (2020).

Zhao, X. & Li, P. Simulating cross-language priming with a dynamic computational model of the lexicon. Biling.: Lang. Cogn. 16 , 288–303 (2013).

Li, P. Lexical organization and competition in first and second languages: computational and neural mechanisms. Cogn. Sci. 33 , 629–664 (2009).

Zhang, X., Yang, J., Wang, R. & Li, P. A neuroimaging study of semantic representation in first and second languages. Lang. Cogn. Neurosci . In press, https://doi.org/10.1080/23273798.2020.1738509 (2020).

Caldwell-Harris, C. L. Emotionality differences between a native and foreign language: implications for everyday life. Curr. Dir. Psychol. Sci. 24 , 214–219 (2015).

Pavlenko, A. Affective processing in bilingual speakers: disembodied cognition? Int. J. Psychol. 47 , 405–428 (2012).

Ivaz, L., Costa, A. & Dunabeitia, J. The emotional impact of being myself: emotions and foreign-language processing. J. Exp. Psychol. Learn. Mem. Cogn. 42 , 489–496 (2016).

Lambon Ralph, M. A., Jefferies, E., Patterson, K. & Rogers, T. The neural and computational bases of semantic cognition. Nat. Rev. Neurosci. 18 (1), 42–55 (2017).

Pulvermüller, F. How neurons make meaning: brain mechanisms for embodied and abstract-symbolic semantics. Trends Cogn. Sci. 17 (9), 458–470 (2013).

Myers, L., LeWitt, R., Gallo, R. & Maselli, N. Baby FaceTime: Can toddlers learn from online video chat? Dev. Sci. 20 , e12430 (2017).

Lytle, S., Garcia-Sierra, A. & Kuhl, P. Two are better than one: Infant language learning from video improves in the presence of peers. PNAS 115 (40), 9859–9866 (2018).

Jeong, H., Li, P., Suzuki, W., Sugiura, M. & Kawashima, R. Neural mechanisms of language learning from social contexts, Manuscript under review (2020).

Dede, C. Immersive interfaces for engagement and learning. Science 323 (5910), 66–69 (2009).

Peeters, D. Virtual reality: a game-changing method for the language sciences. Psychon. Bull. Rev. 26 (3), 894–900 (2019).

Li, P., Legault, J., Klippel, A. & Zhao, J. Virtual reality for student learning: understanding individual differences. Hum. Behav. Brain 1 , 28–36 (2020).

Lan, Y. J., Fang, S. Y., Legault, J. & Li, P. Second language acquisition of Mandarin Chinese vocabulary: context of learning effects. Educ. Technol. Res. Dev. 63 (5), 671–690 (2015).

Hsiao, I. Y. T., Lan, Y. J., Kao, C. L. & Li, P. Visualization analytics for second language vocabulary learning in virtual worlds. J. Educ. Techno. Soc. 20 (2), 161–175 (2017).

Freund, J. et al. Emergence of individuality in genetically identical mice. Science 340 , 756–759 (2013).

Krokos, E., Catherine, P. & Amitabh, V. Virtual memory palaces: immersion aids recall. Virtual Real. 23 , 1–15 (2019).

Johnson-Glenberg, M. C., Birchfield, D. A., Tolentino, L. & Koziupa, T. Collaborative embodied learning in mixed reality motion-capture environments: two science studies. J. Educ. Psychol. 106 (1), 86–104 (2014).

Nature Editorial. STEM education: to build a scientist. Nature 523 (7560), 371–373 (2015).

Zhang, Y. et al. Neural signatures of phonetic learning in adulthood: a magnetoencephalography study. Neuroimage 46 , 226–240 (2009).

Yusa, N., Kim, J., Koizumi, M., Sugiura, M. & Kawashima, R. Social interaction affects neural outcomes of sign language learning as a foreign language in adults. Front. Hum. Neurosci. 11 , 115, https://doi.org/10.3389/fnhum.2017.00115 (2017).

Article   PubMed   PubMed Central   Google Scholar  

Ullman, M. T. The neuronal basis of lexicon and grammar in first and second language: The declarative/procedural model. Biling. Lang. Cogn. 4 (1), 105–122 (2001).

Ullman, M. T. & Lovelett, J. T. Implications of the declarative/procedural model for improving second language learning: The role of memory enhancement techniques. Second Lang. Res. 34 (1), 39–65 (2018).

Goldin-Meadow, S. & Alibali, M. W. Gesture’s role in speaking, learning, and creating language. Annu. Rev. Psychol. 64 , 257–283 (2013).

Ozyurek, A. In Oxford Handbook of Psycholinguistics (eds. Rueschemeyer, S.A. & Gaskell, M.G.) 592–607 (Oxford University Press, 2018).

Ripollés, P. et al. The role of reward in word learning and its implications for language acquisition. Curr. Biol. CB 24 (21), 2606–11 (2014).

Dörnyei, Z. In Individual Differences and Instructed Language Learning (ed. Robinson, P.) 137–158 (John Benjamins, 2002).

MacIntyre, P. D. & Legatto, J. J. A dynamic system approach to willingness to communicate: Developing an idiodynamic method to capture rapidly changing affect. Appl. Linguist. 32 (2), 149–171 (2011).

Download references

Acknowledgements

The authors wish to thank the generous support to the research reported in this article by the US National Science Foundation’s Integrative Neural and Cognitive Systems (NCS) program (NCS-1533625; BCS-1633817) and by a Faculty Startup Fund from the Hong Kong Polytechnic University to PL and the MEXT KAKENHI Grant of Japan (#18K00776) to HJ. We thank Peter Hagoort for his helpful comments on an earlier version of this article.

Author information

Authors and affiliations.

Department of Chinese and Bilingual Studies, Faculty of Humanities, The Hong Kong Polytechnic University, Kowloon, Hong Kong, China

Graduate School of International Cultural Studies & Department of Human Brain Science, Institute of Development, Aging, and Cancer, Tohoku University, Aoba-ku, Sendai, Japan

Hyeonjeong Jeong

You can also search for this author in PubMed   Google Scholar

Contributions

P.L. and H.J. conceptualized the article. P.L. provided the overall framework of the article, and P.L. and H.J. both reviewed the literature and wrote the article.

Corresponding author

Correspondence to Ping Li .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Li, P., Jeong, H. The social brain of language: grounding second language learning in social interaction. npj Sci. Learn. 5 , 8 (2020). https://doi.org/10.1038/s41539-020-0068-7

Download citation

Received : 17 December 2019

Accepted : 29 May 2020

Published : 19 June 2020

DOI : https://doi.org/10.1038/s41539-020-0068-7

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Brain gray matter morphometry relates to onset age of bilingualism and theory of mind in young and older adults.

• Xiaoqian Li
• Kwun Kei Ng
• W. Quin Yow

Scientific Reports (2024)

Knee flexion of saxophone players anticipates tonal context of music

• Nádia Moura

npj Science of Learning (2023)

Psychological Impact of Languages on the Human Mind: Research on the Contribution of Psycholinguistics Approach to Teaching and Learning English

Journal of Psycholinguistic Research (2023)

RETRACTED ARTICLE: IoT Based Multimodal Social Interaction Activity Framework for the Physical Education System

• Liang Zhuang
• Ching-Hsien Hsu
• Priyan Malarvizhi Kumar

Wireless Personal Communications (2022)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Cognition and Young Learners’ Language Development

• Living reference work entry
• First Online: 20 October 2020
• Cite this living reference work entry

• Yuko Goto Butler 2  

Part of the book series: Springer International Handbooks of Education ((SIHE))

140 Accesses

1 Citations

Because young language learners (YLLs) are in the midst of cognitive development while they are developing their languages, it is important for educators of YLLs to understand children’s cognitive abilities in relation to their linguistic development in order to develop appropriate pedagogy for them. This chapter offers a current understanding of major cognitive abilities that are associated with YLL’s language learning. After sketching main theoretical approaches in general cognitive development, the chapter discusses major research findings concerning the potential influence of bilingualism in cognition/metacognition in the three most studied domains; namely, executive function, theory of mind, and metalinguistic awareness. The chapter also covers emerging neuropsycholinguistic research as well as studies concerning greater use of digital technology by children and its potential impact on cognition. The chapter pays special attention to the kinds of measurements that researchers have used to capture YLLs’ cognitive abilities, and the roles of linguistic inputs and other environmental factors (e.g., socioeconomic status). Although research indicates that there is certainly an association between bilingualism and cognition, there are also substantial variabilities in that association across studies. The underlying mechanisms explaining the nature of the relationships and developmental trajectories are not totally clear. The chapter concludes with suggestions for future research.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Institutional subscriptions

Adesope, O. O., Lavin, T., Thompson, T., & Ungerleider, C. (2010). A systematic review and meta-analysis of the cognitive correlates of bilingualism. Review of Educational Research, 80 (2), 207–245. https://doi.org/10.3102/0034654310368803 .

Article   Google Scholar  

Alibali, M. W., & Sidney, P. G. (2015). The role of intraindividual variability in learning and cognitive development. In M. Diehl, K. Hooker, & M. J. Sliwinski (Eds.), Handbook of intraindividual variability across the life span (pp. 84–102). New York: Routledge.

Google Scholar  

American Psychological Association. (2019). Children, youth, families and socioeconomic status. Retrieved from https://www.apa.org/pi/ses/resources/publications/children-families

Anastasi, A., & Cordova, F. (1953). Some effects of bilingualism upon the intelligence test performance of Puerto Rican children in New York City. Journal of Educational Psychology, 44 , 1–19. https://doi.org/10.1037/h0061639 .

Anderson, D. R., & Pempek, T. A. (2005). Television and very young children. American Behavioral Scientist, 48 (5), 505–522. https://doi.org/10.1177/0002764204271506 .

Apel, K. (2014). A comprehensive definition of morphological awareness. Topics in Language Disorders, 34 (3), 197–209. https://doi.org/10.1097/tld.0000000000000019 .

Barac, R., Bialystok, E., Castro, D. C., & Sanchez, M. (2014). The cognitive development of young dual language learners: A critical review. Early Childhood Research Quarterly, 29 (4), 699–714. https://doi.org/10.1016/j.ecresq.2014.02.003 .

Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have a “theory of mind”? Cognition, 21 , 37–46. https://doi.org/10.1016/0010-0277(85)90022-8 .

Barr, R. (2013). Memory constraints on infant learning from picture books, television, and touchscreens. Child Development Perspectives, 7 (4), 205–210. https://doi.org/10.1111/cdep.12041 .

Barrouillet, P. (2015). Theories of cognitive development: From Piaget to today. Developmental Review, 38 , 1–12. https://doi.org/10.1016/j.dr.2015.07.004 .

Belpaeme, T., Vogt, P., van den Berghe, R., Bergmann, K., Göksun, T., de Haas, M., … Pandey, A. K. (2018). Guidelines for designing social robots for second language tutoring. International Journal of Social Robotics , 10 , 325–341. https://doi.org/10.1007/s12369-018-0467-6 .

Berken, J. A., Gracco, V. L., & Klein, D. (2017). Early bilingualism, language attainment, and brain development. Neuropsychologia, 98 , 220–227. https://doi.org/10.1016/j.neuropsychologia.2016.08.031 .

Best, J. R., & Miller, P. H. (2010). A developmental perspective on executive function. Child Development, 81 , 1641–1660. https://doi.org/10.1111/j.1467-8624.2010.01499.x .

Bialystok, E. (1994). Analysis and control in the development of second language proficiency. Studies of Second Language Acquisition, 16 , 157–168. https://doi.org/10.1017/s0272263100012857 .

Bialystok, E. (2018). Bilingualism and executive function. In D. Millder, F. Bayram, J. Rothman, & L. Serratrice (Eds.), Bilingual cognition and language: The state of the science across its subfields (pp. 283–305). Amsterdam: John Benjamins.

Chapter   Google Scholar  

Bialystok, E., & Senman, L. (2004). Executive processes in appearance-reality tasks: The role of inhibition of attention and symbolic representation. Child Development, 75 (2), 562–579. https://doi.org/10.1111/j.1467-8624.2004.00693.x .

Bialystok, E., Majumder, S., & Martin, M. (2003). Developing phonological awareness: Is there a bilingual advantage? Applied Psycholinguistics, 24 , 27–44.

Borghi, A. M., & Fini, C. (2019). Theories and explanations in psychology. Frontiers in Psychology, 10 , 598. https://doi.org/10.3389/fpsyg.2019.00958 .

Bosseler, A. N., Taulu, S., Pihko, E., Mäkelä, J., Imada, T., Ahonen, A., & Kuhl, P. K. (2013). Theta brain rhythms index perceptual narrowing in infant speech perception. Frontiers in Psychology, 4 , 690. https://doi.org/10.3389/fpsyg.2013.00690 .

Burns, T. C., Yoshida, K. A., Hill, K., & Werker, J. F. (2007). The development of phonetic representation in bilingual and monolingual infants. Applied Psycholinguistics, 28 , 455–474. https://doi.org/10.1017/s0142716407070257 .

Bus, A. G., Takacs, Z. K., & Kegel, C. A. (2015). Affordances and limitations of electronic storybooks for young children’s emergent literacy. Developmental Review, 35 , 79–97. https://doi.org/10.1016/j.dr.2014.12.004 .

Butler, Y. G. (2017). The role of affect in intraindividual variability in task performance for young learners. TESOL Quarterly, 51 (3), 728–737. https://doi.org/10.1002/tesq.385 .

Butler, Y. G. (2018). Gaming and young learners. In S. Garton & F. Copland (Eds.), The Routledge handbook of teaching English to young learners (pp. 305–319). Oxford: Routledge.

Butler, Y. G. (2019). Teaching vocabulary to young second- or foreign-language learners: What can we learn from the research? Language Teaching for Young Learners, 1 (1), 4–33. https://doi.org/10.1075/ltyl.00003.but .

Butler, Y. G., & Liu, Y. (2019). The role of peers in young learners’ English learning: A longitudinal case study in China. In M. Sato & S. Loewen (Eds.), Evidence-based second language pedagogy: A collection of instructed second language acquisition studies (pp. 145–167). New York: Routledge.

Calkins, S. D., & Bell, M. A. (2010). Introduction: Putting the domains of development into perspective. In S. D. Calkins & M. A. Bell (Eds.), Child development at the intersection of emotion and cognition (pp. 3–13). Washington, DC: American Psychological Association.

Carlson, S. M., & Meltzoff, M. (2008). Bilingual experience and executive functioning in young children. Developmental Science, 11 , 282–298. https://doi.org/10.1111/j.1467-7687.2008.00675.x .

Carlson, S. M., & Moses, L. J. (2001). Individual differences in inhibitory control and children’s theory of mind. Child Development, 72 (4), 1032–1053. https://doi.org/10.1111/1467-8624.00333 .

Chapman, M. (1987). Piaget, attention capacity, and the functional limitations of formal structure. Advances in Child Development and Behaviour, 20 , 289–334. https://doi.org/10.1016/s0065-2407(08)60405-x .

Chen, X., & Schwartz, M. (2018). Morphological awareness and literacy in second language learners: A cross-language perspective. Reading and Writing, 31 (8), 1685–1694. https://doi.org/10.1007/s11145-018-9866-1 .

Cheung, H., Mak, W. Y., Luo, X., & Xiao, W. (2010). Sociolinguistic awareness and false belief in young Cantonese learners of English. Journal of Experimental Child Psychology, 107 , 188–194. https://doi.org/10.1016/j.jecp.2010.05.001 .

Clark, D. B., Tanner-Smith, E. E., & Killingworth, S. S. (2016). Digital games, design, and learning: A systematic review and meta-analysis. Review of Educational Research, 86 (1), 79–122. https://doi.org/10.3102/0034654315582065 .

Conboy, B. T., & Kuhl, P. K. (2011). Impact of second-language experience in infancy: Brain measures of first- and second-language speech perception. Developmental Science, 14 (2), 242–248. https://doi.org/10.1111/j.1467-7687.2010.00973.x .

David, A., & Wei, L. (2008). Individual differences in the lexical development of French–English bilingual children. International Journal of Bilingual Education and Bilingualism, 11 (5), 598–618. https://doi.org/10.1080/13670050802149200 .

Davidson, D., Raschke, V. R., & Pervez, J. (2010). Syntactic awareness in young monolingual and bilingual (Urdu–English) children. Cognitive Development, 25 , 166–182. https://doi.org/10.1016/j.cogdev.2009.07.003 .

de Bot, K. (2017). Complexity theory and dynamic systems theory: In celebration of Diane Larsen-Freeman. In L. Ortega & Z. Han (Eds.), Complexity theory and language development (pp. 51–58). Amsterdam: John Benjamins.

de Bruin, A., Treccani, B., & Sala, S. D. (2015). Cognitive advantage in bilingualism: An example of publication bias? Psychological Science, 26 (1), 99–107. https://doi.org/10.1177/0956797614557866 .

de Cat, C., Gusnanto, A., & Serratrice, L. (2018). Identifying a threshold for the executive function advantage in bilingual children. Studies in Second Language Acquisition, 40 , 119–151. https://doi.org/10.1017/s0272263116000486 .

Deák, G. O. (2014). Interrelationship of language and cognitive development (overview). In P. Brooks & V. Kampe (Eds.), Encyclopedia of language development (pp. 284–291). Thousand Oaks: Sage Publications.

Devine, R. T., & Hughes, C. (2014). Relations between false beliefs understanding and executive function in early childhood: A meta-analysis. Child Development, 85 (5), 1777–1794. https://doi.org/10.1111/cdev.12237 .

Diamond, A. (2006). The early development of executive functions. In E. Bialystok & F. I. M. Craik (Eds.), Lifespan cognition: Mechanisms of change (pp. 70–95). Oxford: Oxford University Press.

Diamond, A. (2013). Executive functions. Annual Review of Psychology, 64 , 135–168. https://doi.org/10.1146/annurev-psych-113011-143750 .

Diaz, V., & Farrar, M. J. (2018). The missing explanation of the false-belief advantage in bilingual children: A longitudinal study. Developmental Science, 21 . https://doi.org/10.1111/desc.12594 .

Dixon, L. Q., Chuang, H., & Quiroz, B. (2012). English phonological awareness in bilinguals; a cross-linguistic study of Tamil, Malay and Chinese English-language learners. Journal of Research in Reading, 35 (4), 372–392. https://doi.org/10.1111/j.1467-9817.2010.01471.x .

Donaldson, M. (1978). Children’s minds . Glasgow: Fontana.

Engel de Abreu, P., Cruz-Santos, A., Tourinho, C. J., Martin, R., & Bialystok, E. (2012). Bilingualism enriches the poor: Enhanced cognitive control in low-income minority children. Psychological Science, 23 (11), 1364–1371. https://doi.org/10.1177/0956797612443836 .

Erickson, L. C., & Thiessen, E. D. (2015). Statistical learning of language: Theory, validity, and predictions of a statistical learning account of language acquisition. Developmental Review, 37 , 66–108. https://doi.org/10.1016/j.dr.2015.05.002 .

Eviatar, Z., Taha, H., & Schwartz, M. (2018). Metalinguistic awareness and literacy among semitic-bilingual learners: A cross-language perspective. Reading and Writing, 31 (8), 1869–1891. https://doi.org/10.1007/s11145-018-9850-9 .

Fan, S. P., Liberman, Z., Keysar, B., & Kinzler, K. D. (2015). The exposure advantage: Early exposure to a multilingual environment promotes effective communication. Psychological Science, 26 (7), 1090–1097. https://doi.org/10.1177/0956797615574699 .

Farhadian, M., Abdullah, R., Mansor, M., Redzuan, M., Gazanizadand, N., & Kumar, V. (2010). Theory of mind in bilingual and monolingual preschool children. Journal of Psychology, 1 , 39–46. https://doi.org/10.1080/09764224.2010.11885444 .

Ferguson, C. J., & Donnellan, M. B. (2014). Is the association between children’s baby video viewing and poor language development robust? A reanalysis of Zimmerman, Christakis, and Meltzoff (2007). Developmental Psychology, 50 (1), 129–137. https://doi.org/10.1037/a0033628 .

Foursha-Stevenson, C., & Nicoladis, E. (2011). Early emergence of syntactic awareness and cross-linguistic influence in bilingual children’s judgments. International Journal of Bilingualism, 15 (4), 521–534. https://doi.org/10.1177/1367006911425818 .

Gajdamaschko, N. (2015). Vygotsky’s sociocultural theory. In J. Wright (Ed.), International encyclopedia of the social & behavioral sciences (2nd ed.). Amsterdam: Elsevier. https://doi.org/10.1016/B978-0-08-097086-8.23203-0 .

Goetz, P. (2003). The effects of bilingualism on theory of mind development. Bilingualism: Language and Cognition, 6 , 1–15. https://doi.org/10.1017/s1366728903001007 .

Gooch, D., Thompson, P., Nash, H. M., Snowling, M. J., & Hulme, C. (2016). The development of executive function and language skills in the early school years. Journal of Child Psychology and Psychiatry, 57 (2), 180–187. https://doi.org/10.1111/jcpp.12458 .

Gopnik, A., & Bonawitz, E. (2015). Bayesian models of child development. Wiley Interdisciplinary Reviews: Cognitive Science, 6 , 75–86. https://doi.org/10.1002/wcs.1330 .

Grundy, J. G., & Timmer, K. (2017). Bilingualism and working memory capacity: A comprehensive meta-analysis. Second Language Research, 33 (3), 325–340. https://doi.org/10.1177/0267658316678286 .

Halford, G. S., & Andrews, G. (2015). Information-processing models of cognitive development. In U. Goswami (Ed.), The Wiley-Blackwell handbook of childhood cognitive development (pp. 697–722). Malden: Wiley.

Halford, G. S., Wilson, W. H., & Philips, S. (1998). Processing capacity defined by relational complexity: Implications for comparative, developmental, and cognitive psychology. Behavioral and Brain Sciences, 21 , 803–831. https://doi.org/10.1017/s0140525x98001769 .

Kail, R. (1986). Sources of age differences in speed of processing. Child Development, 57 , 969–987. https://doi.org/10.2307/1130372 .

Koda, K. (2005). Insights into second language reading: A cross-linguistic approach . Cambridge: Cambridge University Press.

Book   Google Scholar  

Kovács, A. M. (2009). Early bilingualism enhances mechanisms of false-belief reasoning. Developmental Science, 12 (1), 48–54. https://doi.org/10.1111/j.1467-7687.2008.00742.x .

Kozulin, A. (2015). Vygotsky’s theory of cognitive development. In J. Wright (Ed.), International encyclopedia of the social & behavioral sciences (2nd ed.). Amsterdam: Elsevier. https://doi.org/10.1016/B978-0-08-097086-8.23094 .

Krizman, J., Marian, V., Shook, A., Skoe, E., & Kraus, N. (2012). Subcortical encoding of sounds is enhanced in bilinguals and relates to executive function advantages. Proceedings of the National Academy of Sciences of the United States of America, 109 (20), 7877–7887.

Kuhl, P. K. (2010). Brain mechanism in early language acquisition. Neuron, 67 (5), 713–727. https://doi.org/10.1016/j.neuron.2010.08.038 .

