The German Infinitival Passive: A Case for Oblique Functional Controllers?

Following the standard LFG assumption that a functional controller --- unlike an anaphoric controller --- must bear a term GF and must be present at f-structure, one must assume that the German Infinitival Passive Construction (IPC) involves Anaphoric Control, at least for subject equi verbs. However, an Anaphoric Control analysis of the IPC with equi verbs that select a dative object fails to account for the availability of split antecedents since under Anaphoric Control split antecedents should be possible, but they are not. Instead, I pursue a Functional Control analysis of the German IPC --- thus accounting for the availability of split antecedents. Equi verbs which do not license the IPC, namely accusative object equi, can be analyzed via Anaphoric Control since they prohibit split antecedents. For subject equi verbs, then, the Functional Control analysis of the IPC will require two modifications to be made to LFG's standard approach to Functional Control, both of which I will claim are independently motivated. First, one must allow a non-term argument (namely OBLagent) to be a functional controller; something LFG has previously rejected. Second, the implicit OBLagent argument of passives must be represented at f-structure since this is, I claim, a functional controller. Evidence from binding facts suggests this is required anyway. Finally, allowing non-term functional controllers actually permits an alternative account of Visser's Generalization which also captures its (partial) non-application in German.