Verbal Valency in the MT Between Related Languages

The paper analyzes the differences in verbal valency frames between two related Slavic languages, Czech and Russian, with regard to their role in a machine translation system. The valency differences are a frequent source of translation errors. The results presented in the paper show that the number of substantially different valency frames is relatively low and that a bilingual valency dictionary containing only the differing valency frames can be used in an MT system in order to achieve a high precision of the translation of verbal valency.