Recently, immersive virtual reality (VR) environments for language learning have garnered interest from researchers and practitioners alike, based on their realistic imitations of environments where target languages might be used for "authentic-like" interactions (with other learners, bots and avatars). However, the majority of research thus far has focused on university-level students' perceptions of VR and there is a critical need for research examining how these environments can be used effectively with kindergarten to grade 12 students. This article discusses a pilot study that aims to fill this gap in research through a three-month-long VR study conducted with 10- to 11-year-old students in Spain (n = 24) studying English as a foreign language (EFL). VR lessons were taught using the VR language learning platform Immerse and Oculus Quest 2 VR headsets. The data corpora consist of recordings of the VR activities, comparable non-VR classroom activities and teacher-directed student output (pre-VR and post-VR activities focused on the target language). Fragments of interactions during activities in both VR and the classroom were selected and transcribed. The Common European Framework of Reference for Languages: Learning, Teaching, Assessment Companion Volume (CEFR-CV) illustrative descriptors were then used to map the participants' language performance to the CEFR-CV scales. Findings indicate that production of the target language is less controlled and less accurate in VR compared to face-to-face "traditional" oral activities in the classroom. However, there are more examples of spontaneous language use, more examples of mediation between the learners and even moments of higher than expected levels of production and comprehension, equivalent to English language level B1 descriptors in VR.