Literacy development in children with language impairments

There is currently ample evidence that children with SLI are poor at reading and spelling due to their oral language impairments, and in particular due to their phonological deficits (e.g. Catts et al. 2002, see also 2.3). As Fraser and Conti-Ramsden (2008) note “the influence of phonological awareness on reading accuracy is incontrovertible” (p. 553). In fact, numerous studies have proven that phonological awareness (PA) is a good predictor for subsequent literacy outcomes (Castles & Coltheart, 2004). Moreover, there is evidence that literacy development contributes to the development of PA. Stackhouse and Wells’ (1997) psycholinguistic model illustrates this reciprocal relationship of causality between phonological awareness and literacy development (Image 3). At the heart of this model lies the assumption that children establish a speech processing system by implicitly manipulating oral language (Image 4). “Children receive information of different kinds (e.g. auditory, visual) about an utterance, remember it and store it in a variety of lexical representations (e.g. phonological, semantic) within the lexicon, then select and produce spoken words (p. 8, ib). This system serves as the basis for the development of both speech and literacy. Consequently, the speech perception deficit of children with SLI, leading to difficulties in sound discrimination, will result not only in deficient phonological representations and in problematic PA, but also will impact on their subsequent literacy development as these early phonological representations are essential for learning the mappings between sounds and letters.

However, SLI is a heterogeneous condition, varying in severity, persistence and pattern of linguistic deficits and accompanying cognitive difficulties. Children with SLI are at high risk for subsequent deficits in reading and spelling, yet at least some of them learn to read and spell like typically developing children (Catts et al. 2005; Kelso et al. 2007; Zourou et al. in press). The aim of our study (chapter 6) was to identify the factors, linguistic and cognitive, that differentiate children with SLI with good reading and spelling outcomes from those in risk of important difficulties in literacy. Despite the large body of research that has been conducted, few studies have specifically examined the types of written language difficulties experienced by children with SLI by simultaneously investigating reading decoding and spelling skills. We conducted three studies, two short-term longitudinal studies (one in children with SLI and one in children with NLI) and one comparative study in children with NLI, SLI and resolved-SLI. The first two studies allowed us to explore how children with language impairments evolve in time. We expected that literacy development would have a positive effect on language impairments and in particular on their phonological awareness deficits. The main findings of these two studies were 1/ very poor performance in non-word repetition (tables 29 and 35), task considered as a good measure of phonological working memory (PWM) capacities and 2/ poor performance in reading and in particular in spelling (tables 30 and 36). Concerning their oral language performances, findings were not the same on the two populations, SLI and NLI children. Children with SLI showed as expected an improvement in phonological and vocabulary measures (table 26). Almost half children of the SLI group, showed no particular difficulty in phonology, vocabulary and morpho-syntax. On the other hand, children with NLI showed considerable persistence of their oral language impairments even at the age of 11 years (table 34).

Following these two short-term longitudinal studies, we conducted a comparative study to study the factors that differentiate children with language impairments in their written language outcomes. We studied factors such as the severity and persistence of language impairments, verbal WM skills and nonverbal intelligence (NVIQ), by comparing a group of children with SLI (N=10), a group of children with NLI (N=10) and a group of children with resolved-SLI (N=10). The latter were identified in a longitudinal study (over a period of 20 months) of language impairments in French-speaking school-aged children (Zourou et al. in press). During the last assessment, while they were about 8 years-old, these children showed a considerable improvement of their oral language impairments, thus they consisted our resolved-SLI (r-SLI) group. The findings from this study support that 1/ severity of the language impairment (SLI > NLI), 2/ persistence of the language impairment during school years (r-SLI > NLI), 3/ nonverbal intelligence (SLI> NLI) and 4/ performances in nonword repetition (SLI > NLI), are reflected on reading and spelling outcomes (tables 40, 41 and 42). PA is a good predictor of subsequent reading outcomes in typically developing children, however, is seems insufficient in the differentiation of children with language impairments. The four factors identified in the comparative study could allow researchers and clinicians to predict subsequent literacy profiles in children with language impairments, as they seem to be related to reading and spelling outcomes.

In the light of the above, several remarks can be made. The first one concerns the severe difficulties French-speaking children with language impairments encounter in literacy development. Even children who no longer present oral language impairments presented significant difficulties in spelling when compared to TD children (all measures used were from standardized batteries). Thus, we support the idea defended by Bishop and Clarkson (2003) that written language is a window into residual language deficits. Better performances in reading than in spelling are, however, not typical of Anglo-Saxon children. French language presents a particularity compared to English in that it is less consistent in spelling (sound-to-letter mappings) that in reading (letter-to-sound mappings) explaining why French-speaking children with SLI show more difficulties in spelling (Seymour et al. 2003). The second one concerns the distinction between children with NLI and children with SLI. It has been argued that the cut-off score of 85 on nonverbal IQ is not very informative as many children with language impairments just above and below this cut-off point can show very similar linguistic profiles (Tomblin & Zhang, 1999). The findings from our comparative study support this idea. In fact, the SLI subgroup showed better outcomes in reading and spelling than the NLI subgroup, who presented more severe WM deficits and low nonverbal IQ, but both groups presented similar profiles in oral language measures. Thus, we argue that all three subgroups, NLI, SLI and r-SLI, represent a continuum of language impairments and not distinct disorders; the NLI subgroup is regarded as being at the lower limit and the r-SLI subgroup as being at the upper limit of this continuum. The third concerns the need to identify clinical markers of language impairments and the use of a nonword repetition measure as such. Throughout our studies, we showed that children with language impairments present important difficulties in repeating non-words. Recently, Claessen et al. (2009) defended the idea that poor performance in non-word repetition tasks may reflect poor encoding skills, poor distinctness of phonological representations or even poor output motor programs (image 4). Thus, in line with these authors we argue of the need to develop experimental tasks, requiring no verbal output from the child, allowing us to assess the integrity of phonological representations. Gathering information about the quality of a child’s underlying phonological representations as well as their PA may allow us to fine subgroup children and allow more specifically targeted intervention programs, based on this knowledge.