The present study investigated the effectiveness of recasts and prompts on the acquisition of the English third-person singular form and the mediating role of cognitive style on the effects of feedback. One hundred and seventy-five college students from four intact classes were assigned to four groups: form-focused instruction with recast (FFI-recast), FFI with prompt (FFI-prompt), FFI, and control. The group embedded figures test (Witkin et al. in Rev Educ Res 47:1-64, 1977) was adopted to test learners' cognitive style (field dependence/independence). The results show that the FFI-prompt group outperformed the FFI-recast group and the control group on the immediate post-test; the FFI-prompt group also achieved significantly higher scores than the other groups on the delayed post-test in the written test. However, no significant difference was found among groups in the text-completion test. Regression analyses reveal that in the text-completion test, field dependence/independence mediates the effect of recasts on the immediate post-test.