ScholarWorks@성균관대학교

초록

This study investigates how social registers—age, gender, and region—are reflected in Korean corpus data through collostructional analysis. Unlike conventional frequency-based approaches, this method measures the strength of association between lexical items and specific constructions, allowing for a more precise account of linguistic variation. We apply collexeme analysis— an early version of collostructional analysis—to a large spoken corpus of middle-aged and elderly Korean speakers representing five dialect regions. By treating age, gender, and region as variables, we identify words that show strong associative strength with each register. The results reveal distinct lexical sets tied to social characteristics, providing empirical evidence that Korean speakers’ language use systematically reflects their identities. Our findings further demonstrate that collostructional analysis can be extended to capture social variation, broadening its applicability beyond purely structural studies. This work contributes to Korean computational sociolinguistics by offering methodological insights, empirical evidence, and an illustrative example of data-driven research in an underrepresented language.

키워드