Festschrifts

The COMP-trace phenomenon has long resisted easy explanation or even consensus as to whether it is ultimately due to the syntax or something else. Two analyses are examined here, one based on a principle of syntax (Anti-locality) and the other based on a principle of sentence planning (Principle of End Weight; PEW). Three cases are presented in which the Anti-locality analysis predicts that a COMP-trace effect will arise, while the PEW analysis does not. In all three cases, involving inversion in matrix clauses, inversion in embedded clauses, and clauses headed by prepositional complementizers, the COMP-trace effect does seem to occur, suggesting that the Anti-locality analysis is correct. This result is compared to earlier evidence that suggested that the PEW analysis was empirically superior, and a new way of understanding these results is proposed that is compatible with the evidence presented here in favor of Anti-locality.