Crossover, Chain Formation, and Unambiguous Binding

Abstract

This paper addresses the question of what derives (strong) crossover effects. Investigating the behaviour of different movement types with respect to crossover (on the basis of data mainly from German, but also from Italian and English), I conclude that crossover effects show up with wh-movement, clitic (i.e., head) movement, scrambling, and topicalization, whereas they do not occur with A-movement (as in passive and raising constructions) and dative movement. Thus, the correct descriptive generalization seems to be that Case-driven movement does not induce crossover effects, whereas other movement types do. Given this state of affairs, it turns out that Rizzi’s (1986) analysis of crossover effects, which involves a local binding constraint on chain formation, is empirically too strong — it does not allow Case-driven movement across a co-indexed item. Moreover, the distribution of crossover effects and of improper movement is strikingly similar; improper movement, too, only occurs with movement types which are not Case-driven. This calls for a unified approach. The chain formation approach, however, cannot be generalized so as to cover improper movement. I will argue that the Principle of Unambiguous Binding (PUB), developed in Müller and Sternefeld (1993) as a constraint against improper movement, accounts for the distribution of crossover effects straightforwardly, if it is understood in a strictly representational sense.KeywordsHead MovementDative MovementMovement TypeLinguistic InquiryChain FormationThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.