themselves as victims in order to justify their domination of marginalized groups can point to research as evidence that they, too, are oppressed. Similarly, those seeking to say everyone is equally oppressed can use the equivalence constructed by hate speech definitions to prove their point. Indeed the phrase “military dogs” has been used in precisely this way, to present the military as a victim, unfairly criticized and in need of support.2 The third problem is conceptual: To look for “counter-speech” can suggest that one must look for an antecedent trigger, to which the speech is a directly countering response. This is itself a kind of semantic formula that lends itself to a research process: find speech defined as “hatespeech” and then seek out responses, which can be, as a result, defined as counter-speech. But this formula is too limiting; the research process would be likely to miss too much. In Myanmar there is a broad awareness that there is much speech targeting certain marginalized groups, whether or not people name this fact as “hate speech” or rude language, insults, verbal abuse, harassment, etc. There is also much speech that is motivated by awareness of this context, even if it does not openly highlight such a motivation. Calls for inter-faith coexistence, for example, appear for a reason: they are a response to a situation in which such coexistence seems fragile and there are groups known to be working against it. For this reason, we sought to study counterspeech as something that includes both direct responses to instances of hate speech and also speech that takes stances on behalf of marginalized groups, even where it was not linked in any direct way to an instance of hate speech. To be clear, consideration of these three risks does not mean that “hate speech” and “counter speech” are not useful categories for researchers; this paper will discuss both. The point is only that we sought to be aware of the way that the categories with which we began our research could shape our findings. Rather than begin by defining categories of speech and then seeking examples of speech that fit the criteria of the categories, we instead decided to begin by focusing on groups that are often targeted by this speech. In this paper we thus discuss some of the ways that members of marginalized religious, ethnic, and LGBTQ communities manage the fact that they live in a world where speech that they find concerning or objectionable is widespread, whether or not they call this “hate speech” or respond directly. Our aim thus is to bring within the same frame, on the one hand, examples that would obviously meet the semantic criteria for 2 For example, in a recent music video and song produced by the celebrity Yone Lay, the singer laments those who say “military dogs,” equating them with bringing the downfall of the nation. This is a small part of a song that works to present business figures, the USDP, and military as unfairly maligned and harassed by the NLD; the effect is to claim “both sides” are equally victims and equally at fault and the military is a neutral and necessary savior. In order to accomplish this, the insult “military dogs” must be understood as equal to insults directed at the NLD. This construction of equivalence mirrors the one produced by “hate speech” definitions. For a discussion of the song and reactions by audiences in Myanmar, which roundly mocked the film and prompted some of the musicians involved to disavow their knowledge of its overall message, see “Singers in Myanmar and Vietnam Pay Price For Hitting Sour Notes With Political Views,” Radio Free Asia, 11th August 2020 <https://www.rfa.org/english/news/myanmar/singers-08112020203701.html> 4