Young children's developing sensitivity to discourse continuity as a cue for inferring reference.

Children encounter many opportunities for word learning where a novel word (e.g., "chinchilla") coincides in time with the presence of its referent (e.g., a parent pointing at a fuzzy rodent). These two ingredients are not always paired simultaneously, but they sometimes still occur in succession within a discourse. We investigated children's ability to apply their knowledge of discourse structure to infer the referent of a novel word in the absence of social cues such as pointing and eye gaze. In Experiment 1A, we introduced 2- to 6-year-old children and adults to two novel toys and described each using two sentences. We embedded the introduction of a novel label ("Have you seen a toma before?") between the two sentences about one of the toys, with no cues implying the label's referent other than its position in the discourse. Children older than 3 years and adults were more likely to attribute the label to the toy whose descriptions surrounded the naming event. In Experiment 1B, we tested whether participants made their selections based on temporal associations-choosing the toy that was described closest in time to the naming event-rather than inferences about discourse. Participants heard the novel label introduced after the two descriptions of a toy rather than embedded between them. Both children and adults responded close to chance in this experiment, indicating that temporal proximity alone did not guide their selections. Together, these results suggest that children can use discourse position to make inferences about reference in word learning situations.