Doing the right thing.

In this paper we examine a phenomenon which has proved difficult to account for using standard assumptions of current syntactic theory, namely extraposition. Starting with Rosenbaum (1967), all analyses we are aware of (with the exception of Emonds (1976)) treat extraposition as adjunction of a phrase to some right peripheral projection of a matrix sentence. This adjunction may either be analyzed as base generated (Koster (1978), Culicover & Rochemont (1990), Webelhuth (1989)) or derived by A’-movement (Reinhart (1980), (1983), Baltin, (1982), (1983), (1984), Müller (1994)). Several proposals have been made wrt. the attachment site of extraposed clauses and the trigger of extraposition (cf. Stowell (1981), von Stechow & Sternefeld (1988), Kiss (1993), Truckenbrodt (1994)). Extraposition as movement seems to contradict many of the well-established principles of generative grammar: While A’-movement to the left is unbound, extraposition is far more local. Only leftward movement must respect NP-islands extraposition may violate them. The base generation approaches can do away with these