Automated learning analytics and adaptive testing in foreign language courses in higher education as a tool for personalizing learning trajectories

Abstract

The article is devoted to the integration of automated learning analytics and computer-adaptive testing in university English language courses as a mechanism for personalizing learning amid high heterogeneity of academic groups in non-linguistic majors. It examines the transition from linear methods to a modular course architecture capable of reconfiguration based on learners' digital footprints in the LMS, where feedback and assessment are treated as means of navigation and support, rather than merely control. The implementation on the Moodle platform with enhanced event logging and adaptive task presentation is described; adaptation relies on models of modern testing theory (particularly Rasch and Birnbaum approaches) and a calibrated bank of 4500 test items, refined as responses accumulate. Analysis of logs (over 3,5 million events), attestation results, and content interaction metadata is supplemented by statistical procedures (correlations, clustering of learning strategies, group comparisons). It is shown that personalized trajectories enhance productivity: with less time in the system, students complete more modules and demonstrate higher average scores, as well as lower result variability, interpreted as leveling of preparation without loss of motivation among strong learners. Improvements in psychometric characteristics of adaptive testing are identified: expansion of difficulty range, increase in discriminability, reduction in standard error of measurement, and shortening of test length with increased reliability. Deferred control after three months records higher vocabulary retention, sharp reduction in the share of systematic grammatical errors, and decreased test anxiety, linked to individualized repetition and prevention of knowledge gap accumulation. Pedagogical implications are emphasized: reallocation of time toward communicative practices, strengthening of the teacher's diagnostic role, as well as the importance of algorithm transparency and data ethics for the proper application of personalized assessments.