The extent to which written CF plays a role in learners acquiring the target language is a question that has received a lot of attention over the last 20-30 years. This thesis, by drawing on a cognitive view, continued with that focus, exploring not only the efficacy of written CF on the improved accuracy of learners but also the extent to which working memory and phonological short-term memory may moderate the effects of different types of feedback. The study was undertaken with 100 university students in Iran. Firstly, a quasi-experimental study was used, with a pre-test, treatment, immediate and delayed post-tests, to investigate the effectiveness of four types of written CF (direct CF, direct CF plus revision, metalinguistic explanation, metalinguistic explanation plus revision) on a complex linguistic structure, the English passive voice. Additionally, the learners’ working memory was measured using a reading span test using DMDX software and their phonological short term memory was measured using a non-word span test. The findings showed that a single episode of written CF improved accuracy immediately and over time (after two weeks) for all the experimental groups, but not for the control group. They also revealed that direct CF relatively led to more improved accuracy compared to the other groups. Additionally, the results showed that the non-revision groups (i.e., direct CF and metalinguistic explanation) had greater accuracy in the immediate post-test, but that the accuracy of the revision groups (i.e., direct CF plus revision and metalinguistic plus revision) was retained over a longer period of time. The results regarding the moderating effect of working memory and phonological short term memory revealed that (1) working memory moderated the impact of the metalinguistic explanation and combined metalinguistic explanation groups (i.e. metalinguistic explanation and metalinguistic explanation plus revision) and the combined metalinguistic and direct CF groups both immediately and over time; and (2) working memory moderated the direct CF plus revision and combined revision groups (metalinguistic explanation plus revision and direct CF plus revision) only in the long term. Additionally, phonological short-term memory negatively moderated the impact of direct CF plus revision only in the long term. These results support the role that explicit knowledge in the context of writing can play in L2 development because the learners sustained the explicit knowledge gained from written CF on a complex structure (e.g. the English passive voice) in new texts over time. Additionally, the findings suggest that more explicit types of written CF may be more effective than less explicit types of written CF for targeting a complex structure in a single processing episode.