earticle

영어

The purpose of this study is to find which statistical test shows better results in collocating patterns of Maritime English vocabulary, along with exploring optimal window spans. We use frequency, hypothesis testing such as log-likelihood, z-test, and t-test, and information theory-based statistical methods such as Mutual Information and Mutual Information 3, looking at some samples of all the significant collocates of the top several keywords extracted from the Maritime English corpus against the BNC Baby as a reference corpus. To find a proper ‘cut-off’ span position, we observe average frequency of collocates of near synonyms like maritime/marine,ship/ships,andvessel/vessels. Our analysis of collocational patterns is both statistical and linguistic because we reclassify significant collocations into selected collocates intuitively judged by Maritime English teachers. A study of collocational patterns involving statistical and linguistic analysis will ultimately lead to a better understanding of how the Maritime English corpus works.