Distributed collaborative teams increasingly rely on online tools for interaction and communication in both social and task-oriented goals. Measuring and modeling these interactions along different dimensions can help understand, and better design for, distributed collaboration. Affect is one such dimension that can play a crucial role in the dynamics, creativity, and productivity of distributed groups. We contribute an adaptation of the grounded theory methodology as a flexible and extensible

more »

... eans for constructing a taxonomy of affect in text-based online communication. Such a taxonomy can serve as an analytic lens for the continued investigation of the role of affect in creative collaborative endeavors as mediated by communication technology. We describe our modified grounded theory approach and then validate our method by constructing a taxonomy with data from chat logs collected during a longitudinal study of a multi-cultural distributed scientific collaboration. face-to-face communication [17] . A large portion of this communication takes the form of text, including emails, text messages, and instant messaging chats. These online, text-based forms of communication have become ubiquitous and constitute one of the most important means of contact between members of many distributed groups. Synchronous online chat differs from other online text-based communication media, such as email, in several significant ways. Since conversations can take place in real time, they capture some of the synchronicity that is associated with face-to-face or voice communication. This synchronicity can greatly enhance the effectiveness and efficiency of this mode of communication by allowing for real-time interaction. However, unlike other real-time modes of communication, text-based chat also has benefits associated with asynchronous communication. Messages can be replied to at the convenience of the correspondents or as dictated by the circumstances of the tasks being performed. Additionally, all of the messages can be logged, providing a persistent record of the conversation. In these regards, this kind of text-based online chat can offer positive aspects of both synchronous and asynchronous forms of communication. The trace created by using text-based chat communication to mediate creative problem solving can be studied to better understand collaboration. Affect and mood influence creative performance both in individual and collaborative environments [1, 13, 26] . Here, we use Russ's definition of affect, 'a feeling or emotion as distinct from cognition ' [26]; affect is thus more pervasive than the interrupting neurophysiological experiences of emotions [21] . The expression of affect still plays an important role in these text-based forms of communication, but it takes on forms that are distinct from those found in face-to-face communication [22] . Affect-laden words, emoticons, special abbreviations, deformed spellings, punctuation, and interjections are just a few of the many ways in which the expression of affect has been adapted to text-based forms [15, 17, 18] . These signals, embedded in a detailed trace over time, can help measure the quality and quantity of affect expression in text-based communication between members of a distributed group. The utility of this measurement for further analyses depends on how robust the analytic lens is to the effects that the specific communication medium has on the character of affect expression. Existing taxonomies of affect and emotion focus primarily on classifying psychophysiological responses to internal states and environmental factors [4, 11, 22] . There are several conflicting theoretical models of affect and emotion in multiple fields. Examples include the dimensional models of Russell [27], the emotion wheel of Plutchik [22] , and the distinction between basic and complex emotions [4, 11] . Emotion is often measured via facial expression, vocal features, and body posture as the physical expression of emotion (rather than focusing on internal state) [10] .