Silent interactions: Spanish TP-ellipsis and the theory of island repair

Abstract

This paper centers on the nature of Spanish TP-ellipsis (e.g., Juan fue al cine y María también , lit. ‘Juan went to the cinema and Mary also’), paying special attention to its behavior in contexts of long extraction of the remnant of the elliptical site. The long extraction data provide further evidence that TP-ellipsis in Spanish behaves as Clitic Left Dislocation. In particular, TP-ellipsis is sensitive to adjunct and relative islands. These facts lead us to the question of why other cases of TP-ellipsis, namely, sluicing, can indeed repair islands. I show that an approach such as the one in Merchant (Linguistics and Philosophy 27: 661–738, 2004, Variable island repair under ellipsis, Cambridge University Press, 2008) is not able to derive this pattern and propose an alternative solution. Assuming that ellipsis and copy deletion form a natural class (Chomsky, A minimalist program for linguistic theory, MIT Press, 1993, Chomsky, The minimalist program, MIT Press, 1995), I argue that the presence or absence of island repair effects is derived as a matter of identity. Basically, a phrasal copy is elided under the same mechanism that applies in normal cases of ellipsis. A copy that is elided in a given syntactic cycle is not computed for the identity condition on ellipsis in a later cycle. By contrast, if a copy cannot be elided in a syntactic cycle, it has to be computed for identity reasons at the ellipsis cycle. Under the assumption that indefinites and traces of wh -phrases are identical for the purposes of identity, it follows that ellipsis under wh -sluicing can feed elision of the wh -copies in the elliptical gap when there is an explicit indefinite in the correlate. This is never the case with Spanish TP-ellipsis because the copies of a left dislocated remnant never have an identical correlate in the antecedent.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 77 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.