摘要：This quasi-experimental study investigated the effect of two types of Corrective Feedback (CF) in task-based grammar instruction (TBGI). After administering a Nelson test (for the intermediate) sixty participants out of one hundred were selected from the intact classes. These sixty participants were randomly assigned to three groups: recast, metalinguistic, and no-feedback group respectively each comprising of twenty participants. After that a test, developed by the researcher inquiring the conditionals and wish statements, was given to the participants in order to assure that the participants did not have a prior knowledge of these target structures. After the treatment was given to each group, the posttest was given in order to measure the effectiveness of the instructional approaches in each group. The results of the one-way ANOVA and the t-test revealed that first of all both CF types were effective in TBGI and secondly between the two CF types metalinguistic feedback was more effective than recast. The results of this study are considered to be useful in methodological issues related to error correction techniques, and teacher educators training pre-service or in-service teachers.
