Treebank Data and Query Tools for Rare Syntactic Constructions

This paper reports on the use of the treebank query tool TüNDRA, which is able to process large treebanks of the size needed for rare syntactic constructions such as the Zwischenstellung of finite auxiliaries in German subordinate clauses. The TüPP-D/Z, an automatically annotated treebank with 11.5M sentences, contains a total of 92 examples of this construction, which need to be hand-filtered from corpus queries that produce a significant number of false positives. The corpus findings about the Zwischenstellung shed new light on the usage of this construction in contemporary German that contradict previous claims put forth in the linguistics literature.