Learning verbs under referential uncertainty: The role of referential and syntactic contexts.

摘要

Previous research on early verb learning has focused largely on how children acquire verbs in highly informative contexts. However, children often hear verbs in a variety of less informative circumstances. They hear verbs in the presence of multiple potential referents, and they hear verbs in the absence of relevant scenes. This dissertation investigated whether children can learn anything about verbs under these circumstances of referential uncertainty. The first two sets of experiments (Chapters 2 and 3) examined situations in which children heard a novel verb in a sentence that was consistent with more than one referent within a scene. Experiments 1 and 2 (Chapter 2) showed that 31-month-olds can encode information about the set of potential referents for a novel verb under these circumstances. They can then integrate this information with additional scene information, using cross-situational consistency to identify the verb's referent. Experiments 3 and 4 (Chapter 3) showed that 28-month-olds can also identify a verb's referent by integrating information across multiple sentence structures. Thus, children who heard a novel verb used in the causal alternation realized that the verb had to describe an action on an object that produced an effect (e.g., caused-motion). The final set of experiments (Chapter 4) focused on situations in which children hear a verb in the absence of a relevant referential. 28- month-olds acquired useful combinatorial information about a novel verb simply by listening to its use in sentences and later retrieved this information to map the verb to an appropriate referent.;Taken together, these studies show that children can form lexical entries for verbs, even when they are uncertain about the verb's referent. They can then attach to this entry facts about the set of potential referents for the verb, the sentences structures in which the verb occurs, and facts about the verb's arguments. Children identify a verb's referent by integrating information across different referential scenes, integrating information across sentence structures, and integrating previous sentential information with later referential information.