ABSTRACT
Our proposal starts from the Overlay model of disfluency comprehension, which assumes that listeners identify a reparandum only after encountering a repair. The Overlay model explains why a reparandum lingers in a listener's representation of an utterance, but it does not capture anticipatory processing. We then develop a different model, based on a Noisy-Channel framework, which assumes a more active process for resolving repair disfluencies. On this model, the listener might predict a repair when the speaker becomes disfluent, or even identify a reparandum if a word seems inconsistent with the speaker's intention. We argue further that predictions during disfluency processing operate via the same mechanism involved in the processing of contrast, in which listeners use contrastive prominence to generate alternates. We conclude by arguing that the Overlay and Noisy-Channel approaches to disfluency processing are not inconsistent, but rather both explain important aspects of disfluency processing in complementary ways.
Disclosure statement
No potential conflict of interest was reported by the authors.