Ever since the appearance of Hornstein(1999)'s movement theory of control which reduces control to raising, control for or against movement theory has been one of hot and disputatious topics in recent linguistic studies. In this study I put focus on investigating two opposite approaches: one is based on control under movement and the other one is concerned with control against movement. The former is a purely syntactic account that involves movement, originally proposed by Hornstein(1999, 2000) and developed by Boeckx & Hornstein(2003), and the latter is based on the two different theories such as the standard theory of Landau(2003) that allows PRO and Control Module, and the semantic account of Culicover & Jackendoff (2001) which says that the thematic structure has a quite precise effect on the choice of controller. Despite the desirable unification achieved by Hornstein(1999, 2000)'s movement theory of control, I contend that some semantic account should be at least in part responsible for the proper analysis of control on a par with Culicover & Jackendoff(2001)'s proposal. In this regard, I wish to show that only a syntactic analysis is not enough to capture the generalization of control structures and some consideration of semantic aspects is crucial. |