One important pedagogical skill for language teachers is giving task instructions. Research in this area has largely employed Conversation Analysis to investigate instructions in face-to-face classrooms. Studies have investigated different components of instructions (Markee, 2015), their interactive nature (Somuncu & Sert, in press), and teachers’ verbalisations of written instructions (Ha & Wanphet, 2016). Combining grounded theory (Strauss & Corbin, 1998) and multimodal (inter)action analysis (Norris, 2004), this paper explores the key components of experienced language teachers’ webcam-mediated task instructions and the deployment of embodied and online resources during these sequences. The paper builds on Satar & Wigham (2017) into webcam-mediated instructions that showed teacher-trainees’ use of semiotic resources to allocate roles, introduce key vocabulary, and mark different instructions stages for a role-play task. Here, we examine a semi-controlled corpus comprising three English as a Foreign Language lessons given by three experienced online teachers to pairs of learners for the same convergent information-exchange task. To explore multimodal instruction-giving practices, our presentation illustrates how one teacher achieves modal density through combining online (Google docs and text-chat) and embodied resources (gaze, gestures, and facial expressions) to clarify key task information and suggest a potential way into task.