DEVELOPMENTAL DISABILITIES RESEARCH REVIEWS 17:160-169(2011)

EARLY DELAY AND SPECIFIC LANGUAGE IMPAIRMENT

Jayne Moyle, Stephanie F. Stokes,* and Thomas Klee Department of Communicadon Disorders, University of Canterbury, New Zealand

Early language delay (ELD) is a warning sign that may presage the one of the challenges in early identification of this disorder. presence of a later language impairment (LI). In order to allow more tar- geted identification and earlier intervention for LI, better diagnostic The focus of this article is the developmental pedod in which measures for are needed. Development of accurate predictive/ a core is identified. Language disorders in diagnostic models requires consideration of a set of complex interrelated children are usually identified at four-to-five years of age. cjuestions around definition, causality, and theories of Lis. A multifacto- However, there are indications that SLI might be detectable rial model of and LI is essential to increase the earlier than the age at which it is usuaUy diagnosed. accuracy of prediction. This article examines what is known about LI in the preschool years and language delay in toddlers, and examines these in relation to the Procedural Deficit Hypothesis (Ullman and Pierpont, LATE TALKING AND LATER LANGUAGE [2005] Cortex 41:399-433] and the Statistical Learning Account (Stoices IMPAIRMENT et al., [2012a] J Lang Hear Res; Stolces et al., [2012b] J Late onset of talking has long been thought to be an Lang 39:105-129) to suggest a new framewori< for characterizing ELD to better assist prediction of later LI. ©2013 Wiley Periodicals, Inc. early manifestadon of later LI [Olswang et al., 1998; Thal, Dev Disabil Res Rev 2012; 17:160-169. 2000] and has been referred to vadously as specific expressive language delay (SELD), early language delay (ELD), and late language emergence (LLE). It is idendfied between about 18 Key words: late talkers; preschool specific language impairment and 30 months of age, with the first indicators being early vo- cabulary delay, slow vocabulary growth, and/or delayed onset of word combinadons. For research purposes, this group of hildren with language disorders expedence difficulties children has been identified on the basis of parent-report with which may impact adversely on measures such as the Language Development Survey [LDS; Csocial-emodonal development, academic achievement, Rescorla, 1989], the MacArthur Communicadve Develop- and vocational opdons if the disorder persists beyond the pre- ment Inventory: Words and Sentences [CDI:WS; Fenson school pedod [Clegg et al., 2005]. Dudng this pedod, a et al., 2007], and the Ages and Stages Questionnaire Commu- language impairment (LI) in the absence of significant sensory, nicadon scale [ASQ; Bdcker and Squires, 1999]. Various psychiatdc, neurological, or intellectual impairment/disorder is operational definitions have been used to define the target commonly referred to as specific language impairment (SLI). population, including (1) fewer than 50 words of expressive While the profile of specific linguisdc abilities (e.g., in gram- vocabulary or no word combinadons by 24 (± 6) months of mar, semantics, phonology, and pragmadcs) vades across age on the LDS [e.g., Rescorla, 1989; Klee et al, 1998]; (2) children with SLI and across dme, English-speaking children <10th percentile on the expressive vocabulary section of the with SLI generally exhibit problems with grammatical mor- CDLWS [e.g., Reilly et al, 2007]; and (3) 1 SD or more phology relative to children of the same age without SLI below the mean on a composite score from the ASQ Com- [Leonard, 1998]. In an effort to understand this vadabihty across municadon scale [Zubdck et al., 2007]. These deñnidons linguistic abilities, researchers have identified subgroups of LI. typically result in between 10 and 20% of children being One attempt to delineate subgroups of children with identified as late talkers (LTs). SLI using cluster analysis found three underlying linguistic fac- tors based on phonology/articulation, semandcs/syntax, and While these cdteda are commonly used, they appear to be pragmatics [Tomblin et al, 2004]. These factors dissociate to at best only moderately indicadve of ongoing language disorder. some degree to create different subgroups. Further, semantic/ This is likely due to the wide vadation in the development of syntactic disorders may be expressive only or both receptive early language skills such as vocabulary and grammar. Late talk- and expressive. While there is as yet no consensus regarding ing has been associated with poorer developmental outcomes how many subgroups exist, it is clear that a child's subgroup membership and developmental profile can change over dme [Conti-Ramsden and Botdng, 1999]. There is evidence to 'Correspondence to: Stephanie F. Stokes, Department of Communication Disor- suggest that these vadations in how the disorder manifests ders, University of Canterbury, Private Dag 4S0(), Christchurch, 8140. New Zea- itself may be charactedsdc of the core language disorder as it land. E-maiL [email protected] Received 11 September 2012; accepted 5 October 2012 unfolds over time, rather than separate diagnostic entides View this article online in Wiley Online Library (wileyonlinelibrary.coni). [Bishop, 1994]. The heterogeneity of the SLI populadon is DOI: 10.1002/ddrr.lllO

