This dissertation proposes a novel analysis of German determiner sharing. Determiner sharing refers to non-constituent ellipsis constructions in which omission of a determiner or quantifier is parasitic on another ellipsis, commonly verbal gapping (McCawley 1993). Omission of a determiner is impossible without gapping. On the basis of novel German data, I propose an analysis in which determiner sharing is the result of a ''conspiracy'' of two independent processes, (clausal) ellipsis and split topicalization.
I first provide an empirical basis by presenting a description of determiner sharing in German which, to the best of my knowledge, is the first formal observation of determiner sharing in that language. Based on previous literature and three acceptability judgment studies, I formulate empirical generalizations that characterize this construction: (i) determiner sharing is dependent on ellipsis, (ii) the noun with the missing determiner must be the initial element in its conjunct, (iii) the shared elements need not form a constituent, and (iv) elements that occupy a low position in the nominal spine cannot be shared. Crucially, determiner sharing is not only possible in gapping, but also in stripping contexts. One of the contributions of this thesis is the detection of more contexts which allow determiner sharing.
I then argue that gapping in German must be analyzed as a clausal ellipsis, and that the remnants of gapping are part of an A'-movement dependency, based on evidence from e.g., island sensitivity, P-stranding, the impossibility of cross-conjunct binding, word order of particles, and case morphology on the noun with the missing determiner. Split topicalization is a type of movement that separates a noun phrase from other DP-internal elements, and moves it to the left periphery. I show that split topicalization and determiner sharing overlap significantly in A'-movement properties and the types of DP-elements that can be involved in both constructions.
A detailed step-by-step derivation illustrates how the simultaneous application of ellipsis and split topicalization can generate determiner sharing and account for the empirical generalizations. A move-and-delete approach to gapping can derive the parasitism of determiner sharing naturally: if a noun undergoes split topicalization and moves to a position higher than the elided phrase, its determiner will be left behind in the ellipsis site and consequently deleted. Omission of a determiner is therefore an accidental result of the joint application of split topicalization and (clausal) ellipsis. Viewing gapping as clausal ellipsis allows us to unify it with other constructions, such as stripping. This makes the theory of determiner sharing less complex: unifying gapping and stripping as clausal ellipsis allows us to apply the same analysis of sharing to seemingly distinct ellipsis phenomena. In this way, this thesis contributes to making the discussion of ellipsis phenomena less construction-specific.
I argue that the success of this ''conspiracy'' account of determiner sharing serves as an argument for movement-based approaches to ellipsis, and against in-situ ellipsis analyses, thereby contributing to an ongoing debate in the research of ellipsis (see e.g., Griffiths & Struckmeier 2021} and contributions in Günes & Lipták 2022).