There is a strong case to be made for using telecollaboration in foreign language education: learners are engaged in regular, semi-authentic interaction; can develop strategies for learner independence (O'Rourke, 2007), be exposed to opportunities for negotiation of meaning (Kötter, 2003) whilst at the same time exchanging with 'real informants' of the target culture (O'Dowd, 2013). This qualitative study sets out to observe the role of multimodality in intercultural incomprehension episodes (IIEs) during webconference-supported teaching. These episodes are defined as opportunities in the online interaction (surprises, questions, miscommunications) for the participants to engage in periods of negotiation of meaning. They arise because the participant of one culture does not understand or misunderstands the member(s) of the other culture. During webconference-supported teaching, the immediacy of the visual images provided by the webcam may make a particular contribution to intercultural telecollaboration: the visual mode may mean that participants are unable to avoid or ignore awkward subjects (O'Dowd, 2006) but may also allow complementary information to be provided to help resolve IIEs. The pedagogical context for the study is the telecollaborative project ISMAEL (InteractionS and Multimodality in lAnguage LEarning) during which 12 trainee teachers of French as a foreign language met for online sessions in French with 18 undergraduate Business students from an Irish university. The participants met for seven 40-minute online sessions in 2013 via the webconferencing platform Visu (Bétrancourt, et al., 2011). Each online session was thematic and focused on Business French. A research protocol was designed around this pedagogical context. Data produced during the learning project itself were collected (webcam videos, textchat messages, audio recordings of collective feedback session with the trainee teachers, reflective reports), as well as data produced uniquely for the research project (observation notes, post-course questionnaires and interviews). Participation in the research study was voluntary-all 12 trainee teachers (ten females, two males) and 12 students (eight females, four males) gave permission to use their data. All data were structured into a LEarning and TEaching Corpus (Guichon et al., 2014).