Responses to Cantonese A-not-A questions by Cantonese-English bilingual children

Aims and objectives/purpose/research questions: This study investigates mismatched responses to Cantonese A-not-A questions produced by Cantonese-English bilingual children. Mismatches are hypothesized to result from overgeneralization and cross-linguistic influence. We examine if there are effects of language dominance on children’s production of mismatches. Design/methodology/approach: This corpus-based study uses data from eight Cantonese L1 children and eight Cantonese-English bilingual children from the Hong Kong Cantonese Child Language Corpus and Hong Kong Bilingual Child Language Corpus, respectively. Data and analysis: Children’s production was sampled at 3-month intervals from age 2;0 to 3;0. Responses to A-not-A questions were classified into (1) target response, (2) mismatch, or (3) others. Mixed-effects logistic regression models were used to identify the effect of different variables. Findings/Conclusions: Two types of mismatches are observed: copular mismatch with hai6/m4hai6 ‘be’/‘be not’ and existential mismatch with jau5/mou5 ‘have’/‘have not’. Copular mismatches are attributed to cross-linguistic influence and existential mismatches to overgeneralization. No consistent effect of language dominance is found. Originality: Although mismatched responses to A-not-A questions were noticed in two previous studies, this is the first study which investigates the reasons and factors behind this phenomenon. Significance/implications: This study provides evidence supporting Hulk & Müller’s hypothesis from a novel domain. It demonstrates that cross-linguistic influence can occur from a weaker language to a dominant language in a vulnerable domain. It also shows how syntactic and semantic models help predict non-target forms in bilingual children’s production.