The Role of Mobile Computing in Adaptive Testing for English Language Learners: Personalizing Assessment to Improve Outcomes

Adaptive testing is considered a critical part of personalized assessment, especially for ELLs, since it assesses learners by changing the difficulty level of questions that are being presented to learners based on learner performance. Research studies in the past utilized rule-based or machine learning models, which could not fully capture the richness of linguistic variations and contextual understanding, therefore performing sub optimally. To overcome these limitations, this study introduces a novel BERT-based adaptive testing framework. BERT uses a bidirectional approach to understand the entire context of learner responses, which helps in more accurate assessment and personalization. The proposed method outperforms existing models in key metrics such as accuracy (98.6 %), showcasing its ability to dynamically adjust question difficulty while providing contextually aware feedback. The system operates efficiently to provide real-time, personalized assessments through a mobile interface. It is scalable and accessible for widespread use. It addresses the shortcomings of earlier approaches by improving test accuracy, learner engagement, and overall educational outcomes.