Some (but not all) movement types systematically violate islands
Dominique Sportiche
December 2020

Because some movement dependencies - e.g. question formation - obey islands constraints, it is customarily assumed that all movement dependencies obey islands. This article concludes that this generalization is incorrect: using island independent movement diagnostic tools (namely reconstructability = displaced interpretation) to diagnose movement, it demonstrates that some movement dependencies at least, namely French DP-CL(itic) L(eft) D(islocation), systematically violate (strong) islands. This conclusion requires modifying Phase Theory to preserve successive cyclicity but let islands be in principle permeable to movement, and to distinguish non island violating movement such as question formation, relativization from island violating movement (DP-CLLD) in a principled way. To allow movement, the proposals initiated by Rizzi (1997) that phasal peripheries be finely structured are exploited so that phase edges are taken to be large enough to include several phase peripheral positions, including, systematically, Topic positions allowing multiple hosts. Access to these positions, thus licensing island violations, is limited to phrases overtly marked as topical by a doubling (resumptive) clitic. Such a view may be extended to allow a different conceptualization of successive cyclic movement than is customary, in terms of a succession of topicalizations and focalizations.
Format: [ pdf ]
Reference: lingbuzz/005607
(please use that when you cite this article)
Published in:
keywords: movement, island, phase theory, successive cyclicity, resumption, clitic left dislocation, french, syntax
Downloaded:157 times


[ edit this article | back to article list ]