English Language MOOCs (LMOOCs) employ multimodal resources to enhance second language learners' engagement and motivation. This study examined the multimodal instructional discourse in English LMOOCs in a Chinese MOOC platform, focusing on the rhetorical relations between the linguistic mode, termed verbiage, and two non-linguistic modes, namely facial expressions and gestures. Employing Rhetorical Structure Theory (RST) as its theoretical framework, this study involved a multimodal discourse analysis of 12 English LMOOCs. Specifically, it analyzed and compared the distribution of facial expressions and gestures, and rhetorical relations between verbiage and facial expressions and between verbiage and gestures in six nationally accredited quality LMOOCs and six regular LMOOCs without such national accreditation. The results revealed a significant association between course type and the use of the two non-linguistic modes. In addition, "Elaboration," "Emphasis," and "Preparation" were the three relations identified in two types of modal synergy: Verbiage + Facial expressions (i.e., V + FEs) and Verbiage + Gestures (i.e., V + Gestures), the latter also containing "Restatement" relation. A significant association between course type and the distribution of rhetorical relations was only identified in the V + Gestures but not in the V + FEs. This study contributes important insights into how linguistic and non-linguistic modes work together for meaning-making in LMOOCs and provides evidence for the applicability of RST in analyzing multimodal online teaching. Implications for practitioners are finally addressed.