On Reducing Control to Movement

Hornstein (1999, 2001) proposes a particular version of the movement theory of control (MTC) in which movement is driven by the need of a DP to “receive” a θ-feature from the predicate it merges with. I argue here against the use of θ-features in syntax on various grounds, the most important of which is that, being fundamentally semantic in nature, they belong properly to the bare output conditions of the C-I interface. Assuming the most recent version of the minimalist program (Chomsky 2000, 2001a, 2001b), I propose instead a version of MTC closer in spirit to that of Bowers (1973[1986], 1981), based on the idea that Merge is driven by the need to satisfy interpretable or uninterpretable c-selection features. I then show that given standard minimalist constraints on the operation of Agree, the MTC is not only possible but necessary. After discussing and refuting two important arguments that have been advanced against the possibility of movement to a θ-position, I conclude by showing that the proposed version of the MTC is only compatible with a purely derivational theory of syntax in which interpretation is immediate and there is no intermediate level of LF.