This study provides a corpus-based, data-driven synthesis of research on AI-assisted English language assessment published between 2015 and 2025. Using a corpus of 275 peer-reviewed articles indexed in the Web of Science Core Collection, the study applies an integrated text-mining approach to identify dominant research themes and their structural relationships. The findings indicate that the field has been strongly anchored in automated writing evaluation and automated essay scoring, with writing assessment serving as the central research axis. At the same time, recent studies have increasingly emphasized feedback provision, learner engagement, human–AI comparison, and issues of transparency and fairness. Automated speaking assessment based on ASR technologies has also emerged as a growing but still underrepresented area. Overall, the results suggest an increasing thematic orientation toward learner-oriented and feedback-driven assessment practices. The study offers an empirical knowledge map of AI-assisted English language assessment and provides implications for the development of human–AI hybrid assessment systems and responsible use of AI in language assessment.

I. Introduction
II. Theoretical Background
    1. Technological Evolution of AI-assisted English Language Assessment
    2. Review Studies and Methodological Gaps in AI-assisted EnglishAssessment
III. Research Method
    1. Selection of Included Studies
    2. Procedure
    3. Data Analysis
IV. Results
    1. Word Frequency Analysis
    2. Bigram Text-Mining Analysis
    3. Correlation Matrix Analysis
    4. Topic Modeling Analysis
    5. CONCOR Cluster Analysis
V. Discussion
    1. From Writing-Centered Automation to Learner-Oriented Assessment
    2. Human–AI Comparison and Normative Concerns
    3. Post-LLM Paradigm Shift in Assessment Research
VI. Conclusion
References

• Yena Lee(CHA University)
• Hye Rang Om(Korea University) Corresponding author