Poster F73, Tuesday, March 28, 8:00 – 10:00 am, Pacific Concourse
Behavioral and Neural Evidence for the Effects of Verb Bias and Syntactic Surprisal on Sentence Processing
Kathryn Bousquet1, Tamara Swaab1, Debra Long1; 1University of California, Davis
Verb bias, the frequency with which a verb appears in a given syntactic structure, can facilitate syntactic parsing. Previous behavioral studies have reported increased processing costs when verb bias is incongruent with the syntactically preferred structure of a sentence, such that syntactically complex structures (sentential clauses; SC) – are easier to read than simple structures (direct objects; DO) when the verb occurs more frequently with the complex SC structure (Wilson & Garnsey, 2009). However, these studies have assumed that the strength of a verb’s bias is constant across contexts. In the current study, we manipulated DO/SC sentences to be temporarily ambiguous (e.g. the goalie confirmed/confessed the defeat with real heartbreak; the goalie confessed/confirmed the defeat was really heartbreaking) and modeled syntactic surprisal with the Roark parser (Roark, 2001) to examine the behavioral and neural effects of verb bias at the disambiguating region. In a self-paced reading experiment, we found significantly slower reading times in the disambiguating region when verb bias was incongruent with the syntactic structure. In addition, there was a significant effect of syntactic surprisal. In an auditory ERP experiment, we found a P600 to the critical disambiguating word when verb bias was incongruent for both DO and SC structures. Together, these results suggest that the initial syntactic parse is driven by verb bias rather than syntactic complexity. Furthermore, the syntactic surprisal effect suggests that verb bias changes incrementally with sentence context, and that sentence processing is influenced by the overall verb bias as well as a context-specific bias.
Topic Area: LANGUAGE: Syntax