Paper short abstract:

A self-repair is one of disfluency which inevitably occurs in real-time utterance production. In this presentation, I will introduce the idea of classifying self-repairs in Japanese monologue into five types. Almost 6,000 examples of self-repairs were extracted from Corpus of Spontaneous Japanese.

Paper long abstract:

In spontaneous speech, the production of utterance is largely affected by real-time constraints. A real-time constraint is one that is imposed on the speaker in that the speaker "must conduct the linear and real-time production of linguistic form that has well-formed syntactic structure" (cf. Levelt 1989). In such situations, the fluency of utterances can be lost, and various disfluencies emerge, such as long pauses, filled pauses, pronunciation errors, self-repairs, cut-offs, inversions, and insertions.

Such disfluencies inevitably occur in real utterance production. However, in actuality, even if a disfluency is included in a certain utterance, in most cases, this does not cause the listener to have difficulties in understanding. This fact suggests the existence of a meta-linguistic strategy that is shared among the participants.

In this presentation, I will introduce the idea of classifying self-repairs in Japanese monologue into five types; "Articulatory repairs (R1)," "Repetition (R2)," "Lexical/syntactic repairs (R3)," "Additional repairs (R4)," and "Paraphrase (R5)". R1 is assigned to mispronunciation and its immediate repair. R2 is assigned to simple repetition of words or phrases. R3 is assigned to the selection error of lexical items and its repair. R4 is assigned to insufficient information and its supplement. And R5 is assigned to an intended paraphrase of words or phrases. Almost 6,000 examples of self-repairs were manually extracted from the Corpus of Spontaneous Japanese, and classified into five types. These are classifications based on morphological differences and the functions they perform within discourse.

As long as disfluencies inevitably emerge as a result of real-time constraints, both the speaker and the listener consider their emergence a precondition, and it can be thought that they manage speech by preparing a strategy to adequately process these disfluencies. If certain regularities and patterns can be recognized in such a strategy, this composes an important part for the description of "speech grammar." In other words, by categorizing disfluencies into several types and analyzing/describing their morphological/functional characteristics, this can contribute to the clarification of a mechanism that supports the management of speech communication.