This paper opens the “black box” of real-world deliberation by using text-as-data methods on a corpus of transcripts from the constitutionally mandated gram sabhas, or village assemblies, of rural India. Drawing on normative theories of deliberation, we identify empirical standards for “good” deliberation based on one’s ability both to speak and to be heard, and use natural language processing methods to generate these measures. We first show that, even in the rural Indian context, these assemblies are not mere “talking shops,” but rather provide opportunities for citizens to challenge their elected officials, demand transparency, and provide information about local development needs. Second, we find that women are at a disadvantage relative to men; they are less likely to speak, set the agenda, and receive a relevant response from state officials. And finally, we show that quotas for women for village presidencies improve the likelihood that female citizens are heard.