This paper is mainly concerned with problems arising from the extraction phenomena in English. Recently, Bouma, Malouf and Sag (BMS (2001) have proposed the so-called Dependent Realization Constraint to allow a lexical head to be realized as either a local dependent or a nonlocal dependent. Though this constraint has theoretical applications in that it enables us to provide a unified account of extraction dependencies such as subject, complement and adverb extraction, it appears not to have been subjected to full empirical testing. Specifically, there are at least two problems which the unified account by BMS (2001) should solve in order to be seen as a desirable approach to extraction dependencies. Hence, this paper shows which applications can be problematic for this approach and suggests that in those cases, a new constraint-based analysis be used instead.

• 조세연 | Sae Youn Cho