This study investigated the effects of direct and indirect WCF on students' revision accuracy as well as on new pieces of writing over time. Intermediate level pre-faculty university students were divided randomly into four groups: direct, underlining only, underlining metalinguistic, and a control group. They produced two texts from two different picture prompts and revised these over a two-week period. They also produced two new texts three and four weeks after the last treatment. Both WCF significantly reduced learner errors in subsequent revision tasks. All three treatment groups also made significant reduction of errors after 3 weeks of CF treatment.