We report the results of an experimental and a corpus study on free relative constructions with the case ambiguous German wh-pronoun 'was' ('what'). Case conflicts in German free relative constructions lead to reduced acceptability, though not necessarily unacceptability. The aim of this study is to clarify whether the case conflict is only a surface phenomenon, or whether a conflict in the abstract case features only is sufficient to yield reduced acceptability/frequency. The results support the former point of view.

A second result of our corpus study must be explained in terms of abstract case, however: The overall frequency of the most problematic conflicting case pattern is unexpectedly low. This can be interpreted as an effect of syntactic markedness on processes of speech planning.

The results of a further experiment about conflicting case requirements on the objects of coordinated transitive verbs confirm the position defended in this paper: here, we have a truly syntactic case conflict and no degraded acceptability is observed. Case conflicts can only be resolved if they are morphological, not if they are syntactic.