Overextensions in early language comprehension: evidence from a signal detection approach

Abstract

ABSTRACTTwo studies of children's early language comprehension were performed using the signal detection paradigm introduced by Thomas et al. (1981). Experiment 1 replicated the original findings of Thomas et al., demonstrating that 13-month-olds understand a word that a parent reported they would understand and did not understand a word that a parent reported they would not. In Experiment 2, the procedure was modified to test for overextensions in early language comprehension. On half of the trials, children saw an array of objects including a referent for a ‘known word’. On the other half, the array included an object perceptually similar to the referent of the known word, but which itself was not an appropriate referent (e.g. an apple instead of an orange). Though the children demonstrated an understanding of the known word, they also overextended that word to the inappropriate referent. Mechanisms for such overextensions are discussed, as is the general relationship between comprehension and production.