Kuhl, P. K. (2011). Social mechanisms in early language acquisition: Understanding integrated brain systems supporting language. In J. Decety & J. Cacioppo (Eds.), The handbook of social neuroscience (pp. 649–667). Oxford: Oxford University Press.

Kuo, L., & Anderson, R. C. (2008). Conceptual and methodological issues in comparing metalinguistic awareness across languages. In K. Koda & A. M. Zehler (Eds.), Learning to read across languages: Cross-linguistic relationships in first- and second-language literacy development (pp. 39–67). New York: Routledge.

Ladas, A. I., Carroll, D. J., & Vivas, A. B. (2015). Attentional processes in low-socioeconomic status bilingual children: Are they modulated by the amount of bilingual experience? Child Development, 86 (2), 557–578. https://doi.org/10.1111/cdev.12332 .

Lantolf, J. P. (Ed.). (2000). Sociocultural theory and second language learning . Oxford: Oxford University Press.

Larsen-Freeman, D. (2011). A complexity theory approach to second language development/acquisition. In D. Atkinson (Ed.), Alternative approaches to second language acquisition (pp. 48–72). London: Routledge.

Lawson, G. M., Hook, C. J., & Farah, M. J. (2018). A meta-analysis of the relationship between socioeconomic status and executive function among children. Developmental Science, 21 . https://doi.org/10.1111/desc.12529 .

Lewis, M. D. (2000). The promise of dynamic systems approaches for an integrated account of human development. Child Development, 71 , 36–43. https://doi.org/10.1111/1467-8624.00116 .

Liberman, Z., Woodward, A. L., Keysar, B., & Kinzler, K. D. (2017). Exposure to multiple languages enhances communication skills in infancy. Developmental Science, 20 . https://doi.org/10.1111/desc.12420 .

Liu, D., Wellman, H., Tardif, T., & Sabbagh, M. (2008). Theory of mind development among Chinese children: A meta-analysis of false-belief understanding across cultures and languages. Developmental Psychology, 44 , 523–531. https://doi.org/10.1037/0012-1649.44.2.523 .

Loizou, M., & Stuart, M. (2003). Phonological awareness in monolingual and bilingual English and Greek five-year-olds. Journal of Research in Reading, 26 (1), 3–18. https://doi.org/10.1111/1467-9817.261002 .

MacLeod, C. M. (2015). Attention: Beyond Stroop’s (1935) colour-word interference phenomenon. In M. W. Eysenck & D. Groome (Eds.), Cognitive psychology: Revising the classic studies (pp. 60–70). London: Sage.

Marchman, V. A., Martínez, L. Z., Hurtado, N., Grüter, T., & Fernald, A. (2017). Caregiver talk to young Spanish–English bilinguals: Comparing direct observation and parent-report measures of dual-language exposure. Developmental Science, 20 . https://doi.org/10.1111/desc.12425 .

Mascolo, M. F. (2015). Neo-Piagetian theories of cognitive development. In J. Wright (Ed.), International encyclopedia of the social & behavioral sciences (2nd ed.). Amsterdam: Elsevier. https://doi.org/10.1016/B978-0-08-097086-8.23097-3 .

Miller, P. H. (2014). Piaget’s theory: Past, present, and future. In U. Goswami (Ed.), The Wiley-Blackwell handbook of childhood cognitive development (pp. 649–672). Malden: Wiley.

Miyake, A., & Friedman, N. P. (2012). The nature and organization of individual differences in executive functions: Four general conclusions. Current Directions in Psychological Science, 21 (1), 8–14. https://doi.org/10.1177/0963721411429458 .

Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. (2000). The unity and diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis. Cognitive Psychology, 41 , 49–100. https://doi.org/10.1006/cogp.1999.0734 .

Mohades, S. G., Struys, E., Von Schuerbeek, P., Mondt, K., Van De Craen, P., & Luypaert, R. (2012). DTI reveals structural differences in white matter tracts between bilingual and monolingual children. Brain Research, 1435 , 72–80. https://doi.org/10.1016/j.brainres.2011.12.005 .

Moll, L. C., Amanti, C., Neff, D., & Gonzalez, N. (1992). Funds of knowledge for teaching: Using a qualitative approach to connect homes and classrooms. Theory Into Practice, 31 (2), 132–141. https://doi.org/10.1080/00405849209543534 .

Murakami, A. (2016). Modeling systematicity and individuality in nonlinear second language development: The case of English grammatical morphemes. Language Learning, 66 (4), 834–871. https://doi.org/10.1111/lang.12166 .

Nathanson, A. I., Aladé, F., Sharp, M. L., Rasmussen, E. E., & Christy, K. (2014). The relation between television exposure and executive function among preschoolers. Developmental Psychology, 50 (5), 1497–1506. https://doi.org/10.1037/a0035714 .

Nguyen, T., & Astington, J. W. (2014). Reassessing the bilingual advantage in theory of mind and its cognitive underpinnings. Bilingualism: Language and Cognition, 17 (2), 396–409. https://doi.org/10.1017/s1366728913000394 .

Oatley, K., Parrott, W. G., Smith, C., & Watts, F. (2011). Cognition and emotion over twenty-five years. Cognition and Emotion, 25 (8), 1341–1348. https://doi.org/10.1080/02699931.2011.622949 .

Paap, K. R., Anders-Jefferson, R., Wason, L., Alvarado, K., & Zimiga, B. (2018). Bilingual advantages in inhibition or selective attention: More challenges. Frontiers in Psychology, 9 . https://doi.org/10.3389/fpsyg.2018.01409 .

Paradis, J. (2011). Individual differences in child English second language acquisition: Comparing child internal and child external factors. Linguistic Approaches to Bilingualism, 1 , 213–237. https://doi.org/10.1075/lab.1.3.01par .

Peal, E., & Lambert, W. E. (1962). The relation of bilingualism to intelligence. Psychological Monographs, 76 , 1–23. https://doi.org/10.1037/h0093840 .

Pearl, L., & Goldwater, S. (2016). Statistical learning, inductive bias, and Bayesian inference in language acquisition. In J. L. Lidz, W. Snyder, & J. Pater (Eds.), Oxford handbook of developmental linguistics . Oxford: Oxford University Press. https://doi.org/10.1093/oxfordhb/9780199601264.013.28 .

Piaget, J. (1952). Origins of intelligence in the child (trans: Cook, A.). New York: International Universities Press. (Original Universities Press work published 1936).

Pierce, L. J., Genesee, F., Delcenserie, A., & Morgan, G. (2017). Variations in phonological working memory: Linking early language experiences and language learning outcomes. Applied Psycholinguistics, 38 , 1265–1300. https://doi.org/10.1017/s0142716417000236 .

Raviv, L., & Arnon, I. (2018). The developmental trajectory of children’s auditory and visual statistical learning abilities: Modality-based differences in the effect of age. Developmental Science, 21 . https://doi.org/10.1111/desc.12593 .

Reid Chassiakos, Y., Radesky, J., Christakis, D., Moreno, M. A., & Cross, C. (2016). Children and adolescents and digital media. Pediatrics, 138 (5), e20162593.

Rolla San Francisco, A. R., Carlo, M., August, D., & Snow, C. (2006). The role of language of instruction and vocabulary in the English phonological awareness of Spanish-English bilingual children. Applied Psycholinguistics, 27 (2), 229–246. https://doi.org/10.1017/s0142716406060267 .

Romberg, A. R., & Saffran, J. R. (2010). Statistical learning and language acquisition. Wiley Interdisciplinary Reviews: Cognitive Science, 1 (6), 906–914.

Romeo, R. R., Leonard, J. A., Robinson, S. T., West, M. R., Mackey, A. P., Rowe, M. L., & Gabrieli, J. D. E. (2018). Beyond the 30-million-word gap: Children’s conversational exposure is associated with language-related brain function. Psychological Science, 29 (5), 700–710. https://doi.org/10.1177/0956797617742725 .

Sabourin, L., & Vīnerte, S. (2015). The bilingual advantage in the Stroop task: Simultaneous vs. early bilinguals. Bilingualism: Language and Cognition, 18 (2), 350–355. https://doi.org/10.1017/s1366728914000704 .

Saito, D., Tanabe, H., Izuma, K., Hayashi, M., Morito, Y., Komeda, H., … Sadato, N. (2010). “Stay tuned”: Inter-individual neural synchronization during mutual gaze and join attention. Frontiers in Integrative Neuroscience , 4 . https://doi.org/10.3389/fnint.2010.00127 .

Schwarz, M., & Shaul, S. (2018). Towards a better understanding of the link between executive functions and pre-academic abilities. Written Language and Literacy, 21 (2), 238–268. https://doi.org/10.1075/wll.00016.sch .

Schwarz, M., Taha, H., Assad, H., Khamaisi, F., & Eviatar, Z. (2016). The role of emergent bilingualism in the development of morphological awareness in Arabic and Hebrew. Journal of Speech, Language, and Hearing Research, 59 , 797–809. https://doi.org/10.1044/2016_jslhr-l-14-0363 .

Shahaeian, A., Peterson, C. C., Slaughter, V., & Wellman, H. M. (2011). Culture and the sequence of steps in theory of mind development. Developmental Psychology, 47 (5), 1239–1247. https://doi.org/10.1037/a0023899 .

Šimleša, S., & Cepanec, M. (2015). Development of executive functions during childhood. In J. Wright (Ed.), International encyclopedia of the social & behavioral sciences (2nd ed.). Amsterdam: Elsevier. https://doi.org/10.1016/B978-0-08-097086-8.23083-3 .

Šimleša, S., Cepanec, M., & Ljubešić, M. (2017). The role of executive functions in language comprehension in preschool children. Psychology, 8 , 227–245. https://doi.org/10.4236/psych.2017.82013 .

Smith, M. E. (1939). Some light on the problem of bilingualism as found from a study of the progress in mastery of English among preschool children of non-American ancestry in Hawaii. Genetic Psychology Monographs, 21 , 121–284.

Sorge, G. B., Toplak, M. E., & Bialystok, E. (2017). Interactions between levels of attention ability and levels of bilingualism in children’s executive functioning. Developmental Science, 20 . https://doi.org/10.1111/desc.12408 .

Strouse, G. A., O’Doherty, K., & Troseth, G. L. (2013). Effective coviewing: Preschoolers’ learning from video after a dialogic questioning intervention. Developmental Psychology, 49 (12), 2368–2382. https://doi.org/10.1037/a0032463 .

Sun, H., Yin, B., Amsah, F., & O’Brien, B. A. (2018). Differential effects of internal and external factors in early bilingual vocabulary learning: The case of Singapore. Applied Psycholinguistics, 39 (2), 383–411. https://doi.org/10.1017/S014271641700039X .

Sun, H., Loh, J. Y., & Charles, A. C. (2019). Motion and sound in animated storybooks for preschooler’s total fixation time and mandarin language learning: An eye-tracking study with Singaporean bilingual children. AERA Open, 5 , 2), 1–2),19. https://doi.org/10.1177/2332858419848431 .

Takanishi, R., & Le Menestrel, S. (2017). Promoting educational success of children and youth learning English: Promising futures . Washington, DC: The National Academies Press.

Thomas-Sunesson, D., Hakuta, K., & Bialystok, E. (2018). Degree of bilingualism modifies executive control in Hispanic children in the USA. International Journal of Bilingual Education and Bilingualism, 21 (2), 197–206. https://doi.org/10.1080/13670050.2016.1148114 .

Valian, V. (2015). Bilingualism and cognition. Bilingualism: Language and Cognition, 18 (1), 3–24. https://doi.org/10.1017/s1366728914000522 .

Verga, L., & Kotz, S. A. (2013). How relevant is social interaction in second language learning? Frontiers in Human Neuroscience, 7 . https://doi.org/10.3389/fnhum.2013.00550 .

Vogt, P., de Hass, M., de Jong, C., Baxter, P., & Krahmer, E. (2017). Child-robot interactions for second language tutoring to preschool children. Frontiers in Human Neuroscience, 11 . https://doi.org/10.3389/fnhum.2017.00073 .

Vulchanova, M., Baggio, G., Cangelosi, A., & Smith, L. (2017). Editorial: Language development in the digital age. Frontiers in Human Neuroscience, 11 . https://doi.org/10.3389/fnhum.2017.00447 .

Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes (Eds. and trans: Cole, M., John-Steiner, V., Scribner, S., & Souberman, E.). Cambridge, MA: Harvard University Press. (Original work published 1930).

Vygotsky, L. S. (1986). Thought and language (Ed. and trans: Kozulin, A.). Cambridge, MA: MIT Press. (Original work published 1934).

Wellman, H. M. (2014). Developing a theory of mind. In U. Goswami (Ed.), The Wiley-Blackwell handbook of childhood cognitive development (pp. 258–284). Malden: Wiley.

White, L. J., & Greenfield, D. B. (2017). Executive functioning in Spanish- and English-speaking head start preschoolers. Developmental Science, 20 . https://doi.org/10.1111/desc.12502 .

Woodard, E., & Rodman, A. (2007). Socio-economic status as a mediating factor in the relationship between bilingualism and the cognitive development of pre-school children . Paper presented at the annual meeting of the National Communication Association 93rd annual convention.

Wouters, P., van Nimwegen, C., van Oostendorp, H., & van der Spek, E. D. (2013). A meta-analysis of the cognitive and motivational effects of serious games. Journal of Educational Psychology, 105 (2), 249–265. https://doi.org/10.1037/a0031311 .

Yoshioka, J. G. (1929). A study of bilingualism. Journal of Genetic Psychology, 36 , 473–479.

Zelazo, P. D., & Müller, U. (2014). Executive function in typical and atypical development. In U. Goswami (Ed.), The Wiley-Blackwell handbook of childhood cognitive development (pp. 574–603). Malden: Wiley.

Zelazo, P. D., Müller, U., Frye, D., & Marcovitch, S. (2003). The development of executive function. Monographs of the Society for Research in Child Development, 68 (3), 1–115.

Zhang, Y., Song, H., Liu, X., Tang, D., Chen, Y., & Zhang, X. (2017). Language learning enhanced by massive multiple online role-playing games (MMORPGs) and the underlying behavioral and neural mechanism. Frontiers in Human Neuroscience, 11 . https://doi.org/10.3389/fnhum.2017.00095 .

Zimmerman, F. J., Christakis, D. A., & Meltzoff, A. N. (2007). Associations between media viewing and language development in children under age two years. The Journal of Pediatrics, 151 , 364–368. https://doi.org/10.1016/j.jpeds.2007.04.071 .

Download references

Author information

Authors and affiliations.

Educational Linguistics Division, University of Pennsylvania Graduate School of Education, Philadelphia, PA, USA

Yuko Goto Butler

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Yuko Goto Butler .

Editor information

Editors and affiliations.

Language Department, Oranim Academic College of Education, Kiriat Tivon, Israel

Mila Schwartz

Rights and permissions

Reprints and permissions

Copyright information

© 2020 Springer Nature Switzerland AG

About this entry

Cite this entry.

Butler, Y.G. (2020). Cognition and Young Learners’ Language Development. In: Schwartz, M. (eds) Handbook of Early Language Education. Springer International Handbooks of Education. Springer, Cham. https://doi.org/10.1007/978-3-030-47073-9_2-1

Download citation

DOI : https://doi.org/10.1007/978-3-030-47073-9_2-1

Received : 29 June 2020

Accepted : 29 June 2020

Published : 20 October 2020

Publisher Name : Springer, Cham

Print ISBN : 978-3-030-47073-9

Online ISBN : 978-3-030-47073-9

eBook Packages : Springer Reference Education Reference Module Humanities and Social Sciences Reference Module Education

• Publish with us

Policies and ethics

• Find a journal
• Track your research

Login | Register

• Editorial Team
• Editorial Policies
• Author Guidelines
• Reviewer Guidelines
• Statistics and Methods
• Research Integrity
• Full Policies

Language Development Research: An Open-Science Journal

Read author guidelines (PDF) before submission, and ensure your article includes a standalone Data, Code and Materials Availability Statement.

Science is for everyone. We set up Language Development Research (ISSN 2771-7976) because we don't believe in locking articles behind paywalls, in charging taxpayers and universities to publish research they've already funded, or in privileging papers that are "exciting" over those committed to scientific rigour. We believe that open science is better science. We uphold the highest standards of research integrity . We insist on open data and materials , and commit to publishing every article that is judged by our peer review process to meet our criteria for methodological and theoretical rigour .

We invite submissions of empirical and theoretical investigations of children's language development : typical and atypical, mono-, bi- and multi-lingual, spoken, signed, or written. We are also interested in the exploration of any topic or population relevant to language development, broadly construed (e.g., second language learning, artificial language learning, adult psycholinguistics, computational modeling).

Fiercely independent, we are answerable to no one except the scientific community and our 30-member strong Editorial Board of respected researchers . Please browse our articles  (below on this page), learn more about the journal and its editorial policies , and, when you're ready, submit your article .

Whether or not children can understand metaphor (e.g., ‘John is a  lion ,’ conveying that John is strong and brave) has been a matter of contention in the developmental literature for several decades (for reviews, see Gibbs, 1994; Pouscoulous, 2011; Vosniadou, 1987; Winner, 1988/1997). Much of the research on metaphor comprehension in children was conducted during the 1970s and 80s and suggested that children’s abilities with figurative language are only attained in early adolescence (Asch & Nerlove, 1960; Winner, 1988/1997; Winner, Rosenstiel, & Gardner, 1976). Some more recent research suggests that the lack of understanding found by early studies was linked in part to the complexity of the tasks used and attests instead to an early metaphorical ability emerging during the preschool years (Deamer, 2013; Pouscoulous, 2011; Özçalışkan, 2005). 

Despite the recent resurgence of interest in studies of metaphor development, the field is missing a coherent account of how metaphor comprehension abilities develop throughout childhood. This gap may be due to methodological differences, but there could also be theoretical conflicts. For instance, the idea that children go through a prolonged literal stage, processing language literally regardless of context, persists in some recent accounts of figurative language development (Levorato & Caccari, 1995, 2002). This position, however, is at odds with other findings in pragmatic development, which highlight children’s early pragmatic competence (Matthews, 2014). There is a clear need to integrate insights from children’s metaphor and figurative language comprehension into the broader research context of pragmatic development .

The aim of the special issue is to bring together researchers who work on  metaphor development across the lifespan from different theoretical perspectives and methodologies.  We especially encourage papers that link theoretical and empirical research, for instance by conducting theoretically informed empirical studies or by empirically testing and comparing the predictions of existing pragmatic theories of metaphor comprehension development. We also welcome methodological papers.  We invite authors to submit their original research manuscripts via  https://ldr.lps.library.cmu.edu/submissions/  (for author guidelines, see:  https://ldr.lps.library.cmu.edu/site/authorguidelines/ ). When submitting, please be sure to select  Metaphor Comprehension Special Issue  in the Article Type/Section box.

Important dates

Submission deadline: 1 May 2024 

(Action) Editors

Ingrid Lossius Falkum & Mary Beth Neff

Featured Articles

David Pagmar, Kirsten Abbot-Smith, Danielle Matthews

Predictors of children's conversational contingency

Evelina Leivada, Elliot Murphy

A demonstration of the uncomputability of parametric models of language acquisition and a biologically plausible alternative

alejandrina cristia, Marisa Casillas

Non-word repetition in children learning Yélî Dnye

Linda Kelly, Elizabeth Nixon, Jean Quigley

It’s Your Turn: The Dynamics of Conversational Turn-Taking in Father-Child and Mother-Child Interaction

Khazar Khorrami, Okko Räsänen

Can phones, syllables, and words emerge as side-products of cross-situational audiovisual learning? - A computational investigation

Karla K McGregor, Ronald Pomper, Nichole Eden, Timothy Arbisi-Kelm, Nancy Ohlmann, Shivani Gajre, Erin Smolak

Children’s language abilities predict success in remote communication contexts

Disa Witkowska, Laura Lucas, Maria Jelen, Hannah Kin, Courtenay Norbury

Development of complex syntax in the narratives of children with English as an Additional Language and their monolingual peers

Giulia Bovolenta, Emma Marsden

Expectation violation enhances the development of new abstract syntactic representations: evidence from an artificial language learning study

Nicola Dawson, Yaling Hsiao, Alvin Wei Ming Tan, Nilanjana Banerji, Kate Nation

Features of lexical richness in children's books: Comparisons with child-directed speech

Ben Ambridge

Language Development Research Editorial: Why do we need another journal?

Mitja Nikolaus, Eliot Maes, Jeremy Auguste, Laurent Prévot, Abdellah Fourtassi

Large-scale study of speech acts' development in early childhood

Audun Rosslund, Julien Mayor, Gabriella Óturai, Natalia Kartushina

Parents’ hyper-pitch and low vowel category variability in infant-directed speech are associated with 18-month-old toddlers’ expressive vocabulary

Samuel David Jones, Madeline Dooley, Ben Ambridge

Passive sentence reversal errors in autism: Replicating Ambridge, Bidgood, and Thomas (2020)

Victoria Knowland, Mohreet Rauni, Gareth Gaskell, Sarah Walker, Elaine van Rijn, Courtenay Norbury, Lisa-Marie Henderson

Sleep behaviour in children with developmental language disorder

Benjamin Edward deMayo, Danielle Kellier, Mika Braginsky, Christina Bergmann, Cielke Hendriks, Caroline Frances Rowland, Michael C Frank, Virginia Marchman

Web-CDI: A system for online administration of the MacArthur-Bates Communicative Development Inventories

Maxime Alexandra Tulling, Ailís Cournane

Wishes before ifs: mapping “fake” past tense to counterfactuality in wishes and conditionals

Natalia Kartushina, Nivedita Mani, Aslı Aktan-Erciyes, Khadeejah Alaslani, Naomi J. Aldrich, Alaa Almohammadi, Haifa Alroqi, Lucy M. Anderson, Elena Andonova, Suzanne Aussems, Mireille Babineau, Mihaela Barokova, Christina Bergmann, Cara Cashon, Stephanie Custode, Alex de Carvalho, Nevena Dimitrova, Agnieszka Dynak, Rola Farah, Christopher Fennell, Anne-Caroline Fiévet, Michael C Frank, Margarita Gavrilova, Hila Gendler-Shalev, Shannon P. Gibson, Katherine Golway, Nayeli Gonzalez-Gomez, Ewa Haman, Erin Hannon, Naomi Havron, Jessica Hay, Cielke Hendriks, Tzipi Horowitz-kraus, Marina Kalashnikova, Junko Kanero, Christina Keller, Grzegorz Krajewski, Catherine Laing, Rebecca A. Lundwall, Magdalena Łuniewska, Karolina Mieszkowska, Luis Muñoz, Karli Nave, Nonah Olesen, Lynn Perry, Caroline Frances Rowland, Daniela Santos Oliveira, Jeanne Shinskey, Aleksander Veraksa, Kolbie Vincent, Michal Zivan, Julien Mayor

COVID-19 first lockdown as a window into language acquisition: associations between caregiver-child activities and vocabulary gains

Iris Broedelet, Paul Boersma, Judith Rispens

Distributional learning of novel visual object categories in children with and without developmental language disorder

Martin Fortier, Danielle Kellier, María Fernández-Flecha, Michael C Frank

Ad-hoc pragmatic implicatures among Shipibo-Konibo children in the Peruvian Amazon

Matt Hilton, Katherine E. Twomey, Gert Westermann

Face time: Effects of shyness and attention to faces on early word learning

Robert Fromont, Lynn Clark, Joshua Wilson Black, Margaret Blackwood

Maximizing accuracy of forced alignment for spontaneous child speech

Anika van der Klis, Rianne van Lieburg, Lisa Lai-Shen Cheng, Clara Cecilia Levelt

Pauses matter: Rule-learning in children

Joshua Hartshorne, Yujing Huang, Lauren Skorb

Some puzzling findings regarding the acquisition of verbs

Ben Ambridge, Stewart McCauley, Colin Bannard, Michelle Davis, Thea Cameron-Faulkner, Alison Gummery, Anna Theakston

Uninversion error in English-speaking children’s wh-questions: Blame it on the bigrams?

