초록 열기/닫기 버튼

This study explores the application and implementation of artificial intelligence in the field of Korean language learning. The application of AI for Korean language learning has been discussed since 2017 and has made progress over 2020 and 2021. Moreover, enhanced AI pronunciation evaluation and dialogue systems have led to the development of chatbots for practicing Korean conversation and are continuously being upgraded. To examine this progress, we analyzed three applications ‘세종학당 AI 선생님’, ‘트이다’, and ‘한컴 지니K.’ We have confirmed that the AI-applied Korean learning applications incorporate sound recognition or synthesis technology, and the accuracy of pronunciation evaluation was high. However, for dialogue processing, only closed conversations based on a given scenario are provided and a free conversation with an AI agent is not yet possible. For AI evaluation and grading, and feedback, we examined the case of LCAT in the field of English learning. Based on the results of the study, we propose the following as the key factors when developing AI-applied Korean learning and evaluation tools. First, the technology for natural pronunciation and intonation realization of the sound synthesis engine is required. Second, AI for practicing speaking must be trained with a collection of data optimal for its purpose. Third, for AI language evaluation and auto-grading, a model development strategy that suits the situation and the purpose of use must be devised. Especially, the Korean language education institutions and educators must actively participate in the data construction and refinement process as the main consumers. Finally, UX research focusing on the AI-service using learners is required. Diversification of future service concepts, research on appealing persona, and the implementation of gamification to provide learning motivations must follow. Public institutions and industries are paying much attention to implementing AI technology in Korean learning. The support of the education industry and academia would lead to the development of advanced AI learning and evaluation tools in the near future.