On not recognizing disfluencies in dialogue

This paper tests the hypothesis that listeners miss dis uencies or fail to transcribe them accurately because dis uencies interfere with the normal relationship between speech sound and linguistic context in human spoken word recognition. In a word-level gating experiment 16 listeners heard a total of 56 dis uent utterances selected from a corpus of spontaneous speech, 56 length-matched uent controls, and 56 uent foils. The proportion of words never recognized was greater in dis uent utterances than in controls. The failures clustered around the point where the dis uency interrupted the utterance, ocurring particularly within the reparanda, but were not found at corresponding locations in uninterrupted controls. Repetition dis uencies, where preand post-interruption portions might easily be construed together, allowed more successful word recognitions than recast dis uencies, where reconstruction of a single intended utterance would be di cult, if not impossible. The results have implications both for understanding human speech recognition and for improving the robustness of ASR systems.