Brian Dillon PhD, Assistant Professor at UMass-Amherst, will present a talk titled The ambiguity advantage and syntactic processing: Evidence from speeded acceptability judgments
"There remain many unresolved theoretical debates about the fundamental nature of syntactic comprehension. Some theories hold that syntactic analysis of linguistic input proceeds in a serial, depth first fashion (Frazier, 1979; Traxler et al., 1998; among many others), others hold that comprehenders analyze sentence structure in a breadth-first fashion, keeping active multiple distinct syntactic descriptions of the input in parallel (MacDonald et al., 1994; Levy, 2008; Logacev & Vasishth, 2012). One important empirical finding that has been taken as strong evidence for the serial view is the ambiguity advantage effect, the finding that ambiguous sentences are read more quickly than closely matched unambiguous counterparts. However, the theoretical interpretation of this finding has been challenged (Levy, 2008; Logacev & Vasishth, 2012). In this talk I present initial work in collaboration with Caroline Andrews (UMass) and Matt Wagers (UCSC) that uses speeded acceptability judgment methodology to investigate the ambiguity advantage effect. We replicate the critical ambiguity advantage effect in a judgment task context, showing that it obtains in experimental measures other than eye-tracking while reading. However, analyses of reaction times and computational models of participants' confidence ratings provide evidence against the prevailing interpretation of the ambiguity advantage effect. Instead, our initial results appear to be most compatible with models of the ambiguity advantage effect in which multiple syntactic representations of the input are maintained in parallel (Levy, 2008; Logacev & Vasishth, 2012)."
For more information on Dr. Dillon, visit http://www.umass.edu/linguistics/member/brian-dillon
All are welcome!