For the problems of lacking situation truth, few culture hints, and not strong cross-culture shifting in university English teaching rooms, this research builds a virtual reality based immersion English teaching frame which contains culture hints, branch conversations, partner together existing, and reflection feed back. One 12-week quasi-experimental research was carried out at one university in Guangxi, China. This research has 126 first-year undergraduate students who study majors that are not English disciplines. We have carried out the evaluation through the combination of Intercultural Communication Competence (ICC) Scale, oral task score, behavior record, Immersion and Cognitive Load Scale, and 4-week post-test. The outcomes have indicated that the ICC total score of the experimental group has a rise from 3.21 to 4.06, with the post-test score being maintained at 3.91; as for the control group, its post-test examination score is 3.58. The experimental group’s oral task scores had an increase from 71.8 to 85.6, average reaction time was decreased to 1.65 s, and correction success rate had a rise to 77.3%. The results which belong to specific scenarios show that the highest transfer retention rates have been observed in the scenarios of project negotiation and host family dinner, hence academic seminar Q&A is more significantly influenced by the load of conceptual retrieval. The cutting experiment further showed that the cultural hint layer and branch conversations have the most prominent contribution to cross-cultural suitability, interaction density and conversion efficiency. This research proves that the effect of VR in English immersion teaching is dependent on the cooperative arrangement of high-quality scene building, interaction design, and cognitive load management, and hence offers a repeatable assessment route for the change of immersion teaching and the building of scene banks in college English classes.
