On the difficult gaining of the corpus data for the phonetic research

The paper presents the corpus Monolog, its applications and also its possibilities and restrictions. This corpus was built as a source for analyses of pronunciation in Czech mass media and its mission is to fill partly the gap in the field of the Czech spoken corpora. The paper also shows the results of two analyses concerning the standard pronunciation of Czech: 1) the occurrence of glottal stops in sequences of non-syllabic prepositions + words with vowels in the initial position; 2) the pronunciation of consonant groups sh (i.e. in words like shánění, shoda)