Artificial Intelligence (AI) in elementary language education: A systematic review

Background: Language acquisition during early childhood is foundational for cognitive, social, and academic development. While pedagogical theories have long guided early language education, the integration of Artificial Intelligence (AI) at this level remains underexplored. Although AI has the potential to deliver personalized, adaptive, and engaging learning experiences, existing studies often lack theoretical grounding, generalize across educational phases, and overlook the developmental needs of young learners. As a result, there is a significant knowledge gap regarding how AI supports foundational language skills in authentic elementary education settings. Purpose: This study conducts a systematic review, guided by the Diffusion of Innovation Theory, to address two research questions: (1) What are the main language education aspects addressed by AI in elementary education research? and (2) What are the predominant AI technologies and applications used, and how do they support language learning outcomes? The review maps current trends, identifies underrepresented areas, and offers practical insights for research, policy, and practice Methods: Following PRISMA guidelines and guided by the PICOS framework, a structured search was conducted across six major databases to identify peer-reviewed studies published between 2013 and 2025. 31 studies met the inclusion criteria and were analyzed through quantitative content analysis and thematic synthesis. Measures to reduce bias included dual independent screening, inter-rater reliability (κ = 0.75), and quality appraisal. Results: Findings show a strong focus on speaking, literacy, and vocabulary, with limited attention to grammar, listening comprehension, and sign language. Pedagogical uses of AI centered on enhancing motivation, assessment, and personalized interaction. Conversational AI and natural language processing were frequently used, while advanced modalities like adaptive learning systems and biometric feedback were underutilized. Most studies employed qualitative or mixed methods with medium sample sizes and lacked grade-level specificity. Implications: The review highlights the need for developmentally appropriate, pedagogically coherent, and ethically sound approaches to AI integration. It calls for research that addresses overlooked areas and adopts more rigorous methodologies. Policymakers must ensure ethical, inclusive, and equitable AI use in schools, supported by infrastructure investment and teacher training to maximise AI’s potential in elementary language education.