That-trace effects and resumption – How Improper Movement can be repaired

The ungrammaticality that can be observed in subject extraction from that-clauses in English, known as the that-trace effect, has attracted much attention in generative grammar. Until today, most of the writing takes it for granted that the effect is directly connected to the placement or the role of the subject. In this article we will show that this is likely to be wrong. We will show on the basis of data from German that the that-trace effect emerges as the result of TOPIC EXTRACTION via the specifier of the complementizer that (SpecCP). Since subjects are very likely to serve as topics, and the grammar of English does not allow much variation in using other constituents as topics, it is understandable that the that-trace effect was taken to be a subject effect. The gist of our explanation of the that-trace effect is that constituents which have been moved to the topic position, more precisely to the position of an aboutness-topic, do not have a feature for contrastiveness, and that precisely such a feature is needed in long extraction to value a subfeature of the intermediate complementizer. 1 This requirement allows essentially only constituents from the rhematic part of the clause to move long distance.