The present work adopts a derivational, incremental, phase-based theory of syntax, with the elementary operation Merge at its center, as it has been developed by Chomsky and others within the minimalist program. Against this background, the main goal of this monograph is to develop an approach to the syntax of German that also envisages another primitive operation Remove that is a complete mirror image of Merge: Whereas Merge brings about structure building (both in the form of basic phrase structure generation, and in the form of movement), Remove leads to an elimination of structure. Merge and Remove obey the same constraints, among them the Strict Cycle Condition.

This short chapter focusses on general concepts involved in Remove-based analyses: conflicting structure assignments, short life cycle effects, and displacement without movement. In addition, it illustrates how the new approach may ultimately reconcile the seemingly incompatible trends towards simpler syntactic representations, on the one hand, and highly complex cartographic representations, on the other.

In this chapter, I pursue two main goals. First, I argue for a new empirical generalization: An external argument in German passive constructions is accessible from positions below it but inaccessible from positions above it. The evidence for downward accessibility comes from control into adjunct clauses, secondary predicates, and complement clauses, binding of reflexives and reciprocals, and disjoint reference effects. In contrast, the evidence for upward inaccessibility comes from long-distance binding in impersonal passives and standard passives, accessible subjects for control infinitives, criterial movement constraints, minimality of movement effects, and intervention for anaphoric binding. Second, I present a new theory of passivization from which this generalization can be derived: The elementary operation Remove accounts for both accessibility and inaccessibility of external arguments in the passive in German, by correctly predicting a short life cycle. After this, the chapter addresses the question of how variation in the area of passivization can be accounted for in the new model. Next, there is a brief extension of the analysis to adjectival passives, invoking external Remove. The chapter concludes with a discussion of alternative approaches that either maintain strict accessibility or postulate strict inaccessibility, as well of hybrid approaches.

This chapter introduces the operation Remove. The starting point is the question of how to account for conflicting structure assignments in syntax. After excluding the standard means of syntactic movement for certain cases, several predecessors and alternatives of Remove are discussed (among them tree pruning, S-bar deletion, and exfoliation). In addition, the concept of coanalysis is critically evaluated. The core of the chapter is devoted to introducing Remove as an elementary operation that is the complete mirror image of Merge in that it triggers structure removal rather than structure building, and that it obeys exactly the same restrictions (with respect to triggers, strict cyclicity, etc.). On this basis, the different effects that Remove has for removal of phrases versus removal of heads are illustrated. Some general consequences are discussed next, concerning short life cycle effects, incompatibilities with other constraints (in particular, this holds for the Projection Principle), and semantic interpretation.