EN
This study presents new ways of solving old problems with inventarising the German '-bar' adjectives in the emerging Large German-Czech Dictionary: The authoress shows how contrastive analysis should be supplemented by a corpus-based view using new corpus-linguistic tools (Co-occurrence Analysis, Self-Organizing Maps, SOM). The high number of deverbatives is due to high productivity of this word-formation model. This causes problems with selecting an optimal number of dictionary entries. Results of contrastive analysis of German-Czech adjectival counterparts help to discern between easily formed Czech counterparts with '-telny' and more difficult equivalent types which require the lexicographer's special attention: s/he has to face many problems with equivalency of near synonyms (-bar/-lich). Usually, these are either degraded to absolute synonyms or omitted both in monolingual and bilingual works. A more adequate equivalency can be reached by translating collocations selected using cooccurrence analysis. However, a new corpus-linguistic tool (SOM) presents a topographic scheme which leads the lexicographer toward a conjunction of cooccurrence profiles which indicate typical discourse and cooccurrence partners.