John Hale, Associate Professor, Dept of Linguistics, Cornell University

 

John Hale, Associate Professor, Dept of Linguistics, Cornell University

Sage 4101

September 25, 2013 12:00 PM - 1:30 PM

Abstract:                       

Hearing speech or reading text, we put unconsciously fit words together into combinations – structures – that enable us to understand.  Since the 1970’s, psycholinguists have amassed evidence that this process is ‘incremental’ in the sense that it proceeds more or less in synchrony with the stream of words.  This talk presents models of incremental parsing based on Allen Newell’s problem space idea.  In this conception, steps of the comprehension process relate to individual grammar rules, and search control builds on previous experience.  Viewing parsing this way helps to reframe some classic conflicts in the theory of sentence processing:  top-down vs. bottom-up, parallel vs. serial, garden-path vs. ‘normal’.  Applying Newell and Rosenbloom’s chunking theory of learning to these sorts of models yields a simple explanation for ubiquitous surprise effects in human language processing.

 

What a Rational Parser Would Do B

What a Rational Parser Would Do A1

Surprise Effects in Language as a Consequence of Chunking

Add to calendar
Share|