Merge and Move as costly operations

This article discusses the question whether the operation Merge ( = external merge) is inherently more economical than the operation Move ( = internal merge) and therefore blocks the application of the latter operation, as claimed by Chomsky (1995:226). A positive answer to this question implies that the derivation takes a numeration (or a lexical array) as its point of departure — if Merge is always preferred to Move, and the computational system has free access to the lexicon, movement would never apply, because any feature could be checked by merging it with some element taken from the lexicon directly. However, if Merge does not automatically block Move, it may be the case that the notion of a numeration is superfluous and can therefore be eliminated. Of course, we do need some criterion in order to decide which derivations are in competition, but it might be the case that this is simply a matter of meaning, as suggested by Grimshaw (1997), who claims that only semantically equivalent structures are part of the reference set ( = candidate set in OT terminology). We will argue that the answer to the question whether Merge is inherentl