The present study aimed at investigating the feasibility and constraints in applying an AI chatbot to English education. For this purpose, seven tasks were presented to the upper (n=27) and lower (n=26) groups (college students vs. high school students) using Mitsuku, a representative chatbot, as a way to utilize the chatbot in English education. The participants chatted with Mitsuku to accomplish these tasks. By analyzing the data from the task performance, this study showed that Mitsuku s utterances were two times more than the students utterances,and the difference was larger at the lower group. It also turned out that the text easability of Mitsuku s utterances corresponded to the grades K2 and K3, and 90% or more of the words making up Mitsuku s utterances were included within the first 3,000 word level. In the case of the lower group, it was confirmed that the shorter the sentence length was, the larger the amount of utterance, and the higher the similarity of conceptual meaning between adjacent paragraphs, the more similar to human utterance. In the case of the upper group, the participants seemed to believe that the higher the diversity of the use of the verbal forms, the more the contribution to the effectiveness of English education. In addition, the upper group showed that the lower readability level was, the larger the amount of utterance. As a result, the findings from this analysis could contribute to the educational use of chatbot and the development of a new chatbot as a reference.
II. Literature Review
Ⅳ. Results and Discussions
Ⅴ. Conclusions and Implications