The present dissertation examines the impact of (computerised) adaptive corrective feedback, that is, feedback dynamically adapting to learners’ abilities, in English as a second/foreign language (L2) and explores the ways to maximise this impact. The study was inspired by the sociocultural perspective on development, which had implications for the interpretation of the results, including those obtained through statistical data analyses. The dissertation comprises five articles and a synthesis. In the first article, a positive effect of adaptive corrective feedback on the learners’ ability to formulate L2 English wh-questions is established. The second article explores how learners’ beliefs about corrective feedback guide their performance on dynamic assessment and reflection on it, and how reflection on experience with dynamic assessment mediated in social interaction helps to transform these same beliefs. The results suggest that due to their beliefs, some of the participants skipped the feedback they believed to be useless, but also illustrates how the learners, whose utterances were mediated by the interviewer, other learners in the interview, their teacher’s voice and feedback practices, and experience of dynamic assessment, began appropriating beliefs about corrective feedback that were jointly constructed by the participants in the interaction. Studies reported on in articles three and four aim at addressing the issue of lack of research on L2 English word derivational knowledge (to an extent), the latter being the assessment target in the study reported on in article five. This final article reports on a case study that builds upon the previous studies conducted as a part of my doctoral research project and studies whether generalisations made based on the other two studies add to the usefulness of adaptive corrective feedback in dynamic assessment of L2 word derivational knowledge. The available evidence for the validity of the computerised system and the dynamic test of learners’ ability to formulate wh-questions with auxiliaries is presented in a separate chapter of the synthesis. The theoretical importance of the study lies above all in that it presents quantitative evidence for the beneficial role of corrective feedback provided within learners’ Zone of Proximal Development. The findings also suggest that learners’ performance during computerised dynamic assessment is mediated not only by the adaptive corrective feedback per se but also by their beliefs about corrective feedback and expectations of what a test should look like, their beliefs being, thus a potential threat to validity of computerised dynamic tests but also suggests a way these can be accounted for. As regards practical implications, the findings suggest that the assessment/tutoring system created in the course of the study, or a similar one, using a similar approach to mediating learners’ performance, can be used in the classroom.