This study examines emotion regulation strategies in written digital discussions revolving around controversial issues. Twenty-five undergraduate students, placed in five study groups, took part in written digital discussions. Two groups were chosen to participate in the study. Participants were interviewed and were asked to read the transcript of the digital conversation they took part in, while referring to all conversation turns. They were asked to explain their own, as well as others' reasoning regarding emotion expression and emotion intensity levels. Ninety-three interpretations of participants' turns were made during the interviews. We compared the ways composers labeled their own emotions and intensity levels, with the ways in which other participants' recognize these emotions, in order to assess the correlations between them. We report on several emotion recognition strategies that were found and point to the idiosyncratically rich but lacking in common ground nature of emotional social language. We highlight the gaps between composers' emotion labeling and others' emotion recognition. The study offers new insight regarding emotional communication in CSCL settings, claiming that despite poor correlation rates and lack of shared emotional language, participants were indeed able to communicate emotionally. In CSCL settings, emotions function as a dialogic instrument enabling people to relate to each other by fostering closeness and establishing relations.