ABSTRACT:

Title:

Why Control is Movement and Why It Isn't

Coppe van Urk

Utrecht University

In recent Minimalist theorising, two influential proposals concerning the nature of control have emerged. The first of these derives control through movement (Hornstein 1999; Boeckx & Hornstein 2004, 2006a, 2006b), while the second retains the PRO-based analysis of G&B theory (Landau 2000, 2004, 2006, 2008). In this talk, I examine some cross-linguistic properties of control central to this debate. I conclude that there are convincing arguments for both theories. Effects such as backwards control, Case-sharing, non-distinctness and Shortest Move restrictions strongly suggest movement-derived control is at least an option in natural languages (Polinsky & Potsdam 2002, 2006; Alexiadou et al. to appear; Boeckx & Hornstein 2006b; Fujii 2006; Rooryck 2007). On the other hand, many languages allow Case independence, non-equivalence and Shortest Move violations in obligatory control configurations (Landau 2000, 2004, 2006, 2008; Bobaljik & Landau 2009). If these facts are to be taken seriously, either a sophisticated featural account is necessary (e.g. Landau 2000 et seq.) or it has to be concluded that both types of control can be licit. I argue here for the latter option, on the basis of strict, but predictable, correlations between different control properties and a number of empirical problems with Landau's account. I call this perspective the OCM/OCA Hypothesis.

OCM/OCA Hypothesis: A subset of control relations are derived through movement and a subset through antecedency.

It turns out that, under this view, the empirical complexity of control across languages falls out straightforwardly. I explore a formalisation of this hypothesis based on the assumption that the infinitival subject position is deficient for Person licensing (Bianchi 2003; Sigur∂sson 2008). In this system, PRO receives Person features in binding, while a lexical subject has to raise into a higher clause.