The paper presents the results of an ongoing research of Twitter discussions on inter-ethnic conflicts. The case of Biryulevo bashings is already thoroughly analyzed by the research group; in this paper, we develop a qualitative method for finding patterns of emotional (irrational) argumentation, the patterns that would link emotion and argumentation. Our pilot empirical study is based on 306 tweets of 7 top non-media users; the tweets are analyzed qualitatively for the spectrum of emotions for conflict cases – sadness/sorrow, surprise/fear, anger/indignation, as well as for irony and call to action for complete reconstruction of the chains of emotions. We find that there are two patterns in tweeting strategies – broadcasting the development of the conflict and ‘broadcasting’ emotions of individuals that reflect collective ones. The total number of expressed emotions was counted, followed by reconstructing chains of emotions chronologically. There are three groups of patterns: starting and ending with neutral informing; starting and ending with one certain emotion, no clear pattern with different emotions in the start and in the end. Despite we find only a few sustainable patterns, we argue that further computational research can be done to connect the type of a top user and his/her emotional patterns.