Designing a chatbot to teach English: Transforming learning spaces

Research output: Contribution to conferencePapers


BACKGROUND: The potential of natural language processing tools such as chatbots in learning a second or foreign language has not yet been fully realised (Dale, 2016; Fryer et al., 2019). Language learning occurs through interaction with classmates and teachers, which provides comprehensible input, feedback on output and the opportunity to modify output (Mackey, 2012). Chatbots can serve as an always-on source of interaction. In this study, a chatbot was designed and deployed using Dialog flow Messenger. It acts as a conversational interface for English learners and is available on our website or as a mobile application (app). Language learning is most effective when it is immediate and contextual, so this chatbot not only ‘chats’ but also guides learners to appropriate language study resources. It puts students in control of their learning by answering specific questions (e.g., How do I refer to sources in written texts and oral presentations?, Where can I find IELTS materials?, What is the present perfect?); providing customised learning options; and allowing students to learn at their own level, pace, and at a time of their choosing. This study investigates English language students’ attitudes, beliefs, and perceptions towards using a customised chatbot to enhance language learning.
METHODS: Participants consisted of students and teachers (n=4) from a compulsory English for Academic Purposes course at the largest public university in Hong Kong. In this study, the students (n=76) had access to the chatbot for two months. Data on attitudes, beliefs and perceptions regarding the chatbot were collected using a questionnaire and focus groups; transcripts were collected to identify questions put to the chatbot. This mix of methods was chosen to provide rich, detailed, and complex accounts of users’ experiences.
RESULTS: Findings from the questionnaire and focus groups are consistent with previous research that reported positive learning experiences with chatbots (Fryer & Carpenter, 2006; Lin & Chang, 2020). Of the student participants, 85% said it was an enjoyable learning experience and 71% said it answered their questions and directed them to suitable study resources. Transcripts revealed that the most common questions were about IELTS materials (e.g., Where can I find a sample Task 1 essay?, How do I get a high score on Task 2 of the speaking part?) and subject specific questions about assignments (e.g. How many in-text citations are required?, How do I write an integral citation?). In the interviews, a majority of the participants (students and teachers) highlighted the detailed, concise answers/feedback and suggestions as beneficial and informative for language learning.
CONCLUSIONS: The results indicate the chatbot provided rapid access to rich language learning content and encouraged students to find solutions to their questions through self-directed study. These results show that chatbots are a valuable, viable tool for language acquisition. Copyright © 2020 The Education University of Hong Kong (EdUHK).
Original languageEnglish
Publication statusPublished - Nov 2020
EventThe International Conference on Education and Artificial Intelligence 2020 (ICEAI 2020) - , Hong Kong
Duration: 09 Nov 202011 Nov 2020


ConferenceThe International Conference on Education and Artificial Intelligence 2020 (ICEAI 2020)
Abbreviated titleICEAI 2020
Country/TerritoryHong Kong
Internet address


Kohnke, L. (2020, November). Designing a chatbot to teach English: Transforming learning spaces [Online]. Paper presented at The International Conference on Education and Artificial Intelligence 2020 (ICEAI 2020), Hong Kong, China.


  • Chatbot
  • Natural language processing
  • Language learning
  • Self-directed


Dive into the research topics of 'Designing a chatbot to teach English: Transforming learning spaces'. Together they form a unique fingerprint.