LingLunch : Sanghee J. Kim

Sanghee J. Kim (University of Chicago)
Knowing what's important during language comprehension

Be it a long congressional hearing, or a short casual conversation, one of the core aspects of language is that certain parts of an utterance are more central to the main point whereas other parts are less so. The current work focuses on this facet of language, the division of main (primary) vs. subordinate (secondary) discourse information. I investigate the way this distinction is represented in working memory during language comprehension through a series of behavioral experiments. Findings show that the working memory architecture is indeed sensitive to the main vs. subordinate distinction. But more importantly, they suggest that the division in working memory cannot simply be represented as a static, binary feature representation (e.g., [+/- main]), but instead it requires a dynamic representation of discourse. Using a question-based framework to represent discourse, I argue that taking the approach of incremental-update of “discourse questions” suitably explains it.