Posters

GenAI-powered pronunciation practice: Adapting ASR-based lessons for voice-activated conversational assistants

Authors: William Gottardi (Universidade Federal de Santa Catarina) , Rosane Silveira (Brasil) , Walcir Cardoso (Concordia University)

Abstract

Recent years have witnessed constant and substantial improvements in GenAI-powered virtual assistants, providing new opportunities for L2 pronunciation practice. Building on these developments, this study explores the pedagogical affordances of Voice-Activated Conversational Assistants (VACAs) for L2 pronunciation teaching and learning by designing prompts for three VACA-based pronunciation lessons developed for Google Gemini Live. Grounded in established frameworks in CALL and L2 pronunciation pedagogy, the lessons were designed using prompt engineering techniques, namely: defining a persona, establishing constraints, creating a clear context, specifying a response format and feedback focus, and providing predefined sentences for the pronunciation practice. Based on these insights, the proposed VACA‑based pronunciation lessons are meant to complement teacher‑led instruction, while also enabling learners to practice independently.

Keywords: pronunciation teaching, generative AI, voice-activated conversational assistant, computer-assisted language learning

How to Cite: Gottardi, W. , Silveira, R. & Cardoso, W. (2026) “GenAI-powered pronunciation practice: Adapting ASR-based lessons for voice-activated conversational assistants”, Pronunciation in Second Language Learning and Teaching Proceedings. 16(1). doi: https://doi.org/10.31274/psllt.22791