In minimalist theory of syntax, it is a standard assumption that movement is triggered by feature checking between the attractor (or checker) and the attracted (or moved/checked). Under this assumption, however, the motivation for the intermediate steps in interclausal movements, especially in the so-called "long" wh-movement, does not seem to be very clear. In order to account for the successive cyclicity of wh-movement, Chomsky (2000, 2001a, b) and others assume that the head of every phase-inducing category has some uninterpretable feature, which triggers successive cyclic movement. On the other hand, Boskovie (2002) and others propose that successive cyclicity has no direct relation with feature checking, but that the requirement for intermediate steps in interclausal movement is due to some Subjacency-like locality condition, e.g., the Minimal Chain Condition of Chomsky and Lasnik (1993). Neither of these two approaches, however, is successful in accounting for the successive cyclicity of wh-movement. The feature-based approach of Chomsky (2000, 2001a, b) and others assumes some unmotivated feature, e.g., [uwh]; the constraint-based approach of Bodkovid (2002) and others must assume "look-ahead," which is to be avoided by the phase-by-phase model of the minimalist framework. We suggest two possible directions to pursue: one with top-down structure-building, and the other with Greed-like properties of the extracted element itself, laying more emphasis on the latter approach.

• 박승혁 | Sung Hyuk Park