Statistical Approach in Relation to the Study of Compatibility

The extraction of collocations is one of the urgent tasks in the processing natural language, the results of which are important and in demand in various fields of applied linguistics. Our study compares three statistical measures carried out on a subset of The English Trends Corpus. The corpus has a huge methodological potential, so its use in foreign language lessons will allow students to cope with the solution of many educational and scientific problems, significantly expand their understanding of language as a changing phenomenon. Automatically extracted verb collocations in which the semantically main component was the search query word take were considered. During experiments studied the differences between statistical metrics, and also an attempt was made to find the most effective of them. The Pearson rank correlation coefficient between the measures was calculated. The results obtained show that some measures show similarity and relative interchangeability.