Title: Now you see it, now you don’t: What linguistic illusions tell us about human language processing

Presenter: Associate Professor in Linguistics, Department of English, William & Mary
cell.wm.edu/parker/

Abstract:
Our ability to use and understand language typically proceeds smoothly, implicating a sophisticated set of neurocognitive mechanisms for coordinating multiple pieces of linguistic information across different time scales. But those mechanisms are not infallible. Sometimes, we fail to accurately process even the simplest of grammatical relations like subject-verb agreement during communication, giving rise to so-called “linguistic illusions”. In this talk, I will show that by studying where these sorts of errors occur (and where they don’t), we can better understand the inner workings of the language processing system, much in the same way that optical illusions reveal the workings of the visual system. Drawing on insights from behavioral experiments and computational modeling, I advance two claims. First, I argue that linguistic illusions reflect errors in how we access information about a sentence from working memory, rather than the result of shallow processing or the use of superficial/good-enough representations, as some have suggested. Second, I argue that the observed behavioral profiles (e.g., differences in reading times and judgments) reflect not how quickly we retrieve this information from memory, as standardly assumed, but rather how quickly we can integrate the retrieved information back into the processing stream to guide interpretation. Together, these claims contribute to a more nuanced understanding of how we encode, access, and interpret linguistic information in real-time. To conclude, I will discuss recent work from our lab showing how differences across populations in susceptibility to linguistic illusions can inform broader discussions on race and education.

Bio:
Dan Parker is an Associate Professor of Linguistics at the College of William & Mary, where he has been since he received his Ph.D. from the University of Maryland in 2014. He is also Co-Director of the Computational & Experimental Linguistics Lab (CELL). His research combines theoretical, experimental, and computational methods to study how we encode and access linguistic information in working memory.