Meaning merger: Pragmatic inference, defaults, and compositionality

Abstract

Kecskes: Linguistic underspecification of utterance content is widely accepted across different frameworks, including the neo-Gricean approaches (cf. Horn 2005; Levinson 2000) and relevance theory (Carston 2002, 2005; Sperber & Wilson 1986, 1995). There is also an agreement that if linguistic underdeterminacy is given, pragmatic inference is required if a hearer is to recover a speaker's meaning successfully. In your Default Semantics, you reject the idea of underdetermined semantic representation, and offer an alternative approach in which semantic representation is established with the help of intentions in communication. This means that intentions “intrude” into the semantic representation, and the semantic and pragmatic components are interwoven. What are the advantages of this one-level semantics as opposed to the modular view? It can be argued that, in a way, Default Semantics is also a modular approach because intentions can be considered preverbal thoughts generated in the “conceptualizer” and linguistically shaped in the “formulator” using Levelt's terminology (Levelt 1989, 1999). Do you agree with this line of thinking?