We had completed work in two cities when COVID-19 disrupted our plan: a total of ten one-onone interviews in Yangon and eight focus-group discussions in Lashio3. The interviews and focus-group discussions used a set of question prompts to structure the conversations while striving to allow a fairly free flow discussion. Firstly, the interviews commenced by asking the participants about their experiences with situations in which they encountered verbal acts that they would consider abusive, discriminatory, or hate speech. After that, questions encouraged participants to discuss the ways they managed such situations, including whether or not they sought to counter the speech directly and how. The interview then concluded with questions about their general perceptions of “hate-speech” as a category and how people can, do, or should respond. In response to disruptions from COVID-19, we decided to re-allocate our time to gathering cases of counter-speech on Facebook. Cases consisted of posts on Facebook and the trailing comment threads that they anchored. In total, we analyzed 17 case studies of Facebook posts with comment threads ranging from 118 to 7,000 comments, with an average of approximately 1,800.4 These cases were chosen from over a hundred posts that were considered over a period of three months (April-June 2020). In general, we sought out posts that included both speech directed at one of the marginalized groups we had decided to focus on, and responses of some kind. As a result, we made studies of comments on, broadly, three kinds of original anchoring posts: 1) Anchor posts that would easily meet the criteria of “hate speech”; 2) Anchor posts that were themselves a kind of counter-speech, that is, posts that seemed motivated in some way by the situation of the marginalized population, and attempted to address this marginalization; 3) Anchor posts from news organizations, with news articles related to the marginalized population in some way. Selecting cases to focus on was an iterative process, in which ideas and questions prompted by analysis of initial cases also suggested other kinds of cases to seek out. It was this approach, for example, that clarified for us the third risk that research on counter-speech can encounter and prompted us to expand our study from a standard type, in which the original post is hate speech and responses appear in the comments, to other kinds of anchor posts that would not meet any possible definition of “hate speech” but included counter-speech in the comments. In analyzing the cases, we looked at the setting of the comments (i.e. the nature of the anchoring post and the identity of the posting account); the sequencing of comments (i.e. the ways in which a comment responded to other comments, to the original post, and/ or to other potential audiences); and the cultural resources involved in the comments (i.e., the assumptions and beliefs that the comments relied upon to make sense or might otherwise link to). 3 See annex 1 for complete list of interviewees’ identities 4 See Annex 2 for list of posts that were selected 6