4.3 Extracting Specialized Vocabularies from the Three Business-Oriented
Vocabularies
In order to extract specialized vocabularies and to observe the characteristics of each one,
two statistical measures, LLR and MI, were applied to identify the specialized words in each
of the three corpora, using the methodology described in Chujo & Utiyama (2004) and
Utiyama et al. (2004). These measures identify words whose frequency is significantly higher
in a small text of interest, i.e., the BNC dialogues, Business Eigo, and the TOEIC tests, rather
than in a large reference corpus, i.e., the BNC HFWL mentioned earlier.