People move in characteristic ways during conversation and these movements correlate with their level of particpation. For example, speakers normally gesture significantly more than listeners. These visible, overt movements are normally analysed using full body video or motion capture. Here we explore the potential of a 'minimal' approach to sensing these participatory movements in part of the natural environment of everyday interactions; chair seat covers. Using custom built fabric sensors we test whether we can detect people's involvement in a conversation using only pressure changes on the seats they are sitting in. We show that even from this impoverished data we can distinguish between talking, backchanneling and laughter; each state is associated with distinctive patterns of pressure change across the surface of the chair. We speculate on the possible applications of this new, unintrusive form of social sensing for architecture, performance and augmented human interaction.