EN
The article discusses the problems connected with the homonymy of Czech deverbal surnames which must be solved during automatic annotation in electronic corpora. Solutions for improvement of these procedures, based on orthographic and punctuation criteria, on the list of corresponding surnames, and especially on syntactic features of surnames, are suggested in the paper. The degree of success of the steps proposed for disambiguation of the surnames will be very high; it can be estimated around 95 – 99 %.