Janet Yougi Bang, George Kachergis, Adriana Weisleder, Virginia Marchman

An automated classifier for periods of sleep and target-child-directed speech from LENA recordings

Lena V. Kremin, Amel Jardak, Casey Lew-Williams, Krista Byers-Heinlein

Bilingual children’s comprehension of code-switching at an uninformative adjective

Ana Maria Gonzalez-Barrero, Rodrigo Dal Ben, Hilary Killam, Krista Byers-Heinlein

Word learning in 14-month-old monolinguals and bilinguals: Challenges and methodological opportunities

Elspeth Wilson, Kate Cain, Catherine Davies, Jenny Gibson, Holly Joseph, Ludovica Serratrice, Margreet Vogelzang

Children’s development of conversational and reading inference skills: a call for a collaborative approach

Elizabeth Swanson, Michael C Frank, Judith Degen

Syntactic adaptation and word learning in children and adults

An official website of the United States government

Here's how you know

Official websites use .gov A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS A lock ( Lock A locked padlock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Speech and Language Developmental Milestones

On this page:

How do speech and language develop?

What are the milestones for speech and language development, what is the difference between a speech disorder and a language disorder, what should i do if my child’s speech or language appears to be delayed, what research is being conducted on developmental speech and language problems.

• Your baby's hearing and communicative development checklist

Where can I find additional information about speech and language developmental milestones?

The first 3 years of life, when the brain is developing and maturing, is the most intensive period for acquiring speech and language skills. These skills develop best in a world that is rich with sounds, sights, and consistent exposure to the speech and language of others.

There appear to be critical periods for speech and language development in infants and young children when the brain is best able to absorb language. If these critical periods are allowed to pass without exposure to language, it will be more difficult to learn.

The first signs of communication occur when an infant learns that a cry will bring food, comfort, and companionship. Newborns also begin to recognize important sounds in their environment, such as the voice of their mother or primary caretaker. As they grow, babies begin to sort out the speech sounds that compose the words of their language. By 6 months of age, most babies recognize the basic sounds of their native language.

Children vary in their development of speech and language skills. However, they follow a natural progression or timetable for mastering the skills of language. A checklist of milestones for the normal development of speech and language skills in children from birth to 5 years of age is included below. These milestones help doctors and other health professionals determine if a child is on track or if he or she may need extra help. Sometimes a delay may be caused by hearing loss, while other times it may be due to a speech or language disorder.

Children who have trouble understanding what others say (receptive language) or difficulty sharing their thoughts (expressive language) may have a language disorder. Developmental language disorder  (DLD) is a language disorder that delays the mastery of language skills. Some children with DLD may not begin to talk until their third or fourth year.

Children who have trouble producing speech sounds correctly or who hesitate or stutter when talking may have a speech disorder. Apraxia of speech is a speech disorder that makes it difficult to put sounds and syllables together in the correct order to form words.

Talk to your child’s doctor if you have any concerns. Your doctor may refer you to a speech-language pathologist, who is a health professional trained to evaluate and treat people with speech or language disorders. The speech-language pathologist will talk to you about your child’s communication and general development. He or she will also use special spoken tests to evaluate your child. A hearing test is often included in the evaluation because a hearing problem can affect speech and language development. Depending on the result of the evaluation, the speech-language pathologist may suggest activities you can do at home to stimulate your child’s development. They might also recommend group or individual therapy or suggest further evaluation by an audiologist (a health care professional trained to identify and measure hearing loss), or a developmental psychologist (a health care professional with special expertise in the psychological development of infants and children).

The National Institute on Deafness and Other Communication Disorders (NIDCD) sponsors a broad range of research to better understand the development of speech and language disorders, improve diagnostic capabilities, and fine-tune more effective treatments. An ongoing area of study is the search for better ways to diagnose and differentiate among the various types of speech delay. A large study following approximately 4,000 children is gathering data as the children grow to establish reliable signs and symptoms for specific speech disorders, which can then be used to develop accurate diagnostic tests. Additional genetic studies are looking for matches between different genetic variations and specific speech deficits.

Researchers sponsored by the NIDCD have discovered one genetic variant, in particular, that is linked to developmental language disorder (DLD), a disorder that delays children’s use of words and slows their mastery of language skills throughout their school years. The finding is the first to tie the presence of a distinct genetic mutation to any kind of inherited language impairment. Further research is exploring the role this genetic variant may also play in dyslexia, autism, and speech-sound disorders.

A long-term study looking at how deafness impacts the brain is exploring how the brain “rewires” itself to accommodate deafness. So far, the research has shown that adults who are deaf react faster and more accurately than hearing adults when they observe objects in motion. This ongoing research continues to explore the concept of “brain plasticity”—the ways in which the brain is influenced by health conditions or life experiences—and how it can be used to develop learning strategies that encourage healthy language and speech development in early childhood.

A recent workshop convened by the NIDCD drew together a group of experts to explore issues related to a subgroup of children with autism spectrum disorders who do not have functional verbal language by the age of 5. Because these children are so different from one another, with no set of defining characteristics or patterns of cognitive strengths or weaknesses, development of standard assessment tests or effective treatments has been difficult. The workshop featured a series of presentations to familiarize participants with the challenges facing these children and helped them to identify a number of research gaps and opportunities that could be addressed in future research studies.

What are voice, speech, and language?

Voice, speech, and language are the tools we use to communicate with each other.

Voice is the sound we make as air from our lungs is pushed between vocal folds in our larynx, causing them to vibrate.

Speech is talking, which is one way to express language. It involves the precisely coordinated muscle actions of the tongue, lips, jaw, and vocal tract to produce the recognizable sounds that make up language.

Language is a set of shared rules that allow people to express their ideas in a meaningful way. Language may be expressed verbally or by writing, signing, or making other gestures, such as eye blinking or mouth movements.

Your baby’s hearing and communicative development checklist

Birth to 3 months, 4 to 6 months, 7 months to 1 year, 1 to 2 years, 2 to 3 years, 3 to 4 years, 4 to 5 years.

This checklist is based upon How Does Your Child Hear and Talk ?, courtesy of the American Speech–Language–Hearing Association.

The NIDCD maintains a directory of organizations that provide information on the normal and disordered processes of hearing, balance, taste, smell, voice, speech, and language.

Use the following keywords to help you find organizations that can answer questions and provide information on speech and language development:

• Early identification of hearing loss in children
• Speech-language pathologists

For more information, contact us at:

NIDCD Information Clearinghouse 1 Communication Avenue Bethesda, MD 20892-3456 Toll-free voice: (800) 241-1044 Toll-free TTY: (800) 241-1055 Email: [email protected]

NIH Publication No. 00-4781 September 2010

*Note: PDF files require a viewer such as the free Adobe Reader .

• Research article
• Open access
• Published: 05 April 2019

Paths to language development in at risk children: a qualitative comparative analysis (QCA)

• Kate Short   ORCID: orcid.org/0000-0002-2022-0620 1 , 4 ,
• Patricia Eadie 2 &
• Lynn Kemp 3  

BMC Pediatrics volume  19 , Article number:  94 ( 2019 ) Cite this article

12k Accesses

9 Citations

38 Altmetric

Metrics details

Childhood language development is related to long term educational, employment, health and social outcomes. Previous research identifies a complex range of risk and protective factors which result in good and poor language outcomes for children, however children at risk are an underrepresented group in these studies. Our aim is to investigate the combinations of factors (paths) that result in good and poor language outcomes for a group of 5 year old children of mothers experiencing adversity.

This mixed methods study utilised longitudinal data from a randomised control trial of sustained home visiting (MECSH) to determine the language outcomes in at risk children. Mothers were randomly assigned to a comparison group at entry to the study (prior to child’s birth). Their children who were retained at entry to school completed language assessments ( n  = 41) and were participants in this study. Influence of 13 key factors derived from the extant literature that impact language development were explored. Regression was used to determine the six key factors of influence and these were used in the Qualitative Comparative Analysis (QCA). QCA was employed to examine the necessary and sufficient conditions and paths affecting language development linked to good and poor language outcomes. A post hoc analysis of the risk and protective paths to good and poor language outcomes was also conducted.

Thirteen distinct pathways led to good language outcomes and four paths to poor language outcomes in five year old at risk children. A variety of condition combinations resulted in these outcomes, with maternal responsivity, toddler development and number of children in the home being key. High and low maternal education influenced both good and poor language development.

Conclusions

The paths to good and poor language outcomes were different and complex. Most paths to a good language outcome involved protective factors, though not always. In addition, paths to poor language more often involved risk factors. The varied patterns of risk and protective factors point to the need for interventions across the first five years of life in both health and education for families which can respond to these risk and protective patterns.

Trial registration

The original RCT was retrospectively registered in the ANCTR: ACTRN12608000473369 .

Peer Review reports

Language use in children, the development of understanding and expression of words, grammar and discourse is one of the key and most complex developmental skills acquired through childhood, with far reaching affects through life. Good language development in children is important for effective long term academic, social and economic participation in society [ 1 , 2 , 3 ]. Poor language outcomes are evident from school entry [ 4 , 5 ] and influenced by social determinants of health such as socioeconomic status (SES) and maternal education [ 6 ]. For this reason, language development is a focus of both public health and early childhood policy and practice [ 7 ].

Language development and at risk children

Difficulties in language and communication place a burden on the child, parent and society both financially [ 8 ] and in quality of life [ 9 ]. One in five children have poor language development at 4 years [ 10 ] and there is a social gradient to the prevalence of language difficulties [ 11 ]. Children experiencing social disadvantage (as defined by census based measures of disadvantage e.g. income, suburb, home ownership, parental employment) are twice as likely as other children to have communication difficulties at entry to school [ 12 ] and language difficulties are part of both the cause and consequence of long term disadvantage. For some children, language difficulties lead to a sequelae of cascading negative effects such as poor literacy and social participation, increased academic failure, disengagement from school, instances of juvenile incarceration, a variety of mental health difficulties, generally poorer health, reduced employment and/ or increased relationship breakdown in adulthood [ 2 , 3 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 ]. Often these outcomes are clustered in children who come from low socioeconomic households and have parents who present with risks which result in these children being exposed to more difficult circumstances in their childhood [ 22 ]. This group we will refer to as children who are at risk. At risk here means an exposure to a combination of risk factors that have been shown to affect child development such as: low socioeconomic status, limited resources parental capacity and /or physical needs such as housing, experiencing mental health and drug and alcohol difficulties in the home, child maltreatment and domestic violence amongst other stressors and threats [ 23 ].

Risk and protective factors for language development

Utilising a bioecological model of development [ 24 ], research has established some of the key child, maternal and environmental influences on language development in the early years. These include (but not exclusively) maternal factors such as education [ 25 ], mental health [ 26 , 27 , 28 ] and responsivity [ 29 ], a family history of communication difficulties [ 30 ]; child factors such as birth weight [ 31 ], toddler development [ 32 ] and gender [ 33 ] and environmental factors such as being read to from an early age [ 34 , 35 ], numbers of children in the home [ 36 , 37 , 38 ], attending sufficient good quality childcare [ 39 ] and SES [ 40 , 41 , 42 ]. Debate continues as to whether these factors are mediators or causal in language outcomes [ 43 ].

Though we know these risk and protective factors impact language development, longitudinal cohort studies have consistently shown individual factors on their own represent only small amounts of the variability in language skills [ 30 , 44 , 45 , 46 ]. It has been found there is a compounding effect of multiple risk factors on vocabulary development [ 47 ]. However, how these risk and protective factors combine and impact on each other in language development has been less studied. In a recent example, Baydar and colleagues (2014) investigated the impact of a combination of multiple maternal and environmental family factors on vocabulary development in Turkish children. The responsivity of low SES mothers supported children’s vocabulary development only when the mothers were not depressed. Investigations of risk “clusters” for language development have also emerged. In a longitudinal study of Australian children, those with a risk profile related to speaking a language other than English made fast gains in language through the school years if they had few other risks. However, if they had a number of risks, both the English and non-English speaking children performed poorly. Those with many risks for poor language at the end of the study performed poorest. [ 47 ]

However, not all children with risks end up with language difficulties. What protects against poor language has received some limited attention and reveals some key factors. Turkish children’s vocabulary development was protected in families of depressed mothers who were economically distressed if they were surrounded by a supportive family and community [ 27 ]. Other studies of population and impaired cohorts have found being regularly read to, attending early childhood education, participating in play and the child’s prosocial skills at 4 years were all protective [ 7 , 48 ].

Interventions targeting risk and protective factors

As there are so many possible sources of influence on language development, both proximal and distal to the child, determining the most influential conditions will help services target and create public health preventative interventions [ 4 , 35 ]. Some conditions such as book reading can be changed through interventions or are manipulable. The environmental impacts that may be manipulable in interventions play an important role in the early years of language acquisition and provide an opportunity to prevent language difficulties or change the trajectory of development for some children [ 12 , 35 , 49 ]. Brofenbrenner and Morris [ 50 ] outline there is an interdependence between conditions. Influencing one of these conditions can have an effect on others, these factors being both producers and products of development. There is research, community and governmental interest in targeting low SES groups, where children with more manipulable risks appear to be concentrated and there are some promising interventions documented [ 12 , 51 ] . However, it is unclear, how one or many of these risk and protective conditions should be targeted in interventions to create best language outcomes. Further research is required to explore the combinations of risk and protective factors in at risk children to help develop more tailored interventions for this population.

Statement of the problem

Large prospective cohort studies which unpack key conditions that predict future language abilities are numerous (see Law, Dennis (43) for a review). The factors relate in complex, non linear ways with each other and a number of them interact and reinforce each other [ 25 ]. Considering just one of these and trying to “partial out” its influence is conceptually difficult. Qualitative Comparative Analysis (QCA) is a method designed to help unpack these complex relationships, however it is a method which to our knowledge, is untested in the realm of child development. QCA is a mixed method standing between qualitative and quantitative methodologies and has been employed to help answer complex health policy questions [ 52 ]. Blackman et al. (2013) explains QCA as “… particularly apt for producing evidence about how to tackle complex policy problems that have the character of ‘wicked issues’ … These are issues that pose significant challenges for intervention because of interdependencies between causes.” (pg. 127). Language development in at risk children is one such “wicked issue”. In this study we use QCA to explore the specific combinations of risk and protective conditions important for good and poor language outcomes in a low SES, culturally and linguistically diverse group of 5 year old Australian children at risk of compromised child development. We hypothesize that different factors combine in complex ways to create good and poor language outcomes.

This mixed method, prospective cohort study was nested within the Miller (subsequently Maternal) Early Childhood Sustained Home visiting (MECSH) randomised control trial [ 53 ]. MECSH explored the effectiveness of sustained, nurse home visiting provided to women experiencing adversity from pregnancy until their child was two years old We conducted secondary, quantitative and qualitative analysis of a range of data collected over the 5 ½ years of the study, exploring whether there were multiple paths to good and poor language outcomes. This study comprises of data from the comparison arm of the RCT as they represent a population non-intervention group who received usual care. Usual care at the time meant a mother received a home visit by a child health nurse within 2 weeks of giving birth with the offer if subsequent visits to a well child clinic if the mother chose to attend. All study participants gave written informed consent prior to entering the study.

Recruitment

208 low SES, mothers experiencing adversity were recruited from the Liverpool Hospital, New South Wales. These women were assessed on routine psychosocial assessment at their first presentation to the hospital antenatal clinic prior to the child’s birth. They were eligible for the study if they lived in a particular socio economically disadvantaged area and presented with one or more risk factors for poor maternal or child outcomes. These risks included: mental health problem or disorder (past or current); teenage parent; late antenatal care (after 20 weeks); current substance misuse; history of or current domestic violence; history of abuse as a child; lack emotional or practical support; major stressors in the last 12 months; current probable distress (as indicated by an Edinburgh Postnatal Depression Scale [ 54 ] score of 10 or greater). On consent, participants were randomised, 111 to treatment and 97 to comparison, trial number: ACTRN12608000473369 [ 53 ]. At 5 years (entry to school), 86 children (41%) were retained, 82 of these were assessed on the Wechsler Preschool and Primary Scale of Intelligence III (WPPSI-III) Australian Edition [ 55 ], and 41 of these children from the comparison group were eligible for the current study (see consort diagram Fig.  1 ).

Recruitment and retention for the MECSH RCT and this study

Eligibility for this study included: the child completing a cognitive assessment (WPPSI-III) at end of the first term of formal school entry around 5 years of age (mean = 65 months, SD 4.3) by those in the original comparison group. QCA is a methodology suited to medium n sample sizes, with n between 10 and 50 being ideal [ 56 ], thus this sample is appropriate for this methodology. For this study, the retained comparison group participants ( n  = 41) were compared to the original comparison group ( n  = 97) (see Table  1 ). There has been an attrition of 59%, and there was one significant difference between the groups: significantly more single mothers were retained than there were in the original cohort ( p  = 0.022).

QCA method, theoretical and methodological framework

The QCA process is iterative (thus qualitative in nature) but structured in its method and relies on the principles of Boolean logic and set theory [ 57 ]. It compares the empirical evidence of individual cases with all theoretically possible combinations of risk and protective factors (paths) that lead to an outcome. Through a process of logical minimisation, only those conditions that clearly differentiate good vs poor outcomes are retained in the final explanatory models.

In QCA, 3 key phenomena further our knowledge of what causes an outcome: conjunctual causation, equifinality and causal asymmetry. [ 58 ]. Conjunctual causation refers to the specific combinations of causes lead to a specified outcome. Prior evidence and theory help determine the combinations of risk and protective conditions to place in cumulative risk models. QCA then uses empirical cases to investigate which conditions and in risk or protective mode combine to result in an outcome. In QCA this result is called a path or causal recipe [ 57 ]. There may be a variety of these causal recipes which result in the same outcome. This leads to the second important and novel contribution QCA provides - the notion of equifinality [ 57 ]. Equifinality refers to an expectation that there can be multiple paths that lead to an outcome. QCA may thus move us closer to understanding the multiple combinations of risk and protective factors which result in and good and poor language outcomes. In turn, understanding these complex paths may help us develop more effective interventions.

Finally, investigating causal asymmetry is an important methodological task of QCA. Causal asymmetry assumes good and poor language outcomes are not the result of paths that are the direct opposite of each other.

The two mathematical concepts of necessity and sufficiency are investigated in QCA and are essential to understanding results. Each variable, called a condition in QCA, may be necessary or sufficient (or neither) for an outcome. For a condition to be necessary it must be present for the outcome to come about. Necessity gives clear instructions for intervention. However, few conditions are usually necessary. More common is that conditions are sufficient. That is, when a condition is combined with one or more conditions it is causal for the outcome [ 56 ]. Examination of the sufficient condition combinations will help develop more targeted language interventions for at risk children and families.

QCA processes

QCA has two main steps: 1. Qualitative analysis of cases when the data set is cleaned and calibrated (the process of deciding the rules by which presence or absence of the condition is determined for the outcome and each condition) and 2. Quantitative stage when there is comparison across cases and all possible paths to an outcome are configured through Boolean minimisation. There are two possible methods of QCA crisp set (csQCA) and fuzzy set (fsQCA). In csQCA, all conditions are binomially split, into the “presence” or “absence” of the condition where as in fsQCA the more continuous data is retained. Crisp set was chosen for this first study of QCA with child language, as it allows exploration of the most simple path combinations [ 57 ]. The dichotimisation is part of the data analysis and is outlined for each condition in Table  2 and detailed in Additional file  1 .

Creating, cleaning and calibrating the dataset

Outcome: language status at school entry (see table 2 ).

Language outcome was based on the combination of both functional and standardised test performance in English, administered in the first year of formal schooling. The standardised test performance was determined by utilising the language quotient on the WPPSI-III [ 55 ] verbal subscale (VIQ) which contained 3 subtests of language skills: Vocabulary, Information and Word Reasoning subtests administered at the children’s schools by a registered psychologist. A standard score of 85 or higher was set as good language outcome, 84 and below as poor language outcome. Functional performance was determined by teacher perception of children’s language skills: teachers rated the child’s expressive (spoken) and receptive (understanding) language skills separately in comparison to their peers on a 4 point Likert scale: 1. Much less competent 2. Less competent than others 3. As competent as others 4. More competent than others. The points from these teacher ratings were totalled and each child received a score (possible range 2–8) for their language skill at school and a score of 5 or higher was considered good language outcome, 4 and below for poor language outcome. For the QCA, a child was considered to have the outcome of Poor Language (PL) at 5 years if they fulfilled criteria for poor language in either measure. There were 8 children with PL (19.5% of the sample). All others were considered Good Language outcome (GL: n  = 33) users. Further details of the creation and calibration of this outcome are available in the supplementary file (Additional file 1 ).

Conditions (see Table 2 )

The conditions for the QCA were selected if they were: (1) identified in the literature as salient to child language outcomes from birth - entry to school; (2) represented child, maternal and or environmental risks and protections; (3) able to be divided for analysis in a logical method with at least 25% of the sample meeting criteria for binary categories [ 56 ] and (4) available in the MECSH dataset. This resulted in 13 conditions (see Table 2 ) which were: Child : gender (G); toddler development (D) and behaviour (B). Maternal : education (ME); antenatal distress (AD); chronic distress overtime (CD); responsivity in infancy and toddlerhood (RS). Environmental : Socioeconomic Status (SES); number of children in the home (CH); child read to more than three times a week over time (RD); two years of more of early childhood education prior to starting school (ECE); Language spoken: Language other than English (LOTE) or English.

Detailed discussion of how the conditions were operationalised and calibrated is outlined in Additional file 1 . A range of parent reported survey data, child and parent assessments and coding of videos were used to create the categories that defined the conditions. All conditions were then binomially cut, with the cut point informed by literature, standardised score/test manual recommendations or natural divisions in the data. Some conditions were simple compilations of data with clear cut points, for example the standardised score from the Bayleys MDI was used for the child toddler development (D) condition and cut at 1 SD below the mean (Poor development). Some conditions were more complex to establish and a variety of longitudinal data were used to create them. For example, to form the condition maternal responsivity (RS), data from two different sources at two different time periods were used. Quality of maternal responsivity was operationalized combining a home based analysis using the responsivity subscale score from the HOME Inventory [ 59 ] and NICHD rating of maternal-child interaction in play sample videos conducted in the clinic [ 60 ]. The mean of each of the play sample scores and the HOME responsivity rating were set as cut points and any child below the mean on either one of the measures received a score of zero (condition absent) for responsivity.

