On the basis of an interdisciplinary approach informed by ethnographic methods, and computer-mediated discourse analysis, this paper sets out to explore the relationship between multimodal and multilingual practices in Computer-Mediated English as a Lingua Franca (CMELF) communication on Facebook. As the analysis of some naturally occurring interactions shows, CMELF users exploit multimodal resources in novel ways to achieve mutual understanding in intercultural communication. For instance, when a target term in English is unknown to the subject utterer, a corresponding picture can be searched for on the Internet in their own L1, to be then embedded in the CMELF utterance and fill the lexical gap. Multimodal resources, however, can also be employed as part of pragmatic strategies to proactively avoid misunderstanding: emojis, for example, can be used to flag words which are infrequent and therefore potentially problematic for the intended multicultural audience, while pictures can gloss these visually, thus enhancing utterance clarity.