Presentation

The AI Chatbot, Gliglish, and Potential Pronunciation Learning

---

Abstract

This exploratory analysis investigates the potential of Gliglish, an AI chatbot designed for second language learning, to support pronunciation learning. The study focused on Gliglish's ability to respond to requests for pronunciation information and activities across five major pronunciation learning activity types: explanation, perception, prediction, controlled production, and guided production, as well as feedback provided during communicative role-plays when pronunciation errors were introduced. Results indicated that Gliglish struggled at times to understand prompts, provide sufficient explanations or examples, and to provide accurate information, but performed best when prompted to provide activities for controlled production. Furthermore, the analysis of feedback revealed Gliglish transcribed segmental errors less than half of the time (36.36% for consonants and 45.45% for vowels) and even less frequently for suprasegmental features (7.69%). Further, Gliglish rarely indicated pronunciation issues in its responses in subsequent turns. The findings suggest essential areas for improvement, emphasizing the need for enhanced feedback mechanisms and the incorporation of various activity types.

Keywords: emerging technology, AI chatbots, pronunciation teaching, feedback

How to Cite:

McCrocklin, S., & Colclasure, Q. (2025). The AI chatbot, Gliglish, and potential pronunciation learning. In J. M. Levis, M. Duris, S. Sonsaat-Hegelheimer, & I. Na (Eds.), Proceedings of the 15th Pronunciation in Second Language Learning and Teaching Conference (pp. 1-13). Iowa State University, September 2024. https://doi.org/10.31274/psllt.18418