© 2013 Wiley Periodicals, inc. such as SLI, dyslexia, spectrum velopmental delays [indicated by PREDICTING LATER SLI disorders, and attention deficit hyperac- nonverbal IQ scores below 85] have We turn our attention now to fac- tivity disorder [Buschmann et al., 2008]. not been included. However, this prac- tors that may contribute to an ELD (and In general, children who start out as high tice is questionable for several reasons. thus, later talker status), as weU as a pre- achievers in language will continue to Firstly, the assumption behind this school LI. A range of hnguistic, achieve weU throughout childhood appears to be that chUdren with low demographic, medical, genetic, and envi- whereas those who start off slowly often, nonverbal scores do not have the cogni- ronmental factors have been explored as but not always, have poorer outcomes. tive underpinnings to support normal predictors of later language abiHty However, the majority of children who language development; that is, that an [Zubrick et al., 2007; EUis and Thai, start off slowly wiU accelerate in their de- undetemiined general factor which 2008; ReiUy et al., 2009; Stokes and Klee, velopment to come within normal range caused low nonverbal achievement wUl 2009a]. While late emergence of expres- over the next 18 months [Paul, 1996; have also caused the problem with lan- sive vocabulary alone is not a strong Rescorla and Roberts, 1997]. In compar- guage development. However, groups of predictor of language abiHty at 4 or 5 ison with the 50—70% recovery rate from chUdren with low nonverbal abiHty but years, other variables have some, but lim- late talking, a diagnosis of SLI at age 4 has language in the nomial range have been ited, predictive value. Language a smaller degree of recovery in the short- identified [Rice et al, 2004]. The lan- comprehension and gestural use have term. Bishop and Edmundson [1987] guage abiHty of these children did not been shown to have some predictive value found that of children with SLI at age 4 differ from age-matched controls. Sec- in smaU sample studies [e.g.. Thai et al., identified by a narrative measure, only ondly, it is debatable whether or not the 1991; Thal and Tobias, 1992]. Of the 10 10% had resolved by age 5—6. In a longi- language of chUdren with SLI is qualita- LTs included in these studies, six were tudinal study with the same sample, tively different from those with more "late bloomers" (i.e., chUdren who had Stothard et al., [1998] found that chUdren general delays. Rice et al. [2004] caught up with their age-matched peers) who had resolved by age 5-6 differed reported quaHtative differences while by 30-35 months, while four were stUl from controls at ages 15-16 years on Toniblin et al. [2004] reported only delayed in language production. Exami- measures of phonological processing and quantitative differences. Children on the nation of their gesture use and language Hteracy. However, those who had SLI or borderline between these two diagnoses comprehension at 18 months showed that a general delay at age 5—6 had significant shift between them over time, pointing LTs who continued to have poorer lan- impairments in aU aspects of spoken and to a degree of overlap in these cHnical guage skiUs had lower comprehension written language functioning at out- groups [Vig et al, 1987], leading some scores, and failed to use gestures as com- come. This illustrates the changing researchers to prefer the term primary pensation for reduced expressive nature of language delay and disorder language impaimient (PLI) or language vocabulary abiHty, than the late bloonien. across childhood. As a group, children impainnent (LI) rather than SLL In this Ellis and Thai [2008], in a brief review of with an early vocabulary delay faU within revie'w, we adopt the temi SLI as there the relationship between ELD and later the range of typical development by 4—5 is StiU significant debate about comorbid- LI, suggested that there were three key years but then delays in grammatical de- ity and specificity of language-related Hkely predictors, a family history of lan- velopment become apparent. disorders. Thirdly, studies of other de- guage delay or impaimient, delays in Conversely, there is some evidence that a velopmental disorders are Hkely to be receptive as weU as expressive vocabulary, portion of children start out in the nor- informative with respect to the defini- and the lack of communicative gestures mal range for language but slow in tion of SLI. Boundaries between (the latter is supported by a longitudinal development, and faU in the disordered disorder types may not be distinct. For study [Rowe et al., 2012]). To these they range at a later point [Dale et al., 2003; example, an overlap between dyslexia add a history of otitis media, parent con- Rice et al., 2008]. Predicting which LTs and SLI has been postulated, with the cern, high risk environments (e.g., will go on to manifest a significant lan- shared variance thought to be due to an ), and delays in symboHc play or guage difficulty at age 4-5 poses a underlying phonological processing dis- social skUls. Mother's education [DoUa- diagnostic chaUenge (see Rescorla, this order [Catts et al., 2004]. In addition, ghan et al., 1999], a family history of issue), with part of the problem being the categories of autism spectrum disor- speech/language delay [Bishop et al., what exactly constitutes SLI. der and SLI may overlap to some extent 2003], the child's sex, and early neurobio- because of an underlying auditory per- logical growth [Zubrick et al., 2007], While the term specific is part of ceptual processing difficulty [Oram birth order/number of sibHngs [Stokes, the definition, non-linguistic skills are Cardy et al., 2008]. There is a moderate 1997], a history of ear infections or parent also known to be impaired, with chU- rate of comorbidity (or arguably overlap) concem [Klee et al., 2000], socio-emo- dren having depressed nonverbal IQ with SLI and childhood apraxia of tional development [Bomstein et al., scores [Johnston, 1994], and deficits in speech, attention deficit hyperactivity 1998], and cognitive development [Price hypothesis testing [Karrihi et al, 1984], disorder (ADHD), and autism. Diagnos- et al., 2003], have been shown to have cross-modal processing [Montgomery, ticians attempt to identify clusters of some predictive value, in some, but not 1993], voice processing abilities [Creu- chUdren who fit these categories; how- aU, studies. Evidence for these factors as sere et al., 2004], sustained attention ever there are always individuals who do sole predictors is not consistent, and, as [Finneran et al., 2009], processing not fit clear diagnostic categories, and we show below, a mukifactorial model of capacity and speed [Leonard et al., who sit at the boundary between types. prediction is likely to be the way forward 2007], working memory [Montgomery Children initiaUy identified as LTs may for improving our abiHty to predict the et al, 2010], and motor coordination later be diagnosed with SLI, dyslexia, consequences of ELD. [HiU, 2001]. This begs the question of ADHD, autism spectrum disorder whether or not children with more (ASD), or general inteUectual disabiHty, general developmental delays are moving in and out of diagnostic catego- A MULTIFACTORIAL MODEL included in studies predicting SLL ries as they mature. As we work toward a greater Commonly, children with general de- understanding of factors that contribute

DEV DISABIL RES REV- EARLY LANGUAGE DELAY AND SLI • MOYLE ETAL. 161 Genotvpe uiatwn: ial intemction, íanguage

Neurobkílogv

median tans: ^ psydmimgaistic g, etc.

Fs-^áKhsoúal developmeiK Motor development

..{ ComprehensiiKn (receptive language) [_ j Production {expressive language) i

Languatge contefit: Language fonn: ' ' Language use: | Language nKïdality: I Seniaitics firocHau/orií, «tj I | Graamtar (symax, morphologf}, | ¡ Spoken language, sign la . ^ Wicmotagrií fpAoOBmes, prosooW I "~ i writt«! l:»ipiage Figure 1 Processes, components, and modalities of language development (from Klee and Stokes, [2011]).

to late talking and SLI, we must con- to have a family history of language and Vernes et al. [2008] reported that var- sider multiple interrelated factors leaming disorders than children with iants in CNTNAP2, a regulated [Desmarais et al., 2008]. KJee and non-affected language development by FOXP2, were related to poor per- Stokes [2011] presented a multifactorial [Zubrick et al., 2007; ReiUy et al., formance on language measures in model of language development that 2009]. They are also more likely to school-aged children with SLI. Further, draws on Gottlieb's model of epigénesis score lower as a group on linguistic and relevant here, Whitehouse et al. [GottUeb, 2007], and Bishop and related measures [Flax et al., 2009] and [2011], in a study of 1,149 2-year-old Snowling's [2004] model of relation- to have SLI [Choudhury and Benasich, children, suggested that variations in ships between observed behavior, 2003]. In addition, boys are more likely CNTNAP2 may influence early lan- cognitive processes, neurobiology, and to be LTs than girls [Zubrick et al., guage development. Whitehouse et al. etiology in development (see Fig. 1). 2007; ReiUy et al., 2009]. Boys show [2011] proposed that this genetic var- The model considers the roles of slower expressive vocabulary growth iance could increase the risk of genetic, environmental, neurobiological, patterns than girls before 24 months, but development of SLI, but added the ca- and cognitive factors in language devel- there is no gender difference in overall veat that this was unlikely to occur in opment. Here, we consider how these vocabulary growth [Rowe et al., 2012]. isolation of other genetic or environ- factors might serve as predictors of later The search for specific mental influences [see Bishop (2009), SLI. For example, while the genotype causing neurodevelopmental disorders for a review]. At this stage, for clinical is the starting point for development, has proven more complex than initiaUy purposes, aU we can say is that a family identical genotypes can result in differ- expected. The concept of a "language history of language or learning disorders ent phenotypic outcomes in response to gene" is an appealing but simplistic should be considered a risk factor when differences in the environment (for one. Early studies of the genotype of predicting forward from an ELD, but a example, the variations seen in mono- the KE family, who have a high rate of family history alone is not sufficient as a zygotic development). childhood apraxia of speech in their predictive factor. family, yielded evidence that the FACTORS POTENTIALLY FOXP2 gene was deficient for affected Environmental Factors PREDICTIVE OF SLI family members [Vargha-Khadem et al., While genetic factors are critical 1998]. Recent research suggests that in the phenotypic outcome of disorders, Genetic Factors mutations of FOXP2, a transcription environmental stimulation also plays a Genetic factors play an important factor gene that encodes a regulatory powerful role. Children learn language role in the development of SLI. Verbal protein, are rare and that it is not such from social interaction with other peo- ability seems to be one of the most her- mutations that are relevant for SLI, ple. Research into the effects of the itable of cognitive traits, as demonstrated rather, it the endpoint of FOXP2 regu- environment on by adoption studies [Plomin et al., lation, the target neural circuitry that is has focused on two factors, conversa- 1997]. Twin studies have shown that relevant [Fisher and Scharff, 2009]. tional styles of parent-child interactions monozygotic have higher rates of FOXP2 regulates regions of the brain and variations in linguistic input. The concordance of language disorders than thought to play major roles in the role each plays is likely to differ as a dizygotic twins, indicating the impact of cognitive mechanisms employed in lan- function of age and also at the extreme genes beyond that of the environment guage learning, including the cerebral ends of the ability and environment [Dale et al., 2003]. LTs are more likely cortex, basal ganglia, and thalamus. spectra.

