The role of focus and context in the derivation of scalar inferences

Abstract

This dissertation aims to make a contribution to the on-going discussion on what factors determine the derivation of Scalar Inferences (SIs). In Neo-Gricean pragmatics, SIs are inferences which arise due to the informative expectations a particular utterance creates (Vicente, 2015). Specifically, the use of a less-than-maximally informative lexical term (e.g. some) is taken to mean the negation of the more informative one (e.g. all). But, how does the hearer derive the SI? Is it provided by the grammar, is it instead provided by the inferential mechanism, or both? Is it always derived? Following the relevance-theoretic account of utterance interpretation and building on the ideas of Sperber & Wilson (1995), I examine the effect of focus on SIs, since it has been claimed to influence and even condition their derivation (cf. Van Kuppevelt, 1996; Van Rooij, 2002). I show that 1) focus is decisive for the derivation of cognitive effects in general (among which the SI may or may not be included), and that 2) contrastive focus has a stronger influence on SI derivation than non-contrastive focus, as it instructs the hearer to consider alternatives to the focused element, forcing him to consider and discard them. However, I will argue that there is no guarantee that the SI will be derived when the scalar term is in this position. In line with the predictions of Relevance Theory, and contrary to what Neo-Gricean and Grammatical Theories would predict, contextual relevance can be shown to be the key factor for SI derivation, the same as when deriving any other pragmatic inference. All these claims are consistent with experimental research (e.g. Zondervan, 2010; Chevallier et al., 2008) and with the predictions the currently popular Question Under Discussion theory makes. I provide examples in English, Spanish and Basque in order to show that these inferences are not driven by language-specific grammatical rules but by the inferential mechanism instead, universal across languages.