The Modeling Hypothesis and child bilingual codemixing

According to one explanation of child bilingual codemixing(the modeling hypothesis), bilingual children's rates of mixing are related to rates of mixing in the input addressed to them. An assumption of this hypothesisis that bilingual children are sensitive to codemixing in the input and that they can adjust their own rates on-line in accordance with the input. Despite its widespread appeal, evidence concerning its validity has been largely inconclusive. The assumption is largely noncontro versial in the case of older bilingual children, as evidenced by their adoption of the patterns of codemixing of the speech communities in which they live. However, it is not clear whether young bilingual children have the cognitive and linguistic capacities implicated by this assumption. The present study sought to examine this assumption directly. Six French-English bilingual children(average age 2;4 years) were recorded during play sessions with an assistant who engaged in relatively low(15%) or relatively high rates(40%) of mixing on three separate occasions. The results indicate that these children were sensitive to the language choices of their interlocutors and that they were able to adjust their rates of mixing accordingly; further, they appeared to do this by matching their language choice with that of their interlocutors on a turn-by-turn basis.