Conceptual arguments for ellipsis as null exponence
Craig Sailor, Trinity College Dublin
Wed, 3/29 · 4:30 pm—6:00 pm · via Zoom
Program in Linguistics
While many take ellipsis to involve “deletion at PF”, surprisingly few attempts have been made to identify the actual mechanism(s) responsible for its characteristic silence. I pursue this matter assuming a strictly modular architecture, which turns out to significantly constrain the hypothesis space. I argue that phonological deletion accounts should be excluded on theoretical grounds (they are inherently anti-modular: the phonology cannot interpret syntactic features such as [E]) as well as on empirical grounds (they make the wrong predictions: ellipsis can disrupt allomorph selection, it can salvage ineffability, etc.). The remaining alternative for elliptical silence to arise post-syntactically, i.e. “at PF”, requires it to take effect at Vocabulary Insertion (VI), an increasingly popular position.
In the second half of the talk, I argue in support of this conclusion, but against its prior implementations: under strict modularity, the syntax cannot be in the business of telling a post- (i.e., non-) syntactic operation when or how to do its job, as existing non-insertion proposals assume. Thus, what is needed is a strictly modular theory in which elliptical silence arises due to VI’s successful application, rather than its failed application—in other words, one in which VI inserts silence. I sketch such a theory here, adopting the assumption that VI can target non-terminal nodes, overwriting previously-inserted exponents. The result is a theory in which ellipsis is, essentially, a dramatic case of portmanteau suppletion.
Craig Sailor is a research fellow at Trinity College Dublin within the Centre for Language & Communication Sciences. His research deals with the question of how syntactic structures get externalized, considering issues such as modularity, grammatical timing, and the division of labor between syntax and phonology. He is especially interested in ellipsis — the principled non-externalization of a syntactic constituent — as it provides a unique testing ground for competing theories of the syntax-phonology tradeoff.