This paper reports on a study that investigated the impact of two types of computer-mediated corrective feedback on the development of adult learners' L2 knowledge: (1) corrective feedback that reformulates the error in the form of recasts, and (2) corrective feedback that supplies the learner with metalinguistic information about the nature of the error. High intermediate and advanced adult learners of English (n=23) from an intact class at a Swedish university were randomly assigned to one of three conditions (two feedback conditions and one control) and were randomly paired with English native speakers. During task-based interaction via text-chat, the learners received focused corrective feedback on omission of the zero article with abstract noncount nouns (e.g., employment, global warming, culture). Computer-delivered pretests, posttests and delayed posttests of knowledge (acceptability judgments) measured learning outcomes. Results showed no significant advantage for either feedback type on immediate or sustained gains in target form knowledge, although the metalinguistic group showed significant immediate gains relative to the control condition. (Contains 4 figures, 6 tables and 8 notes.)