In this paper, we analyze interactions with tablets in a specific setting: during a children’s festival, the Lyon public library proposes a game session on a large tablet placed against a wall, small children and their (grand)parents participate in the presence of a library staff member, acting as activity leader. The accomplishment of actions on the touch screen requires a fine-grained motricity and corporeal control in order to shape specific hand configurations and body movements. Adults support children’s physical efforts and provide assistance through scaffolding practices: showing the right gestures, accompanying children’s arms and hands during a gesture realization, producing instructions and words of encouragement. Relying on a fine-grained interaction analysis, we propose a multimodal study of the development of children’s gestures on the tablet. Various learning phases can thus be tracked on a microgenetic scale.