In a local derivational (phase-based) approach to syntax, instances of resumption in German (long-distance) relativization constructions with an empty operator and a complementizer wo ('where') must be analyzed in terms of movement. Against this background, these constructions raise three problems for syntactic analysis: (i) a locality (backtracking) problem (“How can the information that a resumptive pronoun occupies the base position be made accessible on the moved item at later stages of the derivation, where it is required?”); (ii) a problem for movement theory (“How can movement in these resumption constructions circumvent an island?”); and (iii) a last resort problem (“Why does movement in these resumption constructions have to cross an island?”). I will propose a specific solution to problem (i) in terms of buffers that temporarily store minimal pieces of syntactic information; this approach will then be shown to automatically cover problems (ii) and (iii) as well.

The article investigates the conditions under which the w-relativizer was appears instead of the d-relativzer das in German relative clauses. Building on Wiese 2013, we argue that was constitutes the elsewhere case that applies when identification with the antecedent cannot be established by syntactic means via upward agreement with respect to phi-features. Corpuslinguistic results point to the conclusion that this is the case whenever there is no lexical nominal in the antecedent that, following Geach 1962 and Baker 2003, supplies a criterion of identity needed to establish sameness of reference between the antecedent and the relativizer.

In this paper I discuss the remnant movement construction in German with respect to two properties i) the nature of the unbound trace that is located inside the remnant i.e. either as an “adjunct-type” or as an “argument-type” trace (but see below for a refinement of this distinction) and ii) the landing site of its antecedent either in a low νP-internal position or in a higher position in or above TP. The data show that unbound traces are always licensed when the antecedent is in a low νP-internal position whereas only unbound argument traces are licensed in the higher position. It will be argued that the formal features of a DP may license the unbound copy of an argument in the remnant movement construction. No similar operation is available for unbound adjunct-type traces. Assuming a phasal approach, I will argue that no unbound copy at all is created in the case of νP-fronting with an antecedent in the low position thereby explaining why “unbound traces” are always licensed when the antecedent is in a low position.

This paper proposes a typology that classifies Bantu languages according to the function of so-called object markers in their grammar. Three types of language are distinguished: in Type 1, object markers are agreement markers; in Type 2, object markers are pronominal clitics; in Type 3, object marking is a reflex of A-bar movement of the corresponding object. The paper further demonstrates that the three types of language also behave in systematic ways with respect to the possibility of object marking in object relative clauses: in Type 1 languages, object markers are never required, in Type 2, object markers are never possible, and in Type 3, object marking is always compulsory in object relatives. Finally, the paper presents a syntactic analysis of object marking in Type 3 languages, which explains the obligatory occurrence of object markers in right dislocation and relative clause constructions in these languages.

The syntax of object-related anaphora in German has been subject to a controversy that includes the acceptability status of crucial example sentences, and the relevance of systematic empirical research for syntactic theory. In this paper, the relevant data and results are recapitulated. I show that some results from an experimental study by Featherston & Sternefeld (2003) are confounded by a problem first observed by Jackendoff (1992). Once this problem has been taken into account, it turns out that an observation about object related binding of reciprocals with the verb vorstellen ('to introduce'), thus far treated as exceptional, is the best kind of evidence to answer the question whether object-related anaphors obey the obliqueness hierarchy of case forms. To fully understand the patterns of German reflexivization, it is necessary to take into account patterns of complex reflexives formed with intensifiers like selbst . It is claimed that the intensifiers serve different purposes in different contexts. In the case of complex reflexives bound by dative objects, their interpretation is governed by pragmatic principles.

As shown earlier by Gärtner (2002), linked trees, the graphs used by Phrase-Linking Grammar (Peters & Ritchie 1981) to capture (unbounded) dependencies, can be cyclic under the special condition that two “displaced” constituents end up as sisters of each other. Such “PLG-loops” closely match the particular kind of crossing dependency familiar from Bach-Peters sentences. We will show how PLG-loops allow implementing Bach-Peters configurations in line with the movement-based approach to binding by Kayne (2002). The resulting structures correspond to QR-derived adjunction structures of the kind introduced by May (1985).