"Discourse features", syntactic displacement and the status of contrast

Contrast and its syntactic correlates, various contrastive focus and contrastive topic movements, are investigated from the perspective of a hypothesis constraining the set of formal features active in the computational system (C_HL). I propose a Strong Modularity Hypothesis for Discourse Features, according to which no discourse notion can be encoded by formal features. In contrast to currently prevalent cartographic approaches, it claims that only truth-conditional notions may constitute formal features active in the C_HL. Movements corresponding to non-truth-conditional notions, such as notions of information structure, must thus be interface phenomena, rather than driven by a feature-checking mechanism. To test this hypothesis, the paper investigates (a) the so-called contrastive focus movement, well-known from Hungarian, involving exhaustive identification, and (b) a distinct class of widely attested contrast-related movements - contrastive topic and contrastive focus movements - that involve a closed set whose members are explicit in the context, and have no entailment of exhaustivity. The distinct types of discourse-related, and in particular contrast-related, movements analyzed are argued to be due, respectively, to (a) an independent quantificational operator of the C_HL, such as the truth-conditional maximality operator motivated for Hungarian, or (b) interface effects, such as accommodation of nuclear stress assignment or facilitation of the mapping of syntactic representations to information structure.