A feature-checking analysis of Japanese scrambling

This paper explains several unique aspects of scrambling. Using the scrambling feature (Σ-feature) proposed by Grewendorf & Sabel (1999), I discuss (a) what derives the difference between A-scrambling and DP-movement, (b) why scrambling shows radical reconstruction properties, while an operator–variable relation is established in wh -questions, and (c) why we find several types of scrambling. I propose a Feature Interpretation Principle, which states that a feature checking relationship established in derivation is preserved at LF. I show that the Feature Interpretation Principle, together with the multiple feature-checking parameter of Ura (1996) and the nature of heads, explains most of the unique properties of scrambling.