Korean Linguistics - Volume 18, Issue 2, 2022

The present study explores the applicability of Natural Language Processing (NLP) techniques to investigate child corpora in Korean. We employ caregiver input and child production data in the CHILDES database, currently the largest and open-access Korean child corpus data, and apply NLP techniques to the data in two ways: automatic Part-of-Speech tagging by adapting a machine learning algorithm, and (semi-)automatic extraction of constructional patterns expressing a transitive event (active transitive and suffixal passive). As the first empirical report on NLP-assisted analysis of Korean child corpora, this study is expected to reveal its advantages and drawbacks, thereby opening the window to furthering corpus-mediated research on child language development in Korean. Implications of this study’s findings will also contribute to research practice regarding developmental studies on Korean through child corpora, ensuring the reproducibility of procedures and results, which is often lacking in previous corpus-based research on child language development in Korean.

This study explores the speech level shifts from non-honorific to honorific situation by analyzing Korean spoken corpora of conversations between native speakers of Korean through a discourse-analytic framework. Although non-honorific styles (e.g., panmal) are used as frequently as honorific ones (e.g., contaysmal), little attention has been paid to non-honorific expressions in Korean, let alone to the speech level shift between non-honorific and honorific. Analyzing interactions between native speakers of Korean revealed that, depending on the change of their stance in response to the interlocutor’s utterance, speakers dynamically switched their speech levels not only within the boundary of non-honorifics or honorifics, but also across the dichotomous categories, that is, they changed from non-honorific to honorific speech styles. In general, speakers employed the speech level elevation from non-honorific to honorific when indexing a confrontational stance toward a topic and an object, or when upgrading an epistemic stance in naturally occurring interactions. The findings of this study encourage researchers to actively construct and use corpora of authentic/naturalistic conversations to explore the dynamicity of the speech level shift and its functions, which in turn contributes to developing instruction materials that reflect the dynamicity of the Korean honorific system.

How language learners of Korean acquire knowledge on postpositions has been a long-standing research question in Korean language pedagogy due to their polysemous nature. The present study investigates the nature of input involving the locative function of the postposition -ey, one of the representative polysemous postpositions in Korean, through the frequency of its occurrence, types of verbs co-occurring with -ey, and keyness analysis. Sejong written and spoken corpora and two types of textbooks (eight volumes for each type: two volumes for four proficiency levels) for language learners of Korean are analyzed. Results show that iss- ‘to be/exist’ predominantly occurred with locative -ey in the Sejong corpora and a few verbs occupied a large proportion of the total usage. On the other hand, the most frequent verb was ka- ‘to go’ in all proficiency levels of the textbooks, with the exception of the fourth level of the second-type textbook. This suggests that, while the Sejong usage highlights its existential role, -ey for indicating destination is widely emphasized in the textbooks. Since the purpose of language learning is to learn the structure and usage of the target language, this study’s findings can offer guidance in setting and building pedagogical goals and directions.

This study aims to identify patterns of verb conjugation errors that learners of Korean manifest and the factors that influence these errors through an analysis of an error-annotated learner corpus. For this purpose, a paradigmatic relations-based description of language acquisition was proposed. The predictability of each conjugation class was estimated by way of entropy, a tool for measuring predictability. Using entropy allowed us to compare the regularity of each class in detail. The results showed that there are 332 verb conjugation errors that can be classified into three types of errors: errors with vowel endings, errors with lower entropy, and errors with higher entropy. The frequency of the first two types suggests that learners make errors when producing frequently used conjugated forms and with more predictable classes. Considering this study’s reproducibility and the reliability of its procedures and the results, its findings are expected to make a substantial contribution to the study of error analysis using error-annotated learner corpora.