This presentation will focus on the production of gestures by teacher trainees when they are interacting online with distant language learners during a desktop videoconferencing interaction. Previous research has provided a mixed picture of the use of the webcam in pedagogical situations: while it is seen as a useful medium to provide complementary information in communicative breakdown (Buckett, Stringer and Datta, 1999), the limited access to visual cues provided by the webcam is felt as limiting (Zähner, Fauverge and Wong, 2000) or even disruptive (Chanier and Ciekanski, 2008). Because the constraints brought by the webcam shape online communication, research is needed to understand how teachers manage with their own image (paralinguistic cues) and develop strategies to adjust their teaching. In line with the domain of Gesture Studies (McCafferty & Stam, 2008), the present qualitative study aims to understand how teacher trainees use the affordances of the webcam to produce nonverbal cues that may be useful for mutual comprehension. The data are comprised of (1) dynamic screenshots of 40-minute pedagogical interactions (what the learners can see on their screens), (2) films that were shot during the interactions (the context of the interaction invisible to the learners), and (3) stimulated recall sessions during which the same trainees were asked to explain their choices concerning framing and the production of paralinguistic cues (nods, gestures, smiles, etc.). The participants are three French teacher trainees who are being trained to teach online via a multimodal interface over a seven-week period via a telecollaborative project with Irish learners of French. This study will concentrate on the production of gestures that are in the frame and out of the frame and will try to determine the functions of both categories. We will then explore how future French teachers adapt their gestures and what their intentions are when doing so. We contend that attention to framing, that is what the online teacher chooses consciously or unconsciously to include in the restricted field of the webcam frame, is one crucial aspect of the contextualization process at work in an online synchronous interaction (Jones, 2004).