Statistical Methods for Extracting Collocations from a Text Corpus

The article deals with statistical methods for extracting collocations on the example of the English Web 2020 corpus (enTenTen20). The authors considered three types of association measures: t-score, MI-score and Log Dice, which determine the strength of association between collocates based on the frequencies of their own and joint occurrence in the corpus. The importance of corpus linguistics technology in the study of various language aspects is especially emphasized.