An HPSG-Analysis for Free Relative Clauses in German

At the moment there is no theory for free relative clauses in German in the framework of Head-driven Phrase Structure Grammar (HPSG). From GB literature on the subject it is known that free relative clauses behave partly like noun phrases. They can fill argument positions of verbs. And although they are finite sentences, they are serialized like noun phrases in the German Mittelfeld. The function free relative clauses can take is not restricted to complements. Depending on the properties of the relative phrase, free relative clauses can be modifiers as well. I will argue that free relative clauses project to a category that is tightly related to the category of the relative phrase.As Ingria90 has shown, assignment of different cases in the relative and the matrix clause poses problems for grammars that rely on unification alone. In the following paper I will show that his subsumption based account is incompatible with standard assumptions in the HPSG framework. The set-based approach of DK97, which is similar in many respects to Ingria's approach, will also be discussed. It will be shown that some of the problems of the subsumption based account are still present in the set-based approach. I will provide a different solution to the problem that relies on an additional case feature for the case form of NPs. It is projected from the relative phrase and is not affected by case requirements of the verb.In general, there are three possibilities to describe the projections of free relative clauses: firstly, the direct projection of a phrase from the relative phrase and a finite sentence, secondly, an empty head or a unary projection that projects a relative clause and thirdly, a lexical rule that changes the subcategorization frames of governing heads in a way that they subcategorize for relative clauses. I will argue for the unary schema and discuss the alternatives.