Understanding and analyzing chat in CSCL as reading's work

Synchronous communication using text chat—often combined with a shared whiteboard—is increasingly used in CSCL. This form of interaction and learning in small online groups of students presents novel challenges, both for the participating students and for researchers studying their work. Chats differ from talk-in-interaction since the composition, posting and visual inspection of text and graphical objects by any given actor is not observable by the other participants. These structural constraints on the organization of interaction require that actors deploy alternative procedures for achieving what turn taking achieves in talk-in-interaction. This paper describes how communication is organized in text chat, where postings have to provide instructions on how they are to be read. This organization is contrasted with turn taking in face-to-face communication. The notion of “reading’s work” provides a guiding thread, which is explicated. Synchronous text chat seems to offer a particularly effective medium for joint learning activity, but it presents challenges for students using it because the group sense-making strategies to which they are accustomed in faceto-face interactions either do not work or work differently in the online setting, for reasons to be discussed. Similarly, sources of evidence normally available to educational researchers are not available from synchronous online interactions. Student groups develop methods of building their postings to convey instructions for how they are to be read. Chat researchers as well as chat participants must learn to pay heed to such instructions. As we will see, such instructions often involve the sequentiality of postings and the references among those postings (Sarmiento & Stahl, 2008). We have explored the use of text chat for discussions of mathematics in small groups of students for the past six years in the Virtual Math Teams Project (VMT) (Stahl, 2009). Although we have reported on the technology design (Stahl, 2008) and on various case studies (e.g., Stahl, 2006; Zhou, Zemel & Stahl, 2007), this paper is our first attempt to describe the systematic workings of text chat, building on previous publications that began to clear the ground for this. We call this description a “simplest systematics” in analogy to the seminal paper on turn taking in talk by the pioneers of ethnomethodology and conversation analysis (Sacks, Schegloff & Jefferson, 1974), which we take as our inspiration. Interacting Through Text Chat In CSCL online chat systems like VMT, participants can engage with each other in a variety of ways. Rather than interact through emergent talk and observable embodied action, they: exchange text postings through chat technology, post text or graphic elements on a virtual whiteboard and/or use referential tools provided by the system. Interaction in VMT involves actors using computer hardware and software in ways that allow for the production of shared, displayed representations or virtual objects possessing various features that allow these objects to serve as the means by which participants interact. Participants are represented in various ways in VMT in terms of various conventions and practices of action identification. These representations—i.e., naming conventions and displays, avatars, authored messages, posted graphical objects, etc., as well as various changes in the appearance of objects or the state of the system—provide documentary evidence of actor presence (Zhao, 2003) and engagement with the system. It is these same resources that are put to work to constitute social interaction among actors in a chat. In other words, it is through the mediated exchange of what can be seen as locally relevant textual and graphical resources that chat participants organize and constitute their interaction. The problem that chat participants face in task-directed chats of the sort we inspect is to coordinate their participation to collectively and collaboratively perform the task with the technical resources available in the hardware and software and with the textual and graphical resources they construct as relevant to their ongoing tasks. As it happens, this is a challenging problem that involves the management of and allocation of attention across multiple interface areas of the chat system and the ability to produce domain relevant artifacts (text messages, graphical artifacts, etc.) for inspection by others participating in the exchange of such artifacts. Because these systems are designed in ways that allow participants to produce and inspect visual artifacts in particular ways, a natural question arises as to the nature of interaction that emerges in such environments. How do these interactions differ from talk-in-interaction? Speech exchange systems, like face-toface conversation, telephony, video conferencing, etc., exploit and are constrained by the technical affordances of speech production. As Sacks, Schegloff & Jefferson (1974) described, speech exchange systems rely on the affordances of the technology of talk to organize social interaction. The sequential organization of face-to-face conversational speech exchange is a product of the fact that actors are co-present to each other in an embodied way, which necessitates taking turns at listening and speaking. Thus actors allocate opportunities to speak and to listen in various ways such that one speaker speaks at a time and they repair problems of intelligibility that arise from mishearings, poorly produced speech and overlapping speech. Chat environments, on the other hand, are not speech-exchange systems at all, but rather systems of interaction that involve the display and inspection of visual artifacts, including texts (Garcia & Jacobs, 1999). The sequential organization of the production of visual artifacts is observable, available and documentable—and is something to which chat participants orient in their ongoing engagement in and through chat. However, the sequential organization of chat is not based on the same considerations that govern the sequential organization of talk-in-interaction (Garcia & Jacobs, 1999). One obvious difference is that overlap can happen in talk but cannot happen in most kinds of chat systems. Overlap is a phenomenon of talk in interaction. Problems of hearability, problems related to the allocation of turns in talk, problems that provide for repair in talk in interaction simply do not occur in chat. Different kinds of interesting troubles with respect to the intelligibility of postings can and do occur in chat, but these have to do with sequential placement of postings and other displayed graphical artifacts. It is because of this and other differences in the technical production possibilities afforded by chat systems that we feel compelled to provide the beginnings of a simplest systematics of online chat and to describe some of the ways that interactions through online chat differ from interactions through speech.