Artificial intelligence integration in three iOS pronunciation apps: ELSA Speak, Loora, and Vocal Image.

Claims of integrating artificial intelligence (AI) into mobile applications for pronunciation training date back to at least 2011 with the iOS app T Accent (Arivoc Education International, 2011), which used automatic speech recognition (ASR) to provide "Goodness of Pronunciation" (GOP) ratings (Witt & Young, 2000). AI has advanced significantly since 2011, most noticeably with the 2022 release of OpenAI's ChatGPT, which made chatbots powered by generative AI more widely available. AI has led to new applications in pronunciation apps that can evaluate pronunciation and integrate communicative role-play activities. This article examines ASR and chatbot integration in three iOS apps: ELSA Speak, Loora, and Vocal Image. Feedback provided by these apps is frequently inaccurate and often limited to consonant and vowel sounds. This article cautions teachers and learners about the current limitations of these apps and provides recommendations for incorporating AI-powered tools into today's pronunciation classrooms. [ABSTRACT FROM AUTHOR]