Early verb representations

This dissertation investigates the lexico-syntactic representations that children form when learning new verbs. Three questions are asked: (1) how do children integrate multiple sources of evidence from the environment when determining the meanings and syntactic properties of novel verbs? (2) are representations formed in such a way that they can be immediately deployed by the online sentence comprehension system? and (3) can the way in which children learn verbs inform theoretical approaches to argument structure? Young children exploit several cues from the environment to help them learn verbs, and they do so rapidly, immediately mapping new words onto event categories. I present three experiments demonstrating that 3--5 year-old children use both the sentence a novel verb is presented in and the properties of the event highlighted by a preceding event to form a new verb representation. Children were presented with locative events, compounds of a pour -type and a fill-type event, and were tested to see whether they had isolated the manner or result as encoded by the novel verb. When the linguistic and event cues converged on the same interpretation, or when these cues were presented in isolation, children successfully picked out the correct component. The linguistic representations these children formed were also immediately accessible by the online parsing system, yielding anticipatory eye movements to the expected direct object of the verb during listening. When the cues pointed to different interpretations, children were at chance in choosing one, and showed no anticipatory eye movement behavior. These results demonstrate that children construct new verb representations by integrating multiple sources of evidence, and even one experience with a verb can yield a robust enough representation for immediate access to the parsing machinery. A review of theoretical approaches to argument structure concludes that although verb learning studies have been quite valuable in mapping out the evidential sources used by the learner and their developmental time course, verb learning studies like this one do not lend substantial support to particular theoretical claims about how argument structure is represented in the grammar. Future studies must attempt to close this gap between the disciplines