The authors present several language phenomena well known and described in English that remain unidentified by Czech translators due to the traditional description of language. The study demonstrates a restricted perception of grammatical categories in translations and shows the advantages of a corpus-driven approach.