162 DEV DISABIL RES REV- EARLY LANGUAGE DELAY AND SLI • MOYLE ET AL. Some aspects of parent-child guistic environment, SES, and language out, there are few studies of how these interaction have been found to facilitate ability may be mediated by genetic physiological differences relate to lan- language development [Yoder and influences. guage function, and prior to Badcock Warren, 2004]. Several studies have There seems to be a differential et aL, they were Hmited to investigations examined whether the interactions effect of genes and environment on lan- of the KE faiTuly [e.g., Vargha-Khadem between LTs and their parents are dif- guage development toward the extreme et al., 2005]. Nonetheless, these studies ferent in a way that may negatively low ends of the spectrum in environ- do point to areas of the brain where impact on child language development. ment and ability. It is undisputed that neurolinguistic processing may be differ- However, the interaction styles of environmental deprivation at the ent for children with SLI, and parents of typically developing [TD] extreme end of the spectrum (poverty potentially, children w^ho are LTs. children and parents of LTs appear to or severe neglect) is linked with be more similar than different. Paul and depressed cognitive and language devel- COGNITIVE MECHANISMS Elwood [1991] reported that parents of opment [Turkheimer et al, 2003]. In a LTs have a larger gap between their large . Dale et al. [1998] Perception mean length of and their reported that the genetic contribution Measures of perception have been child's language, •which may make lan- accounted for 73% of the variance in found to correlate with language devel- guage more difficult to learn. Vigil vocabulary development for children in opment. Speech perception plays an et al. [2005] found that parents of TD the lowest 5% of language ability at 2 important role in language acquisition. children used more responses, expan- years. For the entire sample, 25% of the Speech perception skills, including pho- sions, and self-directed speech than variance was accounted for by genetic neme discrimination and learning to parents of LTs; however these behaviors variables rather than environmental inhibit processing of non-critical sound could be reflective of higher language factors. contrasts, in infancy have been found to abilities in TD children, which elicit correlate with later language develop- different conversational styles, rather Unfolding Neurobiology ment [Conboy et al., 2008]. The than style differences causing LLE. Having seen that genetic factors assumption is that the sooner infants are Although the amount of linguistic input play a large role in late talking and SLI able to discriminate between their experienced by children from different outcomes, we turn now to considering native language phonemes, the sooner social groups has been shown to have a the neurobiology of language. Signifi- they can map phonological patterns and dramatic impact on the size of child- cant differences in the neurobiology of begin to learn words. Behavioral and ren's developing lexicons [Hart and children with SLI have been demon- electrophysiological measurements, such Risley, 1995], the more subtle differen- strated. For recent reviews, see Webster as event related potentials (ERP), have ces in the features of parent—child and Shevell [2011] and Badcock et al. been used to show that infants' native interactions of late talking and TD chil- [2012]. Brain structures in children language speech discrimination abilities dren are not likely to play a causal role with SLI have been found to differ in measured at seven months predict lan- in developmental language disorders. volume and symmetry firom those of guage outcomes at 14—30 months. Nonetheless, it appears that the TD children. Cohen et al. [1989] also Those with better native speech dis- quality and quantity of linguistic input found differences in gyrification. Vol- crimination had significantly higher in the early years impacts on children's ume differences, relative to TD peers, language abilities whereas those with linguistic achievements. Hart and Risley include a smaller Broca's area [Gauger better non-native phonetic discrimina- [1995] discussed the importance of the et al., 1997], reduced surface area in the tion had poorer language abilities (see cumulative effect of linguistic input primary auditory cortex of the left Kuhl, [2010], for a review). Children over time, with children with more hemisphere [Leonard et al, 2002], and who progress more quickly toward enriched input having larger vocabula- increased gray matter in the right peri- neural commitment to native sounds ries at every time point. The main sylvian region and the occipital petalia contrasts progress faster than those who factor associated with the level of lin- [Soriano-Mas et al., 2009]. Typically retain the ability to discriminate guistic input in this study was socio- the dominant, left hemisphere language between non-native contrasts for longer economic status (SES). Subsequent areas are larger in volume compared [Kuhl, 2010]. Children who advance in authors [e.g.. Hoff, 2003] proposed that with the corresponding regions of the phonetic abilities faster should begin to variation in linguistic input was medi- right hemisphere. However, for chil- detect phonotactic patterns and words ated by SES; however two recent large- dren -with SLI the hemispheres may not more readily, leading to faster develop- scale population studies found little rela- differ in volume, or may show a right- ment of vocabulary and subsequent tionship between SES and LLE ward dominance. Reduced asymmetry grammar. In a series of ERP studies, [Zubrick et al, 2007; Reilly et al, of the perisylvian structures has been Molfese et al. found that infants who 2009]. Conversely, SES has been associ- found [Plante et al., 1991], particularly could discriminate consonants better ated with SLI in 4-6 year old children in the temporal [Ors et al, 2005] and between phonemic boundaries had bet- [Stanton-Chapman et al., 2002; Reilly planum temporale regions [Gauger ter language development at ages 3, 5, et al, 2010]. However, Plomin et al. et al, 1997; Leonard et al, 2002]. At and 9 years old [Molfese et al., 1999]. [1997] found that children adopted at odds with the latter finding is one In addition, Tsao et al. [2004] found birth became increasingly siinilar to report of normal left—right asymmetry that infants who could discriminate their biological parents in cognitive and in children with SLI [Preis et al., 1998]. between two non-native vowels at six verbal abilities fi-om infancy through months of age had better language These volumetric (and possibly adolescence than to their adoptive development at 13, 16, and 24 months. asymmetric) differences become impor- parents, confirming the importance of Newman et al. [2006] found that in tant in their relationship vfith functional genetic influences on language abilities. TD infants, the ability to detect a word language organizadon and performance. It appears that associations between lin- in a stream of connected speech However, as Badcock et al. [2012] point

