Despite continued popular support for democratic ideals, democratic regimes throughout the world are under threat. Fragmentation of the public sphere, and resource scarcity has hindered governors and publics in their efforts to communicate with one another. Advances in computing now provide opportunities to make sense of the abundance of information available, with the potential for identifying and mapping argument structures and disagreements as they emerge at scale. This talk discusses how natural language processing and data science techniques for multimodal analyses, from text, audio and video, can model deliberative discourse, analysing arguments, facts and emotion to support inclusive and informed political deliberation.