Bare argument ellipsis and information structure

Abstract

This dissertation presents a cross-linguistic study of three elliptical predicate constructions: (a) stripping, (b) negative-contrast, and (c) yes/no ellipsis, which are all argued to fall under the scope of a more general type of ellipsis, Bare Argument Ellipsis. From an interpretive point of view, in all three constructions, the constituent that is present in the second conjunct ('the remnant') has a characteristic information role. In yes/no ellipsis, the remnant functions as a contrastive topic whereas in stripping and negative-contrast the remnant is a focused constituent. The latter two constructions are further differentiated with regard to the semantic characteristics of Focus. Based on the assumption that Focus is not uniform, it is shown that stripping involves narrow information focus whereas negative-contrast involves contrastive focus. From a syntactic point of view, I argue that Bare Argument Ellipsis involves overt movement of the remnant to the left periphery of the clause, followed by IP deletion. The PF-deletion approach is extended to all three constructions. Following Rizzi's (1997) split-CP hypothesis, it is proposed that the remnant in yes/no ellipsis moves to TopP, a functional projection in the left periphery of the clause that encodes contrastive topics, by the process of Clitic Left Dislocation. Contrastive topicalization of the remnant forces narrow focus on the polarity marker. Regarding stripping and negative-contrast, it is argued that the semantic difference between narrow information and contrastive focus is directly related to the focus projection that hosts the remnant. Following recent proposals that Focus should be split into several projections, I show that the remnant in negative-contrast ends up in F₁P, a focus projection marked for contrastiveness whereas the remnant in stripping moves to a lower F₂P, which simply encodes new information.