DEV DISABIL RES REV • EARLY LANGUAGE DELAY AND SLI • MOYLE ET AL 163 correlated with language outcomes at 2 this review (see Baddeley [2012] and variations that were irrelevant to their and at 4-6 years of age. This abihty was Courage and Cowan [2009], for over- native language. Infants' native speech not related to general intelligence meas- views). Here we briefly review why discrimination was also related to their ures, but correlated with vocabulary working memory skiUs are relevant for concurrent receptive vocabulary skiUs. and grammar. language learning, adopting Baddeley's The authors suggested the role of inhi- model. In this account, working mem- bition in development of neural Speed of Processing ory, a cognitive system that "mediates commitment to native speech contrasts. Given the rapid transient nature between perception, long-term mem- Studies have measured neural ac- of the acoustic speech signal, it is logical ory, and action" [Baddeley, 2012, p. tivity in the frontal regions associated that efficient processing of brief audi- 25] is considered to be a four-compo- with cognitive processes such as atten- tory temporal stimuh may be an nent cognitive system comprised of a tion, working memory, and associative important cognitive process underlying central executive, an episodic buffer, a learning in relation to predicting lan- the development of language. Difficul- phonological loop, and a visuo-spatial guage. This brain activity, known as ties with rapid auditory temporal sketchpad. All but the latter are relevant resting frontal gamma power, at 16, 24, processing have been associated in some here. The central executive is an atten- and 36 months of age significantly cor- children with SLI [TaUal et al., 1985; tion system, eontroUing directed, related with sentence stmcture scores at Bishop et al., 1999], and may poten- focused, and sustained attention, and 4 and 5 years of age [Gou et aL, 2011]. tiaUy play a contributing role in the attention shifts, while inhibiting irrele- In a similar study, children with a fam- disorder. Studies have found correla- vant information. The episodic buffer ily history of language disorders have tions between rapid temporal processing acts as a temporary store for chunks of been found to have lower gamma of acoustic signals and early language infomiation that hnk working memory, power density functions [Benasich development. For example, Benasich perception, and long-term memory. et al., 2008]. This research indicates et al. [2008] found that children with a The phonological loop is believed to be that better cognitive control and work- family history of language disorders necessary for leaming new word-form ing memory skills at critical points in were not as efficient at detecting very to referent mappings (learning new rapid stimuh (70 ms between stimuli) as development may facilitate language words). This sub-system acts as a tem- those without a family history. The dif- development. porary store of newly encountered ferences in latency of response to these phonological fomis that are linked to rapid signals at six months of age pre- Phonological Short-Term Memory long-term memory with increased ex- dicted language outcomes at 24 Phonological short-term memory months. These correlations have been perience. The loop is susceptible to (PSTM) is a linguistic processing com- found to hold through ages 3 and 4 for rapid decay and capacity limitations, ponent of working memory. PSTM is cognitive and language measures both of which improve with age and usuaUy measured by repetition of a set [Choudhury and Benasich, 2011]. language experience. of nonsense words (nonword repetition; Executive functioning and lan- NWR), increasing in complexity from The speed of spoken word recog- guage usuaUy show moderate one to four or five syllables in length. nition has also been investigated as a correlations in the 3-6 year age range; There are several published measures factor in language development. Fernald however this is thought to be due to the available [e.g., see Gathercole et al., et al. [2006] demonstrated that efficient role language plays in helping children 1994; DoUaghan and CampbeU, 1998]. processing of spoken language correlates regulate their attention and behavior PSTM is beheved to play an important with concurrent and later vocabulary purposefuUy, rather than vice versa role in early language development development in TD infants and tod- [Carlson, 2005; Hughes and Ensor, [Gathercole and Baddeley, 1990]. A dlers. In addition, speed of spoken 2007]. Working memory skiUs are reduced capacity to store and process word recognition at 25 months of age strongly linked to vocabulary leaming in incoming phonological information leads contributes unique variance to language preschool children and deficits in work- to weaker or incomplete phonological and cognitive skiUs at age 8 years ing memory have been identified in representations of words, which are then [Marchman and Fernald, 2008]. Fernald children with SLI [e.g., Gathercole and more difficult to access and retrieve for and Marchman [2012] investigated the Baddeley, 1990]. Deficits in complex production, resulting in slowed vocabu- processing speed of LTs in relation to working memory, inhibition, and atten- laiy acquisition [Gathercole and their language outcomes 18 months tion skiUs are associated with SLI at a Baddeley, 1990]. The ability to recaU later. They found those who started out group level throughout early-to-middle sequences of words rehes principaUy on in the lowest quintile at 18 months of childhood [AUoway and Gathercole, PSTM [Baddeley et al., 1998]. There is age, but had faster processing speeds, 2006; Im-Bolter et al., 2006; Marton, substantial evidence to suggest that a h- were more hkely to show accelerated 2008]. mitation in PSTM could be a factor in vocabulary growth over the next 18 language disorders [Graf Estes et al., months, compared with LTs with less The use of working memory as a predictor factor is in its early stages of 2007]. Within the SLI population, few efficient processing. Their data are con- children have PSTM abihties within the sistent with a model of cascading investigation. However, there is some research on this related to predicting nomial range; however, when it does influences of processing on later devel- occur, better PSTM ability is correlated language outcomes in young children. opment. This is a promising factor to with higher language and hteracy per- Conboy et al. [2008] used behavioral consider in the prediction of language formance [Gathercole and Baddeley, measures such as retrieving objects hid- disorders in LTs. 1990; Botting and Conti-Ramsden, den or placed in boxes to estimate 2001; AUoway and Gathercole, 2006]. development of executive fiinction. Working Memory They found that in 11-month-old Recent research has indicated that A complete review of \vorking infants, scores on these tasks were cor- LTs also perform poorly on a test of memory models is beyond the scope of related with the abihty to ignore speech NWR relative to their age-matched

164 DEV DISABIL RES REV- EARLY LANGUAGE DELAY AND SLI • MOYLE ET AL. peers. For example. Stokes and Klee prefrontal cortex, basal gangha, and cer- Given the observed deficits in vis- [2009a,b] found that NWR was ebellum, is responsible for the learning ual procedural learning, verbal strongly related to expressive vocabulary and storage of rule-based behaviors, declarative learning, and NWR, which size in 2-year-old children. In addition, including some motor [e.g., bike-ri- together are indicative of deficits in Chiat and Roy [2008] reported that ding] and language behaviors (e.g., procedural, declarative, and working early NWR skills at age 2 and 3 were syntax and phonology). The declarative memory, Lum and Bleses [2012] sug- predictive of morphosyntactic develop- memory system, anchored in the medial gested that a pdmary deficit in the ment at age 4 and 5. It should be noted temporal lobe network, particularly the latter may underhe observed deficits in though that LTs with very few words hippocampus, is responsible for the the former. They showed that the poor generally will not participate in a learning and storage of arbitrary (non- performance of school-aged children NWR task. Stokes and Klee [2009a,b] rule-based) events, facts, and life exped- (mean age 7.6 years) with SLI on a reported that children who did not ences, such as form-referent mappings verbal declarative memory task w^as cooperate for NWR testing scored sig- in word learning. mediated by poor working memory nificantly lower on cognitive and UUman and Pierpont [2005] pro- abilities. Overall, there is an emerging language tests than children who did posed the Procedural Deficit view that school-aged children with cooperate. Noncompliance notwith- Hypothesis (PDH) as an account for SLI show memory deficits and difficul- standing, NWR is promising as a SLI, suggesting that SLI adses from a des with statistical learning [e.g., Evans predictive measure of later SLI for LTs. deficit in the procedural memory sys- et al., 2009].There is also evidence of a tem and underlying brain structures. (In strong association between implicit sta- THEORETICAL ACCOUNTS addidon to the above sections on tistical leaming and grammatical abihty The overview of the multifacto- and neurobiology, see UUman in TD pre-school children [Kidd, rial model is reflective of our and Pierpont [2005] and Ullman [2001] 2012]. theoretical perspective of SLI and the for an overview of the neviral correlates relationship between status of SLI in relation to the PDH account.) IMPLICIT STATISTICAL and subsequent development of LI. We In this account, a grammatical LI LEARNING OF LANGUAGE IN adopt a domain general, rather than a reflects difficulty with learning mle- TODDLERS domain specific perspective. Domain based language knowledge (syntax, The findings on the role of specific theodes include a specific morphology, and phonology) due to implicit learning in SLI prompted grammadcal deficit [van der Lely, 2005] deficits in the memory system responsi- Stokes et al. [Stokes, 2010; Stokes and a delay in setting the parameters of ble for rule-based leaming, the et al., 2012a, 2012b] to explore the grammatical system [Rice et al., procedural memory system. whether implicit learning deficits or dif- 1995]. However these grammatical the- Recent empidcal evidence sup- ferences could be detected in very odes are challenged by evidence of ports the PDH, highlighting the young children who are LTs and hence nonlinguistic impairments. Conse- difficulties that children with SLI have at dsk of SLI. This research focused on quently, we do not discuss domain with both nonverbal and verbal proce- probable statistical cues to language specific theories (or linguistic theodes) dural learning tasks. Tombhn et al. leaming dudng infancy. The language in this review. Domain general theories [2007] reported poor learning by ado- input that infants hear is often a contin- implicate poor temporal auditory proc- lescents with SLI on a visual procedural uous string of language without word essing [Tallal et al., 1985; Bishop et al., learning task (the Sedal Reaction Time boundades (thedogischasingthecat). This 1999], reduced verbal working memory task), as did Lum et al. [2010] in their continuous string would seem to pro- capacity, either specific to phonological study of 7-year-olds with SLI. Ho'w- vide a significant challenge to infants processing [Gathercole and Baddeley, ever, Lum et al. [2010] also reported who need to 'crack the code" of lan- 1990; Chiat, 2001] or more broadly for significance between group differences guage learning [Kuhl, 2004]. To linguistic processing [Ellis Weismer and on a verbal declarative memory task overcome this challenge, infants take Evans, 2002; Montgomery and Evans, (the number of semantically unrelated advantage of stadstical (probabihstic) 2009; Baird et al, 2010;], a generalized word pairs recaUed in a list task), sug- cues in the input that help in the iden- slow speed of processing [EUis Weismer gesting that memory deficits may tificadon of words [Saffran, 2003; and Hesketh, 1996], and weaknesses in underhe lexical learning in SLI. Sai&an and Graf Estes, 2006]. These the procedural memory system [Ullman As Lum and Kidd [in press] cues include cross-syUable transitional and Pierpont, 2005]. indicated, while the procedural mem- (sequential) probabilities of sound sequences [Aslin et al., 1998; Thiessen The domain general theodes ory system is principally concerned et al, 2005] and recurdng clusters of emphasize a cascading effect of lower with learning information that is se- syUables [Swingley, 2005]. None of this level processing difficulties impacting quential and probabilistic/statistical learning occurs with feedback and it on the development of higher level (e.g., grammar), and the declarative requires repeated or sustained input, all skills. Investigations of how children memory system is principaUy con- of which are the haUmarks of proce- with SLI leam in tasks that recmit the cerned with the association between dural leaming. The question that Stokes procedural memory system have con- stimuli to store a memory (e.g., map- et al. addressed was whether there tdbuted to the current focus on ping a heard phonological form to a might be statistical cues in the language learning mechanisms in children with referent in the visual field), the two input at the lexical, rather than sub-lex- SLI. Ullman's [2001] declarative-proce- systems interact. Further, working ical, level that might influence dural model of language postulated two memory and the declarative and pro- vocabulary acquisition in toddlers. Such separate (but possibly interactive) mem- cedural memory systems also interact, lexical cues would include phonological ory systems. It is postulated that the given shared neurological substrates, neighborhood density and word procedural memory system, anchored in particularly the prefrontal cortex [Lum frequency. the interactive network compdsing the et al., in press].

