Text-to-video AI for linguistically responsive teaching: educators' use and reflections

Purpose This study examines how AI-powered text-to-video (T2V) tools can support culturally and linguistically responsive teaching (CLRT) for English Language Learners (ELLs). Against the backdrop of growing linguistic diversity and limited bilingual educators, the study explores the potential of T2V tools to bridge the gap between student needs and teacher preparedness. Design/methodology/approach The study used a two-phase qualitative case design. Phase One involved functional testing of T2V tools (VEED.oi and CapCut) to assess their capabilities in generating visuals, audio, captions, and multilingual contents, particularly in freemium modes. Phase Two engaged 18 preservice teachers from a Midwestern US university and a third-grade teacher from a linguistically diverse Northeastern district in co-creating multilingual read-aloud videos for bilingual learners. Data included demonstration videos, screenshots, and annotated reflections. Technical and thematic analyses examined the T2V tools' functionality, identified key affordances, and explored teachers' perceptions of CLRT alignment, and support for differentiated instruction. Findings The analysis showed that VEED.io and CapCut provide features, such as multilingual voiceovers, captions, and avatars, that can support CLRT for ELLs. Participants created read-aloud and phonics lessons reflecting students' languages and cultures, with preservice teachers reporting a mindset shift toward viewing AI as a supportive aid rather than a replacement. However, challenges included translation inaccuracies, limited support for less commonly spoken languages, Western-centric avatar designs, and restricted access to premium features. These findings highlight both the promise and constraints of T2V tools for advancing CLRT in multilingual educational contexts. Research limitations/implications This study was limited to two T2V AI tools, VEED.io and CapCut, and only their freemium versions, which restricted access to advanced features and constrained evaluation of full functionalities. Data relied primarily on participants' reflective notes, which, while insightful, did not capture the depth of experiential engagement or nuanced perspectives. Future studies should include a wider range of T2V AI tools, premium versions, and validated instruments for assessing usability and pedagogical impact. Incorporating interviews, focus groups, and K-12 student perspectives would provide richer insights into the effectiveness of T2V AI tools in multilingual and culturally responsive classroom contexts. Practical implications The findings highlight practical implications for teacher education, instructional design, and technology integration. VEED.io and CapCut show how accessible T2V tools can support CLRT in resource-constrained settings. For educators without linguistic expertise, these tools provide valuable means to deliver lessons that engage diverse learners. The positive mindset shifts among preservice teachers suggest that integrating T2V AI into training enhances digital pedagogical skills and confidence in differentiated instruction. However, to maximize impact, structured training and critical guidance are essential to ensure educators use these tools thoughtfully and align them with sound pedagogical practices. Originality/value This is one of the first empirical studies to examine the use of T2V AI tools specifically for multilingual education within teacher preparation. Unlike existing research that focuses on T2V AI tools for social media production, this work centers on how T2V AI tools can support CLRT in education. The study brings in technical findings and original empirical experiences from educators. It offers a foundation for practical pathways for implementing T2V AI tools in education.