Vocabulary size prediction of Croatian texts

Preliminary research on the vocabulary size of Croatian lexical corpora shows that the distribution of types is regular and that deviations of the calculated values are within theoretically acceptable limits. The research also brought us to conclusion that Zipf´s law in Croatian language is not applicable because the lexical density is different, i.e. the proportion of types and tokens in different languages is different and the parameters of that proportion need to be calculated for every language separately.