Submove: Towards a Unified Account of Scrambling and D-Linking

Abstract

In this paper we suggest a parallelism between D-linked Wh-phrases and long-distance scrambled elements, and argue for a specific syntactic step operative in both constructions. The parallelism resides in similar behavior of the two constructions with respect to Superiority, discourse effects, semantic vacuity, islands, and clitic doubling. We suggest that this parallelism can provide insight into the nature of both scrambling and D-linking. Our proposal views D-linking and scrambling as not related to agreement-checking, but as taking place purely for discourse purposes expressed in the left periphery of the clause. The absence of agreement-checking in the C-domain enables us to understand the 'peripheral' properties of these two seemingly unrelated constructions. Both target a position inside the C-domain without operator behaviour. The justification for two distinct projections comes from the necessity to motivate movement on the one hand, and the unavailability of the usual phi-features/agreement relations connected to movement on the other. 1 Introduction This paper explores a largely ignored, but we believe highly revealing, parallel behavior between D-linked Wh-phrases and scrambled elements (in particular long-distance scrambling as found in Japanese). The reasons we believe this parallelism can provide insight into the nature of both scrambling and D-linking are the following. Firstly, neither movement of D-linked Wh-phrases nor long-distance scrambling obey the Superiority Condition. Secondly, both operations carry obvious discourse effects. Thirdly, both have been argued to be semantically vacuous. Fourthly, both D-linked Wh-phrases and scrambled elements are not obviously sensitive to island effects. Finally, D-linked Wh-phrases are accompanied by clitic doubling in some languages, something that has also been noted for scrambling. We would like to claim that the parallel behavior just noted is the result of a similar process: the absence of phi-feature checking by the relevant moving element. If D-linked Wh-phrases move at all, it is for clause-typing or topicalization purposes, much like it is possible to front associates of expletives. We would also like to argue that viewing D-linked Wh-phrases as sitting in a different position, plausibly a topic projection, is the result of the absence of phi-feature checking. Such Wh-phrases have recently been argued to target TopP, within a more articulate Comp-domain à la Rizzi (1997). The obvious relation to topic rather than a focus/operator position can also be seen in instances of Chinese Wh-fronting. Essentially the same operation takes scrambled elements out of their phi-feature checking clause into a higher, non-agreement-related (topic) position. Cedric Boeckx and Kleanthes K. Grohmann 2 The role of agreement-checking or absence thereof in the C-domain enables us to understand the 'peripheral' properties of two seemingly unrelated constructions: (long-distance) scrambling and D-linked Wh-fronting. Both target—when moved—a position inside the C-domain without operator behaviour. The justification for two distinct projections, at least a Topic Phrase, comes from the necessity to motivate movement on the one hand, and the unavailability of the usual phi-features/agreement relations connected to movement on the other.