Retour vers liste

Détail de la contribution

Auteur: Hélène DELAGE

Co-Auteur(s): Ulrich FRAUENFELDER, Psycholing., Univ. de Genève, Suisse

Relation between working memory and complex syntax in SLI

Abstract/Résumé: A leading account of specific language impairment (SLI) explains the linguistic deficits observed in terms of limitations in nonlinguistic cognitive systems (such as attention or memory). Our goal here is to explore the relationship between working memory and complex syntax in SLI. Previous studies on children with SLI have already observed deficits in working memory (eg. Majerus et al., 2009), while others have documented poor performance on syntactically complex sentences (eg. Hamann et al., 2007). However, very few have linked these two components directly, and this is precisely our goal here. Following Jakubowicz (2007, 2011), we assume that limitations in working memory capacity persist in children with SLI and that this incomplete maturation of memory limits the processing of complex sentences. In contrast, such memory limitation disappears with age in typical development, enabling the child to process increasingly complex sentences. To investigate this hypothesis, we evaluated 48 typically-developing French-speaking children aged 5-12 compared to 28 children with SLI aged 5-14. In accordance with the working memory model of Baddeley (1986) and the findings in Barrouillet & Camos (2001, 2007), we assessed 1) the phonological loop using three “simple span” tasks (forward digit span, repetition of words and non-words) and 2) verbal working memory (associating the phonological loop and the central administrator) using three “complex span” tasks (backward digit span, counting span and running span). The children’s comprehension and repetition of complex utterances were assessed by means of tasks (sentence-image matching and immediate repetition) in which we varied sentence complexity (according to type of embedded clause and depth of embedding). Analyses of samples of the children’s spontaneous language were used to evaluate the complexity of their syntax in a natural context (MLU, rate of subordination and of deep embedding). Our results confirm an increase in working memory capacities and in complex syntax both across age within groups and across groups, with significantly lower scores for the SLI group. We also found a close link between working memory capacities and the mastery of complex syntactic structures, with different patterns for the two groups of children. Indeed, whereas the complex spans (and particularly the counting span task) accounted for a significant part of the variance in syntactic comprehension and in spontaneous production of complex sentences in control children, the simple spans (and in particular the system of retention of serial order) were the most predictive variables for syntactic measures of children with SLI. The theoretical and clinical implications of these results will be discussed.