Variations in the amount and nature of early language to which children are exposed have been linked to their subsequent ability (e.g. Huttenlocher, Haight, Bryk, Seltzer & Lyons, iggi; Hart & Risley, 1995). In three computational simulations, we explore how differences in linguistic experience can explain differences in word learning ability due to changes in the development of semantic category structure. More specifically, we manipulate the amount of language input, sentential complexity, and the frequency distribution of words within categories. In each of these simulations, improvements in category structure, are tightly correlated with subsequent improvements in word learning ability even when the nature of the input remains the same over time. These simulations suggest that variation in early language environments may result in differences in lexical proficiency by altering underlying cognitive abilities like categorization.

Comprehending language describing typical events, such as going to a baseball game or playing in the snow, involves activating general knowledge of the type of event described. Research has demonstrated that event knowledge guides the generation of expectations for upcoming words as sentences are processed in real time, word-by-word. This dissertation begins by examining the activation of concepts that are related to the described event but are not expected to appear in the language (i.e., linguistically unexpected concepts). In an event-related brain potential (ERP) experiment, participants read short stories that contained unexpected words related or unrelated to the described event. Unexpected words related to the described event elicited a reduced N400 ERP component relative to unexpected words unrelated to the described event. This result indicates that event knowledge activation during comprehension is not limited to only this concepts expected to appear in the unfolding sentence. A subsequent ERP experiment utilized visual-half field presentation of critical words to examine asymmetries across the cerebral hemispheres in the activation of linguistically unexpected event knowledge elements. Unexpected words related to the described event elicited a reduced N400 with left visual field (right hemisphere) but not right visual field (left hemisphere) presentation, suggesting a crucial role of the right hemisphere in activating unexpected event knowledge elements. A third experiment investigated the temporal dynamics of event knowledge activation by monitoring eye movements over arrays of images as participants listened to short stories. When a story first established an event context, unmentioned images depicting concepts related to the described event immediately attracted visual attention. As the story elaborated on the event, specifically highlighting one of the images related to the event, comprehenders shifted visual attention to this image and also to a distractor image that had not drawn visual attention when the event context was first established. This finding suggests that comprehenders considered the potential relevance of each image to the unfolding event description at each point in time. Together, these studies advance our understanding of how event knowledge is engaged during language comprehension.