Currently, online language teaching is experiencing global traction and there is growing body of research dedicated to the study of synchronous online pedagogical interactions (Develotte et al., 2008; Hampel & Stickler, 2012; Kozar, 2016; Guichon & Tellier, 2017). Several studies center their attention on comparing the communicative potentials of audioconferencing and videoconferencing conditions. Researchers have explored effective communication in videoconferencing-supported teaching (Develotte, et al., 2008; Cappellini, 2014; Yamada & Akaroli, 2009). However, some researchers (Yanguas, 2010; Guichon and Cohen, 2014) found that online language teaching and learning through audioconferencing could potentially improve and / or facilitate L2 acquisition or communication by eliminating the opportunity of relying on co-verbal devices. In other words, the importance of webcam was less crucial than expected, and there were no notable quantitative differences between these two communicative modes. Other studies have focused on online teachers’ online teaching skills and competencies (Guichon, 2013; Cappellini & Combe, 2017), which were defined as techno-semiotic-pedagogical by Guichon (2009). One of these competencies is delivering task instructions during synchronous sessions. Nonetheless, when engaged in online teaching through videoconferencing, a challenge encountered both by teachers and students is to activate their video cameras, impacting social presence and interaction (Sederevičciūtė-Pačiauskienė et al., 2022; Castelli &Sarvay, 2021), as well as L2 learning opportunities (Cohen & Wigham, 2019). This study aims to explore online foreign language teaching, focusing particularly on situations where two modes of interaction coexist within the same group: videoconferencing and audioconferencing. We will examine three French L2 tutoring sessions conducted in 2022 between trainee teachers at a French university and learners of B1 CECRL level at a Taiwanese university. In each session, one student interacted via through videoconferencing whilst the other through audioconferencing. Building on previous studies (Satar & Wigham, 2020, Satar & Wigham, 2023), we explore the instruction-giving phase for a role-playing task, in which videoconferencing and audioconferencing conditions coexist. In this study, we utilize multimodal (inter)action analysis (Norris 2004, 2019) to investigate how online trainee teacher deliver instructions for the same task to their learners when videoconferencing and audioconferencing conditions coexist. We will illustrate the composition of both instruction-giving higher-level actions and lower-level actions in the repeated task, comparing these two actions. We then explore exploring the similarities and / or differences in certain lower-level actions. Our analysis allows us to highlight trainee teachers’ approaches to instruction delivery when videoconferencing and audioconferencing conditions coexist.