Morphological Disambiguation of Multiword Expressions and Its Impact on the Disambiguation of Their Environment in a Sentence

This study concerns the impact of the collocation/phraseme disambiguation component within the complex system of the rule-based morphological disambiguation of Czech. This system constitutes one of the two main disambiguation subsystems that are responsible for the morphological disambiguation of the corpora of synchronic Czech within the Czech National Corpus project. We will show that although the part of texts constituted by collocations/phrasemes (generally multiword expressions – MWEs) is relatively small and consequently the errorfree morphological disambiguation of MWEs covers only a small portion of textual material, such perfectly disambiguated fragments in sentences help to improve the disambiguation of the rest, non-MWE part of sentences.