Can the metrical structure of Italian motivate focus fronting?

This paper addresses the issue of the interplay between discourse-related properties, syntax, and prosody in Italian. How this interplay between prosody, syntax, and information structure is to be conceptualized is lively debated in the literature. The theoretical relevance of this issue is not marginal, because it gives rise to several core questions with regard to the architecture of the grammar. According to the cartographic approach (Rizzi 1997; Cinque and Rizzi 2008; and much related work), discourse-related properties are encoded as active features in the syntax, and focus fronting is an instance of feature-driven syntactic movement. Reinhart (1995, 2006) proposes a radically different view: focus and d-linking are encoded at PF (Phonetic Form). Building on Reinhart\u27s work, Szendr?i (2001, 2002) develops a model in which PF information is directly available at the conceptual-intentional interface. Under this approach, discourse-related phrasal movement is not feature driven, but takes place to repair potential mismatches at the PF-LF interface. Focus movement in Italian is analyzed as prosodically motivated by the need to align focus with main prominence and to destress given information (in the sense of Schwarzschild 1999). The key component of this analysis is that postfocal elements in Italian are extraprosodic and destressed. The aim of this paper is twofold. First, we want to address the issue of the metrical representation of postfocal and given constituents in Italian. Second, we want to ascertain on empirical grounds the prosodic assumptions underpinning the stress-based approach to focus movement. On the basis of a production experiment, we propose an analysis of the metrical structure of Italian according to which phrasing and head assignment apply exhaustively. We show that postfocal elements, though given, are assigned phrase-level metrical heads by virtue of default syntax-prosody mapping rules. Accordingly, we claim that Italian fails to destress given information, and that rightmostness of prosodic heads is violated when focus does not occurs in sentence-final position: postfocal constituents are neither extraprosodic nor destressed. These conclusions strongly undermine the stress-based approach. On the basis of a comprehension experiment, we further support the validity of our analysis showing that the distribution of phrase-level metrical heads and boundaries in postfocal contexts are used by listeners in sentence comprehension