Relative clause production as a vulnerable linguistic feature of Developmental Language Disorder in Cantonese

Abstract

Developmental Language Disorder (DLD; termed earlier as “Specific Language Impairment; SLI) is a disorder primarily affecting linguistic abilities and language development in children. It is a language impairment that is not secondary to hearing loss, emotional and behavioral problems, severe intellectual deficits, and clear neurological problems. Difficulty with relative clauses (RCs) is robustly documented in children with DLD/SLI cross-linguistically. Cantonese is an important language in debates regarding acquisition and processing of RCs, given its unique typological properties. Yet, to date there has been no published research on the syntactic competence of RCs in Cantonese children with DLD/SLI. This study investigated whether RCs are particularly vulnerable in Cantonese-speaking children with DLD relative to their typically developing age peers. As a SVO language, Cantonese is unusual in placing the RC before the head noun, resulting in processing demands competing in opposite directions: Cantonese subject RCs are arguably less costly to process in light of structurally-oriented constraints like shorter structural fillergap distance, but more costly in terms of longer linear filler-gap distance and its non-canonical VOS order. As such, structural and linear based processing demands work in opposite directions to both favor and disfavor subject RC processing. For children with DLD having limitations in working memory, resolving such competing processing demands becomes particularly challenging. Forty children (22 DLD; 18 Typically Developing (TD); age-matched) aged between 6;6–9;7 were tested using a sentence repetition task (Frizelle & Fletcher, 2014) to assess production of a wide range of relativized positions: Subject(S), Agent(A), Patient(P), Indirect Object(IO), Oblique(OBL) and Genitive(GEN). Additionally, we compared two relativisation strategies: classifier RCs versus RCs with the RC marker ge3. Our first results confirmed that RCs are indeed vulnerable in Cantonese children with DLD. The DLD group was significantly worse than their TD age peers in producing all structure types, except P- RCs. S-RCs were the easiest and GEN- RCs the most challenging to produce for both DLD and TD groups. Unlike English, German and Mandarin, there was lack of a robust Agent over Patient advantage. Ongoing analyses will compare the relativisation strategies and the error patterns in DLD versus TD groups.We discuss our findings by considering how language specific properties affect acquisition outcomes in Cantonese-speaking children with and without DLD. Comparing the two relativisation strategies also allows testing certain structurally oriented vs. processing-based accounts of DLD/SLI, e.g. the Computational Grammatical Complexity account (Van der Lely, 2005) vs. the limited processing capacity account (Montgomery & Evans, 2009).