Chapman, C. 1 , Hackl, M. 2 & Kucerova, I. 1
1 McMaster University
2 Massachusetts Institute of Technology
Previous psycholinguistic work on filler-gap dependencies demonstrates that the left-to-right incremental parser is sensitive to the syntactic dependency holding between a wh-filler and its gap position. However, little work has investigated how the parser might resolve constructions in which a phrase must be interpreted in a distinct structural position (i.e., in its logical form, or LF, position) from where it appears on the surface. The interpretation of three different types of DPs (namely, anaphors, pronouns and proper names) provide a tool to investigate LF in real-time. Using self-paced reading, we examined how these DPs are processed in sentences where the anaphor/pronoun linearly preceded its antecedent. Results suggest that the parser searches for an antecedent as soon as it finds an unbound anaphor (Principle A) but that no such search occurs for pronouns (Principle B). Greater processing difficulty is also incurred when names are first introduced compared to pronouns, which can be explained by current models of variable binding: pronouns can enter the derivation with an index whereas an index needs to be created for names to serve as binders. We propose a processing model which predicts when processing difficulty will arise based on the current semantic theories.