jats:titleAbstract</jats:title> jats:pThis article reports on a mixed methods study that investigated the extent to which written corrective feedback (WCF) contributes to L2 learners’ acquisition of regular past tense in English and the cognitive processes that underpin the corrective feedback provided. The study adopted a pretest-posttest-delayed posttest design and involved 113 intermediate-level Chinese learners of English who were assigned to four conditions: indirect WCF, direct WCF, metalinguistic WCF, and control. A picture description task and a grammaticality judgement test were used to measure gains in the target structure. To explore how learners process feedback, nine learners, three from each treatment group, were selected to produce think aloud protocols. The study found that all three types of WCF had a positive effect on the picture description task, though none of them had a clear impact on the grammaticality judgement test. In addition, indirect WCF was found to have an advantage over direct and metalinguistic WCF on the delayed posttest, when compared to the control group. Think aloud data suggested that indirect WCF induced deeper cognitive processes than the other two kinds of feedback, which may account for the superiority of indirect WCF.</jats:p>