Item

Abstract

In my doctoral dissertation, I demonstrate that, upon encountering disfluencies such as uh and uhm, listeners may predict that the speaker will refer to a relatively more complex object. This suggests that listeners draw inferences about what caused the speaker’s production difficulty. The effect of speaker identity on these inferences remains debated: do listeners draw inferences based on what the listener himself would find difficult to refer to (egocentric account), or what the listener assumes the particular speaker at hand would find difficult (perspective-taking account)? In this talk, I will examine the studies that have investigated how knowledge about the speaker’s identity affects the processing of disfluency, describe some of my own work in this area, and propose a novel way of discriminating between the two accounts.