Comments on Czech Morphological Tagset

In the area of natural language processing the appropriate morphological annotation is necessary. In this paper we offer some comments on the Czech morphological tagset as used in the analyzer Majka that has been developed in NLP Centre (CZPJ) both for academic and commercial purposes relatively recently. We try to argue that the existing approach to the morphological annotation is not in agreement with the language reality and that the used solutions are motivated rather technically than theoretically. We think that it makes sense to consider some changes of the presently applied annotation principles that might, if applied, to improve the annotation accuracy.