1 Variation in long-distance dependencies Ankelien Schippers & Jack Hoeksema, Rijksuniversiteit Groningen 1 1. Introduction This article concerns the variation and change in long-distance (LD) dependencies in Dutch. Both historical and contemporary corpus data dealing with these constructions is discussed. Recently, it has been argued that LD-movement constructions are formed based on a fixed formula (cf. Dbrowska 2004, 2008; Verhagen 2005, 2006). The reason for assuming this is that corpus data demonstrate that LD wh-questions show very limited variation in the domain of the matrix clause, and hence seem to conform to a general template. In this paper, we present a number of counterarguments against such an analysis. The outline of this article is as follows. First, the four types of LD-movement constructions that are central to this paper are treated. Next, the data discussed in Dbrowska (2004, 2008) and Verhagen (2005, 2006) is presented as well as some of the main claims these authors put forward. Subsequently, we present our own corpus data, which we argue forms evidence against the analogy analysis of LD-movement constructions. We point out some of the factors we believe cause the limited variation in the matrix clause of LD wh-questions, and round off with a general conclusion. 2. LD-movement constructions LD-movement has been at the heart of generative grammar over the past few decades. Traditionally, four types of constructions are considered to involve this kind of A’-movement, namely wh-questions, relatives, topicalization constructions and comparatives (cf. Chomsky, 1977). These constructions are illustrated in (1) – (4), respectively (1) Wh-movement [ CP Who do you think [ CP John will kiss t who ]] (2) Relativization [ CP That is the girl who I think [ CP John will kiss t who ]] (3) Topicalization [ CP The girl I think [ CP John will kiss t the girl ]] (4) Comparatives [ CP John has kissed more girls [ CP than OP I think Peter did t OP ]] Especially within generative frameworks, LD-movement is seen as a productive rule in which an element is moved from a subordinate clause into a higher clause. For example, in (1), the wh-phrase who, which is the object of the subordinate verb, has moved to the left periphery of the matrix clause. The reason for treating these constructions as one and the same is that they behave alike in many respects. In all cases, movement leaves behind a gap, and proceeds in intermediate steps. Furthermore, all four constructions are sensitive to the same kind of interveners. 1 Corresponding author: Ankelien Schippers (a.schippers@rug.nl)