The Impact of AR-Enhanced AI Apps on Vocabulary Retention through Spatial and Environmental Learning

Abstract

This paper investigates the pedagogical impact of augmented reality (AR) enhanced artificial intelligence (AI) applications on vocabulary retention in second language acquisition, with a particular focus on the mechanisms of spatial and environmental learning. As mobile AR hardware has matured and on-device large language models have become commonplace, a new generation of language-learning applications now overlays target-language vocabulary onto the learner's physical surroundings in real time. Drawing on cognitive load theory, dual coding theory, embodied cognition, and the encoding specificity principle, this paper synthesizes recent empirical literature and theoretical frameworks to argue that AR-AI vocabulary learning produces measurably stronger retention than conventional flashcard-based or screen-only mobile applications. The paper examines how the binding of lexical items to physical locations, multisensory input, and contextually adaptive AI feedback collectively scaffold long-term memory. It also addresses methodological challenges in the field, equity and access concerns, and directions for future research, including longitudinal studies, neuroimaging-based confirmation of memory mechanisms, and the implications of fully ambient learning environments. The paper concludes that AR-AI applications represent a meaningful pedagogical shift rather than a novelty, but that their effectiveness depends on principled instructional design grounded in established theories of memory and learning.