Artificial Intelligence and Virtual Reality in Teaching of Language Phraseology: Cognitive-Digital Model and Its Effectiveness
TL;DR
The proposed cognitive-digital model, which implements a sequential integration of metaphorical modeling, generative AI, and VR, is an effective tool for the deep acquisition of foreign language phraseology, as demonstrated by the example of Spanish.
Abstract
The purpose of the study is the development, theoretical substantiation, and empirical testing of a cognitive-digital model for teaching foreign language phraseology. This model is based on the integration of metaphorical modeling, generative artificial intelligence (AI), and virtual reality (VR). It aims to overcome the cognitive gap between the form and the culturally determined meaning of set expressions, which poses a significant challenge for learners. The research was carried out on the example of Spanish idioms in a university classroom. Materials and methods. The theoretical foundation is a synthesis of cognitive linguistics (conceptual metaphor theory), principles of embodied cognition, and social constructivism. A four-stage methodological model was developed, including: 1) semantic-cultural deconstruction using generative AI (ChatGPT-4, DeepSeek) and specialized prompt templates; 2) immersive visualization of the historical-cultural context in accessible VR environments (CoSpaces Edu, Google Cardboard); 3) contextualized practice of idiom usage with adaptive AI tools (Character.AI for dialogues, DeepL Write for text analysis, ELSA Speak for phonetic practice); 4) socioreflective transfer through a theatrical method with digital reflection tools (Padlet, Mentimeter). Empirical testing was conducted via a pedagogical experiment involving 60 A2-level linguistics students divided into control and experimental groups. Methods included preand post-testing, comparative and correlation analysis, qualitative analysis of the digital footprint (logs of interaction with AI and VR), expert assessment of video recordings of theatrical performances, and data triangulation. Results. A statistically significant superiority of the experimental group over the control group was established across all measured parameters: recognition and understanding of idioms (p < 0.001, Cohen’s effect size d = 1.80), pragmatic choice of appropriate usage (p < 0.001, d = 1.45), and, especially, the task of explaining the conceptual foundations of idioms (p < 0.001, d = 2.30). Qualitative analysis revealed positive correlations between the structure of AI prompts (r = 0.65), active exploratory behavior in the VR environment (r = 0.71), and the depth of competence, facilitating a shift from rote memorization to meaningful mastery of phraseology. Conclusion. The proposed cognitive-digital model, which implements a sequential integration of metaphorical modeling, generative AI, and VR, is an effective tool for the deep acquisition of foreign language phraseology, as demonstrated by the example of Spanish. The scientific novelty lies in the theoretical and methodological synthesis of cognitive linguistics and digital pedagogy, as well as in the development of a comprehensive assessment model based on data triangulation. The practical significance consists in creating teaching modules, prompt templates, and VR scenarios adapted for A2–B1 levels, ready for the implementation in university practice. The identified limitations (digital fatigue, dependence on prompt quality) indicate the need for pedagogical support and preliminary development of AI interaction skills. Research prospects are associated with the development of an integrated software platform, adaptation of the model for other languages, and an in-depth study of digital load optimization.