Each condition was coded for every participant, referred to as a case in QCA. All conditions were constructed in the positive, thus presence of a condition, meant a case was assigned a score of one for that condition and this indicated a notionally protective variable for good language development. Assigning zero meant absence of the condition or risk of lower language development. In all conditions, missing data fields were left unassigned.

The next process was to reduce the number of conditions to that which could be adequately supported by the number of cases, providing good model coverage [ 56 ]. Quantitative methods were utilised to reduce the number of conditions from 13 to the maximum of seven considered to be adequate with the number of empirical observations to maintain good diversity [ 57 ]. Initially correlation between the conditions was conducted, controlling for language outcome on the WPPSI standard score at 5 years. Two factors were highly correlated: behaviour and development. Development was chosen to remain in the set of predictors due to the stronger correlation with language outcome (development: r = 0.490, p  = 0.001; behaviour: r = 0.312, p  = 0.050). Regression was then used to determine those conditions with greatest predictive value for the language outcome at 5 years as measured on the VIQ of the WPPSI–III [ 55 ]. Each condition was individually regressed against the outcome and cut point of p  = 0.1 or less was used for inclusion in further analysis.

Regression revealed seven significant factors predicting language outcome at 5 years for inclusion in the QCA (see Table  3 ): toddler development; maternal education; maternal antenatal distress; maternal responsivity in infancy and toddlerhood; number of children in the home; amount of early childhood education prior to starting school and Language Other Than English (LOTE) being spoken at home. Once the significant predictors were chosen a further three cases were excluded due to missing data. Thus the final group for analysis in QCA was 38 cases.

Path configuration and analysis

Following choice of variables, the conditions were placed in the fsQCA 2.5 software program and two QCA analyses were conducted: one exploring paths to Good Language (GL) outcome and one the paths to the Poor Language (PL) outcome. A truth table was established, which contains all of the condition combinations logically possible. A minimum consistency of 0.75 and coverage of 0.5 were used as boundaries to determine sufficient paths [ 56 ]. Following this, the Quine-McCluskey algorithm which uses Boolean algebra to compute the commonalities between the paths that lead to GL and PL was used. 7This logically reduces the configurations to produce a solution [ 61 , 62 ]. There are two parameters used to reduce rows: 1. Coverage: the empirical relevance of a solution; and 2. Consistency: the extent to which cases sharing similar conditions display the same outcome. Initially necessary condition analysis was conducted to determine those conditions which were essential to the outcome. Parameters of fit were set to determine necessary conditions these were: consistency 0.9 and coverage 0.5 [ 56 ]. The paths were then exposed to sufficiency analysis. This is a more complex analysis to determine if any conditions or combinations of conditions (conjunctual causation) were essential for either the Good or Poor Language outcome. There are three possible solution models to report in QCA. For this study the intermediate solution, a combination of both theory and empirical data will be presented below. This was chosen to allow both the empirical data and theory influence over the final solution. The other two possible solutions (the complex and parsimonious) can be found in the Additional file 1 . Subsequently, further classification of the paths was conducted according to risk and protective components as outlined in Fig.  2 .

Classification of paths by combinations protection and risks conditions

The number of conditions predict the number of possible paths, thus in this study seven conditions predicted 128 different possible paths to Good (GL) and Poor Language (PL) outcomes, of which 32 were created by the empirical data (coverage 25%). Of these, 27 were paths to GL and five paths to PL (see Table  4 ). Most paths had one case however there were six paths with two or more cases. Two of these paths (which represented 4 of 38 cases) had cases with different outcomes (conflicts) (Paths 6 and 7 in Table 4 ). Note, in reporting of the paths an asterisk (*) represents a logical “AND”; upper case codes e.g. “D” for toddler development, indicate the protective form of toddler development and lower case “d” indicates the risk form of the condition. One path D*me*RS*ch*ece*LOTE, had an outcome of both PL (case 36) and GL (case 74) and the other path D*ME*rs*CH*ECE*LOTE also had an outcome of PL (case 73) and GL (case 44). These resulted in truth table conditions of 0.5 consistency which contravened the consistency boundaries and these four cases were removed from further analysis. To check for their influence, analysis were conducted with and without the cases. Excluding these cases resulted in a change of one path, with the addition of one condition (with the above cases Path 9 of GL was: D*AD*RS and without the cases was D*AD*RS*English). There were no changes to the PL solution with removal of the cases. Thus, there was minimal impact of removing the conflict cases, although these cases were explored further to develop understanding of the conflicting outcomes. All further analysis was conducted with these cases removed. As not all paths could be represented by the 38 cases, the empirically unrepresented paths (logical remainders) required consideration in the analysis. Logical remainders for five of the six conditions were set to present (for GL) or absent (for PL) in as indicated by the literature except LOTE / English speaking. This condition was set to neither present nor absent in both the GL and PL QCA. This was done as investigations of the data indicated LOTE speakers were equally present in both PL and GL outcomes. The truth table emerged as outlined in in the additional file (now without Paths 6 and 7). This led to the intermediate solution presented below.

Outcome 1: Good language (GL)

Necessity (see table 5 ).

There were no conditions which fulfilled the set parameters to be considered necessary for GL and thus no condition was essential for GL to result at 5 years.

Sufficiency (see Table  4 and Fig.  3 )

All conditions were kept for all analyses. There were 13 paths in the group leading to GL all of which had strong consistency [ 57 ] indicating the solution strongly relates to the outcome observed. It is notable that 5 of the 13 (Paths 3 & 10–13) paths represented no unique coverage, thus were present however not of high importance to the overall findings and have not been included in later analysis [ 56 ].

Risk and protective conditions and paths to good language outcome

The paths to GL were usually via the presence of no and minimal risk factors (see Fig.  3 ). There was one path of high risk to GL outcomes (Path 1), one of only two paths in which English speaking was sufficient . Eight of 13 paths were protective (Paths 2, 3, 6, 7, 8, 9, 12 & 13). That is, they contained only conditions in the protective mode. Overall, these paths consisted of all of the conditions in different combinations in protective mode: both manipulable (D, RS, ECE) and non manipulable (AD, CH, ME) conditions. All but one of these protective paths had good toddler development (D) as an influential condition (as indicated by the coverage score of 0.83), D was almost necessary for a Good Language outcome but not in all cases. Present in half of these paths was the combination D*ME (+*LOTE in one path).

Paths to the good language outcome

There were four of 13 paths that were Mostly Protective (Paths 4, 5, 10 & 11). That is, they had one risk and this risk was always low maternal education. This risk was also always linked to having two or fewer children in the home (CH). Three of these four paths were also linked to two years or more of centre based early childhood education prior to starting school (ECE).

Most risk factors had no influence on any path to GL. These conditions were neither necessary nor sufficient: Having more than three children in the home (ch), not being responsive in infancy and toddlerhood (rs), having less than two years of early childhood education before starting school (ece) and being antenatally distressed (ad). That is, these risk factors were not influential in the outcome and were logically minimised out of all pathways to GL.

Outcome 2: Poor language (PL)

For PL, poor maternal responsivity was a necessary condition, present in every path and every case however coverage was limited (consistency 1; coverage 0.29) due to the few cases with PL.

Sufficiency (see Table 4 and Fig. 4 )

All conditions were kept for all analyses. The intermediate solution resulted in four paths in the group leading to poor language, all of which had acceptable consistency. This model had good solution coverage (1) and consistency (1). All paths provided some unique coverage.

Paths to the poor language outcome

Risk and protective conditions and paths to poor language

The paths to poor language were via the presence of risk factors and one protective factor (Fig. 4 ). The protective factor was either D or ME. Overall, these paths contained all of the conditions in different combinations in the risk mode: both manipulable (d, rs, ece) and non manipulable (ad, ch, me) conditions. As indicated by its necessary status, all paths contained poor maternal responsivity and three of the four also contained having 3 or more children in the home. The minimum number of risks conditions in each path was three. Both LOTE and English were influential in the paths to PL.

Comparison of good and poor language paths

For this analysis only paths with unique coverage were included (GL Paths 3 & 10–13 were not considered). There was not complete symmetry in the paths to good and poor language outcomes, however there commonly was some when only the conditions were considered. Those conditions most consistently present for GL outcome (Table  6 ) were: good toddler development; higher maternal education; no antenatal distress; fewer children in the home and LOTE. For PL outcome most commonly present conditions were: good and poor toddler development; high and low maternal education; antenatal distress; poor maternal responsivity; three or more children in the home; non optimal amounts of centre based ECE prior to starting school. As evident above and in Table 6 , there were a number of conditions with symmetry in GL versus PL outcome: the number of children in the home, early childhood education, maternal responsivity and antenatal distress. For all of these conditions the protective version was only present when GL resulted and the risk condition was only present with PL. For example good maternal responsivity (RS) was only ever present in the paths to GL and poor maternal responsivity (rs) was only ever present in the paths to PL. However, no other symmetry was present, the combinations of conditions in the paths were not the same. Maternal education and toddler development were two conditions that existed in both the risk and protective mode in both GL and PL outcomes. They also existed together in patterns: for GL, if development existed with maternal education they were both always in either the protective mode (5 paths) or both in the risk mode (1 path). In contrast for PL one was always protective and one always risk for example d*ME or D*me.

This study used a novel mixed method, Qualitative Comparative Analysis (QCA) to explore the risk and protective pathways to good and poor language development in a group of at risk children. This method resulted in 13 different paths to good language (GL) and four to Poor Language (PL) outcomes. Multiple paths to the same outcome (equifinality) were present. That is, this empirical data demonstrates there are a variety of risk and protective pathways to both GL and PL outcomes. Also present was conjunctual causation, that is a variety of condition combinations to the outcomes were present. Overall, there was no one risk or protective factor that was necessary (present in all pathways) for good language outcomes. However, poor maternal responsivity in infancy and toddlerhood was necessary for poor language outcome at five years. Protective factors dominated the paths to GL with risk factors rarely being influential. In contrast, paths to PL outcomes always included poor maternal responsivity and a range of other risk factors. The model showed both causal symmetry and asymmetry, that is some conditions such as maternal responsivity, acted symmetrically. When mothers were responsive this was only influential to GL outcomes (it was protective) and when poor was only influential to PL outcomes (was a risk factor). Other conditions such as maternal education and toddler development functioned assymetrically and combined with other conditions to be influential in both good and poor language outcomes.

Currently we know that at risk children have a higher prevalence of PL than children not experiencing adversity [ 12 ]. The at risk children with GL in this study had factors present that appear to be protective against such poor language outcomes. Consistent with previous research, these children were in part predictable by their lack of influential risks and the presence of protective factors such as good toddler development, fewer children in the home, two years or more of early childhood education prior to starting school and no antenatal distress in the mother. Only five of 13 paths to GL included risk factors, four of which had one risk factor and this was always low maternal education (Paths 5, 6, 10 & 11). This finding provides some nuance about the role of maternal education in language acquisition. It is often found in large cohort studies that high maternal education is an influencing factor in GL outcomes [ 10 , 46 ]. Our findings are no different, however they also explicate the context in which low maternal education can be related to GL outcomes for their child. As Harding, Morris (25) highlighted maternal education is a broad concept which represents a range of maternal skills and parenting practices. In this study, good language development occurred in children of mothers with low education when combined with having fewer children in the home and usually with the child having two years or more of early childhood education prior to starting school. Interestingly the language spoken in the home did not affect this finding – mothers of children with good language at five years spoke either English or another language (more commonly they were LOTE).

Consistent with previous work [ 47 ] the pathways for children with poor language (PL) were notable for many risk factors: specifically this study highlighted the importance of poor maternal responsivity in combination of other influential risk factors. Particularly evident in all cases of PL was the influential effect of poor maternal responsivity. In this study as in previous ones, children of less responsive mothers demonstrate more limited language than children of more responsive mothers [ 29 , 63 , 64 , 65 ]. This study adds other key conditions that combine with poor responsivity to result in these poorer outcomes – in particular more children in the home, antenatal distress and or limited early childhood education. We assert, the compounding effect these conditions which result in less individualised and shaped responses to a child, impacts over time culminating in lower language at school entry. Similar to the pathways for GL, both high and low maternal education was associated with PL at five years. In addition, toddler development at three years, either good or poor, was associated with PL.

Toddler development emerged consistently as important and influential, particularly for GL outcomes. In all of the no risk paths to GL, good toddler development was influential in the outcome over and above other characteristics, reflective of previous findings [ 32 ]. As currently debated in the literature, children’s previous language and developmental outcomes are often good predictors of their later language outcomes but not always [ 4 ] . A large, longitudinal cohort study of children’s language development found 6% of children presenting with language difficulties at four years were not predicted by outcomes at two years and 14% of children with concerning development at two years had typical development at four years [ 10 ]. This study gives us more direction as to the combination of factors which result in the more unpredictable outcome. In Path 1 toddler development was poor and GL also resulted when combined with the influential effects of using a LOTE and low maternal education. Thus good development was not necessary for GL outcomes. As was the case for GL - both good and poor toddler development resulted in PL. It is possible this result may be explained by a missing condition which is hypothesised to influence the later emergence (in the early school years) of language concerns: genetic inheritance [ 66 , 67 ]. Many children born to parents with poor language will also have difficulties: hereditability varies with age and testing method but reveal hereditability estimates of .24–.92, but as twin studies have shown both genes and environment may be involved and there is a complex relationship between these [ 68 , 69 , 70 , 71 , 72 , 73 ] . The evidence is still emerging, however it is thought ‘good’ genes may ‘protect’ against poorer environmental influences and ‘poor’ genes increase risk despite good environments and this may be evident (but untested) in these cases [ 74 ]. Future research requires family history of speech and language difficulties to be collected and included in models.

In QCA the cases of conflict (same pathway but different outcome) provide guidance as to where models may be deficient. There were four conflict cases in this data set Path 6 (cases 36 & 74) and Path 7 (cases 73 & 44) which were removed from the analysis. Their exclusion made minimal difference, however exploring these cases qualitatively is important for understanding how paths vary. The outcome for case 73 (PL at 5 years) may have been different to the paired case (44) due to the difference in English language exposure (there were few risks present in these cases). Though both were LOTE, case 44 reported speaking English as the main language to their child at 2 years of age. At the same age case 73 reported speaking a LOTE. This is consistent with Hoff’s [ 75 ] findings that, the more rich exposures to the dominant language the better the child will be in that language. For cases 36 and 74 a key point of difference was the behaviour score at three years, which was poor for the former and high for the latter. Although toddler behaviour was usually correlated with good development, this was one of the cases where it was not. Including more conditions such as toddler behaviour in analysis could have potentially resolved this conflict, and provided a further important path to PL but could not be supported by the sample size in this study.

Policy and program implications

There are a number of policy and practice implications from our findings. Continued investigation of and investment in targeted treatments which impact on the responsivity of mothers early interactions with infants and toddlers, particularly targeting those mothers who are at greater risk and those with antenatal signs of distress is needed. As this and previous studies show, targeting maternal responsivity alone will not be enough to improve language outcomes, or reduce the incidence of language difficulties [ 76 ]. Rather treatments are needed that have the flexibility to target the range of risks present in this group, such as poor maternal responsivity, and parenting many children in the home, and/or child participation in high quality early childhood education programs. One such promising treatment may be sustained home visiting treatments starting antenatally and continuing until the child is two or three years old [ 77 , 78 , 79 ]. The findings also suggest that policies supporting implementation of long term early childhood education in areas of high disadvantage are helpful for children’s language development. We advocate that at risk children of mothers experiencing adversity require both very early public health interventions such as home visiting and then high quality, long term early childhood education to ameliorate the impacts of adversity.

Limitations

There are a number of limitations in this study. This is case based work, aiming to determine causality within the sample and as such cannot be easily generalised to other populations, particularly as this study had a small sample and only eight cases had poor language. This number was smaller than could be expected in a group of at risk children based on the prevalence of poor language outcomes in the literature. This may have been impacted upon by the presence of initiatives focused on improving community level disadvantage in the study area at the time. Goldfeld, O’Connor (23) have developed a new model of disadvantage for children experiencing adversity, which includes aspects of neighbourhood environments not included in this study (e.g. neighbourhood liveability) that may have shed light on this finding. Additionally, this study population may have been socio economically relatively well resourced when compared to some other studies of children at risk. Although they lived in a low SES area, only 20% of the mothers were on welfare benefits. However, they were psychologically burdened with over 60% of the cohort presenting with either depression and or mental health difficulties and experiencing a range of significant adverse experiences, which have been shown to have impacts on children’s development.

The small sample also limited the diversity of conditions that could be included in the model. Choice of conditions was further limited by secondary analysis of existing data: other variables that are known to influence language acquisition, such as genetic bases or family history of language difficulties were not collected. The limitation in conditions explored is evidenced by the conflict cases discussed above. The decision to do this study as crisp set for clarity in this first exploration of language development with QCA may also have limited some nuance in the analysis.

Future research

Future research of the pathways to good and poor language requires larger samples with purpose-designed/collected conditions. Continuing to use QCA is recommended for its ability to capture and elucidate the complexity of pathways to good and poor language, and hence support a more nuanced discussion about how to facilitate good language development and ameliorate poor language development in at risk children.

This mixed methods study of language development in at risk children demonstrated there are varied pathways to both good and poor language outcomes. Most paths to good language involved protective factors, though not always. Similarly, but not symmetrically, paths to poor language more often involved risk factors. Key to poor language outcomes was poor maternal responsivity combined with other risk factors associated with poorer language development. Other conditions which differentiated the paths to good and poor language were the number of children in the home and toddler development. The complex pattern of factors associated with language outcomes suggests the need for complex interventions which can respond to these varied risk and protective patterns.

Abbreviations

Antenatal Distress

Australian New Zealand Clinical Trials Registry

Mental Development Index

Center for Epidemiologic Studies Depression Scale

The number of Children in the Home

Toddler Development

Early Childhood Education

Edinburgh Post Natal Depression Scale

Good Language

Language Other Than English

Maternal Early Childhood Sustained Home Visiting

Poor Language

Qualitative Comparative Analysis

Randomised Control Trial

Maternal Responsivity

Socio Economic Status

Verbal Intelligence Quotient

Wechsler Preschool and Primary Scale of Intelligence

Conti-Ramsden G, Durkin K, Simkin Z, Knox E. Specific language impairment and school outcomes. I: identifying and explaining variability at the end of compulsory education. International Journal of Language & Communication Disorders. 2009;44(1):15–35.

Article   Google Scholar  

Johnson CJ, Beitchman JH, Brownlie E. Twenty-year follow-up of children with and without speech-language impairments: family, educational, occupational, and quality of life outcomes. American Journal of Speech-Language Pathology. 2010;19(1):51–65.

Article   PubMed   Google Scholar  

Schoon I, Parsons S, Rush R, Law J. Childhood language skills and adult literacy: a 29-year follow-up study. Pediatrics. 2010;125(3):e459–e66.

McKean C, Wraith D, Eadie P, Cook F, Mensah F, Reilly S. Subgroups in language trajectories from 4 to 11 years: the nature and predictors of stable, improving and decreasing language trajectory groups. J Child Psychol Psychiatry. 2017;58(10):1081–91.

Norbury CF, Vamvakas G, Gooch D, Baird G, Charman T, Simonoff E, et al. Language growth in children with heterogeneous language disorders: a population study. J Child Psychol Psychiatry. 2017;58(10):1092–105.

Article   PubMed   PubMed Central   Google Scholar  

Law J, Reilly S, Snow PC. Child speech, language and communication need re-examined in a public health context: a new direction for the speech and language therapy profession. International Journal of Language & Communication Disorders. 2013;48(5):486–96.

Collisson BA, Graham SA, Preston JL, Rose MS, McDonald S, Tough S. Risk and protective factors for late talking: an epidemiologic investigation. J Pediatr. 2016;172:168–74.

Le HN, Gold L, Mensah F, Eadie P, Bavin EL, Bretherton L, et al. Service utilisation and costs of language impairment in children: the early language in Victoria Australian population-based study. International journal of speech-language pathology. 2017;19(4):360–9.

Eadie P, Conway L, Hallenstein B, Mensah F, McKean C, Reilly S. Quality of life in children with developmental language disorder. Int J Lang Commun Disord. 2018;53(4):799-810.

Reilly S, Wake M, Ukoumunne OC, Bavin E, Prior M, Cini E, et al. Predicting language outcomes at 4 years of age: findings from early language in Victoria study. Pediatrics. 2010;126(6):e1530–e7.

Law J, Todd L, Clark J, Mroz M, Carr J. Early language delays in the UK. London: Save the Children; 2013.

Google Scholar  

Law J, Charlton J, Dockrell J, Gascoigne M, McKean C, Theakston A. Early Language Development: Needs, provision, and intervention for preschool children from socio-economically disadvantage backgrounds. Institute of Education-London; 2017.

Clegg J, Ansorge L, Stackhouse J, Donlan C. Developmental communication impairments in adults: outcomes and life experiences of adults and their parents. Language, Speech, and Hearing Services in Schools. 2012;43(4):521–35.

Brownlie EB, Graham E, Bao L, Koyama E, Beitchman JH. Language disorder and retrospectively reported sexual abuse of girls: severity and disclosure. J Child Psychol Psychiatry. 2017;58(10):1114–21.

Article   CAS   PubMed   Google Scholar  

Beitchman JH, Jiang H, Koyama E, Johnson CJ, Escobar M, Atkinson L, et al. Models and determinants of vocabulary growth from kindergarten to adulthood. J Child Psychol Psychiatry. 2008;49(6):626–34.

Brownlie E, Beitchman JH, Escobar M, Young A, Atkinson L, Johnson C, et al. Early language impairment and young adult delinquent and aggressive behavior. J Abnorm Child Psychol. 2004;32(4):453–67.

Beitchman JH, Wilson B, Douglas L, Young A, Adlaf E. Substance use disorders in young adults with and without LD: predictive and concurrent relationships. J Learn Disabil. 2001;34(4):317–32.

Beitchman JH, Wilson B, Johnson CJ, Atkinson L, Young A, Adlaf E, et al. Fourteen-year follow-up of speech/language-impaired and control children: psychiatric outcome. J Am Acad Child Adolesc Psychiatry. 2001;40(1):75–82.

Snowling MJ, Adams JW, Bishop D, Stothard SE. Educational attainments of school leavers with a preschool history of speech-language impairments. Int J Lang Commun Disord. 2001;36(2):173–83.

Hughes N, Chitsabesan P, Bryan K, Borschmann R, Swain N, Lennox C, et al. Language impairment and comorbid vulnerabilities among young people in custody. J Child Psychol Psychiatry. 2017;58(10):1106–13.

Conti-Ramsden G, Durkin K. Postschool educational and employment experiences of young people with specific language impairment. Language, Speech, and Hearing Services in Schools. 2012;43(4):507–20.

Caspi A, Houts RM, Belsky DW, Harrington H, Hogan S, Ramrakha S, et al. Childhood forecasting of a small segment of the population with large economic burden. Nat Hum Behav. 2016;1:0005.

Goldfeld S, O’Connor M, Cloney D, Gray S, Redmond G, Badland H, et al. Understanding child disadvantage from a social determinants perspective. J Epidemiol Community Health. 2018;72(3):223–9.

Bronfenbrenner U. Making human beings human: bioecological perspectives on human development. Thousand Oaks: Sage; 2005.

