6-To-8 Year-Old Children Adjust Their Linguistic Generalizations Based on Distributional Cues

Generalizing the syntactic privileges of words based on limited exposure is a fundamental aspect of language acquisition, but how well children can use distributional cues to guide this process is not fully understood. Here we investigated whether 6- to 8-year-old children in the U.S. could use distributional cues to determine when to extend or restrict generalization in miniature artificial languages. In four experiments we systematically varied the number, density, and overlap of contexts available for nonsense words in the languages. When their contexts overlapped completely, children generalized broadly, extending the same syntactic privileges to all words that occurred in that set of contexts. When overlap was reduced, children were less likely to generalize syntactic privileges. When those gaps persisted across a larger input sample, they restricted generalization even further. Our results suggest that children are highly sensitive to the type and consistency of the distributional cues they receive and can adjust their generalizations accordingly.