ARText: Vocabulary Learning with Personalised and Predefined Keyword-Associations in Augmented Reality

The “keyword method” is a mnemonic technique that can be used for learning foreign vocabulary by associating a word’s meaning with a phonetically similar keyword. For example, the Japanese word for “tree” is “ki”, which sounds like“key” (keyword) so one might imagine “a tree with key-shaped leaves” (association). Research in non-contextualised settings (e.g., on paper or screen) shows that personalised keyword-associations improve retention when learners create and visualise their own associations. Studies also suggest that externalising these associations through images enhances recall, while Augmented Reality (AR) further strengthens retention by anchoring words to real-world objects. However,existing AR studies have only explored predefined keyword-associations with expert-designed visuals, leaving a gap in understanding how personalised and predefined approaches compare in contextualised AR learning. To explore this,we developed ARText, an AR system that visually annotates real-world objects with (i) their corresponding words in both English and the target language,(ii) keywords, and (iii) visual representations of associations, generated using text-to-image synthesis. Participants experienced keyword-associations in both personalised condition (keyword-associations created by users) and predefined condition (keyword-associations designed by experts). The findings indicate that participants preferred predefined keyword-associations. This condition also facilitated faster and more efficient word recall. In this paper, we discuss possible reasons for these outcomes and explore their implications for designing futureAR-based vocabulary learning systems.