Harding JF, Morris PA, Hughes D. The relationship between maternal education and children’s academic outcomes: a theoretical framework. J Marriage Fam. 2015;77(1):60–76.

Frech A, Kimbro RT. Maternal mental health, neighborhood characteristics, and time investments in children. J Marriage Fam. 2011;73(3):605–20.

Baydar N, Kuntay AC, Yagmurlu B, Aydemir N, Cankaya D, Goksen F, et al. "it takes a village" to support the vocabulary development of children with multiple risk factors. Dev Psychol. 2014;50(4):1014–25.

Tough SC, Siever JE, Leew S, Johnston DW, Benzies K, Clark D. Maternal mental health predicts risk of developmental problems at 3 years of age: follow up of a community based trial. BMC pregnancy and childbirth. 2008;8(1):16.

Smith J, Levickis P, Eadie T, Bretherton L, Conway L, Goldfeld S. Concurrent associations between maternal behaviours and infant communication within a cohort of women and their infants experiencing adversity. Int J Speech Lang Pathol. 2018;20(5):516-27.

Reilly S, Wake M, Bavin EL, Prior M, Williams J, Bretherton L, et al. Predicting language at 2 years of age: a prospective community study. Pediatrics. 2007;120(6):e1441–e9.

Barre N, Morgan A, Doyle LW, Anderson PJ. Language abilities in children who were very preterm and/or very low birth weight: a meta-analysis. J Pediatr. 2011;158(5):766–74.

Durand VN, Loe IM, Yeatman JD, Feldman HM. Effects of early language, speech, and cognition on later reading: a mediation analysis. Front Psychol. 2013;4:586.

Law J, Rush R, Parsons S, Schoon I. The relationship between gender, receptive vocabulary, and literacy from school entry through to adulthood. International Journal of Speech-Language Pathology. 2013;15(4):407–15.

Farrant BM, Zubrick S. R. Parent–child book reading across early childhood and child vocabulary in the early school years: findings from the longitudinal study of Australian children. First Lang. 2013;3(33):280–93.

McKean C, Mensah FK, Eadie P, Bavin EL, Bretherton L, Cini E, et al. Levers for language growth: characteristics and predictors of language trajectories between 4 and 7 years. PLoS One. 2015;10(8):e0134251.

Harrison L, McLeod S. Risk and protective factors associated with speech and language impairment in a nationally representative sample of 4- to 5-year-old children. Journal of Speech, Language, and Hearing Research. 2010;53(2):508–29.

Choudhury N, Benasich AA. A family aggregation study: the influence of family history and other risk factors on language development. Journal of Speech, Language, and Hearing Research. 2003;46(2):261–72.

Short K, Eadie P, Descallar J, Comino E, Kemp L. Longitudinal vocabulary development in Australian urban aboriginal children: protective and risk factors. Child Care Health Dev. 2017;43(6):906–17.

Tayler C. The E4Kids study: Assessing the effectiveness of Australian early childhood education and care programs: overview of findings at 2016. Melbourne, Australia: The University of Melbourne: Melbourne School of Graduate Education; 2016. 31/10/2016

Farkas G, Beron K. The detailed age trajectory of oral vocabulary knowledge: differences by class and race. Soc Sci Res. 2004;33(3):464–97.

Nelson KE, Welsh JA, Trup EMV, Greenberg MT. Language delays of impoverished preschool children in relation to early academic and emotion recognition skills. First Lang. 2011;31(2):164–94.

Rice ML, Hoffman L. Predicting vocabulary growth in children with and without specific language impairment: a longitudinal study from 2; 6 to 21 years of age. Journal of Speech, Language, and Hearing Research. 2015;58(2):345–59.

Law J, Dennis JA, Charlton JJV. Speech and language therapy interventions for children with primary speech and/or language disorders. Cochrane Database Syst Rev. 2017, Issue 1. Art. No.: CD012490. https://doi.org/10.1002/14651858.CD012490 .

Reilly S, Bavin EL, Bretherton L, Conway L, Eadie P, Cini E, et al. The early language in Victoria study (ELVS): a prospective, longitudinal study of communication skills and expressive vocabulary development at 8, 12 and 24 months. International Journal of Speech-Language Pathology. 2009;11(5):344–57.

Zubrick SR, Taylor CL, Rice ML, Slegers DW. Late language emergence at 24 months: an epidemiological study of prevalence, predictors, and covariates. Journal of Speech, Language, and Hearing Research. 2007;50(6):1562–92.

Taylor CL, Christensen D, Lawrence D, Mitrou F, Zubrick SR. Risk factors for children's receptive vocabulary development from four to eight years in the longitudinal study of Australian children. PLoS One. 2013;8(9):e73046.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Christensen D, Taylor CL, Zubrick SR. Patterns of multiple risk exposures for low receptive vocabulary growth 4-8 years in the longitudinal study of Australian children. PLoS One. 2017;12(1):e0168804.

Conti-Ramsden G, Durkin K. What factors influence language impairment considering resilience as well as risk. Folia Phoniatrica et Logopaedica. 2015;67(6):293–9.

Johnson SB, Riis JL, Noble KG. State of the art review: poverty and the developing brain. Pediatrics. 2016;137(4):1–16.

Brofenbrenner U, Morris P. Theoretical models of human development. In: Damn W, Lerner RM, editors. Handbook of child psychology, vol. 1; 1998. p. 993–1028.

Assmussen K, Feinstein L, Martin J, Chowdry H. Foundations for life: what works to support parent child interaction in the early years. London: Early Intervention Foundation; 2016.

Blackman T, Wistow J, Byrne D. Using qualitative comparative analysis to understand complex policy problems. Evaluation. 2013;19(2):126–40.

Kemp L, Harris E, McMahon C, Matthey S, Vimpani G, Anderson T, et al. Miller early childhood sustained home-visiting (MECSH) trial: design, method and sample description. BMC Public Health. 2008;8(1):424.

Cox JL, Holden JM, Sagovsky R. Detection of postnatal depression: development of the 10-item Edinburgh postnatal depression scale. Br J Psychiatry. 1987;150:782–6.

Wechsler D. Wechsler preschool and primary scale of intelligence (WPPSI-III). Third edition ed. San Antonio: Harcourt Asessment. Inc.; 2002.

Schneider CQ, Wagemann C. Set-theoretic methods for the social sciences: a guide to qualitative comparative analysis. Cambridge: Cambridge University Press; 2012.

Ragin C. Redesigning social inquiry: set relations in social research. Chicago: University of Chicago Press; 2008.

Book   Google Scholar  

Rihoux B, Ragin C. Configurational comparative methods: qualitative comparative analysis (QCA) and related techniques. California and London Sage: Thousand Oaks; 2009.

Caldwell BM, Bradley RH. Home inventory administration manual: University of Arkansas for Medical Sciences; 2003.

NICHD ECCRN. Child care and mother-child interaction in the first 3 years of life. Dev Psychol. 1999;35(6):1399–413.

Quine WV. The problem of simplifying truth functions. Am Math Mon. 1952;59(8):521–31.

Quine WV. A way to simplify truth functions. Am Math Mon. 1955;62(9):627–31.

Smith J, Eadie T, Levickis P, Bretherton L, Goldfeld S. Predictive validity of verbal and non-verbal communication and mother–child turn-taking at 12 months on language outcomes at 24 and 36 months in a cohort of infants experiencing adversity: a preliminary study. International Journal of Language & Communication Disorders. 2018.

Landry SH, Smith KE, Swank PR, Zucker T, Crawford AD, Solari EF. The effects of a responsive parenting intervention on parent-child interactions during shared book reading. Dev Psychol. 2012;48(4):969–86.

Hudson S, Levickis P, Down K, Nicholls R, Wake M. Maternal responsiveness predicts child language at ages 3 and 4 in a community-based sample of slow-to-talk toddlers. International journal of language & communication disorders. 2015;50(1):136–42.

Zambrana IM, Pons F, Eadie P, Ystrom E. Trajectories of language delay from age 3 to 5: persistence, recovery and late onset. International Journal of Language & Communication Disorders. 2014;49(3):304–16.

Snowling MJ, Duff FJ, Nash HM, Hulme C. Language profiles and literacy outcomes of children with resolving, emerging, or persisting language impairments. J Child Psychol Psychiatry. 2016;57(12):1360–9.

Dale P, Simonoff E, Bishop D, Eley T, Oliver B, Price T, et al. Genetic influence on language delay in two-year-old children. Nat Neurosci. 1998;1(4):324.

Rice ML, Zubrick SR, Taylor CL, Gayán J, Bontempo DE. Late language emergence in 24-month-old twins: heritable and increased risk for late language emergence in twins. Journal of Speech, Language, and Hearing Research. 2014;57(3):917–28.

Rice ML, Zubrick SR, Taylor CL, Hoffman L, Gayán J. Longitudinal study of language and speech of twins at 4 and 6 years: twinning effects decrease, zygosity effects disappear, and heritability increases. Journal of Speech, Language, and Hearing Research. 2018;61(1):79–93.

Hayiou-Thomas ME, Dale PS, Plomin R. The etiology of variation in language skills changes with development: a longitudinal twin study of language from 2 to 12 years. Dev Sci. 2012;15(2):233–49.

Dale PS, Dionne G, Eley TC, Plomin R. Lexical and grammatical development: a behavioural genetic perspective. Journal of child language. 2000;27(3):619–42.

Newbury DF, Fisher SE, Monaco AP. Recent advances in the genetics of language impairment. Genome medicine. 2010;2(1):6.

Rice ML. Language growth and genetics of specific language impairment. International Journal of Speech-Language Pathology. 2013;15(3):223–33.

Hoff E. Interpreting the early language trajectories of children from low SES and language minority homes: implications for closing achievement gaps. Dev Psychol. 2013;49(1):4–14.

Wake M, Tobin S, Girolametto L, Ukoumunne OC, Gold L, Levickis P, et al. Outcomes of population based language promotion for slow to talk toddlers at ages 2 and 3 years: Let’s learn language cluster randomised controlled trial. Br Med J. 2011;343(7821):1–10.

Landry SH, Smith KE, Swank PR, Guttentag C. A responsive parenting intervention: the optimal timing across early childhood for impacting maternal behaviors and child outcomes. Dev Psychol. 2008;44(5):1335.

Olds DL, Kitzman H, Knudtson MD, Anson E, Smith JA, Cole R. Effect of home visiting by nurses on maternal and child mortality: results of a 2-decade follow-up of a randomized clinical trial. JAMA Pediatr. 2014;168(9):800–6.

Olds DL, Kitzman H, Cole R, Robinson J, Sidora K, Luckey DW, et al. Effects of nurse home-visiting on maternal life course and child development: age 6 follow-up results of a randomized trial. Pediatrics. 2004;114(6):1550–9.

Bayley N. Bayley scales of infant development. 2nd ed. Psychological Corp: San Antonio; 1993.

Kemp L, Harris E, McMahon C, Matthey S, Vimpani G, Anderson T, et al. Child and family outcomes of a long-term nurse home visitation programme: a randomised controlled trial. Arch Dis Child. 2011;96(6):533–40.

Download references

Acknowledgements

The authors would like to thank all of the mothers and children who gave up their time over many years to be part of this study. Thanks also to the many research officers and practitioners who collected data in the van and the clinic to name a few: Sheryl Sharkie, Cathy Kaplan, Amanda Kemp, Roberta Chavez, Astrid Toscan, Maretta Coleman, Kat O’Heir, Manal Nasreddine and Brooke Butt. Finally, thanks to Fiona Byrne for her assistance in making the paths clear.

The trial was funded by the Australian Research Council (LP0560285), Sydney South West Area Health Service, the NSW Department of Community Services and the NSW Department of Health. Australian Research Council Linkage Projects Funding (DP0770212).

Availability of data and materials

Most of the dataset(s) supporting the conclusions of this article are included within the article and its additional file. All other datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Author information

Authors and affiliations.

School of Public Health and Community Medicine, University of NSW, Sydney, NSW, Australia

Graduate School of Education, University of Melbourne, Parkville, Victoria, Australia

Patricia Eadie

School of Nursing and Midwifery, Western Sydney University, Sydney, NSW, Australia

Ingham Institute of Applied Medical Research, Liverpool, NSW, Australia

You can also search for this author in PubMed   Google Scholar

Contributions

LK was the trial coordinator and led the conceptualisation and design of the broader study. KS was a research assistant in collection of data, designed this current study as part of her doctoral thesis and conducted data analysis and interpretations. KS, LK, and PE designed the current study, interpreted the results, drafted and approved the final manuscript.

Corresponding author

Correspondence to Kate Short .

Ethics declarations

Ethics approval and consent to participate.

This study received approval from both the Sydney South West Area Health Service (Western Zone) and the University of NSW Human Ethics Committees. Written informed consent was provided by all adult participants as well as from the parents/guardians of the minors included in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Additional file

Additional file 1:.

Paths to language development in at risk children: a Qualitative Comparative Analysis (QCA). (PDF 1990 kb)

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License ( http://creativecommons.org/licenses/by/4.0/ ), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Cite this article.

Short, K., Eadie, P. & Kemp, L. Paths to language development in at risk children: a qualitative comparative analysis (QCA). BMC Pediatr 19 , 94 (2019). https://doi.org/10.1186/s12887-019-1449-z

Download citation

Received : 02 November 2018

Accepted : 05 March 2019

Published : 05 April 2019

DOI : https://doi.org/10.1186/s12887-019-1449-z

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Children at risk
• Language development
• Risk and protective factors

BMC Pediatrics

ISSN: 1471-2431

• Submission enquiries: [email protected]
• General enquiries: [email protected]

Don't Miss Out!

Subscribe to the E-Bulletin for regular updates on research, free resources, solutions, and job postings from WestEd.

Subscribe Now

Upcoming Event

Curriculum development for english language development: planning culturally responsive and sustaining instruction.

California’s English Learner (EL) students deserve effective, culturally sustaining and responsive designated and integrated English Language Development (ELD) instruction. To achieve this, teachers need high-quality materials, ELD Standards-aligned curriculum, and a coherent instructional system . 

This webinar will present a model for designing and supplementing curriculum for culturally sustaining and responsive ELD instruction in the context of professional learning communities.

Content area teachers will receive guidance on how to supplement their curriculum to make explicit the language demands of their content standards. Designated ELD teachers will receive guidance on how to ensure alignment between ELD Standards, assessment, and instruction.

Participants will also learn about a model for developing or supplementing curriculum for integrated and designated ELD, and consider formative assessment uses of English Language Proficiency Assessments for California (ELPAC) task types and the Observation Protocol for Teachers of English Learners (OPTEL).

This webinar is intended for content area and designated ELD teachers, administrators, and professional learning providers committed to achieving equitable outcomes for EL students.

Registration Options

Dr. Christine Snyder is a Research Associate at WestEd. Snyder designs professional learning with the San Diego County Office of Education, performs district English Language Arts (ELA) curriculum reviews, and facilitates the Interim Formative Assessment ELPAC hand-scoring trainings for the California Department of Education (CDE). Through the Region 15 Comprehensive Center at WestEd, Snyder supports the CDE’s OPTEL rollout and ELD Standards webinars. A former designated ELD and ELA teacher, Snyder’s approach to professional learning emerges from her classroom experience. She holds a PhD in Education from Claremont Graduate University and an MA in the Teaching of English from Columbia University Teachers College.

Tony Mora is a District Advisor at the San Diego County Office of Education and a Regional English Learner Specialist for the CDE in Region 9 (San Diego, Orange, and Imperial Counties). He supports districts and charters with Title III plan development, implementation, and evaluation. He also supports local educational agencies (LEAs) with EL/multilingual learner policy, education code, and compliance items (such as California EL Roadmap implementation, reclassification, OPTEL, ELPAC, ELD, and EL Master Plans, and more). He is a former site administrator, district EL resource teacher, ELD coordinator, middle school teacher, and bilingual resource specialist program teacher.

Dr. Julie Webb is a Senior Program Associate on the Assessment for Learning team at WestEd. Webb is a former classroom teacher and reading specialist with National Board Certification in Early and Middle Childhood Literacy. Webb’s role at WestEd includes leading professional learning and providing technical assistance to districts and state departments of education to foster educator learning around evidence-based literacy and language instruction and building assessment literacy. Webb leads research and evaluation projects including school/family engagement for parents and caregivers of multilingual children and professional learning courses promoting simultaneous content knowledge and language development practices for teachers of multilingual students. Webb holds an EdD in Educational and Organizational Leadership and an MA in Curriculum and Instruction with an emphasis on Professional Development.

Dr. Yetunde Akinola is a Research Associate at WestEd, where she supports local education agencies and state departments of education working to achieve equitable systems for all. Her research and professional interests include culturally sustaining and responsive pedagogy, school climate and culture, and addressing educational disparities for Black and Brown communities. A former early childhood teacher, Akinola founded a program in Washington, D.C. that partnered with families to support children, ages 1–5, on a path of long-term success. Akinola is dedicated to family and community engagement, especially home-school partnerships. Akinola holds an MA and PhD in developmental psychology from Howard University.

View More Events

Ask a question, request information, make a suggestion, or sign up for our newsletter.

• WestEd Bulletin
• Insights & Impact
• Equity in Focus
• Areas of Work
• Charters & School Choice
• Comprehensive Assessment Solutions
• Culturally Responsive & Equitable Systems
• Early Childhood Development, Learning, and Well-Being
• Economic Mobility, Postsecondary, and Workforce Systems
• English Learner & Migrant Education Services
• Justice & Prevention
• Learning & Technology
• Mathematics Education
• Resilient and Healthy Schools and Communities
• School and District Transformation
• Special Education Policy and Practice
• Strategic Resource Allocation and Systems Planning
• Supporting and Sustaining Teachers
• Professional Development
• Research & Evaluation
• How We Can Help
• Reports & Publications
• Technical Assistance
• Technical Assistance Services
• Policy Analysis and Other Support
• New Releases
• Top Downloads
• R&D Alert
• Best Sellers
• Board of Directors
• Equity at WestEd
• WestEd Pressroom
• WestEd Offices
• Work with WestEd

Work at WestEd

Princeton University

Princeton engineering, can language models read the genome this one decoded mrna to make better vaccines..

By Scott Lyon

April 8, 2024

Princeton researchers led by Mengdi Wang have developed a language model to home in on partial genome sequences and optimize those sequences to improve function for the development of mRNA vaccines and other therapies. Illustration from Adobe Stock.

The same class of artificial intelligence that made headlines coding software and passing the bar exam has learned to read a different kind of text — the genetic code.

That code contains instructions for all of life’s functions and follows rules not unlike those that govern human languages. Each sequence in a genome adheres to an intricate grammar and syntax, the structures that give rise to meaning. Just as changing a few words can radically alter the impact of a sentence, small variations in a biological sequence can make a huge difference in the forms that sequence encodes.

Now Princeton University researchers led by machine learning expert Mengdi Wang are using language models to home in on partial genome sequences and optimize those sequences to study biology and improve medicine. And they are already underway.

In a paper published April 5 in the journal Nature Machine Intelligence, the authors detail a language model that used its powers of semantic representation to design a more effective mRNA vaccine such as those used to protect against COVID-19.

Found in Translation

Scientists have a simple way to summarize the flow of genetic information. They call it the central dogma of biology. Information moves from DNA to RNA to proteins. Proteins create the structures and functions of living cells.

Messenger RNA, or mRNA, converts the information into proteins in that final step, called translation. But mRNA is interesting. Only part of it holds the code for the protein. The rest is not translated but controls vital aspects of the translation process.

Governing the efficiency of protein production is a key mechanism by which mRNA vaccines work. The researchers focused their language model there, on the untranslated region, to see how they could optimize efficiency and improve vaccines.

After training the model on a small variety of species, the researchers generated hundreds of new optimized sequences and validated those results through lab experiments. The best sequences outperformed several leading benchmarks for vaccine development, including a 33% increase in the overall efficiency of protein production.

Increasing protein production efficiency by even a small amount provides a major boost for emerging therapeutics, according to the researchers. Beyond COVID-19, mRNA vaccines promise to protect against many infectious diseases and cancers.

Wang, a professor of electrical and computer engineering and the principal investigator in this study, said the model’s success also pointed to a more fundamental possibility. Trained on mRNA from a handful of species, it was able to decode nucleotide sequences and reveal something new about gene regulation. Scientists believe gene regulation, one of life’s most basic functions, holds the key to unlocking the origins of disease and disorder. Language models like this one could provide a new way to probe.

Wang’s collaborators include researchers from the biotech firm RVAC Medicines as well as the Stanford University School of Medicine.

The Language of Disease

The new model differs in degree, not kind, from the large language models that power today’s AI chat bots. Instead of being trained on billions of pages of text from the internet, their model was trained on a few hundred thousand sequences. The model also was trained to incorporate additional knowledge about the production of proteins, including structural and energy-related information.

The research team used the trained model to create a library of 211 new sequences. Each was optimized for a desired function, primarily an increase in the efficiency of translation. Those proteins, like the spike protein targeted by COVID-19 vaccines, drive the immune response to infectious disease.

Previous studies have created language models to decode various biological sequences, including proteins and DNA, but this was the first language model to focus on the untranslated region of mRNA. In addition to a boost in overall efficiency, it was also able to predict how well a sequence would perform at a variety of related tasks.

Wang said the real challenge in creating this language model was in understanding the full context of the available data. Training a model requires not only the raw data with all its features but also the downstream consequences of those features. If a program is designed to filter spam from email, each email it trains on would be labeled “spam” or “not spam.” Along the way, the model develops semantic representations that allow it to determine what sequences of words indicate a “spam” label. Therein lies the meaning.

Wang said looking at one narrow dataset and developing a model around it was not enough to be useful for life scientists. She needed to do something new. Because this model was working at the leading edge of biological understanding, the data she found was all over the place.

“Part of my dataset comes from a study where there are measures for efficiency,” Wang said. “Another part of my dataset comes from another study [that] measured expression levels. We also collected unannotated data from multiple resources.” Organizing those parts into one coherent and robust whole — a multifaceted dataset that she could use to train a sophisticated language model — was a massive challenge.

“Training a model is not only about putting together all those sequences, but also putting together sequences with the labels that have been collected so far. This had never been done before.”

The paper, “A 5′ UTR Language Model for Decoding Untranslated Regions of mRNA and Function Predictions,” was published in Nature Machine Learning. Additional authors include Dan Yu, Yupeng Li, Yue Shen and Jason Zhang, from RVAC Medicines; Le Cong from Stanford; and Yanyi Chu and Kaixuan Huang from Princeton.

Related News

Episode 2: That Magic Touch

Episode 3: Haydn Seek

Episode 1: Stanley Jordan Pulls Out All the Stops

Personalizing ChatGPT can make it more offensive, researchers find

Princeton IP Accelerator funding awarded to support promising new technologies

Built for AI, this chip moves beyond transistors for huge computational gains

Mengdi Wang

Bioengineering and Health

Data Science

Related departments and centers.

Electrical and Computer Engineering

Omenn-Darling Bioengineering Institute

Crane Center For Early Childhood Research and Policy

The Crane Center for Early Childhood Research and Policy is a multidisciplinary research center conducting research related to children’s learning and well-being, and seeking to impact early childhood policy and practice.