DEV DISABIL RES REV- EARLY LANGUAGE DELAY AND SLI • MOYLE ET AL. 165 ies are now needed to explore whether LTs who develop SLI, when compared with LTs who normalize, show differ- ent patterns of lexical learning arising from differing learning mechanisms, such as the use of the phonological cues inherent in word fomis, and PSTM abilities.

CLINICAL RECOMMENDATIONS We can draw several cHnical rec- ommendations from this literature review. There is strong evidence to suggest that at least three factors at 2 years of age should be considered in cHnical assessments, a family history of late talking, being male, and being less Neighborhoo i^ (Z-Score) than 85% of predicted birth weight. We suggest here that hnguistic process- Figure 2 Scatterplot of the relationship between neighborhood density and expressive vocabu- lary size (modified from Stokes, 2010). [Color figure can be viewed in the online issue, which is ing skiUs also be assessed. Two-year-old available at wileyonlinelibrary.com.] chUdren with language delay should be screened on a test of NWR, and this should be repeated at six-month inter- Toddlers' first spoken words are and 41% for EngHsh, French, and Dan- vals. We predict that children who influenced by the sound (phonological) ish, respectively [Stokes, 2010; Stokes become late bloomers wiU be the chU- and whole word (lexical) characteristics et al., 2012a, 2012b]. dren who score higher on a test of of words in the ambient language, and NWR than those LTs who will the statistical regularities with which Extended Statistical Learning become language impaired in the pre- they occur [Stokes, 2010]. For example, On the basis of these results. school years. This idea is speculative at toddlers' first words are short words Stokes et al. proposed a theory of present, and awaits research. In addi- that sound similar to many other words Extended Statistical Learning to account tion, we have suggested that children (those that share phonological strings) for slow vocabulary growth. This who begin as LTs and develop into in the ambient language (e.g. cat has 35 theory argues that poor learners have a chUdren with SLI may have quite dif- similar words, or neighbors, such as prolonged period of use of high phono- ferent language leaming mechanisms mat, pat, cap, kit). Such words have logical density as a cue to word from their TD peers. We suggest that high phonological neighborhood den- leaming. This initiaUy successful statisti- language clinicians explore teaching sity (ND). Words that have few cal cue effectively blocks subsequent high and low density words to children phonological neighbors are said to learning of words from sparse neighbor- who are identified as later talkers, in an "reside" in sparse neighborhoods [mouth hoods. The idea of blocking the use of effort to explore chUdren's learning has six neighbors: math, mouse, myth, more effective statistical learning mech- mechanisms. moth, south, mouth (verb)]. Stokes et al. anisms had first been proposed by Ashn [Stokes, 2010; Stokes et al, 2012a, and Newport [2008], although these SUMMARY 2012b] showed that the relationship authors did not suggest specific mecha- between phonological ND and vocabu- At the outset of this article, we nisms, such as using ND as a cue to stated that distinguishing between the lary size was very strong in 2-year-old learning. The question for clinicians children. Figure 2 shows the strength of LTs who do and do not develop a lan- and researchers is why would LTs be guage disorder at 4—5 years of age poses the relationship for Enghsh-speaking learning words from dense neighbor- children. The figure shows that when a diagnostic challenge. SLI is a hetero- hoods? We beheve the answer may lie geneous disorder, with various vocabtolary size is very smaU (lower in differences in working memory abil- than -1.25 SDs from the mean of 0; subgroups and a changing profUe for ities, given that Stokes and Klee each individual across development. 2:-scores), children's words are, on aver- [2009a] found that children with sniaU age, drawn from words in the input Recent research on the genetic and expressive vocabularies perfomied neurobiological influences on language that have very dense phonological poorly on a test of NWR. neighborhoods (more than 2 SDs above development and inipaimient, including the mean ND of 0; z-scores). Poor working memory abilities SLI, suggest that there are Hkely to be may limit the ability to achieve form- genetic variants that are related to indi- The finding is remarkably robust referent mapping of words from sparse vidual differences in neurobiological across . In three languages phonological neighborhoods because development in children that in turn (EngHsh, French, and Danish), children these phonological strings occur infre- contribute to the heterogeneity of SLI. who were poor vocabulary learners had quently in the ambient language input This heterogeneity is one of the reasons expressive lexicons comprised of far [Stokes et al., 2012b]. Hsu and Bishop why it is difficult to identify accurate fe^ver sparse words than TD children of [2011] suggested that deficits in work- predictors of development from late the same age. The amount of variance ing memory abiHty could underpin the talker status to SLI status. Attempts to in expressive vocabulary size accounted abihty to use statistical cues in the input date to identify predictive factors have for by phonological ND was 47%, 53%, in chUdren with SLL Longitudinal stud- at best been moderately successful. In

