It is by now well established that self-organized acts of learning can have important long-term cognitive benefits. The many programs of cooperative learning, while they vary in detail, all agree that participants must actively work together for such benefits to be realized (Cohen, 1990). Further, because cognitive change is revealed through the ways in which learners' sense of ownership and control over their intellectual products is increased over time, the changes are not easily observed as they happen. Students generally work in small, informally organized groups, relying on their own everyday communicative practices. Instances of ongoing productive collaboration often seem fleeting or hard to detect. Apart from relying on such finished products as workbooks or written texts, teachers—who spend much of their time circulating in the classroom and answering questions as they come up—may have difficulty determining whether student groups are making progress. One of the teachers in our research group commented that watching the analyzed videotapes of groups at work enabled her to gain useful insights into how the children go about solving their tasks and how easily adults' questioning can influence students' attempts to integrate what they are learning with what they already know.
Our research on collaborative learning has been guided by questions such as the following: If we are looking for evidence of individual achievement in order to evaluate program success, how can we know we are looking at the right phenomena? Can we rely on evaluations that relate performance on standardized tests to incentive and reward structures? The validity of such measures has been seriously debated for several years (Durán, 1989; Gardner, 1990). Should we instead, as a means of gaining more direct insights into the potential benefits of cooperative learning, examine more closely how collaboration reveals itself in the classroom and analyze what children do in working together and how teachers meet the challenge of entering into informal interactions with student groups? That is the course we intend to take here.
We use the term collaboration to mean collaborative interaction in situations where two or more individuals are demonstrably working together toward the achievement of particular tasks. Recent work on organizational design and small group functioning with adults tells us that collaboration in work settings is regarded as difficult to achieve; it does not just happen when people are put together and required to do a task in unison (Galagher, Kraut, & Egido, 1990). A supportive social milieu and a task infrastructure are required. In this paper, we focus on collaboration as a group phenomenon in which complex tasks are managed through close, step-by-step, apparently casual monitoring by participants of each other's actions, often cued through language. It can be argued that with children, the need to provide support and task infrastructure is even greater than with adults (Tharp & Gallimore, 1988).
The advantages of student-guided learning are described in the recent literature on mathematics instruction, where cooperative programs have come to take an important place among the innovative methods designed to improve performance. They do this by creating a learning environment where students are able to make sense of new information by relating it to what they already know. Consider the following statement from a report by the National Council of Teachers of Mathematics:
"All students engage in a great deal of invention as they learn mathematics; they impose their own interpretation on what is presented, to create a theory that makes sense to them. Students do not learn simply a subset of what they have been shown. Instead, they use new information to modify their prior beliefs. As a consequence, each student's knowledge of mathematics is uniquely personal. (1991, p. 2)"
Brown and Campione (1993) have developed methods of instruction through guided discovery and participation that address some of these issues directly. Their classroom procedures are designed to provide a complex organization of several types of participation frames (Goodwin, 1990). They conceive of instruction as sequences of first small, then larger participatory groups, until the whole class is involved in what is called a "community of learners." Brown and Campione state that "the essence of team-work is pooling expertise," and they point out that in their classrooms, "we also aim at increasing diversity of expertise and interests so that members of the community can benefit from the increasing richness of knowledge available to them" (1993, p.8). They describe their methods of achieving instructional change as the creation of "sub-cultures of expertise," where students can do their own work but where they can share their findings in many ways, from addressing a small group to presenting in front of a whole classroom community. Although the person in the role of expert varies, sequences of presentation, discussion, and deliberation remain central to the activities. Whole class and group discussion sequences are built on a dialogic structure that usually follows a modified Socratic method, where one student is placed in the role of expert, and there is a main speaker and multiple listeners, as well as a single responder within a question-answer format ( Brown & Campione, 1993; Brown & Palinscar, 1989). The authors go on to point out that the difficult art of teaching in such classrooms requires the teacher to judge the situations and times at which adult intervention would provide guidance and not interference. They stress that in order to bring cooperation to the traditional classroom, it is necessary to adjust the balance between teacher-led discussion and group work, so that children can exchange information for themselves.
However, the Brown and Campione science education project has each group choose a single expert who directs the proceedings and so sets up a dominantly one-to-many participation frame. Moreover, much science instruction, like mathematics instruction, relies in part on certain instructional tools, such as numerical representations or geometrical models or laboratory equipment, to assist in conveying information. Students tend to work on problems where the cognitive operations are closely related to pre-established procedures or algorithms, so that the domain of expertise can more readily be delineated. Language arts programs, on the other hand, although able to emphasize different kinds of language genres, such as poems, story telling and writing, plays, play acting, and —for young children—the use of drawing to support writing, ultimately depend on words as the main tools for learning. Both instruction and evaluation of performance rely almost exclusively on talk, so that potentially everyone is an expert. For this reason, it becomes necessary to seek systematic information on how language or talk functions in such programs in actual classroom situations: that is, information about the communicative activities through which teachers teach a subject's content and through which students manage their own learning in order to understand how performance outcomes are achieved.
In cooperative learning environments, small groups of students work together to accomplish specific pedagogical tasks, and teachers act as facilitators (Slavin, 1990). Many of these classrooms rely on relatively informal communicative strategies that are conversational in form but instructional in intent, as teachers scaffold the learning process by restating and expanding on students' contributions (Tharp & Gallimore, 1988). Frequently made arguments in favor of these new informal methods are that they have much in common with the parent-child and peer-group interaction processes that researchers have found to be so important for children's learning and cognitive development (Rogoff, 1990).
One highly significant characteristic of cooperative learning that has received little consideration so far is the shift in participation frame that takes place when students are left alone to work on classroom tasks. In many traditional classrooms, the teacher directs the learning process, guiding the students through the various stages of a task by questioning and evaluating contributions at every stage in the process. Students' participation is usually confined to relatively brief replies that are expected to be responsive to the teacher and that are subject to close monitoring and evaluation by the teacher for their possible relevance (Mehan, 1979). In cooperative learning processes on the other hand, students are free to take their own time to work out their learning strategies, and they rely on peer group processes of the kind associated with home and play environments, both to establish collaboration and to guide their own learning. Support for cooperative learning processes comes from educational psychologists and anthropologists who, following Vygotsky's theories, see learning as an interactive process based in communities of practice in which groups of individuals collaborate in the pursuit of shared goals (Moll, 1992; Wertsch, 1985). It is in such exchanges that learning processes can be made into observable activities. Central to the case studies described here is the fact that everyday informal conversational exchanges play an essential role in group processes where one speaker is given primary rights of speaking, but where participants must compete for the floor and cooperate in achieving shared communicative tasks.