This presentation covers the evolution of AI-powered speaking applications used in English language learning, starting with “read aloud” or pronunciation assessment using speech recognition, to listen and repeat using elicited response, to open response questions using speech-to-text, to dialog-based systems using ChatGPT. The accuracy and efficacy of these systems are discussed including the pedagogy that underpins them and how each impacts student motivation. Data from deployments of these applications in universities in Japan will also be reviewed.

Shunsuke Furuyama is the Director of Speech & AI at EnglishCentral, where he oversees the development and implementation of the company's speech recognition, pronunciation evaluation, and conversational AI technologies. He’s worked in EdTech for over 10 years, starting his career at DMM Eikaiwa.