Why the Binding Theory Doesn't Apply at LF

Abstract

This article argues that the relegation of the binding theory to the C-I interface (LF) is theoretically undesirable and empirically unwarranted. Recent Minimalist research has sought to eliminate the binding theory from UG by reducing its conditions to narrow-syntactic operations (Hornstein 2000, 2006; Reuland 2001, 2006; Kayne 2002; Zwart 2002, 2006; Hicks 2006). This approach remains controversial since the canonical Minimalist binding theory (Chomsky 1993; Chomsky and Lasnik 1993) views the binding conditions as interpretive rules applying at LF, supported by evidence that Condition A interacts with other interpretive phenomena assumed to be determined at LF (Lebeaux 1998; Fox and Nissenbaum 2004). While the interaction of anaphor binding and scope relations in particular is not disputed, I show that it is attributable to factors outside the binding theory, namely the requirement that variables (including anaphors) must be c-commanded by their binders at LF. Deprived of its strongest empirical argument, the LF binding theory can then be picked apart.