Compound vocabulary knowledge development in Mandarin-English bilingual children : a comparison with Monolingual English children

Our study investigated the processing of compound vocabulary of bilingual (BL) Mandarin-English children and their performance in comparison to monolingual (ML) English children. From this study, we sought to determine (a) how the BL children performed in Mandarin compared to English (b) how the BL children performed compared to the ML children, and (c) how background factors, such as language use and vocabulary size affect compound processing. We predicted that the BL children would show an advantage on compound processing tasks over the ML children due to the importance of compounding in word formation in Mandarin Chinese. In addition, we also used performance on picture vocabulary tasks as covariates to take into consideration potential differences in vocabulary size, as BL children often have a smaller vocabulary in each language because of distribution across languages.

Data were collected from 25 BL Mandarin-English children (between 40 to 104 months of age) who were matched within three months to 25 ML English children (between 40 to 105 months of age). Children participated in a compound analogy task, in which they produced novel compounds after a model; and a compound knowledge task, in which they explained real compounds. Comparing performance across languages, results showed that the BL children demonstrated higher performance in the dominant language (English) than in the nondominant language (Mandarin). The BL children were more likely to accurately produce novel compounds, but also more likely to make errors that involved the use of compounds. No significant difference was found in BL and ML performance on compound knowledge tasks. Significant relationships were found for some of the participant characteristics for both the BL and ML children and performance. In particular, age, picture identification, and picture naming performance were correlated with compound performance for the BL participants; performance on the picture identification task and compound processing tasks were correlated with each other for the ML participants. These findings provided limited support for our hypothesis. Future investigations should include BL participants who have a more balanced proficiency in both languages, as well as examine factors that were found to influence ease of compound processing.