The current study investigated how recasts can promote the L2 pronunciation development of word-initial /?/ by Japanese learners of English in relation to two developmental stages of English /?/ acquisition (i.e. change in second formant [F2] ? change in third formant [F3]) as well as four affecting variables (i.e. the amount of recasts and repairs, initial pronunciation levels, explicit articulatory knowledge). In total, 54 Japanese learners of English participated in the study and received 4 hours of form-focused activity. While teachers gave pronunciation-focused recasts to students in the experimental group (n = 29), no recasts were directed to those in the control group (n = 25). According to the results of the rating session, which involved 20 native speaker listeners, their positive perception of /?/ was associated with lower F3 values as the primary cue; lower F2 values were demonstrated to be secondary, and less relevant, for listeners' positive perception of /?/. The results of the ANOVAs showed that whereas the F2 values of both the experimental and control groups equally decreased (i.e. due to the use of the interlanguage strategy), only the experimental group significantly lowered their F3 values (the extent of acquisition). Furthermore, a range of post-hoc analyses found that recast effectiveness was related to the amount of recasts and repairs and initial pronunciation levels.