Exposing and avoiding unwanted inferences in conversational interaction

Utterances give rise to many potential inferences. They can be communicated explicitly or implicitly, they may or may not be intended, and they may or may not even be inferred. In this paper, we focus on how speakers orient to ‘unwanted inferences’: potential inferences (or inferables) that can – but need not – be inferred from what has been said. Such inferences can be ‘unwanted’ in different ways: as secondary propositions that the speaker does not wish to be – or to have been – inferred, or as secondary propositions that the speaker does not wish to be held committed to having communicated. We illustrate the practices or methods by which speakers attempt to divert attention from them, offering a 6-part taxonomy that accounts for: (a) the source of the unwanted inference, and (b) the extent to which the inference in question is exposed or remains embedded in the conversational record. We then present some observations on how meanings of varying degrees of explicitness are drawn upon and negotiated by all parties, evidencing the range of meanings that are available in the minds of speakers that go beyond what the speaker is canonically taken to mean to communicate.