Multimodal Immediate Feedback (Audio/Video + LLM-based Code Reviewers) in Practice-Oriented Disciplines: Causal Analysis of Effects on the Pace of Skill Formation, Metacognitive Monitoring, and Sustained Engagement

Amid the rapid digitalization of higher education and the emphasis on practice-oriented trajectories, traditional feedback formats exhibit declining effectiveness. This study conducts a causal analysis of the impact of an integrated system of multimodal immediate feedback, combining expert audiovisual (A/V) commentary and automated reviews based on large language models (LLM), on key educational outcomes. The aim is to explicate the causal mechanisms through which the synergy of these modalities affects the pace of practical skill formation, the quality of metacognitive monitoring, and learners? sustained engagement. The methodological framework rests on synthetic causal analysis conducted on the basis of a systematic review of empirical studies from 2020?2025. The results indicate that the integrated system implements an effective two-loop architecture: the LLM loop provides rapid tactical correction, accelerating the acquisition of procedural skills, whereas the A/V loop initiates deep strategic reflection, enhancing metacognitive self-regulation. It is shown that this design comprehensively addresses students? basic psychological needs for competence, autonomy, and relatedness, which, in the logic of self-determination theory, serves as a causal driver of sustained engagement. The conclusions support the hypothesis of a synergistic effect that exceeds the sum of the influences of each modality taken separately. The material is of interest to researchers in educational technology, developers of intelligent tutoring systems, and instructors in practice-oriented disciplines.