Answers are grounded in this paper and cite their sources.
Problem statement. The integration of artificial intelligence (AI) into educational practice calls for pedagogically sound solutions to regulate its use in English for Specific Purposes (ESP) instruction in higher education. This study addresses the gap in structured methods for the controlled use of generative neural networks, which has emerged alongside their rapid adoption in professionally oriented language learning. The purpose of this study is to develop, describe and test a three-stage methodology for working with a large language model (LLM) in ESP vocabulary teaching within the context of digitalization of education. Methodology. The study employs a design and development research (DDR) approach combined with a quasi-experimental pedagogical study conducted at Vyatka State University (spring semester 2025). The sample consisted of 24 third-year law students (specialization 40.03.01 Jurisprudence) divided into an experimental group (n = 12) and a control group (n = 12). Methods included theoretical analysis of ESP pedagogy and AI in education literature, pedagogical modelling of a three-stage structure (initial, formative and evaluative stages), and empirical validation through pre-test and post-test assessment using the Mann-Whitney U test. Results. A methodologically grounded three-stage model for acquiring legal lexis with the help of a large language model is proposed. Pilot testing showed a statistically significant improvement in vocabulary acquisition in the experimental group compared to the control group as measured by post-test scores (p 0.05). Qualitative analysis of student interviews revealed three key outcomes: development of metacognitive awareness regarding the accuracy of AI-generated content; acquisition of precise query formulation as a digital literacy skill; and a reinforced understanding of the teacher’s irreplaceable mediating role in verifying domain-specific knowledge. Conclusion. The proposed methodology shows that pedagogically sound use of a large language model does not replace traditional teaching but complements it as a tool for creating learning materials under the teacher’s guidance. Preliminary validation confirms the model’s effectiveness for specialized vocabulary acquisition and its potential for transfer to other ESP domains. This study demonstrates how theoretical design and empirical evidence can be combined in the field of AI-enhanced language instruction, which is important for the didactics of digitalization of education.
No takes yet. Share an insight, caveat, or question.
Геннадьевна et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: