Proceedings of the Annual Meeting of the Cognitive Science Society

Disfluency leads listeners to expect an upcoming reference tounfamiliar objects. In two experiments, we examined if thisexpectation is adapted based on the way disfluency has beenused in the discourse. Participants listened to instructions tolook at an object on a screen containing familiar and novelimages. We manipulated the co-occurrence of disfluency andreference to novel vs. familiar objects. In the predictivecondition, disfluent expressions referred to novel objects, andfluent expressions referred to familiar objects. In the non-predictive condition, fluent and disfluent trials referred toeither familiar or novel objects. Participants’ gaze revealed thatlisteners more readily predicted familiar images for fluent trialsand novel images for disfluent trials in the predictive conditionthan in the non-predictive condition. Listeners adapted theirexpectations about upcoming words based on recentexperience with disfluency. Disfluency is not invariablyprocessed, but is a cue adapted within the local context.