Realizing Corrective Feedback in Task-Based Chatbots Engineered for Second Language Learning

Abstract

Building on the work of customized chatbots for language teaching and learning and the second-language acquisition literature on corrective feedback (CF), this article showcases an innovative practice for building a tailored and task-based chatbot to provide CF. Given that extant chatbots are generally not sensitive to learners' grammatical errors, we illustrate a way to install a CF function by using 'action and parameters' and 'define prompts' options in the chatbot-building platform known as Google Dialogflow (TM). Our study, which included upper-grade English-as-a-foreign language learners in South Korea, demonstrated that customized chatbots could offer CF when students made non-target utterances and elicit learner uptake successfully. Based on our innovation, we then provide directions for pedagogy on chatbot-based language learning.