Quick Links

Our Staff Our Work Partner With Us Diversity, Equity, and Inclusion Current Partners

• News and Events

CRANE RESEARCH FORUM RECAP – Shared Parental Responsiveness and Child Development Among Families with Low Income

Dr. Joyce Lee , assistant professor of social work at The Ohio State University and director of the Child and Family Wellbeing Laboratory, discussed how preschoolers’ behavior and receptive language development are affected by shared parental responsiveness in low-income households.

Responsive relationships are important in young children’s early development, but the ways in which fathers and mothers work together as a system — as well as the role of shared parental responsiveness in child development — are not well understood. In this presentation, Dr. Joyce Lee discussed findings on the effects of shared parental responsiveness between fathers and mothers in low-income households on preschoolers’ developmental outcomes. The specific developmental outcomes studied include children’s behavior problems, prosocial behaviors and receptive language.

Dr. Lee’s research aims to promote child welfare and family strengthening through preventing child maltreatment, supporting positive parenting and promoting the health of children in foster care. Her work is intended to inform child welfare policies and practices to improve children’s health outcomes and strengthen children’s relationships with their family members.

– Read the research brief written by Dr. Lee on this work.

– Check out the full research article for a more in-depth look at the research behind Dr. Lee’s presentation.

• The Graduate School >
• Beyond the Classroom >
• Graduate Professional Development >
• Art of Research >
• Art of Research Winners >

2024 Art of Research Winners

Explore the 2024 Art of Research competition winners below. Work is also on display at the Buffalo Museum of Science throughout May. 

Grand Prize

Monster larvae.

Robert Ditter Postdoctoral Scholar Biological Sciences, College of Arts and Sciences

A depiction of deep-sea Decapod larvae that were linked to their adult form through the use of molecular genetics tools (i.e., DNA Barcoding). Until the advent of modern genetics methods, these organisms were commonly referred to as "monster larvae". Now we know that many of the would-be monsters are actually the larval forms of commercially valuable crustacean species.

Judge's Choice

Best communication of research, magnetic nano eye.

Damalka Balasuriya Graduate Student Chemistry (PhD), College of Arts and Sciences

Explore the world of "Nanoscale," where science and technology delve into the super tiny scale. Imagine a nanometer, so small it's 100,000 times tinier than the width of a human hair. Now, meet the magnetic nano eye, a super tiny nano-sized particle that functions like a perceptive eye in the realm of analytical chemistry, gathering pollutants like an eye captures light. This enables a closer look at our surroundings to unveil hidden secrets of the environment. It possesses a magnetic core and a cover made of silica fibers. The magnetic core, coated in a sleek black hue, spans a mere 300 nanometers in width, imbuing the particle with magnetic power which allows for easy separation of the particles after capturing pollutants from environmental samples. Meanwhile, layers of vibrant blue and luminous yellow representing the fibrous layer span up to 200 nanometers, enhancing the particle's surface area. This enables it to capture and retain pollutants with unparalleled efficiency, allowing the unseen to be seen.

Most Intriguing Composition

Overdrive: the ecological impact of nostalgia and the marketing of memory.

Jonathan Bolt Graduate Student Art (MFA), College of Arts and Sciences

Overdrive: The Ecological Impact of Nostalgia and the Marketing of Memory is a visualization and reflection on the complicated relationship we have to childhood objects that we cherish, and how nostalgia is intertwined in a larger system of marketing tactics and long-term environmental consequences. I specifically focus on the consumption and disposal of toys, drawing from a deep personal attachment to toy cars in my childhood. Experiencing the world through a neurodivergent lens, I saw my college of Hot Wheels cars as an escape from childhood hardships, a world that I can imagine with a narrative all my own. Despite using the world of the object as my escape, I now confront the harsh reality that the objects I cherish are created with crude oil, from the car's paint to the plastic wheels and base. I must take into account that these elements produce material waste in production and shipping, as well as the inevitable time when the toys will be discarded. The degradation of our environment is far more complicated than simply stating we have a problem with consumption. To confront our ecological situation, we must start by taking a look at ourselves. Nostalgia, and our deep yearn for simpler times, is a marketing tactic used to sell us more harmful rubbish, exacerbated by a constant stream of sequels, remasters, and nostalgia-themed brand content. An important step in processing climate change is to think of memory, consumption and nostalgia as something enmeshed in larger environmental issues.

Most Polychromatic

Brain's blueprint: mapping the brain's connectivity using tractography.

Elizabeth Castro Graduate Student Orthopedics (PhD), Jacobs School of Medicine and Biomedical Sciences

Tractography is a technique used in neuroscience and medical imaging to visualize and map connections in the brain. This technique gives researchers and clinicians a roadmap to understand and investigate the intricate connections between different regions of the brain. These connections, called white matter tracts, can be thought of as the "wiring" of the brain. Tractography analysis allows researchers to estimate the presence of these tracts using magnetic resonance imaging (MRI). These tracts play roles in communication between brain regions, the integration of information, the adaptability of the brain, and many other functions involving movement, memory and learning.

In examining these white matter tracts throughout the brain, many diseases and disorders of the brain can be understood. Specifically, tractography has proven useful in researching traumatic brain injury (TBI). TBI is an injury that often involves white matter tract damage, specifically within the corpus callosum (pictured). This has the potential to cause disturbances to brain functions (mentioned above). Information flow can be diminished, the brain's recovery from TBI can be difficult, and regular learning and memory functioning are affected. However, the different ways in which these injuries present in people make it challenging to find patient-specific methods to treat TBIs. Through the lens of tractography, research aims to unveil the mechanisms and consequences of TBIs, paving the way for improved understanding, treatment and recovery. 

Best Live-Action Fieldwork

Gazing through the mist: surveying for invasive species.

Sarah Chang Graduate Student Biological Sciences (PhD), College of Arts and Sciences

Photo taken at sunrise of scientists conducting fieldwork in Tonawanda Creek. Electrofishing surveys are conducted early in the morning to investigate the occurrence of invasive species in Lake Erie and associated riverways. Water conditions in early fall can lead to warm water temperatures evaporating into the cold early morning air.

Best Layered Visualization

Expanding the narrow.

Viyona Chavan Graduate Student Urban and Regional Planning (MS), School of Architecture and Planning

Rooted in my research on Dharavi, India, one of Asia's largest informal settlements, my work sheds light on the remarkable diversity of its residents. Despite facing challenging conditions, the people of Dharavi exhibit a rich tapestry of coexistence, making the most of their surroundings with creativity and resourcefulness.

Picture a slender passage where two people squeeze past each other, yet it morphs into more than just a walkway. It becomes a playground for children, a venue for wedding preparation, a place for water storage, a cozy conversation corner, a spill-out space, and even a modest grocery store. Dharavi's alleys, though physically narrow, unfold stories of resilience, adaptability, and an unwavering sense of community.

The uneven alleyways, with sewer lines running beneath, act as invisible borders between two sides of the street. Paradoxically, these tangible lines symbolize an illusionary divide—a reminder that boundaries are often perceptions rather than realities.

Composed as a digital illustration collage from images captured during my observational field visits, my artwork emphasizes the endless possibilities within seemingly restricted spaces. The message resonates beyond Dharavi, underscoring the universal truth that the size of a space is insignificant when it comes to fostering vibrant life, human connection and perception towards life.

Most Playful Representation of Research

Toast to tomorrow: cheers to micro-valve magic.

Aditya Chivate Graduate Student Industrial and Systems Engineering (PhD), School of Engineering and Applied Sciences

Introducing a groundbreaking marvel of microengineering: the 3D printed micro-valve, smaller than a grain of rice yet mighty in its capabilities. Disguised within the whimsical shape of a champagne flute tower, this tiny wonder harbors a sophisticated multiport selector valve capable of directing matter unidirectionally with unparalleled precision. While its playful exterior may evoke visions of elegant soirées and celebratory toasts, the true magic lies within its intricately designed mechanism.

With the gentle manipulation of this miniature marvel, controlled drug flow becomes not just a possibility but a reality. In the realm of biomedicine, where every drop counts, this micro-valve stands poised to revolutionize drug delivery, ensuring medications reach their intended targets with unparalleled accuracy and efficiency.

Crafted through the cutting-edge art of micro 3D printing, this valve showcases the immense potential of precision engineering on a microscopic scale. So, as we marvel at the whimsical form of the champagne flute tower, let us not be deceived by its playful facade. Within its miniature frame lies a testament to human ingenuity, a symbol of innovation and progress that transcends the boundaries of size and shape. Let us raise a toast to the limitless potential of tiny wonders and the profound impact they hold on our world.

Most Out of This World

A martian dreamscape.

Katelyn Eaman Graduate Student Geological Sciences, College of Arts and Sciences

What comes to mind when you hear geologist? Maybe a khakied individual with a flannel, holding a rock hammer while surveying mountainous lands? Although some of that may still hold true, the vast majority of geoscientists surpass that stereotype. We are modelers. We are coders. We are remote sensing scientists. We are mappers.

A few of us do not even study our home-Earth. As a planetary scientist, I cannot walk upon the ground I study. I cannot feel its surface. I cannot breathe its air. I must rely on distant data. In my research, I investigate ancient channels, carved once by liquid water, on a dormant volcano in the southern highlands of Mars. Using a conglomeration of orbiter satellite imagery and applications in Geographic Information Systems, I map these channels and other erosional surface features along the flanks of Hadriacus Mons.

This is a 3D model of Hadriacus I generated from a blend of multiple digital elevation model data, derived from the Mars Orbiter Laser Altimeter aboard NASA's Mars Global Surveyor spacecraft. It is vertically exaggerated to show the volcano's unique geomorphology. Although, I cannot physically stand in the presence of Mars' rocks, I can dream right?

Most Transformative

Total molar eclipse: blacking out the oral microbiome.

Cole Matrishin Graduate Student Oral Biology (PhD), School of Dental Medicine

Your mouth contains a subset of over 700 species of bacteria! This image shows the selective growth of some of these bacteria under visible and ultraviolet light. Which helps identify "hidden" aspects of diversity and ecology when it comes to characterizing members of a community, such as the oral microbiome.

Our lab focuses on better understanding the interactions between bacteria and their viral predators, known as "phages", and the role these interactions play in the context of the oral microbiome. A common approach when studying phages is to first isolate and identify their bacterial hosts. The circles shown in this image are pictures of a nutrient agar plate on which bacteria were grown from a donor's dental plaque scrapings. The picture of the red plate was taken under visible light, showing the diversity we can observe with the naked eye. While the purple plate is the same as the red plate but was pictured under a 395 nm ultraviolet light. This revealed fluorescent pink bacteria that were not observable with the naked eye. An observation that can aid in their identification and understanding of their ecological impact that would have otherwise been missed.

This image was created in an aesthetic manner to depict the transition of viewing the bacterial community under visible light into ultraviolet light and back to visible light. Thus, creating the appearance of an "eclipse". Highlighting that, as with many things, there are additional layers to better understanding than what is most easily observable.

People's Choice

Shaping bilingual minds: the power of family language policy in biliteracy.

Azzah Alzahrani Graduate Student Learning and Instruction, Graduate School of Education

This photo captures a moment of quiet study, where a young child delves into the world of languages right from their home. This scene, with the girl poised between two linguistic worlds, serves as a powerful metaphor for their balance and integration of multiple cultures and languages in her education. It's a snapshot that represents much more than just a study session; it's a window into the heart of research on Family Language Policy (FLP). The study explores how a family's decision to homeschool in Arabic not only nurtures a child's biliteracy but also weaves a rich tapestry of cultural identity and linguistic skill. By focusing on the daily routines, goals and perspectives of one Arabic-speaking family in the United States. I uncover the profound impact of parental involvement on a child's ability to master two languages. The findings shed light on the intricate dance between teaching and learning, showing how home can be the foundation of lifelong bilingualism. Through this work, I aim to fill a gap in understanding how bilingual families use homeschooling for language and literacy development, offering insights that could guide others on a similar journey.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• HHS Author Manuscripts

Translational research on early language development: Current challenges and future directions

Marjorie beeghly.

Harvard Medical School and Children’s Hospital, Boston

There is a pressing need for the early and accurate identification of young children at risk for language and other developmental disabilities and the provision of timely, age-appropriate intervention, as mandated by Part C of the Individuals with Disabilities Education Act. Research has shown that early intervention is effective for many language impaired children in different etiological groups, and can reduce the functional impact of persistent disorders on children and their families. Yet, the accurate identification of infants and toddlers at risk for language impairment remains elusive, especially for late-talking children without obvious genetic or neurological conditions. In this paper, the need for translational research on basic processes in early language development in typical and atypical populations and the contextual factors that affect them are discussed, along with current challenges and future directions for its successful implementation. Implications of this research for clinical evidence-based practice are also considered.

Of the 12–16% of American children with neurodevelopmental or behavioral disorders ( American Academy of Pediatrics Committee on Children with Disabilities, 2001 ), a majority has delayed or deviant language ( Feldman, 2005 ; Grizzle & Simms, 2005 ). This is especially the case when disorders result from genetic or chromosomal abnormalities such as Down syndrome ( Abbeduto & Murphy, 2004 ; Chapman & Hesketh, 2001 ), fragile X syndrome ( Abbeduto & Hagerman, 1997 ; Dykens, Hodapp, & Finucane, 2000 ), or Williams syndrome ( Bellugi, Wang, & Jernigan, 1994 ; Karmiloff-Smith & Thomas, 2003 ), or from complex syndromes such the autism spectrum disorders ( Happe & Frith, 1996 ; Lord & Paul, 1997 ). Although these disorders are fairly rare, prevalence estimates for preschool-aged children with language impairment with no known genetic etiology and normal nonverbal intelligence (e.g., specific language impairment)are substantially higher, ranging between 2 and 8% (median 5.95%; Feldman, 2005 ). A sizable percentage of late-talking preschoolers with normal nonverbal intelligence will continue to manifest language problems in later childhood ( Grizzle & Simms, 2005 ; Paul, 1996 ), augmenting clinical concerns for this group of children.

Given that language and speech problems are associated with academic, socioemotional, and psychiatric problems ( Catts, 1993 ; Cohen, Davine, Horodezky, Lipsett, & Isaacson, 1993 ; Rice, Hadley, & Alexander, 1993 ), there is a pressing need for early and accurate identification of young children with emergent language problems and the provision of age-appropriate intervention. Moreover, Part C of the Individuals with Disabilities Education Act mandates the early identification and provision of intervention for infants and toddlers with developmental disabilities (birth to age 2)through the development of community-based systems. Studies have shown that early intervention is effective for many language impaired children in different etiological groups ( Guralnick, 1997 ; Leonard, 1998 ; Ramey, Campbell, & Ramey, 1999 ), and can reduce the functional impact of persistent disorders on children and their families. Improvements in sociocommunicative skills are especially well documented.

Unfortunately, accurate detection of true developmental delays or deviations in early childhood remains elusive, especially for late-talking children without obvious genetic or neurological conditions ( Dale, Price, Bishop, & Plomin, 2003 ). This is likely due to the wide range of normal variability in the early stages of typical language development ( Bates, Bretherton, & Snyder, 1988 ; Fenson et al., 1994 ; Goldfield & Snow, 2005 ; Shore, 1995 ) and the fact that many toddlers with delayed language development and normal cognitive and motor development function within normal limits on language assessments by school age ( Leonard, 1998 ; Rescorla, 2002 ). Moreover, children’s early language development in both typical and atypical populations is a dynamic process that is affected by a complex array of transacting factors from multiple levels of influence, such as genetics, gender, temperament, the child’s own skills in other developmental domains (cognitive, motor, socioemotional), and a host of biological and social risk and resilience factors, for example, premature birth, prenatal exposure to substances, parental education, caregiver interactive style, parent–child mutual regulation, bilingualism, and other cultural influences ( Bates, 2003 ; Bates, Bretherton, Beeghly, & McNew, 1982 ; Beeghly, 1997 ; Beeghly & Cicchetti, 1994 ; Cicchetti, 1984 ; Elman et al., 1996 ; Landry, Smith, Miller-Loncar, & Swank, 1997 ; Luthar, Cicchetti, & Becker, 2000 ; Sameroff & Fiese, 1990 ; Spiker, Boyce, & Boyce, 2002 ; Thorpe, Rutter, & Greenwood, 2003 ). This complexity has made it difficult for clinicians to pinpoint the exact nature of the presenting problem and to make unequivocal diagnoses of clinically significant language impairments, especially when children are younger than 3 or 4 years of age ( Leonard, 1998 ; Thal & Katich, 1996 ; Thal, Reilly, Seibert, Jeffries, & Fenson, 2004 ).

Need for Translational Research on Early Language Development

Thus, there is a pressing clinical need for accurate, detailed information on multiple aspects of early typical and atypical language functioning and the diverse cognitive, behavioral, and contextual factors that affect them. In response, the NIH has recently called for collaborative translational (“bench to bedside”) research on basic developmental processes in both typical and atypical populations to better inform and guide clinical practice with children with developmental disorders, including language impairment, and to encourage basic behavioral scientists to seek a further understanding of how behavioral processes (such as language)are altered by developmental disorders. Until recently, however, advancement in collaborative translational language research has been hampered by fragmentation among multidisciplinary fields, related funding constraints, and poor communication among investigators from different disciplines ( McCardle, Cooper, & Freund, 2005 ; Nelson et al., 2002 ; Rice & Warren, 2005b ).

Most extant research on early language acquisition in children with language disorders has focused on a narrow range of language behaviors (e.g., vocabulary or morphosyntax) within one specific clinical group (e.g., Down syndrome, autism, or specific language impairment), describing single deficits relative to various control groups or evaluating within-group profiles of specific linguistic and cognitive skills within a particular group ( Rice & Warren, 2005a ; Rice, Warren, & Betz, 2005 ). Despite this limited focus, this research has provided a wealth of valuable information about the nature and process of those particular aspects of language development in specific populations. Results of profile analyses within different populations have been especially helpful in debunking stereotypes about different etiologic groups and have highlighted methodological caveats for future research, including the need for longitudinal research and the continued need to evaluate multiple dimensions of language (e.g., vocabulary/semantics, grammar/syntax, and pragmatics) in diverse groups. This research has also fueled and sparked continuing debates among researchers regarding the age-old philosophical question of the ontogenetic association between language and cognition (e.g., Elman et al., 1996 ; Karmiloff-Smith & Thomas, 2003 ; Thelen & Bates, 2003 ).

For illustrative purposes, a brief summary of some of the key research findings for four clinical groups of children with delayed or disordered language development that have been investigated intensively (Down syndrome, autism spectrum disorders, Williams syndrome, and specific language impairment) is provided here. This summary is not meant to be exhaustive.

Down syndrome

With a prevalence of approximately 1 in 700 to 1,000, Down syndrome is the most common genetic cause of mental retardation ( Rozien, 1997 ). Children with Down syndrome have been of particular interest to language development researchers because Down syndrome is etiologically distinct and can be identified and studied very early in life. As is true for children with other genetic disorders, children with Down syndrome manifest a unique profile of linguistic, cognitive, affective-motivational, and social abilities in early childhood. Although developmentally delayed, children with Down syndrome vary in the level of their general cognitive functioning from severely retarded to nearly normal ( Chapman & Hesketh, 2001 ). Their delayed yet variable development allows for a more precise examination of the sequences of various aspects of their language and nonverbal development than is possible with typical children with more rapid development ( Cicchetti & Beeghly, 1990 ).

Evidence for both delayed and deviant aspects of early language and nonverbal cognitive development has been reported for children with Down syndrome ( Abbeduto & Murphy, 2004 ; Beeghly, Weiss-Perry, & Cicchetti, 1990 ; Chapman & Hesketh, 2001 ; Fowler, 1990 ; Miller, 1990 ). With increasing age, young children with Down syndrome exhibit increasing linguistic deficits in relation to their nonverbal cognitive abilities ( Chapman, 2003 ; Chapman & Hesketh, 2001 ), including pretend play ( Beeghly, 1997 ; Beeghly, Perry, & Cicchetti, 1989 ; Cielinski, Vaughn, Seifer, & Contreras, 1995 ). However, production delays tend to exceed comprehension delays, and syntactic skills are more severely compromised than lexical or pragmatic abilities ( Beeghly et al., 1990 ; Fowler, 1990 ; Miller, 1990 ). Fowler (1990) has reported that many children with Down syndrome do not progress beyond the early stages of morphological and syntactic development. Strengths and weaknesses have also been reported for cognitive skills associated with language functioning. For instance, children with Down syndrome exhibit deficits in auditory short-term memory relative to visual short-term memory and other cognitive skills ( Chapman, 2003 ).

Of note, specific results for particular domains of functioning within language often vary among studies, depending on the age of the children being evaluated and the specific assessment contexts and dependent measures used. For instance, in an observational study of children with Down syndrome interacting with their mothers during semistructured and unstructured play contexts, children with Down syndrome exhibited pragmatic skills (diversity of speech acts, turn taking skills, connected discourse) that were consistent with their general cognitive abilities ( Beeghly & Cicchetti, 1997 ). In contrast, in an experimental study using a variety of elicitation prompts and dependent measures, children with Down syndrome exhibited both strengths and weaknesses in pragmatic skills ( Abbeduto & Murphy, 2004 ), raising methodological caveats for future studies.

Autism and pervasive developmental disorders

Despite the relatively low incidence of autism (approximately 4–10 cases in every 10,000 live births; Happe, 1995 ), the early language and cognitive development of children with autism and related disorders has been intensively studied during the past 2 decades (see reviews by Baron-Cohen, 1995 ; Cicchetti, Beeghly, & Weiss-Perry, 1994 ; Lord & Paul, 1997 ; Mundy, Sigman, & Kasari, 1990 ; Sigman, 1994 ). Children with autism exhibit a triad of severe impairments, including language/communicative deficits, social cognitive (theory of mind) difficulties, and social behavioral differences ( American Psychiatric Association, 1994 ), as well as wide individual differences in functioning ( Dawson & Castelloe, 1992 ; Lord, Risi, & Pickles, 2004 ).

In prior reviews, investigators estimated that approximately half of children with autism were nonverbal (see Lord & Rutter, 1994 ). However, this estimate appears to be changing, perhaps due to the broader diagnostic criteria currently used for autism and the increasing prevalence of autistic disorders ( Newschaffer, Falb, & Gurney, 2005 ). For instance, in a recent analysis using combined data from several different studies, Lord and colleagues ( Lord et al., 2004 ) estimated that, by age 9, 14–20% of children with autism had no consistent words, 10–14% had “words but not three-word phrases,” 23–35% had “phrases but not fluent,” and 41– 43% had fluent language.

Delays and differences in particular aspects of language have been reported for children with autistic spectrum disorders. Generally speaking, pragmatic skills (e.g., discourse processes, communicating intended meaning to others)are more severely compromised than lexical or syntactic skills ( Lord & Paul, 1997 ), but delays in the latter aspects of language are also present. Tager-Flusberg (2004a) has recently hypothesized that there may be a subgroup of children with autism who exhibit deficits in morphosyntactic skills that are similar to those observed in children with specific language impairment ( Tager-Flusberg, 2004a ). In addition to language deficits, many children with autism exhibit comorbid difficulties in working memory and executive functions such as planning, sequencing, and set shifting, which could interfere with their language and sociocommunicative performance ( Happe & Frith, 1996 ; Joseph, 1999 ).

