This study aims at investigating the role of EFL teachers’ different levels of self-perceived language proficiency and self-efficacy in their corrective feedback (CF) practices. To this end, four EFL teachers were assigned to two groups of ‘High proficient, High self-efficient’ (first group) and ‘Less proficient, Less self-efficient’ (second group) based on their answers to two questionnaires determining self-reported levels of language proficiency and self-efficacy of the participants (Eslami & Fatahi, 2008). Using Lyster and Ranta’s (1997) model, 12 hours of classroom oral interactions were observed, audio-recorded, transcribed and analyzed by the researchers. In this vein, frequency and distribution of different types of CF practiced by different groups were figured out. Generally, findings revealed teachers from the first group tended to correct students’ errors less frequently but more systematically than those from the second group. Not surprisingly, recast was the most frequently used type of CF by the teachers from both groups. However, teachers from the first group provided more percentage of meta-linguistic information than those from the second group. Furthermore, conducting a semi-structured interview indicated that the contexts (accuracy and fluency) in which CF is provided are paid more attention by the first group; that is, teachers from this group better know CF types and have more conscious awareness of contextual factor while practicing CF.
