Well they had a couple of bats to be truthful: Well-prefaced, self-initiated repairs in managing relevant accuracy in interaction

This paper reports a conversation analytic study of well-prefaced, self-initiated repairs in talk-in-interaction. We show that speakers use well-prefacing of self-repairs to manage the credibility of claims in talk. Specifically, well-prefaced self-repairs attend to the relevant accuracy of a turn-so-far by revising it but without retracting it. For example, in the extract from which the title for this paper is taken, a speaker tells an interviewer that in ‘New Zealand the- they for millions of years had no mammals. The- they they only had really birds’. This turns out to be a slightly exaggerated claim, which the speaker self-repairs in the transition space with a well-prefaced statement – ‘Well they had .hhh a couple of batsto be (.) .hhh to be truthful But (.) they had no big mammals. No cats. No (.) dogs. No stoats’. Here, the additional information modifies the claim that there were no mammals (because there were bats) but also maintains the gist of what was said earlier (i.e. there were no large, predatory mammals). Our work has clear resonances with Drew's (2003) analysis of precision and exaggeration in interaction, though where he focussed on recipient-prompted revisions, we focus on self-initiation. Like Drew, we note that participants’ orientations to speaking precisely connect to matters of veracity and accountability.

► This paper is a CA study of well-prefaced self-repairs in interaction. ► Well-prefaced repairs are used to modify rather than retract the trouble source. ► In general, well-prefaced repairs occur as part of a three-part structure. ► We comment on the relationships between form and action, and repair and epistemics.