Review of the book "Morfologie českého slovesa a tvoření deverbativ jako problém strojové analýzy češtiny"

Czech deverbatives are usually derived from verbs by means of mostly very regular patterns. The rules for deverbative derivations have been described by many linguists in many previous works. Nonetheless, those descriptions are not possible to convert directly into a system usable in the field of natural language processing (NLP). They are not detailed enough to cover the whole diversity of morphological alternations in the particular verb forms taken as the basis for the derivations. For automatic processing, it is necessary to present the description down to the slightest detail. Moreover, the rules have to be expressed within a formalism that ensures their exactness and unambiguity. The only widely used system utilizing the regularity of deverbative generation in NLP is the Prague morphological tool, in which the generation rules constitute an intrinsic part. However, the rules are very general, the result of which is a massive overgeneration. In morphological analysis, it does not matter as much, but in other fields of usage, it is naturally better to generate only those words that (may) really exist. Thus, the presented treating deverbatives with respect to their application in NLP is a meritorious achievement.