Collocation Extract : A Tool for Extracting Collocation

Wirote Aroonmanakun · The Journal of English Studies · 2005

Collocation knowledge is necessary for language learners if they want to produce native-like languages. Collocation Extract is a free software tool for searching and ranking word chunks that could be collocations. Three basic statistical methods, namely log-likelihood, mutual information, and chi-square, are used for weighting cllocation candidates. Since these statistics can be used to measure associations only between two elements, pseudobigram transformation is used to estimate statistical values of word chunks with three or more words. This software can also be used to extract technical terms when applied to a corpus of specific domain.

Read the paper · More papers on PaperTik