Séminaire SynSem

Responsable : Andrés Soria-Ruiz (Universidad de Barcelona, LOGOS)
Titre : Factual, metalinguistic, and expressive usage of evaluative expressions. Diagnosing & modelling the distinction.

Abstract: In discussions about evaluative language (good, beautiful, fun), it is customary to (i) draw a distinction between factual and non-factual uses of these terms, and (ii) characterise non-factual uses by recourse to various mechanisms, such as metalinguistic negotiation, or relativist/expressivist semantics—often seen as competing accounts. In contrast to this, I want to push the idea that some non-factual uses of evaluative terms are purely metalinguistic, and some are purely expressive. I draw two consequences of this: first, the right semantics for evaluative terms should mobilize three independent parameters: a factual parameter (a set of possible worlds), a metalinguistic parameter (a set of interpretations), and some species of a relativist/expressivist parameter, such as e.g., hyperplans, judges, or normative standards. Second, the characteristic update that non-factual uses of evaluatives perform on the conversational common ground is best captured by a model combining metalinguistic and expressive information. In such uses, all factual information (i.e., a food’s flavor, a painting’s visual properties) is backgrounded, and an utterance containing an evaluative term may be used to convey metalinguistic information about that term’s usage, expressive information about interlocutors’ responses to the object under consideration, or a combination of both. This model, based on Barker (2002, 2013) and others’, can accommodate the intuition that many disputes involving evaluative terms are metalinguistic without renouncing the indispensability of expressive information.