Published online by Cambridge University Press: 07 April 2016
While recent studies suggest children can use cross-situational information to learn words, these studies involved minimal referential ambiguity, and the cross-situational evidence overwhelmingly favored a single referent for each word. Here we asked whether 2·5-year-olds could identify a noun's referent when the scene and cross-situational evidence were more ambiguous. Children saw four trials in which a novel word occurred with four novel objects; only one object consistently co-occurred with the word across trials. The frequency of distracter objects varied across conditions. When all distracter referents occurred only once (no-competition), children successfully identified the noun's referent. When a high-probability competitor referent occurred on three trials, children identified the target referent if the competitor was absent on the third trial (short-competition) but not if it was present until the fourth trial (long-competition). This suggests that although 2·5-year-olds’ cross-situational learning scales up to more ambiguous scenes, it is disrupted by high-probability competitor referents.
This research was supported by grants from the University of California Merced Graduate Research Council to John Bunce and Rose M. Scott, and by a grant from the Hellman Fellows Fund to Rose M. Scott. We thank Anne Warlaumont for helpful feedback on the manuscript, the staff of the UC Merced Center for Early Cognition and Language for their assistance with data collection, and the parents, children, and students who participated in this research.