On Monday 10/20, Dave Kush (Haskins Laboratories) will be visiting the psycholinguistics Proseminar (LINGUIST712), where he will talk about some of his recent work on the processing of bound-variable anaphora. He will give his talk at 2:30pm, in ILC458. The talk is entitled "Relational Constraints and Antecedent Retrieval: Evidence from Bound-Variable Pronouns.” Abstract follows.
All are welcome!
Interpreting a pronoun during incremental sentence processing typically requires identifying and accessing potential antecedents from memory. According to the grammar, there are two kinds of constraints that determine whether a previously-seen NP can serve as an antecedent for a pronoun: (i) morphological constraints, which enforce feature-match relations between the NP and the pronoun, and (ii) syntactic constraints, which determine the NP's eligibility based on its relative structural position to the pronoun.
In this talk I first discuss reasons to think that cue-based models of antecedent retrieval should be able to implement morphological, but not relational, constraints. I then present results from 2-3 eye tracking experiments that indicate that antecedent retrieval displays unexpected sensitivity to the c-command/scope constraint on bound-variable anaphora. In light of these findings I explore methods for achieving this apparent relational sensitivity within the confines of a cue-based architecture.