Optimization via syntactic amalgam: Syntax-prosody mismatch and copula doubling*

Many linguists have observed the emergence of the nonstandard English construction instantiated by the following sentence: “Now the problem is // is that nobody’s going to invade anybody else’s boundaries”. In this pattern (which we will refer to as ISIS), a clausal complement is preceded by two finite forms of the copula, the first of which is typically prosodically prominent and followed by a major intonational break. While Massam (1999), among others, views ISIS as a variant of Pseudocleft, we see two problems with this approach. First, there are distributional and discourse-pragmatic properties that distinguish ISIS from Pseudocleft, including the referential status of the subject NP and the topic-focus articulation of the clause. We will argue that ISIS, rather than being an instance of the Pseudocleft pattern, is a syntactic amalgam that is closely related to an appositive pattern that we will refer to as Hypotactic Apposition, e. g., That’s the real problem is that you never really know. Second, the Massam analysis fails to explain why a speaker would select ISIS over a simpler and more compositional alternative construction, which we will refer to as Simplex: The problem is // there’s nothing else to buy. Using prosodically labeled data from the Switchboard corpus, we show that this choice involves optimization: Simplex has prosodic defects that ISIS repairs. In Simplex tokens the copula is typically followed by a break, creating misalignment of prosodic and syntactic phrases (Croft 1995; Watson and Gibson 2003); it is also typically prosodically prominent, although function words otherwise receive prominence only by deflection of accent from a discourse-old complement (Ladd 1995). While the Simplex copula performs double duty (as focus marker and as VP head) ISIS allocates these functions to the two respective copulas. Nevertheless, ISIS is far rarer than Simplex. If we view ISIS as a repair strategy, rather than a mere overgeneralization of the Pseudocleft pattern, this fact makes sense. Corpus Linguistics and Linguistic Theory 1 1 (2005), 45 88 1613-7027/05/0001 0045 Walter de Gruyter 46 J. M. Brenier and L. A. Michaelis