LSLT: Rosa Lee (LING)

Although we will continue to offer a Zoom option for LSLT for now, we strongly encourage you to attend in person! The talk is only part of the point of LSLT: it's a good opportunity to meet and chat with students and faculty in other departments. This week we're reinstating the classic LSLT sandwich line (with a few improvements to keep things more sanitary).

Zoom link: https://go.umd.edu/lslt-zoom

Do we speak better than we understand?

Abstract: Previous findings from various measures (EEG, eye-tracking, etc.) suggest that comprehenders are not so good at using argument role information (i.e., who is the agent and who is the patient of the verb) to constrain the processing of subsequent verbs. That is, comprehenders seem to be, at least temporarily, blind to role-reversal errors, as in: …the cop that the thief arrested. In contrast, findings from production studies using the speeded cloze paradigm suggest that speakers are quite good at using the same information to generate role-appropriate continuations. In this talk, I will present some collaborative work digging into the divergence between comprehension and production and discuss the steps we are taking to narrow down the possibilities, in order to improve our understanding of how we use argument role information to guide our expectations in comprehension and production.