166 DEV DISABIL RES REV. EARLY LANGUAGE DELAY AND SLI • MOYLE ET AL. this review, we suggested that it is time Bishop DVM, Carlyon R, Deeks J, et al. 1999. Dale PS, Price TS, Bishop DVM, et al. 2003. Out- to widen the search for predictive fac- Auditory temporal processing impairment: comes of early language delay. I. Predicting per- neither necessary nor sufFicient for causing sistent and transient language difficulties at 3 and tors, so as to include measures of language impainnent in children. J Speech 4 years. J Speech Lang Hear Res 46:544—560. language leaming mechanisms. The last Lang Hear Res 42:1295-1310. Dale PS, Simonoff E, Bishop DVM, et al. 1998. decade has produced a wealth of evi- Bishop DVM, Edmundson A. 1987. Language Genetic influence on language delay in two- dence on the relationships among impaired 4 year olds: distinguishing transient year-old children. Nat Neurosci 1:324-328. memory systems and language behav- from persistent impairment. J Speech Hear Desmarais C, Sylvestre A, Meyer F, et al. 2008. Disord 52:156-173. Systematic review of the literature on char- iors. Very recent work has shown that Bishop DVM, Price TS, Dale PS, et al. 2003. acteristics of late-talking toddlers. Int J Lang this relationship is present even in 2- Outcomes of early language delay II: etiol- Commun Disord 43:361-389. year-old children. Children with ex- ogy of transient and persistent language diffi- DoUaghan CA, Campbell TF. 1998. Nonword pressive language delays have poor culties. J Speech Lang Hear Res 46:561- repetition and child language impainnent. J 575. Speech Lang Hear Res 41:1136-1146. short-term memory abilities and \ve DoUaghan CA, Campbell TF, Paradise JL, et al. suggest that this memory deficit con- Bishop DVM, Snowling MJ. 2004. Developmen- tal dyslexia and specific language impair- 1999. Maternal education and measures of tributes to the use of learning ment: same or different? Psychol Bull early speech and language. J Speech Lang mechanisms that differ from those of 130:858-886. Hear Res 42:1432-1443. TD 2-year-old children. We propose a Bomstein MH, Haynes MO, Painter KM. 1998. EUis E, Thai D. 2008. Early language delay and new framework for thinking about Sources of child vocabulary competence: a risk for language impainnent. Perspect Lang ELD. We propose that ELD is charac- multivariate model. J Child Lang 25:367—393. Leani Educ 15:93-100. Botting N, Conti-Ramsden G. 2001. Non-word Ellis Weismer SE, Evans JL. 2002. The role of terized not only by the quantitative repetition and language development in chil- processing limitations in early identification metric of number of words produced, dren with specific language impairment (SLI). of specific language impainnent. Top Lang or the presence/absence of word coni- IntJ Lang Commun Disord 36:421-432. Disord 22:15-29. binadons, but also by short-temi Buschmann A, Jooss B, Rupp A, et al. 2008. Chil- Ellis Weismer SE, Hesketh LJ. 1996. Lexical memory abilities that impact on child- dren with developmental language delay at 24 leaming by children with specific language months of age: results of a diagnostic work- impainnent: effects of linguistic input pre- ren's learning mechanisms. Improving up. Dev Med Cliild Neurol 50:223-229. sented at varying speaking rates. J Speech our ability to predict SLI outcomes Bricker D, Squires J. 1999. Ages & Stages Question- Hear Res 39:177-190. from late talker status could depend on naires (ASQ): a parent-completed, child-mon- Evans JL, Saffran JR, Robe-Torres K. 2009. the careful combinadon of the most itoring .system, 2nd ed. Baltimore: Brookes. Statistical learning in children with specific Carlson SM. 2005. Developmentally sensitive promising predictors, be they genetic, language impainnent. J Speech Lang Hear measures of executive function in preschool Res 52:321-335. neurobiological, and/or cognitive learn- children. Dev Neuropsychol 28:595-616. Fenson L, Marchman VA, Thal D, et al. 2007. ing mechanisms. Catts HW, Adolf SM, Hogan TP, et al. 2004. The MacArthur-Bates Communicative De- Are specific language impairment and dys- velopment Inventories: user's guide and tech- lexia distinct disorders? J Speech Lang Hear nical manual, 2nd ed. Baltimore: Brookes. REFERENCES Res 48:1378-1396. Femald A, Marchman VA. 2012. Individual dif- Alloway TP, Gathercole SE, editors. 2006. Work- Chiat S. 2001. Mapping theories of developmen- ferences in lexical processing at 18 months ing memory and neurodevelopmental disor- tal language impairment: premises, predic- predict vocabulary growth in typically ders. New York: Psychology Press. tions and evidence. Lang Cogn Process developing and late-talking toddlers. Cliild 16:113-142. Dev 83:203-222. Aslin RN, Newport EL. 2008. What statistical Chiat S, Roy P. 2008. Early phonological and Fernald A, Perfors A, Marchman VA. 2006. Pick- learning can and can't tell us about language sociocognitive skills as predictors of later lan- ing up speed in understanding: speech proc- acquisition. In: Colombo J, McCardle P, guage and social communication outcomes. J essing efficiency and vocabulary growth Freund, L, editors. Infant pathways to lan- Child Psychol Psychiatry 49:635-645. across the 2nd year. Dev Psychol 42:98-116. guage: methods, models and research direc- Choudhury N, Benasich AA. 2003. A family Fmneran DA, Francis AL, Leonard LB. 2009. tions. Hove: Psychology Press, p 15-29. aggregation study: the influence of family Sustained attention in children with specific Aslin RN, SafFran JR, Newport EL. 1998. Com- history and other risk factors on language language impairment (SLI). J Speech Lang putation of conditional probability statistics development. J Speech Lang Hear Res Hear Res 52:915-929. by human infants. Psychol Sei 9:321-324. 46:261-272. Fisher SE, Scharff C. 2009. FOXP2 as a molecu- Badcock NA, Bishop DVM, Hardiman MJ, et al. Choudhury N, Benasich AA. 2011. Maturation of lar window into speech and language. 2012. Co-localisation of abnonnal brain auditory evoked potentials from 6 to 48 Trends Genet 25:166-177. structure and function in specific language months: prediction to 3 and 4 year language FlaxJF, Realpe-BoniUa T, Roesler C, et al. 2009. impainnent. Brain Lang 120:310—320. and cognitive abilities. Clin Neurophysiol Using early standardized language measures 122:320-338. to predict later language and early Baddeley A. 2012. Working memory: theories, Clegg J, HolUs C, Mawhood L, et al. 2005. De- outcomes in children at high risk for lan- models and controversies. Annu Rev Psy- velopmental language disorders—a follow- guage-leaniing impaimients. J Learn Disabil chol 63:1-29. up later in adult life. Cognitive, langtiage 42:61-75. Baddeley A, Gathercole S, Papagno C. 1998. The and psychosocial outcomes. J Child Psychol Gathercole SE, Baddeley AD. 1990. Phonological phonological loop as a language leaming de- Psychiatry 46:128-149. memory deficits in language disordered chil- vice. Psychol Rev 105:158-173. Cohen M, Campbell R, Yaghmai F. 1989. Neu- dren: is there a causal connection? J Mem Baird G, Dworzynski K, Slonims V, et al. 2010. ropathological abnomialities in developmen- Lang 29:336-360. Memory impainnent in children with lan- tal dysphasia. Ann Neurol 25:567-570. Gathercole SE, Willis CS, Baddeley AD, et al. guage impairment. Dev Med Child Neurol Conboy BT, Sommeiville JA, Kuhl PK. 2008. 1994. The children' test of nonword repeti- 52:535-540. Cognitive control factors in speech perception tion: a test of phonological working mem- at 11 months. Dev Psychol 44:1505-1512. ory. Memory 2:103-127. Benasich AA, Gou Z, Choudhury N, et al. 2008. Conti-Ramsden G, Botting N. 1999. Classifica- Gauger LM, Lombardino LJ, Leonard CM. 1997. Early cognitive and language skills are linked tion of children with specific language Brain morphology in cliildren with specific to resting frontal gamma power across the impainnent: longitudinal considerations. J first 3 years. Behav Brain Res 195:215-222. language impaimient. J Speech Lang Hear Speech Lang Hear Res 42:1195-1204. Res 40:1272-1284. Bishop DVM. 1994. Is specific language impair- Courage ML, Cowan N, editors. 2009. The de- Gottlieb G. 2007. Probabalistic epigénesis. Dev ment a valid diagnostic category? Genetic velopment of memory in infancy and child- Sei 10:1-11. and psychoHnguistic evidence. Philos Trans hood. Hove, UK: Psychology Press. Gou Z, Choudhury N, Benasich AA. 2011. Rest- R Soc Lond B Biol Sei 346:105-111. Creusere M, Alt M, Plante E. 2004. Recognition ing frontal gamma power at 16, 24 and 36 Bishop DVM. 2009. Genes, cognition, and com- of vocal and facial cues to affect in lan- months predicts individual differences in munication: insights from neurodevelopmen- guage-impaired and normally-developing language and cognition at 4 and 5 years. tal disorders. Ann NY Acad Sci 1156:1-18. preschoolers. J Commun Disord 37:5-20. Behav Brain Res 220:263-270.

