Abstract
Adult production is influenced by the larger linguistic contexts in which words appear. Children, like adults, hear words in recurring linguistic contexts, but little is known on the effect of that context on their speech. We look at the production of irregular plurals in English (e.g., mice, feet) to argue that children attend to the larger phrases words appear in and make use of that distributional knowledge in production. We assess the role of linguistic context by comparing irregular plurals elicited with a general question (What are all these?) versus a lexically specific frequent frame (e.g., Three blind —). In study 1, 4;6 year-olds produced many more correct irregulars after lexically specific frequent frames (72%) than after a general question (32%). Corpus data on spontaneous speech offered analogous findings: Children did not overregularize irregular plurals after lexically-specific frames. In study 2, we confirm children's sensitivity to the relation between particular words and phrases: A familiar frame (So many) enhanced production (52%) but not as much as the lexically specific frames in Study 1. Children's word production is affected by the larger patterns words appear in. Consequently, studies of lexical and morphological acquisition need to take linguistic context into account when assessing children's abilities. This has implications for models of both lexical and morphological acquisition.
ACKNOWLEDGMENTS
We thank Kimberley Chu for her help in collecting and analyzing the data; the children at Bing for all their cooperation; Virginia Marchman for kindly making some additional data available to us; and Virginia Marchman and Nola Stephens for commenting on an earlier version of the paper.
Notes
1There are other measures of the relation between the noun and the frame that could influence production. In particular, how many different frames a noun can occur with. The effect of preceding linguistic context may be weaker for nouns that appear with many different frames (we thank an anonymous reviewer for this observation). We set out to look at the effect of noun flexibility (how many different frames a noun appears with) and frame predictability (how predictable is the lexically-specific frame given the particular noun) on our results. But both measures were highly correlated with important control variables (noun flexibility with noun frequency, and frame predictability with frame frequency), making it hard to discern what independent effect they had. We think these are important issues for future investigation using items that systematically vary along the relevant dimensions
2We thank an anonymous reviewer for this suggestion.