The time course for structuring spoken utterances

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

With National Science Foundation support, Dr. Zenzi Griffin will conduct three years of psycholinguistic research on the time course for structuring sentences in speech. Predictions come from two contrasting views of syntax. In one view, syntactic structures such as clauses and verb phrases are selected as chunks that guide subsequent word preparation. In the other, the structures result from a sequence of smaller decisions about what to mention next. This research will make use of the tendency of speakers to look at objects in a visual scene while deciding how to refer to each one. Preliminary results suggesting that indecision about what to mention next may be reflected in eye movements. The experiments will test language planning and production in monologues and dialogues in response to visual scenes and to written words. This research addresses how implicit knowledge of syntactic structure is used in speaking. The contrasting views imply different representations and processing mechanisms, which in turn affect conceptualization of how children learn to produce multiword utterances, how the ability is disrupted in some aphasias, and what forms of training can improve performance. Determining how humans use syntactic knowledge to produce sentences allows for more sophisticated comparisons between syntactic processes and other human abilities. In addition, relating eye movements to language production reveals interactions between very different cognitive systems, which may be used to improve language recognition systems in human-computer interaction. Finally, the project will train several undergraduate research assistants to use cognitive research methods.