Although the provision of error correction is common in education, there are controversies regarding "when" correction is most effective and why it is effective. This study investigated the differences between Iranian English as a foreign language (EFL) teachers and learners regarding their perspectives towards the timeline of error correction in the speaking skill. Two main categories of immediate and delayed feedback types were determined, with the immediate category specifically referring to "very soon" and "after speaking" and the delayed one relating to "after activities" and "end of class" sub-categories. To empirically test the hypothesis behind the study, a total of 460 teachers and learners were invited to complete Fukuda's questionnaire. The data obtained were analysed through multivariate analysis of variance (MANOVA), the results of which showed teachers' preferences for immediate correction while the learners' inclined towards delayed correction. Based on these results, specific properties and circumstances in relation to the timing of feedback are proposed to enhance its efficiency in language classrooms.