Pragmatics and Society 3:2 (2012), 294–320. doi 10.1075/ps.3.2.08sid issn 1878–9714 / e-issn 1878–9722 © John Benjamins Publishing Company “Who knows best?” Evidentiality and epistemic asymmetry in conversation Jack Sidnell University of Toronto his essay reviews current work in conversation analysis with an eye to what it might contribute to the study of evidentiality and epistemic asymmetry. Ater a brief review of some aspects of the interactional organization of conversation, I turn to consider the way in which participants negotiate relative epistemic po- sitioning through the use of particular practices of speaking. he analytic focus here is on agreements and conirmations especially in assessment sequences. In conclusion, I consider a single case in which various practices are employed to convey a delicate balance of knowledge and simultaneously to attend to a range of other, non-epistemic, interactional issues. Keywords: Evidentiality, interaction, epistemic asymmetry, conversation analysis, assessments, questions Introduction Recent work in conversation analysis has shown that, in designing their contri- butions to some spate of talk, conversationalists take into account not only the diferential distribution of knowledge among participants but also the diferential distribution of rights and responsibilities to know (Heritage and Raymond 2005, Raymond and Heritage 2006, Stivers 2005, Stivers et al. forthc., among others). In what follows, I review some of that research, attempting to show how it might be relevant to the study of evidentials and evidentiality by linguists. he literature on evidentials indicates that all languages provide a range of resources that allow speakers to convey various degrees of ‘doubt’ and ‘certainty’ via meanings relating to source of knowledge. 1 As is well-known, some languages require a speciica- tion of source of knowledge in many sentence-types as a requirement of general