Squaring the circle

Abstract

Making the distinction between semantics and pragmatics has proven to be a tricky task, leading to several problems that look like Gordian knots, or worse; perhaps semantics and pragmatics are so tangled that separating them is impossible, like squaring the circle. A widespread, plausible, Grice-inspired view of the distinction is threatened by what (Levinson Presumptive meanings. MIT Press/Bradford Books, Cambridge, Mass, 2000) called ‘Grice’s circle.’ Gricean inferences to derive the pragmatic content of the utterance (such as conversational implicatures) require the determination of what is said (also known as the ‘semantic content’ or the ‘literal truth-conditions’); but determining what is said (by processes of disambiguation, precisification, reference fixing, etc.) requires pragmatic inference. In a nutshell, pragmatic inference both requires and is required by the determination of what is said. Thus, there is no way to unravel semantics and pragmatics. In this paper, we will show how to square Grice’s circle. We untie the semantics/pragmatics knot, without using any of Alexander’s methods: slicing it with a sword or removing the (semantic) pin around which it was bound. The approach consists in assuming a minimal but truth-conditionally complete notion of semantic content (Perry Reference and reflexivity. CSLI Publications, Stanford, 2001), which doesn’t constitute what is said by the utterance, but does provide the required input for pragmatic reasoning.