Research in cognitive and affective neuroscience has come to back up language teachers’ observations and intuitions: the interactions produced during playful and immersive activities within educational environments have a positive effect on the cognitive development of learners (Immordino-Yang, 2016). In the 1990s, neuroscientist Varela (1993; 1996) developed the enaction paradigm falling within the embodied cognition theory to define how meaning and knowledge are created in the mind through bodies and emotions. This paradigm has enabled to study scientifically the positive effects of sensory-motor devices such as 3D virtual worlds in educational contexts (Deutschmann & Panichi, 2013; Berti, 2019). Using virtual reality applications like Immersive Virtual Environments (IVEs) which require the use of an avatar (Privas-Bréauté, 2016) creates new relations between perception, cognition and action which support language learning/ teaching (Molle et al., 2020; Peterson, 2017) and helps reshape the contours of interactions and social activities including travels and cultural events in a pandemic context. Language teachers thus need to be trained to this new educational device ( Abendroth-Timmer, 2017; Cicurel, 2011) My research question focuses on the perception of the educational potential of virtual reality and the use of avatars, contemporary metaphors for the learners’/ users’ bodies, which enhance immersion and engagement in communicative situations through engagement, agency, and motivation. I have recently led pedagogical experiments with in-service English teachers to study their perception of VR in the context of language teacher training. This presentation will explore their perception of the creation of new learning environments in virtual reality facilitating language acquisition considering the enaction paradigm and embodiment theories. The data collected along the experiments (made up of the participants’ answers to pre- and post-questionnaires as well as photos and videos) will shed light on the need to train teachers devise educational scenarios based on the accomplishment of tasks (Chen & Kent, 2019).