Examining the origin of 'island effects' in complex sentence comprehension

Voiceover

Main Presenter(s)

Amber Mikkelsen

Abstract or Description

Within experimental syntax, there has been a dispute on whether innate grammatical rules or processing limitations are the source of island effects, a phenomenon in which sentences with certain types of linguistic relationships that span a complex structure (an island) are judged to be ungrammatical. This project seeks to compare how two different types of linguistic relationships are impacted by island structures to inform the debate regarding the source of island effects. One of these relationships, wh-dependencies, is subject to grammatical constraints, while the other, bound variable pronominal dependencies, is not. Both, however, are subject to similar processing costs, making this an ideal test case to adjudicate among theories attributing island effects to grammatical constraints and those attributing such effects to processing limitations. 128 sentences have been constructed by manipulating which of these two dependencies (wh-dependency vs. pronominal dependency) are used, as well as length of dependency (short vs. long) and presence of island (island vs. non-island). Participants will read a context and target sentence and give an acceptability judgement for the target. If a significant super-additive interaction is found in acceptability judgements for only for conditions that violate grammatical constraints, grammatical theories will be better supported. If both dependencies pattern similarly, processing theories may better account for the interaction as only similar processing limitations could contribute to a dip in acceptability for both grammatically licit and illicit stimuli. The results will refine our knowledge of the nature of grammar and its influence on sentence comprehension.