Lea, M., & Spears, R. (1992). Paralanguage and social perception in computer-mediated communication. Journal of Organizational Computing and Electronic Commerce, 2(3), 321-341. doi: 10.1080/10919399209540190.
This paper was referenced by the another one I read in reference to the social identity/de-individuation (SIDE) model of CMC. Turns out there are earlier papers about that model that I should probably follow up on, but this paper does include some interesting findings that may help me to understand it!
They did two studies, the first of which looked at paralinguistic cues in e-mail and how they affected the perception of the communicators, and compared the use and interpretation of paralinguistic cues between novice and experienced users.
The second study got groups to take part in discussions via a text-only computer-conferencing system. The groups were given 4 different conditions: the group identity or the individual identity were highlighted (group salience), and they could see each other or they were in different rooms (de-individuation).
Looking at how they switch which identity is most salient could be useful. They did it in quite straight-forward ways, by manipulating the way the discussion was presented to the group. So group identity was pushed by being told they were being tested as a group, and the instructions referred to individuals as 'group members'. Each user was given a 'group membership number' to be identified by. In the individual identity groups, they were told they were being tested personally, and the instructions referred to 'participants'. Rather than a 'group membership number' they just had a 'participant number'.
They measured frequency of paralinguistic marks (ellipses, inverted commas, quotation marks, exclamation marks. Sequences/combinations double-weighted) and got people to rate each other on person-perception scales. The correlations between paralanguage use and perceived personal attributes were interesting. When subjects were in the same room as each other, the correlations don't really change regardless of the high or low group salience. In the condition where the participants couldn't see each other, high group salience gave positive correlation to the use of paralanguage, where as the low group salience gave a negative correlation. In other words, people who used paralanguage in a group were liked when everyone recognised themselves as part of a group, and disliked when people perceived themselves as individuals.
This paper extended work done in empirical studies from a social identity approach into the computer-mediated communication environment.