Towards Automatic Detection of Applicable Diatheses

The valency behavior (argument structure) of lexical items is so varied that it cannot be described by general rules and must be captured in lexicons separately for each lexical item. For verbs, lexicons typically describe only unmarked usage—the active form—while natural languages allow for certain regular changes in the number, type and/or realization of complementations (e.g. passivization). Thanks to their regularity, such changes may be described in a separate rule component of the lexicon; however, they are typically seen in many but not all verbs and their applicability to a given lexical unit (verb meaning) is not predictable from its valency alone. In this paper, we describe our initial experiments with using a large morphologically annotated corpus of Czech for determining which diatheses are applicable to a given lexical unit.