Harnessing the power of images in CALL: AI image generation for context specific visual aids in less commonly taught languages

Abstract

[EN] This paper explores the application of AI image generation in Computer-Assisted Language Learning (CALL) for Less Commonly Taught Languages (LCTLs). It delves into the potential of text to image generation models in creating context specific visual aids to enhance comprehension and engagement among learners. The integration of AI generated images in a language learning game, Cipher, is discussed, showcasing the benefits and challenges encountered. Learner feedback indicates positive inclinations towards the AI generated images, but also highlights the need for meticulous selection to address biases and stereotypes. Overall, this approach shows promise in creating culturally relevant CALL resources and improving language learning experiences for learners of LCTLs.