MERGE-LABELING, Transfer, and CED Effects

Abstract

Since the postulation of Huang’s (1982) condition on extraction domain (CED), adjuncts and subjects have been standardly assumed to be opaque for extraction. However, the CED is empirically contradicted by cases where certain adjuncts and subjects are not islands to movement. Given this background, we propose that adjuncts in general are immediately transferred to resolve a problematic configuration {XP, YP} where labeling cannot be determined (i.e., problem of projection (POP)). Being sent to Transfer, extraction from adjuncts is barred. However, as uncovered by Truswell (2007), there are exceptional adjuncts in which extraction from them is occasionally permitted. In such cases, we claim that labeling via feature sharing can be achieved in the mode of restructuring or reanalysis (cf. Choe 1988; Wurmbrand 2001). Since labeling is properly determined, sub-extraction from them naturally follows. Concerning the typical subjects that ban movement from them, we argue that it is due to the labeling-wise defectiveness of T in English; movement from Spec,TP to Spec,CP is not allowed owing to the prior Transfer of typical subjects in Spec,TP position. With respect to certain exceptional subjects in which sub-extraction from them is licit, we recourse to the determination of labeling at their landing sites; thus, the island-hood of such subjects is voided. All in all, the present paper deduces CED effects within the system of MERGE through labeling and Transfer.