Collocational Status of the Bigrams in Korean EFL Learners’ Argumentative Writing: A Quantitative Approach

The current study examined the development of collocational knowledge of Korean EFL learners by analyzing a large set of argumentative essays taken from the Yonsei English Learner Corpus. A comparable native-speaker corpus was used as a norm for evaluating the learner texts. Collocational status of the bigrams in the corpora was measured in terms of (i) the strength of statistical association between the component words (MI and t-score) calculated based on a large native-speaker reference corpus and (ii) whether they ever occur in the reference corpus. The overall collocational strength of each text in the corpora was quantified by the mean MI and t-score of the bigrams in the text and the proportion of those that are absent in the reference corpus. These three measures of collocational knowledge were then compared between the native-speaker texts and the learner texts, with the latter grouped by proficiency level to capture the developmental pattern of collocational knowledge. The results showed that different measures reflect different aspects of collocational development and provide a rich source of information that can characterize the collocational knowledge of Korean EFL learners. Practical effects of each measure are also discussed in terms of their methodological implications for corpus-based research on foreign language development.