Intervention in tough-constructions revisited

In this paper, we subject to closer scrutiny one particularly influential recent argument in favour of the long-movement analysis of tough-constructions. Hartman (2011, 2012a, 2012b) discovered that experiencer PPs lead to ungrammaticality in tough-constructions, but not in expletive constructions. He attributes this ungrammaticality to defective intervention of A-movement, a movement step crucially postulated only in the long-movement analysis. He takes this as evidence that tough-constructions are derived via long movement. We make the novel observation that a PP intervention effect analogous to that in tough-constructions also arises in constructions that do not involve A-movement, namely pretty-predicate constructions and gapped degree phrases. Consequently, the intervention effect does not provide an argument for an A-movement step in tough-constructions or for the long-movement analysis, but rather for the base-generation analysis. We develop a uniform account of the intervention effects as a semantic-type mismatch. In particular, we propose that what unifies tough-constructions, pretty-predicate constructions, and gapped degree phrases is that they all have an embedded clause that is a null-operator structure. Introducing an experiencer PP into these constructions creates an irresolvable semantic-type mismatch. As such, we argue for a reassessment of what appears to be a syntactic locality constraint as an incompatibility in the semantic composition.