DEV DISABIL RES REV- EARLY LANGUAGE DELAY AND SLI • MOYLE ETAL. 167 Graf Estes K, Evans JL, Else-Quest NM. 2007. Lum JAG, Conti-Ramsden G, Page D, et al. Reilly S, Bavin EL, Bretherton L, et al. 2009. Differences in nonword repetition perform- 2012. Working, declarative, and procedural The Early Language in Victoria Study ance of children with and without specific learning in specific language impaimient. (ELVS): a prospective, longitudinal study of language impairment: a meta-analysis. J Cortex 48:1138-1154. communication skiUs and expressive vocabu- Speech Lang Hear Res 50:177-195. Lum JAG, Gelgic C, Conti-Ramsden G. 2010. lary development at 8, 12 and 24 months. Hart B, Iiisley TR. 1995. Meaningfiü differences Procedural and declarative memory in chil- IntJ Speech Lang Pathol 11:344-357. in the everyday experiences of young Amer- dren with and without specific language Reilly S, Wake M, Bavin EL, Prior M, Williams ican children. Baltimore, MD: Paul H. impaimient. Int J Lang Commun Disord J, Bretherton L et al. 2007. Predicting lan- Brookes. 45:96-107. guage at 2 years of age: A prospective com- Hill EL. 2001. Non-specific nature of specific Lum JAG, Kidd E. 2012. An examination into munity study. Pediatrics 120:1441-1449. language impairment: a review of the litera- the associations between multiple memory ReiUy S, Wake M, Ukoumunne OC, et al. 2010. ture with regard to concomitant motor systems, past tense and vocabulary in typi- Predicting language outcomes at four years impainnents. Int J Lang Commun Disord caUy developing 5-year old children. J of age: findings from early language in Vic- 36:149-171. Speech Lang Hear Res 55:989-1006. toria study. Pediatrics 126:1530-1537. Rescorla L. 1989. The Language Development Sur- HofF E. 2003. The specificity of environmental Marchman VA, Fernald A. 2008. Speed of word recognition and vocabulary knowledge in vey: a screening tool for delayed language in influence: socioeconomic status affects early toddlers. J Speech Lang Hear Res 54:587-599. vocabulary development via matema] infancy predict cognitive and language out- comes in later childhood. Dev Sei 11:9-16. Rescorla L, Roberts J. 1997. Late talkers at 2: out- speech. Child Dev 74:1368-1378. come at age 3. J Speech Lang Hear Res Marton K. 2008. Visuo-spatial processing and ex- Hughes C, Ensor R. 2007. Executive function and 40:556-566. theory of mind: predictive relations from ages ecutive fiinctions in children with specific 2 to 4. Dev Psychol 43:1447-1459. language impaimient. Int J Lang Commun Rice ML, Taylor CL, Zubrick SR. 2008. Lan- Hsu HJ, Bishop DVM. 2011. Graimnatical diffi- Disord 43:181-200. guage outcomes of 7-year-old children with culties in children with specific language Molfese DL, Molfese VJ, Espy KA. 1999. The or without a history of late language emer- impaimient: is leaming deficient? Hum Dev predictive use of event-related potentials in gence at 24 months. J Speech Lang Hear 53:264-277. language development and the treatment of Res 51:394-407. language disorders. Dev Neuropsychol Rice ML, Tomblin JB, Hoffinan L, et al. 2004. Im-Boker N, Johnson J, Pascual-Leone J. 2006. 16:373-377. Grammatical tense deficits in children with Processing limitations in children with spe- Montgomery JW. 1993. Haptic recognition of SLI and nonspecific language inipainnent: cific language impaimient: the role of exec- children with specific language inipainnent: relationships with nonverbal IQ over time. J utive function. Child Dev 77:1822-1841. effects of response modality. J Speech Hear Speech Lang Hear Res 47:816-834. Johnston JR. 1994. Cognitive abilities of children Res 36:98-104. Rice ML, Wexler K, Cleave PL. 1995. Specific with language inipainnent. In: Watkins RV, Montgomery JW, Evans JL. 2009. Complex sen- language impairment as a period of extended Rice ML, editors. Specific language impair- tence comprehension and working memory optional infinitive. J Speech Hear Res ments in children. Baltimore, MD: Paul H. in children with specific language impair- 38:850-863. Brookes, p 107-120. ment. J Speech Lang Hear Res 52:269-288. Rowe ML, Raudenbush SW, Goldin-Meadow S. Kamhi AG, Catts HW, Koenig LA, et al. 1984. Montgomery JW, Magimairaj BL, Finney MC. 2011. The pace of vocabulary growth helps Hypothesis-testing and nonlinguistic sym- 2010. Working memory and specific lan- predict later vocabulary skill. Child Devt bolic abilities in language-impaired children. guage impairment: an update on the relation 83: 508-525. J Speech Hear Disord 49:169-176. and perspectives on assessment and treat- Saffran JR. 2003. Statistical language leaming: Kidd E. 2012. Implicit statistical learning is ment. Am J Speech Lang Pathol 19:78-94. mechanisms and constraints. Curr Dir Psy- directly associated with the acquisition of Newman R, Ratner NB, Jusczyk AM, et al. chol Sei 12:110-114. syntax. Dev Psychol 48:171-184. 2006. Infants' early ability to segment the SafTran JR, Graf Estes K. 2006. Mapping sound to Klee T, Carson DK, Gavin WJ, et al. 1998. Con- conversational speech signal predicts later meaning: connections between leaming current and predictive validity of an early language development: a retrospective analy- about sounds and learning about words. Adv Child Dev Behav 34:1-38. language screening program. J Speech Lang sis. Dev Psychol 42:643-655. Hear Res 41:627-641. Olswang LB, Rodriguez B, Timler G. 1998. Soriano-Mas C, Pujol J, Ortiz H, et al. 2009. Recommending intervention for toddlers Age-related brain structural alterations in Klee T, Pearce K, Carson DK. 2000. Improving children with specific language impaimient. with specific language leaming difficulties. the positive predictive value of screening for Hum Brain Mapp 30:1626-1636. Am J Speech Lang Pathol 7:23-32. developmental language disorder. J Speech Stanton-Chapman TL, Chapman DA, Bainbridge Oram Cardy JE, Flagg EJ, Roberts W, et al. Lang Hear Res 43:821-833. NL, et al. 2002. Identification of early risk 2008. Auditory evoked fields predict lan- Klee T, Stokes SF. 2011. Language development. factors for language impaimient. Res Dev In: Skuse D, Bmce H, Dowdney L, Mrazek guage ability and impairment in children. Disabil 23:390-405. IntJ Psychophysiol 68:170-175. D, editors. Child psychology and psychiatry: Stokes SF. 1997. Secondary prevention of paedi- Ors M, Ryding E, Lindgren M, et al. 2005. frameworks for practice, 2nd ed. Chichester: atric language disability: a comparison of SPECT findings in children with specific Wiley-Blackwell. p 45-50. parents and nurses as screening agents. Int J Kuhl PK. 2004. Early language acquisition: crack- language impaimient. Cortex 41:316-326. Lang Commun Disord 32:139-158. ing the speech code. Nat Rev Neurosci Paul R. 1996. CUnical implications of the natural Stokes SF. 2010. Neighborhood density and word 5:831-843. history of slow expressive language develop- frequency predict vocabulary size in tod- Kuhl PK. 2010. Brain mechanisms in early lan- ment. AmJ Speech Lang Pathol 5:5—21. dlen. J Speech Lang Hear Res 53:670-683. guage acquisition. Neuron (fl-HYb-lTl. Paul R, Elwood TJ. 1991. Matemal Hnguistic Stokes SF, Bleses D, BasbeU H, et al. 2012a Sta- Leonard CM, Lombardino LJ, Walsh K, Eckert input to toddlers with slow expressive lan- tistical learning in emerging lexicons: The MA, Mockler JL, Rowe LA, WiUiams S, guage development. J Speech Hear Res case of Danish. J Speech Lang Hear Res DeBose CB. 2002. Anatomical risk factors 34:982-988. 55:1265-1273. that distinguish dyslexia from SLI predict Plante E, Swisher L, Vance R, et al. 1991. MRI Stokes SF, Kern S, dosSantos C. 2012b. Extended reading skill in noniial children. Jnl Comm findings in boys with specific language statistical leaming as an account for slow vo- Dis 35:501-531. impaimient. Brain Lang 41:52-66. cabulary growth. J Child Lang 39:105-129. Plomin R, Fulker DW, Corley R, et al. 1997. Leonard LB. 1998. Children with specific lan- Stokes SF, Klee T. 2009a. Faetón that influence Nature, nurture, and cognitive development vocabulary development in two-year-old chil- guage impaimient. Cambridge, MA: MIT from 1 to 16 Years: a parent-offspring adop- Press. dren. J Child Psychol Psychiatry 50:498-505. tion study. Psychol Sei 8:442-447. Stokes SF, Klee T. 2009b. The diagnostic accu- Leonard LB, Ellis Weismer SE, MiUer CA, et al. Preis S, Jïncke L, Schittler P, et al. 1998. Nomial racy of a new test of early non word repeti- 2007. Speed of processing, working mem- intrasylvian anatomical asymmetry in chil- tion for differentiating late talking and typi- ory, and language impairment in children. J dren with developmental language disorder. caUy developing children. J Speech Lang Speech Lang Hear Res 50:408-428. Neuropsychologia 36:849-855. Hear Res 52:872-882. Lum JAG, Bleses D. 2012. Declarative and proce- Price TS, Eley TC, Dale PS, et al. 2003. Genetic Stothard SE, Snowling MJ, Bishop DVM, et al. dural memory in Danish speaking children and environmental covariation between 1998. Language-impaired preschoolers: a with specific language impairment. J Com- verbal and nonverbal cognitive development foUow-up into adolescence. J Speech Lang mun Disord 45:46-58. in infancy. Child Dev 71:948-959. Hear Res 41:407-418.

