This study examines the role of recast frequency and its effectiveness in the acquisition of English articles in a computer-mediated context. Sixty-one preintermediate university language learners in Turkey were randomly divided into four main groups: high frequency recast (HF), low frequency recast (LF), test control, and task control groups. The learners in the HF and LF recast groups completed five and two tasks, respectively, in a video-conferencing environment and received oral recasts on their incorrect use of English articles. Learners in the test control group only took the pre and posttests, and learners in the task control group completed five tasks without receiving feedback on the target structure. The outcome was measured through online picture description and error correction tasks. Findings showed that in the picture description task, learners in the HF group performed significantly better than those in the LF recast group and the control groups. In the error correction task, the results revealed a short-term advantage for learners in the HF group, which faded away in the delayed posttest. Significant correlations were also found between the recast frequency and learners' score improvement in the immediate and delayed picture description tasks but not in the error correction tasks. These results suggest that recast quantity may play an important role in improving learners' accuracy of their oral production.