This thesis investigates a particular word order phenomenon in German, the occurrence of discontinuous NPs (which we refer to as the NP-PP split construction) in order to probe the division of labor between the syntactic analysis and lexical as well as discourse constraints on this construction. We argue that many of the factors which previous literature has tried to explain in terms of syntactic restrictions on movement are in fact derivable from discourse factors.
Building on the empirical characteristics of the NP-PP split construction investigated in part one of the thesis, in part two we investigate two main questions, one concerned with the proper syntactic analysis of the construction, the other with the integration of such a syntactic analysis with an approach to the lexical and contextual factors involved.
Starting with the syntactic analysis of NP-PP split, we reevaluate the empirical basis of the choice between the two syntactic analysis ideas proposed in the literature, extraction and reanalysis, and provide an explicit theory licensing reanalysis-like structures for NP-PP split within the framework of Head-Driven Phrase Structure Grammar (HPSG). Regarding the lexical-semantic effects observable for the NP-PP split construction, we show how an account based on the generative lexicon can be integrated into our syntactic HPSG analysis. Finally, in order to find an explanation for the context effects that show up with the construction, we explore the possible focus-background structures of NP-PP split constructions. The empirical insights from this investigation of discourse requirements are then integrated into our account of NP-PP split by developing an information-structure component for our HPSG fragment of German. Since this information structure component accounts for some of the restrictions traditionally viewed as part of syntax, this thesis can be understood as arguing for a more equal division of labor between syntax theory and a theory of discourse, which has been unevenly balanced as a result of the significant advances in syntactic theorizing in the generative enterprise.