Both the electronic and printed dictionaries contain examples as a part of dictionary entries. The examples illustrate a usage of a word (or phrase) and are very popular with foreign language learners. Therefore, the quality of these examples inevitably affects the quality of the whole dictionary. In modern dictionaries these examples are usually chosen from text corpora, which represent the real (not artificially created) usage of particular expressions. The aim of this thesis is to design and implement an algorithm, which automates (or at least simplifies) the retrieval of suitable examples illustrating a given word or phrase. The output of this thesis should be a system helpful to lexicographers for creating new dictionaries based on text corpora. It is advisable to incorporate it into an existing corpus query system.