Although less well studied, an uneven profile of emergent language and nonverbal communicative skills has also been reported for young children with autism spectrum disorder. In a study of toddlers and preschoolers with autism spectrum disorder using the Mac-Arthur Communicative Development Inventory—Infant Form (CDI; Charman, Drew, Baird, & Baird, 2003 ), children with autism exhibited significant delays in language and gestural communication, relative to typically developing children. However, word comprehension was delayed relative to word production, and the production of early gestures involving shared reference was delayed relative to the production of late gestures involving actions with objects. Considerable variability in language acquisition was also observed, which is consistent with the typical pattern ( Fenson et al., 1994 ).

Others have reported that young children with autism consistently have impairments on early nonverbal symbolic skills that are thought to be precursors and correlates of later language development and theory of mind, that is, joint attention skills, imitation, and pretend play ( Baron-Cohen, 1987 ; Charman, 1997 ; Charman et al., 2000 ; Mundy, Sigman, & Kasari, 1994 ; Phillips, Baron-Cohen, & Rutter, 1992 ; Roeyers, VanOost, & Bothuyne, 1998 ; Yirmiya, Pilowsky, Solomonica-Levy, & Shulman, 1999 ). For instance, in a study of nonverbal communication in young children with autism and children with developmental delays or language impairments using a battery of structured communication tasks ( Stone, Ousley, Yoder, Hogan, & Hepburn, 1997 ), children with autism requested more often but commented less often than the comparison children. Consistent with other research ( Mundy, Sigman, & Kasari, 1993 ), children with autism were less likely to point, show objects, or use eye gaze to communicate, but were more likely to directly manipulate the experimenter’s hand. The autistic group also used less complex combinations of behaviors to communicate. Of note, other research has shown that these nonverbal communicative differences cannot be explained solely by the presence of motor deficits, attention problems, or low sociability ( Phillips et al., 1992 ).

Williams syndrome

Williams syndrome is a rare genetic disorder (incidence between 1/10,000 and 1/20,000 live births) that is caused by a microdeletion of about 20 genes on chromosome 7q11.23 and associated with a highly unusual neuropsychological profile ( Karmiloff-Smith & Thomas, 2003 ; Mervis, 2003 ; Morris, 2004 ). Despite moderate mental retardation (average IQ scores range from 50 to 70), individuals with this disorder have surprisingly complex language skills ( Bellugi et al., 1994 ; Singer, Bellugi, Bates, Jones, & Rossen, 1994 ) and tend to be quite affectively expressive and socially attuned ( Reilly, Klima, & Bellugi, 1991 ). At the same time, these persons have profound deficits in their fine motor and visual–spatial cognitive functioning, which exceed the level of their general cognitive impairment ( Bellugi et al., 1994 ). Some investigators in early studies of this disorder have claimed that the marked language/cognition dissociations observed in individuals with Williams syndrome provide strong support for a “modularity” hypothesis regarding the relationship between language and cognition in ontogeny. However, recent studies evaluating a greater variety of language and cognitive measures at diverse ages have indicated a more complex relationship ( Bates, 2003 ; Karmiloff-Smith & Thomas, 2003 ; Mervis, 2003 ).

Evidence for both delayed and deviant aspects of language has been reported for school-aged children with Williams syndrome. For instance, Robinson, Mervis, and Robinson (2003) reported that 10-year-olds with Williams syndrome relied on their working memory to a greater extent than grammar-matched typical children during language tasks, even when differences in receptive vocabulary skills were covaried. This finding suggests a deviant pattern of language development (see also Karmiloff-Smith & Thomas, 2003 ). However, other research has provided evidence for a delayed pattern of language production. For instance, Zukowski (2004) reported that the errors made by children with autism when producing sentences containing relative clauses and negative questions were similar to those made by typically developing children at younger ages ( Zukowski, 2004 ). Further research evaluating multiple aspects of language in this population is needed to resolve these apparent inconsistencies.

Although understudied, it appears that young children with Williams syndrome also exhibit delays and differences in early language acquisition, when compared to typical children or children in other clinical groups. However, specific results vary. It is generally agreed that children with Williams syndrome are delayed in the onset of their first words and first word combinations, relative to age-matched typical children. Once productive vocabulary emerges, however, toddlers with Williams syndrome have higher expressive vocabularies than age-matched toddlers with Down syndrome, as measured using the MacArthur CDI ( Mervis & Robinson, 2000 ), a pattern consistent with their relative strengths in lexical development later in ontogeny.

A somewhat different pattern of findings was reported for young children with Williams syndrome in another cross-population comparison study using the MacArthur CDI ( Singer et al., 1994 ). In that study, the early language and nonverbal communicative skills of children with William syndrome were compared to those of children with Down syndrome (average chronological age = 12–76 months). Consistent with other research, children in both groups were significantly delayed in the onset of first productive words, relative to typically developing children. However, when children were classified by level of expressive vocabulary (<50 vs. ≥50 words), differences in nonverbal communicative and grammatical skills emerged. Among children producing fewer than 50 words, children with Williams syndrome produced significantly fewer communicative and pretend gestures (e.g., intentional communicative gestures such as pointing and pretend/referential gestures such as holding a phone to the ear) than children with Down syndrome. In contrast, among children producing 50 words or more, children with Williams syndrome showed significant advances in grammatical development relative to the children with Down syndrome.

In a series of studies on the nonverbal communicative precursors to language, Laing et al. (2002) found that preschoolers with Williams syndrome were impaired in triadic communicative processes, which have been linked with the referential uses of language in typical development (see also Mervis et al., 2003 ). Despite higher levels of expressive vocabulary, children with Williams syndrome also exhibited dissociations between comprehension and the production of referential pointing, which could not be explained solely by motor impairment. These findings suggest that the course of early language acquisition for children with Williams syndrome may follow an atypical pathway, which challenges prior claims that children with Williams syndrome have preserved linguistic and social skills ( Laing et al., 2002 ).

Taken together, these findings from studies of early language development in different clinical groups raise several methodological caveats. Investigators in future cross-population studies of language acquisition in young children with language disorders should control for both level of language development and chronological age, and evaluate multiple aspects of language and nonverbal communicative behavior. Moreover, longitudinal studies of the complex, dynamic interplay between different aspects of language and nonverbal communicative skills would be especially helpful to clarify discrepant findings among studies.

Specific language impairment

By definition, children with specific language impairment perform below age expectations on standardized language measures but within normal limits on measures of nonverbal intelligence, and have no hearing loss or known genetic or neurological disorders. Thus, in contrast to the other populations described above, language impairment is the defining feature of this condition ( Leonard, 1998 ). The prevalence of preschool-aged children with specific language impairment ranges from 2 to 8% (median 5.95%), with a higher prevalence among boys, children with a family history of language, speech, or reading disorders, children born prematurely, and children from families with low socioeconomic status (SES; Feldman, 2005 ; Law, Boyle, Harris, Harkness, & Nye, 2000 ; Tomblin, Smith, & Zhang, 1997 ). About a quarter to one-half of language delayed preschoolers with normal nonverbal intelligence will continue to manifest language problems in later childhood ( Grizzle & Simms, 2005 ; Paul, 1996 ).

Despite much research, many questions remain concerning the causes and characteristics of specific language impairment. As is the case with other language disorders, considerable attention has been given to evaluating whether the language deficits observed in this group are delayed or deviant compared to typical children ( Rice & Warren, 2005b ; Wulfeck, Bates, Krupa-Kwiatkowski, & Saltzman, 2004 ). This research has shown that, although multiple aspects of their spoken language are delayed, relative to that observed in age peers, children with specific language impairment appear to have particular difficulty with phonological processing (measured by nonword repetition tasks; Conti-Ramsden, Botting, & Faragher, 2001 ) and specific morphosyntactic features of language (i.e., morphemes that denote grammatical finiteness of a sentence, such as third person singular “–s,” past tense “–ed,” irregular past tense forms, copula “be,” and auxiliary “be” and “do”; Rice et al., 2005 ; Rice, Wexler, Marquis, & Hershberger, 2000 ). That is, children with specific language impairment perform below both age-matched and language-matched peers on these measures, suggesting a deviant pattern of these aspects of linguistic growth ( Leonard, 1998 ; Tager-Flusberg, 2004a ; Wulfeck et al., 2004 ). In contrast, their receptive vocabulary skills and mean length of utterance in morphemes appear to be consistent with that observed in language-matched peers, suggesting a delayed rather than deviant pattern of lexical and general syntactic development ( Rice, 2004 ). However, other research has shown that a subset of children with specific language impairment exhibit a relatively high level of pragmatically inappropriate responses to conversational solicitations that cannot be explained by limited grammar or vocabulary ( Bishop, Chan, Adams, Hartley, & Weir, 2000 ), indicating that specific language impairment is a heterogeneous group.

How specific language impairment manifests during infancy and toddlerhood is unclear. When “late talkers” are followed from early to later childhood, half to three-quarters score in the normal range on language development tests ( Paul, 1996 ; Rescorla, 2002 ), and the others continue to exhibit delayed expressive language development in later childhood (and often develop reading and learning problems). It is unclear what specific factors may underlie such diverse outcomes. Given that specific language impairment is associated with male gender and tends to run in families ( Leonard, 1998 ), genetic etiological factors are likely; however, the specific genetic factors associated with this disorder have not yet been identified ( Rice, 2004 ).

Given the heterogeneity of outcome this group, researchers and clinicians are beginning to reevaluate current diagnostic criteria in this group. For instance, a growing number of studies has reported that children in this group exhibit below-age level performance on specific nonlinguistic skills such as processing capacity and the ability to encode temporal characteristics of auditory stimuli ( Bishop, 1994 ; Tallal, Merzenich, Miller, & Jenkins, 1998 ). This indicates that general omnibus measures of nonverbal intelligence may not be adequate for diagnostic purposes for these children, and suggests avenues for future research in this population ( Rice, 2004 ).

Cross-Population Studies

Given the momentum in genetic and biobehavioral investigations within particular language disorders during the past decade, Rice and colleagues have recently heralded the need for further collaborative cross-population research to document the ways in which language disorders are manifest across clinical conditions (see Rice et al., 2005 ; Rice & Warren, 2004 ). Such research could lead to a deeper understanding of the commonalities and differences across conditions, the specific symptom profiles associated with each condition, and how general intellectual impairment with a specific genetic basis may affect the process of language acquisition in multiple domains. Cross-population comparisons would also facilitate the identification of subtypes of disability within diverse clinical conditions and enhance our ability to identify early in development young children who are likely to develop clinically significant language-learning problems ( Reilly, Losh, Bellugi, & Wulfeck, 2004 ; Thal et al., 2004 ). In turn, the documentation of subtypes and the specific contextual factors associated with individual differences within conditions would augment our understanding of the different levels of risk that may predispose very young children in different groups for persistent language impairment. More generally, such research would contribute to our understanding of the nature and variability possible in human language development and would enhance our understanding of the brain bases of language and language development ( Rice & Warren, 2004 ; Thal et al., 2004 ).

Although cross-population studies of language disorders are relatively rare to date, those that exist have raised some important questions and offered a few surprises. For instance, in a series of comparative studies conducted by a multidisciplinary group of collaborators in San Diego, aspects of children’s language processing and acquisition were evaluated across several clinical populations at different ages. The clinical groups included “late-talking” toddlers, children with early focal brain injuries in either the right or left hemisphere (similar to those observed in adult aphasia patients), children with Williams syndrome, children with Down syndrome, and school-aged children with a diagnosis of specific language impairment, as well as appropriate typically developing controls ( Bates, 2003 ).

For instance, in a longitudinal study comparing the lexical and syntactic development of late-talking toddlers, children with early focal brain injury, and typically developing children from 2 to 3 years of age ( Thal et al., 2004 ), a number of interesting findings were reported. First, children in both at-risk groups exhibited significant delays in both vocabulary and grammatical development at both ages, and exhibited greater variability in language functioning than typical children (see also Kennedy, Sheridan, Radlinsky, & Beeghly, 1991 ). Second, different longitudinal associations between language comprehension and syntactic development were observed in each group, suggesting that different language skills may be organized differently in specific clinical groups. Third, qualitative analyses indicated that the at-risk groups produced the same kinds of errors (despite relative delays). However, the late talkers produced a substantially greater proportion of errors in obligatory contexts than did children with early focal brain injury, a pattern consistent with that observed in older children with specific language impairment.

In several other studies from the San Diego collaborative group, the linguistic and pragmatic skills of school-aged children in different clinical groups were compared ( Marchman, Saccuman, & Wulfeck, 2003 ; Reilly et al., 2004 ; Weckerly, Wulfeck, & Reilly, 2003 ; Wulfeck et al., 2004 ). Results showed that, by school age, children with early focal brain injury performed in the normal range on most language measures, regardless of lesion size or location, and outperformed children with specific language impairment. These findings shed light on the extent and nature of brain plasticity and recovery for language functioning in these groups. In addition, contrary to expectations, specific morphosyntactic measures of the children with specific language impairment with no frank neurological impairment clustered with those of the children with Williams syndrome, who were moderately retarded. Moreover, the performance of children with Williams syndrome on these language tasks was solidly linked to mental age (and often resembled that of children with specific language impairment), suggesting that children with Williams syndrome are not “language savants,” at least for the measures evaluated in these studies, as several investigators in early studies have claimed ( Bates, 2003 ). Reilly et al.’s (2004) results also highlight aspects of language and discourse that may be dissociable.

Contrasting results regarding the morphosyntactic development of children with Williams syndrome have been reported by other researchers ( Clahsen & Almazan, 1998 ), which may reflect, in part, methodological differences across studies. For instance, Rice and others reported that 7-year-olds with Williams syndrome performed at near ceiling level in an elicited production task of finiteness morphemes, in contrast to the lower performance of language-matched children with specific language impairment and typical children ( Rice, 2003 ).

In another series of cross-population studies, Tager-Flusberg and colleagues (see Tager-Flusberg, 2004a , for a review) demonstrated that a subgroup of children with autism with language impairment and language-matched children with specific language impairment exhibited a strikingly similar pattern of language skills. Children in both groups exhibited similarly poor performance and error patterns on tasks of nonword repetition and finite-verb morphology, suggesting that these children may share similar deficits in phonological processing and morphosyntactic knowledge. At the same time, children in both groups exhibited a relative sparing of articulation skills and verbal fluency. Interestingly, neuropathological findings in these groups are consistent with these language findings, suggesting the possibility of a common genetic etiological pathway to language disorder in the two groups ( Tager-Flusberg, 2004a ).

Need for Longitudinal Research

Further longitudinal studies of multiple aspects of language development in different clinical populations of children with language disorders are needed, especially during the period of time in ontogeny when language is acquired ( Bates, 2003 ). Although costly and time consuming, such studies will allow us to evaluate the manner in which specific language behaviors in different domains emerge and change over time within different groups, and their potential linkages at various ages with genetic and neurocortical functions ( Karmiloff-Smith & Thomas, 2003 ; Tallal & Benasich, 2002 ), higher order cognitive and socioemotional processes, and biological and social risk and resilience factors. Longitudinal designs will also allow us to evaluate within-group individual differences in profiles of language, cognitive, and behavioral skills over time, and describe the language phenotypes of subgroups of children more clearly and precisely ( Tager-Flusberg, 2004b ). Such research is critically needed to support and inform emerging studies evaluating the genetic and neurobiological etiology of different developmental language disorders.

To illustrate this need, Lord et al. (2004) noted that investigators in many prior studies of autism have tended to use general measures of verbal functioning (usually as control measures when evaluating the effect of autism on other cognitive or social skills such as theory of mind or central coherence). As a result, little is known about variations in different aspects of language (e.g., phonologic, semantic, morphosyntactic, pragmatic) among verbal children with autism, or (especially) how profiles among these different skills may change from early to later childhood. For instance, it is currently unclear whether children in the autistic spectrum disorders (or subgroups within this category) exhibit unique trajectories of language development from early to later childhood. Lord et al. contended that further understanding of the nature of these trajectories will have important implications for our understanding of the etiology and nature of developmental change within the autistic spectrum disorders, and children’s potential response to treatment within this population.

In a rare long-term longitudinal study, children with autism exhibited an uneven developmental trajectory of their language and cognitive skills from the preschool period to adolescence ( Sigman, 2005 ). Although about one-third of the children with autism made dramatic gains in cognitive and language skills from the preschool to the mid-school period, these gains had diminished dramatically (to less than half the change in their chronological age) by adolescence. Moreover, adolescents with autism and mental retardation exhibited almost no language gains. Notably, significant predictors of children’s language functioning in adolescence were children’s earlier functional play skills, responsiveness to others’ bids for joint attention, and the frequency of requesting behaviors during the pre-school period.

Further longitudinal studies in autism and other clinical populations are needed to shed light on how specific verbal and nonverbal cognitive and behavioral measures may be associated (or dissociated) at various points in time in ontogeny, and how functioning in different cognitive or socioemotional domains may support or derail language functioning in particular conditions or subgroups. An example of this type of longitudinal study comes from Mundy’s programmatic research (see Mundy et al., 1993 , for a review). In a 13-month longitudinal study of children with autism, Mundy et al. (1990) assessed the joint attention skills, pretend play behaviors, and language abilities of young children with autism and language-matched, mentally retarded children. All children were less than 5 years of age and were matched on receptive and expressive language age ( M = 12.5 and 13.6 months, respectively). At the initial observation, joint attention deficits (but not requesting deficits) were observed in the children with autism, before functional or symbolic play deficits were observed in this sample. Of note, joint attention deficits were also observed for the children with autism at the follow-up visit, when play deficits were present. Moreover, children’s initial joint attention (but not requesting) scores significantly predicted their symbolic play abilities at the time of the follow-up visit in the mentally retarded sample and approached significance in the autistic sample.

Mundy et al. (1990) speculated that nonverbal communicative and symbolic skills (i.e., joint attention and symbolic play skills) share a common source of variance and hypothesized that each may contribute to the development of children’s language and social–cognitive and theory of mind abilities. Of note, Baron-Cohen, Allen, and Gillberg (1992) provided corroborative evidence that the joint attention and symbolic play deficits of children with autism are lawfully related. In their autism screening studies, the majority of toddlers who showed impairments in both joint attention skills and symbolic play in the second year received a diagnosis of autism later in childhood. In contrast, none of the toddlers who had deficits in only one domain or who had no deficits were diagnosed with autism.

Taken together, these findings suggest that joint attention skills precede the emergence of language and nonverbal symbolic play skills in children with autism, and may share variance with factors associated with the subsequent emergence of verbal and nonverbal symbolic functioning. Moreover, these results indicate that the type of associations and dissociations among cognitive and socioaffective behaviors associated with language development in children with autism may change in ontogeny.

Longitudinal cross-population studies such as these are necessary if we are to capture the dynamic, changing nature of language learning and its correlates in various clinical groups over time. One-time “static” snapshots of language behavior so often described in prior language studies are inadequate to capture the dynamic, transactional process of early language development and may yield misleading results ( Karmiloff-Smith & Thomas, 2003 ).

In an intriguing treatise, Karmiloff-Smith (2001) argued that a key to understanding developmental disorders lies in the study of the dynamics of development itself, using a neuroconstructivist approach. Rather than focusing on single behaviors and “unidirectional chains” of single behaviors, she contended that investigators of language development in atypical populations need to evaluate the dynamic interplay among multiple contextual factors (e.g., genetics, brain/behavioral relations, and cognitive, behavioral, and environmental factors) on language and related processes from early infancy onward (see also Cicchetti et al., 1991 ; Thelen & Bates, 2003 ). Cicchetti et al. (1991) have further argued that it will be important to assess children’s language and communicative competencies in multiple observational contexts that focus on stage-salient developmental issues (i.e., tasks that require the child to coordinate language, cognition, affect, and behavior)and vary in the degree of structure and challenge placed on the child. The overarching promise in such multifactorial longitudinal research is that it will yield detailed, age-specific information on early language and associated processes that will be useful to clinicians, educators, and other professionals in guiding and timing comprehensive intervention and rehabilitation programs for children exhibiting delayed or deviant language development and their families ( Bates, 2003 ).

Genetic Bases of Language Disorders

Inquiry into the genetic bases of language disorders has gained momentum in the past decade, partly in response to the increasing number of behavioral phenotypic studies in different clinical groups, and partly from the hope that further knowledge of the genetic and environmental underpinnings of complex cognitive disorders can lead to more effective diagnoses and therapeutic interventions ( Smith & Morris, 2005 ). In behavioral genetics studies of early language development in typically developing children, results of multivariate genetic modeling revealed a consistently high genetic correlation between vocabulary and grammar at 2 and 3 years, suggesting the same genetic influences operate for both vocabulary and grammar ( Dionne, Dale, Boivin, & Plomin, 2003 ), a finding that is inconsistent with an “autonomy” hypothesis. In addition, results of crosslagged longitudinal genetic models showed that both lexical and syntactic bootstrapping operate from 2 to 3 years. In other behavioral genetics research, a complex interplay between genetic and environmental influences was associated with the timing and process of early language development ( Spinath, Price, Dale, & Plomin, 2004 ), suggesting the need for complex, dynamic systems approaches in future genetic studies of language disorders ( Thelen & Bates, 2003 ).

Two contrasting neurocognitive approaches (“top-down” and “bottom-up”) have recently been described that hold promise for future studies examining genetic contributions to language disorders. Generally speaking, in a top-down approach, the investigator focuses on an “optimal” phenotypic description of language behavior in a genetic disorder and then attempts to understand its genetic and biological underpinnings, often using electro-physiological brain measures ( Phillips, 2005 ). In contrast, in a bottom-up approach, the investigator focuses on primary etiological pathways for language disorders and attempts to explain complex “outcome phenotypes” in terms of more fundamental, underlying difficulties, such as those observed in early-emerging sensorimotor behavior ( Muller, 2005 ).

Both approaches are likely to be valuable, and each has the potential to identify gene effects at various levels of language processing and to describe specific genetic and neurological pathways contributing to language disorders ( Rice & Warren, 2005b ). For instance, advances in neuroimaging techniques (e.g., event-related potential, functional magnetic resonance imaging [fMRI], and near infrared spectroscopy) during the past 15 years have indicated that neurocortical measures such as these hold great promise for top-down genetic research in developmental language disorders. In studies of typically developing older children and adults, functional neuroimaging data have led to a greater understanding of the ways that dynamic neurocortical systems subserve higher level cognitive processes such as language. Moreover, fMRI studies in older children with specific language impairment or autism have yielded interesting corroborative information regarding behavioral findings in these disorders, with implications for our understanding of brain/behavior relations ( Carpenter et al., 2001 ; Nelson et al., 2002 ).

However, caution is needed in interpreting findings from this literature, given methodological variations in the use of these techniques across laboratories and related validity and reliability issues ( Billingsley-Marshall, Simos, & Papanicolaou, 2004 ). Further development and refinement of these techniques are needed. A major obstacle for the incorporation of these techniques in research on early language acquisition is that some neuroimaging techniques (e.g, fMRI) cannot be used successfully with infants and young children. This is due to technological limitations regarding movement artifacts, as well as ethical problems associated with sedation (see Nelson et al., 2002 ; Phillips, 2005 , for reviews).

