This study endeavors to investigate the effects of metapragmatic instructions: role play and elicited conversation negotiated feedback (RP+ NF, EC+ NF) embedded in Task-Supported Language Instruction (TSLI) on the students’ apology strategy. We used a laboratory-based research design encompassing 75 fifth-semester students of economics major taking English business for international communication in higher educational level. Those students were randomly assigned into one control group and two experimental groups. We utilized a mixed-design repeated measure analysis of variance (RM-ANOVA) to gauge the students’ apology proficiency explicated in a two-test design ie, apology judgment test (AJT) and apology oral test (AOT) in a three test sessions (pre-, post-, and delayed test). The finding reveals that there is a significant difference between among groups in which both experimental groups outperform the control group in post and delayed tests. Also, a significant increase is explicated in both experimental groups from pre to post test, but not in the control group. While RP+ NF provides the most robust of all and stimulates a long term effects with big effect sizes on both test designs, EC+ NF fails to provide a long term effect in AOT.