Semantics of noun/verb conversion in Czech: Lessons learned from corpus data annotation

The paper contributes to the study of conversion in Czech, which in the literature on Czech is rarely understood as an independent word-formation process and therefore not much attention has been paid to its semantics (only as part of research on derivation). An annotation experiment is presented in which semantic relations in 800 conversion pairs are assessed on the basis of corpus concordances by two human raters in parallel. The annotation experiment provides evidence of the applicability of semantic classification of conversion based on meaning paraphrases on large-scale linguistic data, showing that it is possible to reach reasonable inter-annotator agreement. The sources of disagreement, as well as the patterns of polysemy found in the data, provide information about the conceptual relationships between the semantic categories in conversion.