Recent interaction research has shown that corrective feedback may benefit L2 learners to varying degrees, and that certain types of feedback may be more beneficial than others, engendering much discussion and research attempts to investigate the issue. Goo`s (2012) study is one of those empirical attempts conducted in this line of research paradigm. The present study, as a conceptual replication of Goo`s original study, explores this on-going area of feedback research, the relative efficacy of different forms of corrective feedback, and whether/how phonological short-term memory (PSTM) mediates the impact of different feedback moves on the acquisition of the English that-trace filter. Twenty-nine adult high-intermediate Korean EFL learners were randomly assigned to one of three groups: a recast, metalinguistic feedback, and comparison groups. The study employed a pretest-posttest design. One intensive treatment session was implemented. Two dependent variable measures were developed/administered to assess learner performance: a timed grammaticality judgment test and a timed oral production test. Overall results showed that metalinguistic feedback was more effective than recasts at facilitating the acquisition of the target construction. PSTM was not found to be correlated with the effectiveness of recasts and metalinguistic feedback. Different patterns of learner performance were also observed depending upon the mode of dependent variable measure.