This study aimed to compare the effects of self-regulation, goal orientation, and speaking anxiety on speaking performance between metaverse and face-to-face contexts. We randomly assigned 253 Korean middle school students to either metaverse or face-to-face groups for 12 weeks of English-speaking lessons. Before and after the experiment, students completed speaking tests and submitted a post-questionnaire. Structural equation modeling and multi-group analysis revealed that student attributes had a similar impact on speaking performance in both settings. Specifically, self-regulation and mastery goal orientation positively influenced speaking performance, while speaking anxiety had a negative effect, regardless of the context. Furthermore, self-regulation played a mediating role in reducing speaking anxiety in both settings. This implies that metaverse-mediated and face-to-face classes offer similar learning environments, where students can leverage their goal orientation and self-regulation skills to manage speaking anxiety. Ultimately, students can enhance their speaking performance by employing self-regulation strategies and nurturing a mastery goal orientation, irrespective of the learning context.