English speaking is a challenging learning objective for EFL (English as Foreign Language) learners owing to the lack of practicing environments. In this study, Interactive Spherical Video-based Virtual Reality (ISVVR) was adopted as the learning mode to provide realistic sociocultural contexts for English speaking. In addition, an experiment was conducted to investigate the effects of the SVVR approach on the English speaking performances and anxiety of the EFL students with different cognitive styles. A pretest-posttest quasi-experimental design was conducted to investigate the effects of the interaction between cognitive style and multimedia learning mode on students' learning motivation, speaking anxiety, oral presentation, and move analysis. The student sample consisted of two classes of 93 students. All of the students took the Group Embedded Figures Test (GEFT) to identify whether their cognitive style was field dependent or independent. The results showed a significant effect for the interaction between the cognitive styles and the learning modes on the students' oral presentation and speaking anxiety. Their structural moves in oral presentations with the ISVVR learning mode showed more similar spoken genre patterns to those of TED speakers. Findings further support the benefit of incorporating cognitive styles into the ISVVR learning mode for EFL learners.