Language Processing in Conversation
Dr. Sarah Brown-Schmidt *****NOTE---DIFFERENT DAY/TIME*****
Thursday, January 13, 2005, 12:00pm - 07:00pm
University of Rochester, Department of Brain and Cognitive Sciences
My research focuses on the mechanisms by which people produce and understand utterances during the most basic form of language use: interactive conversation. My research combines on-line methods that allow me to monitor production and comprehension on a millisecond scale, with tasks in which naive participants engage in unscripted interactive conversation. By examining interactive conversation, I can explore fundamental questions about how speakers formulate utterances and how speakers and listeners coordinate and communicate their goals and intentions. By using real-time measures, I can test specific hypotheses and observe the rapid processes involved in generating and interpreting utterances.
In first part of this talk, I will describe two large-scale experiments which I used to establish the feasibility of examining real-time language processes during conversation. The first experiment focused on interpretation of referring expressions like the horizontal red block.� I successfully monitored the incremental interpretation of these expressions and moreover, I made a startling discovery-that listeners easily understand underspecified expressions, like the red block, uttered in the context of multiple red blocks. Interlocutors' use of similar referential domains supported the use and interpretation of these less-specific expressions. The second experiment investigated how message-level information is planned during a conversation, focusing on the production of expressions like the big peach.� Analysis of pre-speech eye-movements indicated speakers can update messages on-line as they encounter new relevant information, and can use disfluency to buy extra planning time.
Finally, I will present third piece of work which builds on these results, focusing on how interlocutors use and maintain a representation of what information is mutually known to both partners during a conversation. The insights we gain from this, and related experiments will have important consequences for memory and language representations, will indicate whether language processes are encapsulated from social/cognitive processes, and will determine whether results from standard experiments extend to the most basic form of language use-conversation.
Note: If you would like to receive email announcements about the colloquium series, please contact the Business Office to have your name added to our announce lists at firstname.lastname@example.org.