Meta-representation in utterance comprehension: The case of semantically incomplete expressions

•Utterance comprehension need not have meta-representational results.•Meta-representation may nevertheless specify background assumptions.•Meta-representations need not feature in the comprehension process but only support one or more of its steps.•Comprehension is contrasted with distinctively meta-representational cognitive tasks.

Grice and his followers all share the idea that utterance comprehension is a meta-representational task, i.e. a task in which the hearer relies on what the speaker believes, knows, intends and so on. In this paper, I criticise this view, particularly in the version defended by Sperber, according to which to understand an utterance is to perform an inferential process which has a meta-representation as its conclusion. Against Sperber, I argue that our intuitions as to “what we have understood” of a certain utterance are not distinctively meta-representational. I then examine the case of semantically incomplete expressions, and I defend the view that their comprehension may not rely on meta-representation as much as one may expect. Finally, I argue that the role of meta-representation in utterance comprehension has to be revised: the conclusion of the process is rarely to be expected as meta-representational, even though meta-representation may have a role in framing information about the utterance's background (what the speaker is talking about, what she refers to, etc.).