dc.description | Shatz (1978a) has proposed an initial constraint on children's interpretation of speech acts such that they tend to respond with action to speech addressed to them. A number of methodological improvements over the original test of this Action Bias Hypothesis were included in the present study, most notably the use of a robot to present stimulus sentences. Twenty-eight children were videotaped interacting with their mothers and with the robot in a laboratory playroom at 18 and 24 months of age. Subjects' responses to sentences whose lexical content was systematically manipulated indicated that response mode varied with lexical content; that is, as a group, children tended to act when sentences contained action verbs, and tended to produce informing responses when sentences contained copula verbs. This pattern was found at both ages, and argues against an initial bias to act. Rather, young children are sensitive to linguistic and contextual markers of a speaker's communicative intent. Further evidence of their sensitivity to communicative intent was demonstrated in a series of correlational analyses relating maternal speech characteristics to individual differences in children's response strategies. Maternal use of a specific form-function pair--yes/no prescriptives--was a reliable predictor of response style, even when other predictors were held constant statistically. Thus, children do not begin language learning with a bias to associate speech with action, but are sensitive to the functions for which specific forms are used in their communicative environments. | en_US |