Portal de Periódicos da CAPES

Subject Relative Production in SLI Children during Syntactic Priming and Sentence Repetition

2012; Wiley; Volume: 34; Issue: 34 Linguagem: Inglês

1551-6709

Moreno I. Coco, Maria Garraffa, Holly P. Branigan,

Reading and Literacy Development

Subject Relative Production in SLI Children during Syntactic Priming and Sentence Repetition Moreno I. Coco (mcoco@staffmail.ed.ac.uk) School of Informatics (ILCC) University of Edinburgh 10 Crichton Street Edinburgh, EH8 9AB Maria Garraffa (mgarraff@staffmail.ed.ac.uk) School of Psychology University of Edinburgh 7 George Square Edinburgh, EH8 9JZ Abstract Holly P. Branigan (holly.branigan@ed.ac.uk) School of Psychology, University of Edinburgh 7 George Square Edinburgh, EH8 9JZ memory demands. We investigate this hypothesis by compar- ing SLIC’s and TDC’s production of SRCs in tasks where they are explicitly elicited (in a sentence repetition task) and when they are implicitly elicited (in a picture-description syn- tactic priming paradigm). We use Bayesian Data Analysis to investigate our hypothesis, and design a series of Bayesian models to tackle it. Children with Specific Language Impairment (SLIC) experi- ence difficulties in processing Subject relative clauses (SRC). This has been interpreted as evidence that they lack syntac- tic representations for SRC. Our study investigates the spon- taneous production of SRC in typically developing children (TDC) and SLIC in a structural priming paradigm, and com- pares their performance in a sentence repetition task. We demonstrate that SLIC are much more likely to produce SRC during priming than in sentence repetition; moreover, when primed, their performance matches TDC’s baseline (unprimed) performance. Furthermore, we design two simple unsuper- vised Bayesian models, and predict the developmental group (SLI, TD) and priming condition (Primed, Non-Primed). Over- all, this study shows that SLIC can spontaneously produce SRC when primed, suggesting their impairment is related to working memory, rather than a deficit in syntactic knowledge. Experiment Substantial research has used a syntactic priming paradigm to demonstrate that people use abstract syntactic representa- tions to process language (Bock, 1986; Pickering & Branigan, 1998). In such research, speakers show an increased tendency to use a particular structure after previously encountering the same structure (even with different lexical content). These ef- fects have been have been argued to provide evidence about syntactic representation (Branigan et al., 1995). Recent re- search on TDC has therefore used syntactic priming to pro- vide evidence for the early acquisition of TDC’s syntactic representations, e.g., passive constructions (e.g., Bencini & Valian 2008): If children are more likely to produce a partic- ular structure after previous exposure to it, it implies that they have an abstract representation for that structure, which can be facilitated through residual activation or implicit learning (Chang et al., 2006; Pickering & Branigan, 1998). The for- mer may explain short term priming effects; the latter may explain long term and cumulative priming effects. We argue that a syntactic priming paradigm can similarly be used to examine whether SLIC have access to an abstract representa- tion of SRC, whose availability can be incremented through priming. To do this, we used a Snap priming paradigm (Branigan et al., 2005), in which SLIC and TDC engaged in a card game that involved three elements: 1) listening to the experimenter describe a picture (using either a simple noun or an SRC), 2) describing their own picture, 3) and deciding whether or not the two pictures matched 1 . In the analysis, we assess whether the two groups (TDC, SLIC) differ in their production of SRC (Model 1); and test if any difference depends on the priming condition (i.e., whether the experimenter’s description involved a simple noun or an SRC; Model 2). We expect TDC to have a higher production of SRC than SLIC. However, we expect SLIC to perform better in the primed than in the non-primed condi- Keywords: specific language impairment; language develop- ment; syntactic priming; sentence repetition; Bayesian data analysis. Introduction Subject relative clauses (SRC) such as the cat that’s on the ta- ble are generally early acquired, at around 3 years in typically developing children (TDC; e.g., Crain et al. 1990). However, children with Specific Language Impairment (SLIC) display difficulties in producing subject (and object) relative clauses (Novogrodsky & Friedmann, 2006). Preschool SLIC show a delayed onset of relative clause production, and frequent omission of the complementizer in both elicitation and spon- taneous production (Contemori & Garraffa, 2010). This diffi- culty extends to repetition of sentences involving SRCs. This is particularly interesting, because recent research has sug- gested that in TDC, prior exposure to even difficult struc- tures can facilitate their subsequent production (e.g., Bencini & Valian 2008). Such effects have been identified as mani- festations of syntactic priming, whereby an abstract syntactic representation is facilitated (Bock, 1986). In adults, syntactic priming appears to be implicated in sentence repetition (Pot- ter & Lombardi, 1998). It is therefore striking that SLIC do not show facilitated production of SRC in sentence repetition, as we would expect a benefit from a syntactic priming effect, enhanced by lexical repetition. Previous research has therefore proposed that SLIC do not have a syntactic representation of SRC (Conti-Ramsden et al., 2001). In this paper, we consider an alternative hypothesis, namely that SLIC’s poor performance in sentence repetition does not reflect a lack of syntactic knowledge, but rather a task-specific difficulty, which may be related to working 1 On three quarters of trials, the pictures did not match, and on one quarter - Snap trials - they did match