Another methodological obstacle to genetic research in language disorders is that both the top-down and bottom-up methods require accurate, thorough descriptions of language phenotypes (and related cognitive and behavioral processes) in different populations of children with language delays and disorders. Clearly defined phenotypic measures are also necessary to inform and support cellular and molecular studies that will, in turn, lead to the identification of genes that contribute to the complex pathways affect language development and disorders ( McCardle et al., 2005 ).

Unfortunately, precise phenotypic descriptions of many aspects of language and related phenomena do not currently exist ( Rice & Warren, 2005b ). Part of the problem in coming up with a set of precise, “optimal” measures of language phenotypes in different clinical populations is that investigators from different disciplines have tended to focus on different language phenomena at different ages, with concomitant differences in preferred assessment paradigms and operational definitions of measures. To address and overcome some of these obstacles, Mervis and colleagues ( Mervis & Klein-Tasman, 2004 ; Mervis & Robinson, 2005 ) have provided several helpful methodological caveats for researchers in this field.

First, the most useful dependent measures of “optimal language phenotypes” (at least in initial studies) are those that assess narrowly defined language behaviors, can be used longitudinally with demonstrated reliability and validity, and can be utilized reliably by different investigators in different laboratories. This is especially important in multisite, cross-population comparative studies. In addition, given the growing cultural diversity in the United States, measures should be selected that recently have been normed (or renormed) for age and gender on geographically and racially representative national samples. Measures that are “culture free” and can be used appropriately in multicultural groups are especially desirable. Moreover, particular phenotypic measures should be selected for genetic evaluation that have known genetically related variation (heritability) in different populations, ideally those that have been shown to be consistent over time and within families, as determined in prior behavioral genetics studies ( Mervis & Robinson, 2005 ).

Second, Mervis and Klein-Tasman (2004) provide caveats pertaining to group-matching designs, which are commonly used to identify diagnosis-specific characteristics of children with language disorders. For instance, they describe the alpha levels needed for control variable comparisons, recommend the use of raw or standard scores rather than age equivalent scores when evaluating group differences on developmental matching variables, and advocate the use of sensitivity and specificity criteria to delineate membership in the target and control groups.

Third, Mervis et al. recommend that investigators eliminate from their study samples (or control for as much as possible) subjects with secondary behavioral or psychiatric conditions (e.g., attention deficit/hyperactivity disorder, depression)or environmental and biological risk factors (e.g., prematurity, prenatal substance exposure, teen parenting) that can co-occur with language disorders and confound interpretation in genotype/phenotype analyses. Assurance that target and comparison groups are matched carefully for gender, race/ethnicity, and sociodemographic characteristics (in addition to developmental features) is also of critical importance.

Fourth and finally, collaborative, multisite cross-population studies of language disorders are encouraged, because they can increase sample size in studies of rare disorders (such as Williams syndrome), overcome problems associated with subject attrition, and make the maximum use of the data. Furthermore, ideal collaborations would be multidisciplinary and incorporate input from scholars in genetics, molecular biology, linguistics, neuroscience, and psychology.

However, cross-site communication in multisite collaborative research is not a simple or easy task. To initiate and maintain effective, successful collaborative research, it will be important to coordinate efforts and methods across multidisciplinary groups prior to start up. As detailed by Mervis and Robinson (2005) , key constructs to be measured need to be identified, agreed-upon, and defined, using similar measures, instruments, and assessment paradigms across laboratories. In addition, investigators at different sites should utilize the same design (including specific child ages of assessment and control groups) and statistical analyses, to allow for data archiving and data sharing. Accomplishing these goals will require the implementation of procedures to promote excellent cross-site communication and reliable systems (e.g., a central site) for monitoring quality control at all phases of data collection and reduction.

Individual Differences and the Effects of Biological and Social Risk and Resilience Factors

Within-group variation in language and associated cognitive and behavioral skills is a hallmark feature of many clinical groups of children with delayed or disordered development ( Kennedy et al., 1991 ; Lord et al., 2004 ; Mervis, 2003 ; Tager-Flusberg, 2004a ; Thal et al., 2004 ; Tomblin, Zhang, Weiss, Catts, & Weismer, 2004 ), yet the primary focus in many prior comparative studies has been on describing average group differences in language performance, which can obscure within-group variations. Thus, there is an urgent need for investigators in future translational language research to evaluate individual differences within clinical groups of children with language delays and disorders. This shift in focus will allow for further analysis of data collected in within-group studies and the identification of possible subgroups of children with unique profiles of linguistic, cognitive, and behavioral skills, with implications for both genetic and intervention studies ( Lord et al., 2004 ; Tager-Flusberg, 2004b ).

It will also be important in future translational language studies to evaluate the effects of contextual risk and resilience factors that are known to affect variations in language acquisition. Factors such as child gender, temperament, parenting style, and biological and social risk and resilience factors on children’s language functioning have been shown to affect children’s language performance in both typically developing and various at-risk groups but have been largely ignored in prior research in children with developmental language disorders ( Blacher, Kraemer, & Schalow, 2003 ; Dykens, 2003 ; Mervis & Robinson, 2005 ; Torr, 2003 ). A brief review of this literature is provided here for illustrative purposes.

In typical development, language acquisition usually emerges rapidly and follows a predictable sequence during early childhood. By the end of the preschool period, most typically developing children have mastered the basic components of speech and language development, including a rich, varied lexicon, diversity in semantic and morphosyntactic applications, and pragmatic skills ( Berko Gleason, 2005 ). Moreover, a large body of normative research suggests that there is marked similarity across different ethnic and language groups in the onset and mastery of these skills ( Berko Gleason, 2005 ; Tomasello & Bates, 2001 ).

However, individual differences in the style and rate of early language acquisition are well documented ( Fenson et al., 1994 ; Shore, 1995 ). Robust effects of biological and social contextual factors such as child gender, temperament, birthweight premature birth, caregiver interactive style, culture, and demographics on multiple specific aspects of language development have also been reported, which may account for some of this variation ( Bates et al., 1994 ; Goldfield & Snow, 2005 ; Hart & Risley, 1995 ; Landry et al., 1997 ; Massey, 1996 ; Morisset, Barnard, & Booth, 1995 ; Spiker et al., 2002 ).

Caregiving factors are especially critical to consider, as individual differences in early language acquisition have been strongly linked to both distal and proximal measures of the caregiving environment, such as parental education, maternal psychosocial adaptation, and parental interactive style, including the amount and quality of verbal input to the child ( Messer, 1994 ). For instance, low socioeconomic status is often linked with delayed language acquisition. Results of a large, home-based observational study showed that parents from low SES homes spoke significantly less to their young children than parents from middle-class homes ( Hart & Risley, 1995 ), suggesting that the effect of low SES on children’s language development is mediated by SES-related variations in parental verbal input to children. In other research, variations in caregiving have been linked to children’s language outcomes within low-income samples. For example, in a longitudinal study of African American preschoolers from low-income families, a global measure of maternal responsivity and support in the home environment predicted children’s language and early literacy skills ( Roberts, Jurgens, & Burchinal, 2005 ). Similarly, in another longitudinal study of mostly Caucasian low-income toddlers’ vocabulary development between 1 and 3 years of age ( Pan, Rowe, Singer, & Snow, 2005 ), maternal lexical input and maternal language and literacy skills exerted positive effects, whereas maternal depression exerted negative effects, on toddlers’ lexical outcomes. Moreover, extreme perturbations in the caregiving environment, such as child maltreatment, are strongly associated with delays in children’s lexical, syntactic, and pragmatic development during early childhood, even after controlling for general cognitive status ( Beeghly & Cicchetti, 1994 ; Coster, Gersten, Beeghly, & Cicchetti, 1989 ; Eigsti & Cicchetti, 2004 ).

At a more proximal level, specific variations in maternal verbal input and sensitivity or responsivity to children during mother–child interaction have been linked with the rate of children’s language development, in both typically developing and at-risk groups ( Dale, Greenberg, & Crnic, 1987 ; Dunham & Dunham, 1992 ). For instance, in multivariate longitudinal studies of children’s vocabulary development, Bornstein and colleagues ( Bornstein, 1998 ) showed that mothers’ spontaneous expressive vocabulary to their infants uniquely predicted infants’ comprehension at 20 months. Effects of variations in the socio-affective quality of parent–infant relationship on language outcomes have also been reported. In a meta-analytic review ( van Ijzendoorn, Dijkstra, & Bus, 1995 ), the quality of attachment between infant and parent was strongly associated with the infant’s language development. Similarly, in a study of medically high risk toddlers, additive effects of a secure mother–infant attachment relationship and home stimulation on children’s language competence (especially receptive skills) were observed ( Murray & Yingling, 2000 ).

Several investigators have suggested that variations in maternal interactive style and sensitivity may exert stronger effects on early linguistic development when children are at risk for language delay due to the presence of social or biological risk factors ( Baumwell, Tamis-LeMonda, & Bornstein, 1997 ; Beeghly & Cicchetti, 1994 ). For instance, in Landry’s programmatic longitudinal research on prematurely born infants with varying biological risk characteristics, maternal interactive style characterized by a high prevalence of maintaining (rather than redirecting) of the child’s focus of attention during dyadic interactive tasks was associated with longitudinal gains in children’s language, cognitive, and social functioning in both term and prematurely born children, with larger gains seen in the higher risk preterm groups (see Landry, Miller-Loncar, & Smith, 2002 , for a review). Similarly, in a large longitudinal study of the effects of child care on children’s outcomes, maternal sensitivity during mother–child interactions from infancy through 36 months mediated the effect of chronic maternal depressive symptoms on children’s cognitive and linguistic outcomes ( NICHD Early Child Care Research Network, 1999 ).

Although understudied, similar caregiving effects have been observed in language studies of children with a variety of developmental disabilities. For instance, in experimental intervention studies of children with language delays and with Down syndrome, Yoder, Hooshyar, Klee, and Shaffer (1996) showed that mothers’ responsive linguistic behaviors that were fine tuned to their infants’ language skills predicted gains in their children’s syntactic development. In another study ( Yoder & Warren, 1999 ), maternal verbal responsivity mediated the relationship between children’s prelinguistic intentional communication and later language skills. In a study of the early predictors of language in children with and without Down syndrome ( Yoder & Warren, 2004 ), parental verbal responsivity predicted children’s later productive language above and beyond etiology. Similarly, in a prospective study of hearing mothers and their deaf and hard-of-hearing children ( Pressman, Pipp-Siegel, Yoshinaga-Ito, & Deas, 1999 ), maternal affective sensitivity to toddlers during mother–child interaction was significantly, positively associated with children’s expressive language gain, even after controlling for maternal education, degree of child hearing loss, dyadic mode of communication, and other covariates.

Of course, the direction of effects in these studies is unclear, given that development is dynamic and transactional in nature ( Abbeduto & Murphy, 2004 ; Bates et al., 1982 ). Parents’ psychosocial well-being and interactive style with children are also affected by variations in the child’s own developmental and behavioral status. What is still unclear (and deserves further evaluation) is how children’s developmental and socioemotional behavioral characteristics affect others’ responsivity to them and how this transactional process may indirectly contribute to the quality of child’s language-learning environment over time ( Abbeduto & Murphy, 2004 ).

Gene–environment interactions in language disorders

Several investigators ( Abbeduto & Murphy, 2004 ; Hodapp, 1997 , 2004 ) have suggested that, because factors from multiple levels of influence affect the trajectory of children’s language development, gene–environment interaction effects on language outcomes in clinical groups are especially likely and should be evaluated. An effective way to study this would be to evaluate the ways in which children with different genetic disorders (e.g., Down syndrome, Williams syndrome, or autism)use language with others in social contexts, and how variations in children’s communicative style affect others’ responses to them over time. It is possible that the resulting specific child–environmental interchanges in different groups or subgroups could gradually affect the developmental trajectory of those children’s profile of language skills and alter their phenotypic manifestation of genetically based behavioral predispositions. Putatively, these indirect genetic effects could also affect children’s functioning in other domains (e.g., cognitive processes, socioemotional behavior).

Although intriguing, evaluating gene–environment interactions in translational language research (especially at an individual level) is likely to be complex and difficult. For instance, it remains to be determined which specific aspects of the child’s language and communicative behavior affect others, whether these effects persist over time, and whether these effects are specific to a single disorder or are generalizable to all children with similar profiles in other groups ( Hodapp, 2004 ).

Cumulative risk

Longitudinal research with at-risk groups of children has shown that the presence of multiple risk factors may be stronger predictors of children’s language and other developmental outcomes than single risk factors. In several studies of preterm birth, for instance, infants with lower gestational age and birth weight, along with additional comorbid biological risk factors (e.g., chronic lung disease, white matter brain injury) performed more poorly on tests of language ability than their lower risk counterparts ( Briscoe, Gathercole, & Marlow, 1998 ; Landry, Miller-Loncar, & Smith, 2002 ). Similarly, in a longitudinal study of the effect of prenatal cocaine exposure on low-income children’s language outcomes at 6 and 9.5 years of age ( Beeghly et al., 2006 ), children with both prenatal cocaine exposure and lower birth-weight had more compromised expressive language functioning than exposed children with higher birthweight or unexposed children.

Taken together, this literature confirms that biological and social contextual variables from multiple levels of influence can alter the trajectories of children’s language development, whether children are developing typically, atypically, or at risk for developmental problems. Therefore, investigators in future translational research on language development should measure and evaluate these contextual factors carefully. As Zigler (1971) contended, children with developmental delays or disabilities are children first, and thus investigators need to take an integrated, “whole-person” approach to their study.

Clinical Implications

Let us now return to the clinician’s predicament raised at the beginning of this paper. How can results from longitudinal cross-population research on early language development in at-risk and atypical groups be “translated” into more effective evidence-based clinical practice?

Translational research on early language development has many implications for clinical practice. A few compelling examples are provided here.

First, translational research on atypical language development to date has shown that there are striking individual differences in language acquisition and related cognitive and behavioral processes within different clinical groups and subgroups of children, as well as striking commonalities across conditions. The provision of further detailed information about these cross-population similarities and differences in language profiles and documentation of subtypes within populations would enhance clinicians’ ability to identify young children who are likely to develop clinically significant language-learning problems and would enhance clinicians’ understanding of the different levels of risk that may predispose very young children for persistent language impairment ( Thal et al., 2004 ). Moreover, this information would help clinicians individualize and customize existing programs and intervention strategies for different children and maximize their effectiveness ( Abbeduto & Murphy, 2004 ).

Second, further longitudinal evaluation of the unique trajectories of language development within and across different clinical groups would shed light on how profiles of abilities within or across clinical groups change or remain the same over time ( Lord et al., 2004 ). This research would also identify the specific contextual factors (biochemical, neurocortical, or environmental) that may be triggering these age-related changes ( Rice, 2004 ). This information would allow clinicians to anticipate accelerations and decelerations in specific skills in different subgroups at different ages, and tailor and fine-tune interventions for individual children at different levels of language functioning in developmentally appropriate ways.

Third, translational research documenting the specific cognitive and behavioral precursors of language development (e.g., joint attention, pretend play) in different clinical groups may assist clinicians in identifying more accurately and at earlier ages children who are at risk for persistent language problems ( Yoder & Warren, 2004 ), which would allow these children to receive intervention services as early as possible. Research suggests that intervention services are most effective when received during infancy and toddlerhood ( Guralnick, 1997 ; Leonard, 1998 ).

Fourth, the identification of specific biological and social risk factors that may support or derail children’s language development in different clinical populations would further assist clinicians in the early identification of children at risk for language problems and offer insights into the most effective interventions. Tomblin, Hardy, and Hein (1991) have suggested that information regarding children’s biological and family risk status could help clinicians identify those at greater risk for communicative impairments more clearly. For instance, in their retrospective study of the speech and language status of 662 children between 30 months and 5 years of age, a set of multiple risk factors comprising various family background and birth history variables predicted 55% of children with poor communication skills and 76% of those with normal communicative development ( Tomblin et al., 1991 ).

Regarding the development of effective treatment programs, results from intervention research documenting the efficacy of different types of parental verbal responsivity on children’s language outcomes could be incorporated into ongoing parenting programs to maximize their effectiveness ( Yoder & Warren, 1999 , 2004 ). More generally, reduction of the overall level of risk for individual children and their families and provision of social support at multiple levels may do much to facilitate positive child developmental outcomes ( Dunst, 2000 ; Tronick & Beeghly, 1999 ). For instance, results from intervention research spanning nearly three decades ( Olds, 2006 ) have shown that general and specific prenatal and infancy preventive interventions for low-income families provided by home visiting nurses can significantly improve parental care and are associated with better infant and emotional outcomes, particularly for families at greater risk.

Fifth and finally, longitudinal translational research on the efficacy of different intervention programs could provide valuable information to clinicians regarding the relative efficacy of different types of intervention programs on parenting and child language outcomes in different clinical groups, whether these effects vary for children at different ages, and whether different approaches are needed for families from different socioeconomic and cultural backgrounds. Such research could also shed light on how specific intervention parameters (intensity, duration, comprehensiveness of service) may affect parental adaptation, parenting, and children’s language outcomes in different clinical groups (see, e.g., Hauser-Cram, Warfield, Shonkoff, & Krauss, 2001 ; Warfield, Hauser-Cram, Krauss, & Upshur, 2000 ).

EU and Nigeria boost cooperation on research and learning mobility

Today at the Global Gateway High-Level Event on Education in Brussels, European Commissioner Jutta Urpilainen showcased the European Union’s strong commitment to education and health equity, signing with Didi Esther Walson-Jack, Permanent Secretary of Nigeria’s Federal Ministry of Education, a cooperation agreement on €18 million EU support to enhance research and development capacities for implementing Nigeria’s national plan for the pharmaceutical industry and local production of vaccines and medical technologies.

Commissioner Urpilainen said: “ Economic growth is dependent on an educated, skilled workforce and healthy societies, and investing in strengthening education and health systems worldwide is an integral part of the European Union’s Global Gateway strategy. Our investments in quality education, research and training seek to empower future generations by equipping them with the knowledge, skills and competencies they need in a changing world to tackle global challenges and build prosperity. ”

The European funding signed today under the Team Europe Initiative on Manufacturing and Access to Vaccines, Medicines and Health Technologies in Africa (MAV+) will support the wider enabling environment around Nigeria’s pharmaceutical sector, notably by promoting:

• skills development through education and training
• research and development (e.g. research in artificial intelligence and nanotechnology)
• the digitalisation of essential dimensions of the ecosystem
• a centralised system for forecasting, procurement and distribution of quality medical products
• trade, investment and customs facilitation, intellectual property rights frameworks and conditions, and an enabling environment for preferential trade and investment.

Commissioner Urpilainen also signed 15 Intra-Africa Mobility Scheme projects funded by the EU with €27 million under the flagship Youth Mobility for Africa . The projects will provide learning mobility opportunities for students, trainees and staff across the continent to boost high-level green and digital skills. Nigeria will benefit from six projects:

• CB4EE - Capacity Building for Engineering Education Practice and Research (€1.8 million of EU funding in total, with the participation of the University of Lagos-Unilag)
• CREATE-Green Africa - Climate Research and Education to Advancing Green Development in Africa (€1.8 million of EU funding in total, with the participation of the University of Port-Harcourt)
• GENES II - Mobility for Plant Genomics Scholars to Accelerate Climate-Smart Adaptation Options and Food Security in Africa II (€1.8 million of EU funding in total, coordinated by the Ebonyi State University)
• GREEN STEM - Green, Resilient and Entrepreneurial Science, Technology, Engineering and Mathematics for Africa (€1.8 million of EU funding in total, with the participation of the University of Lagos-Unilag)
• HCE Solutions – Promoting Inclusive Homegrown Clean Energy Solutions for Climate Change Adaptation and Mitigation in Africa (€1.8 million of EU funding in total, coordinated by the Federal University of Technology and with the participation of the University of Nigeria)
• ORPHAN - Mobility for High Skilled Scientists and Entrepreneurs on Orphan Crops in Higher Education for Accelerated Climate Change Solutions in Africa (€1.8 million of EU funding in total, with the participation of the Ebonyi State University)

Commissioner Urpilainen also launched a key initiative of the Youth Action Plan in EU external relations, the Africa-Europe Youth Academy , which will provide opportunities for formal and informal learning and exchanges to young people looking to improve their leadership skills and create networks between Africa and Europe. Finally, Nigeria can also benefit from the regional Team Europe Initiative on Opportunity-driven Skills and Vocational Education and Training in Africa, launched today, which will orient country-level vocational training initiatives towards concrete employment opportunities created by Global Gateway investments.

The Team Europe Initiative on Manufacturing and Access to Vaccines, Medicines and Health Technologies in Africa (MAV+) works with African partners to strengthen their pharmaceutical systems and manufacturing capacity to improve access to quality, safe, effective and affordable health products. It offers a 360-degree approach through the supply side, the demand side, and the enabling environment, and six work streams:

• industrial development, supply chains and private sector
• market shaping, demand and trade facilitation
• regulatory strengthening
• technology transfer and intellectual property management
• access to finance
• R&D, higher education and skills

Education is a powerful mechanism to address inequality and poverty, boosting human potential, opening doors for girls, youth and marginalised groups, and providing a springboard for human connections, debate and democratic values. Education creates an enabling environment for investments in digital and green transformations to succeed, and forms an integral part of the EU’s Global Gateway offer to partner countries.

The European Union is the leading investor in education worldwide. The EU institutions and Member States provide more than 50% of all official development aid to education worldwide. The EU committed to dedicating at least 10% of its international partnerships budget for the period 2021–2027 to education. In the period 2021–2023, the EU’s commitments have amounted to around €3 billion, approximately 13% of the budget.

For More Information

Global Gateway High-Level Event on Education - European Commission

Education - European Commission

Team Europe Initiative on Manufacturing and Access to Vaccines, Medicines and Health Technologies in Africa

Photos gallery of the Global Gateway High-Level Event on Education

Share this page

• Publications
• News and Events
• Education and Outreach

Software Engineering Institute

2023 year in review highlights impact of sei research and development.

April 15, 2024 • Article

April 15, 2024— The Software Engineering Institute (SEI) today released the 2023 SEI Year in Review , a report highlighting some of the SEI’s most noteworthy accomplishments of the 2023 fiscal year. The Year in Review spans the institute’s technical portfolio of research and development in software engineering, cybersecurity, and artificial intelligence (AI) at the intersection of government, industry, and academia.

The articles in the 2023 edition show how the SEI combines its body of knowledge with external material and systems engineering to deliver impact to Department of Defense (DoD) and other U.S. government organizations and end users. Leveraging collaboration, the SEI advanced the state of the art and practice of software and related disciplines.

“AI may get headlines, but the SEI’s other core competencies in cybersecurity and software continue to support our national security mission,” wrote SEI Director and CEO Paul Nielsen in the Year in Review ’s introduction. “AI surely won’t be the last uncertain technology we face in our unique mission to advance the art of software engineering.”

Read the 2023 SEI Year in Review online or download a PDF to learn more about our work in

• generative AI
• vulnerability prioritization and management
• UEFI security
• open source software
• U.S. leadership in software engineering and AI engineering
• support for the Long Range Standoff Program
• AI trustworthiness for warfighters
• large language models for intelligence
• responsible AI
• safety-critical system software architecture

International Conference on Conceptual Modeling ER 2024 Opens Call for Papers