Context and language. Contextual blindness in scalar implicatures computation (ConLan)

Scalar implicature is the mechanism whereby scalar sentences, like existential Some Greeks are philosophers, are interpretatively strengthened via negation of the corresponding stronger alternative, in this case universal All Greeks are philosophers. Scalar implicature was initially explained as part of a conversationalist framework but was soon to receive alternative treatments, including structuralist accounts, mainly motivated by such surprising properties as embeddability. Recently, a grammatical approach has emerged in the literature (Chierchia et al. 2012) that breaks with some crucial assumptions of the conversationalist account and regards scalar implicature as a purely semantic device, part of the spontaneous «logicality» of language. This project launches an investigation into the theoretical foundation and the empirical properties of the contextual blindness principle, assumed in certain versions of the grammatical approach (Magri 2009).

The contextual blindness principle has been submitted to account for the possibility that a scalar implicature is generated even when a contextual contradiction is obtained as a result of the generation. For example, a sentence like Some Greeks come from a warm country can be shown to be interpretatively strengthened, via negation of the alternative sentence All Greeks come from a warm country, even if the two sentences are equivalent given the piece of contextual knowledge that all Greeks come from the same country. In order to account for the observed negation, then, it is concluded that the mechanism must be blind to, i.e. must operate independently from, that specific piece of information. The contextual blindness principle has been subject to much criticism, most of which is pragmatically motivated. For one, it has been argued that the principle falls short in its ability to provide a comprehensive treatment of scalar implicature, for there is no obvious way to extend the reasoning sketched above to other cases of inferences widely regarded as scalar implicature, such as those involving Hirschberg scales. In addition, it has been noted that the principle might be empirically inadequate even to account for the standard cases involving positive quantifiers, for purely logical relations are insufficient to establish the necessary asymmetric entailment relation between the sentences.

This project, which partly builds on previous results of The Strength of Scalar Inference project (SSI), aims at understanding whether a blindness-based response to such criticism would be empirically tenable and could be satisfactorily integrated within a standard theory of the semantics/pragmatics interface. To begin with, an experimental probe will be launched to establish whether a scalar implicature can also be generated when crucial pieces of information, e.g. concerning the emptiness of restrictors or the ordering of alternatives, are missing from the context, i.e. when contextual knowledge entails that they do not hold. Then, the interaction of scalar implicature with other linguistic devices, e.g. presuppositions and exclusives, will be analyzed from a theoretical and formal point of view, so as to provide an evaluation on the pervasiveness of blind interpretive mechanisms within the system of the natural languages.