168 DEV DISABIL RES REV- EARLY LANGUAGE DELAY AND SLI • MOYLE ET AL. Swingley D. 2005. Statistical clustering and the Developmental language disorders: from Venies SC, Newbury DF, Abrahams BS, et al. contents of the infant vocabulary. Cogn Psy- phenotypes to etiologies. Mahwah, NJ: 2008. A functional genetic link between dis- chol 50:86-132. Lawrence Erlbaum. p 53-89. tinct developmental language disorders. N Tallal P, Stark RE, Melhts ED. 1985. Identifica- Tsao F, Liu H, Kuhl PK. 2004. Speech percep- EnglJ Med 359:2337-2345. tion of language-impaired children on the tion in infancy predicts language develop- Vig S, Kaminer RK, Jediysek E. 1987. A later basis of rapid perception and production ment in the second year of life: a longitudi- look at borderline and mildly retarded pre- skills. Brain Lang 26:314-322. nal study. Child Dev 75:1065-1084. schoolers. J Dev Behav Pediatr 8:12-17. Tbal D. 2000. Late-talking toddlers: are they at Turkheinier E, Haley A, Waldron M, et al. 2003. Vigil DC, Hodges J, Klee T. 2005. Quantity and risk? San Diego, CA: San Diego State Uni- Soeioeeonomie status modifies heritability of quality of parental language input to late- venity Press. IQ in young children. Psychol Sei 14:623-628. talking toddlers during play. Child Lang Teach Ther 21:107-122. Thai D, Tobias S. 1992. Communicative gestures in LJllman MT. 2001. A neuroeognitive perspective children with delayed onset of oral expressive on language: the declarative/procedural Webster RI, Shevell MI. 2011. Topical review: vocabulary.J Speech Hear Res 35:1281-1289. model. Nat Rev Neurosci 2:717-726. neurobiology of specific language impair- ment. J Child Neurol 19:471-481. Thai D, Tobias S, Morrison D. 1991. Language Ullnian MT, Pierpont EL 2005. Specific language Whitehouse AJO, Bishop DVM, Ang QW, and gesture in late talkers. J Speech Lang impaimient is not specific to language: the pro- et al. 2011. CNTNAP2 variants affect Hear Res 34:604-612. cedural deficit hypothesis. Cortex 41:399-433. early language development in the general Thiessen ED, HiU EA, Saffian JR. 2005. Infant- van der Lely HJK. 2005. Domain-specific cogni- population. Genes Brain Behav 10: directed speech facilitates word segmenta- tive systems: insight from Grammatical-SLI. 451-456. tion. Infancy 7:53-71. Trends Cogn Sei 9:53-59. Yoder PJ, Warren SF. 2004. Early predictors of Tomblin JB, Mainela-Arnold E, Zhang X. 2007. Vargha-Khadem F, Gadian DG, Copp A, et al. language in children w^ith and without Procedural learning and adolescents with 2005. FOXP2 and the neuroanatomy of . Am J Ment Retard and without specific language impairment. speech and language. Nat Rev Neurosci 109:285-300. Lang Leam Dev 3:269-293. 6:131-138. Zubrick SR, Taylor CL, Rice ML, et al. 2007. Tomblin JB, Zhang X, Weiss A, et al. 2004. Vargha-Khadem F, Watkins KE, Price CJ, et al. Late language emergence at 24 months: an Dimensions of individual differences in 1998. Neural basis of an inherited speech epidemiological study of prevalence, predic- communication skills among primary grade and language disorders. Proc Nati Acad Sei tors, and covariates. J Speech Lang Hear children. In: Rice ML, Warren SF, editors. USA 95:12695-700. Res 50:1562-1592.

DEV DISABIL RES REV- EARLY LANGUAGE DELAY AND SLI • MOYLE ETAL 169 Copyright of Developmental Disabilities Research Reviews is the property of John Wiley & Sons, Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.