There have been two fundamentally different theories of long-distance scrambling in languages like Japanese and Korean: the movement-based theory, according to which long-distance scrambling is an operation that induces filler-gap dependency, and the linearization-based theory, according to which long-distance scrambling alters the linear order of expressions without altering the syntactic structure involved. In this article, the results of some questionnaire studies regarding long-distance scrambling in Japanese are presented, and it is argued that they provide some reasons to favor the linearization-based theory over the movement-based theory. The findings presented here indicate that there are at least two distinct kinds of dislocation in natural language syntax and that an analysis in terms of complex domain formation involving partial compaction is an adequate description of one of them.
(S. Yatabe, "Evidence for the linearization-based theory of long-distance scrambling in Japanese," in Linda Uyechi and Lian Hee Wee, eds., Reality Exploration and Discovery: Pattern Interaction in Language & Life, CSLI, Stanford, 2009, pp. 271-286.)