Language, Action and Thought

People have clear and spontaneous intuitions about whether a given sentence is acceptable within a language. ‘What did Peter eat ravioli with?’ is an acceptable English sentence but ‘What did Peter eat ravioli and?’ is not. These intuitions form the empirical basis for almost all linguistic theory, because they are assumed to provide a window on grammar and grammaticality. Here I describe how acceptability intuitions are at root not intuitions about grammar as such, but rather possible relevance. Precisely, when a speaker treats a sentence as acceptable, what they are effectively revealing is that there is some possible intended meaning for which the sentence could be an optimally relevant use of the linguistic items—commonly called ‘constructions’—that are common knowledge in the population. Being grammatical typically minimises cognitive effort and hence helps to optimise relevance, but this is not what intuitions of acceptability are ultimately about. This analysis explains several distinctive and otherwise unusual features of acceptability intuitions in a unified way, and breaks longstanding debates between formalist and functionalist approaches to grammar. More broadly, I show how treating relevance as foundational, and pushing that insight to its logical conclusions, can reshape our descriptions and explanations of basic linguistic phenomena. Grammars are the macro-accretion of many micro-moments of optimal relevance in communication.