Beyond Replacement: How Large Language Models Influence Dictionary Usage Patterns Among Chinese English Learners

Abstract This study investigated how Large Language Models (LLMs) influence Chinese English learners’ dictionary usage patterns. Through a mixed-methods approach combining questionnaire surveys (n = 608) and semi-structured interviews (n = 17), the findings reveal that LLMs reconstruct the language learning tool ecosystem through clear functional divisions rather than simple replacement patterns. Results demonstrate that LLMs predominantly serve complex language tasks including translation, writing assistance, and grammar correction, while traditional dictionaries maintain competitive advantages in providing authoritative information, precise definitions, and structured vocabulary learning tools. Learners have developed sophisticated task-oriented selection strategies, following a ‘dictionaries for discrete knowledge acquisition, LLMs for integrated language application’ usage pattern that maximizes learning efficiency. Beyond behavioural adaptations, this study identified significant demographic stratification in tool adoption, with age, educational background, and English proficiency level significantly influencing usage patterns. The research further revealed cognitive paradigm shifts in language learning conceptualization, exposing tensions between instrumental utility and cultural acquisition perspectives. These findings suggest two critical directions for future lexicographic development: (1) intelligent integration combining authoritative content with interactive AI capabilities, and (2) specialized personalization addressing domain-specific and scenario-based learning needs through enhanced functionality and user-centred design.