Understanding Educational Computational Artifacts Across Community Boundaries

Abstract

Viewed within its activity system, learning is a social process in which artifacts – whether physical, digital or linguistic – play central roles. Even individual learning is immersed in contexts of collaborative learning, in which communities define structures of meaning, goals of research, distribution of tasks and audiences for new knowledge. The field of Computer Supported Collaborative Learning (CSCL) is devoted to designing and evaluating artifacts such as communication media and digital simulations that foster learning by groups in schools. The artifacts in question must be understood by three communities: their designers, their users and their researchers. As meaningful physical objects, artifacts by definition both provide persistence across the three communities and require interpretation by each community. The first community designs into the artifact meaningful affordances that must be properly understood in practice by the second community. To evaluate the success of this undertaking, the third community must interpret the designed affordances and also interpret the users’ practical understandings of these.

As researchers, we take a detailed look in this paper at how a group of middle school students understands a digital simulation as they collaboratively struggle to use it to solve a scientific task. We focus on a particular moment of collaboration lasting 17 seconds that was particularly hard to understand from the transcript. A micro discourse analysis of this moment demonstrates that the students were engaged in making visible to each other the structure of references within their discourse that had become problematic for them as a group engaged in collaborative learning within a classroom activity structure. In making their learning visible to themselves, they made it visible to us as well. Furthermore, they made visible the central affordance of the artifact, that had until then eluded them and caused their group confusion.

The students constructed a shared understanding by making explicit the references from their discourse that had created confusion when different students had constructed divergent interpretations. To make their learning visible to us as researchers, we deconstruct the references within their discourse. The meaning that the participants constructed is analyzed as constituting a network of semantic references within the group interaction, rather than as mental representations of individuals. No assumptions about mental states or representations are required or relevant to the researcher’s analysis. Collaborative learning is viewed as the interactive construction of this network. Shared understanding consists in the alignment of utterances, evidencing agreement concerning their referents.

The world, situation or activity structure in which the students operate consists of a shared network of references among words and artifacts. To design new artifacts for these worlds, designers must understand the nature of these referential networks, build artifacts that fit into and extend these networks in pedagogically desirable ways, and provide tasks and social practices that will lead students to incorporate the artifact’s new references meaningfully into their shared understandings. Researchers who understand this process can analyze the artifact affordances and the situated student discourse to assess the effectiveness of CSCL technologies. The theory sketched here implies a methodology for CSCL design, practice and research that goes beyond the scope of this paper; here we will focus on the concrete, empirical discourse analysis to illustrate how students collaboratively constitute the referential networks in which they interact and comprehend collaboratively.

Methodological Introduction

Computational artifacts such as scientific simulations, productivity software, organizational knowledge repositories and educational systems are designed by one community (e.g., software developers, educators, domain experts or former employees) for use by another (end-users, students, novices or future employees). The two communities typically operate within contrasting cultures; their shared artifacts must cross cultural boundaries to be effective. Diversity among interacting communities of practice leads to many of the same issues and misunderstandings as cultural diversity among traditional communities.

A computational artifact embodies meaning in its design, its content and its modes of use. This meaning originates in the goals, theories, history, assumptions, tacit understandings, practices and technologies of the artifact’s design community. A user community must activate an understanding of the artifact’s meaning within their own community practices and cultural-historical contexts. Given the diversity between the design and user communities, the question arises: how can the meaning embodied in a computational artifact be activated with sufficient continuity that it fulfills its intended function? A further question for us as researchers is how we as members of a third community can assess the extent to which the designers’ intentions were achieved in the students’ accomplishments.

This paper investigates the process of meaning-activation of computational artifacts through an empirical approach: It conducts a micro-ethnographic analysis of an interaction among middle school students learning how to isolate variables in a computer simulation. The analytic affordances designed into the computational simulation of rocket launches were activated through the involvement of the students in a specific project activity. Their increasing understanding of the artifact’s meaning structure was achieved in group discourse situated within their artifact-centered activity.

This micro-ethnographic analysis is a scientific enterprise, like viewing under a microscope the world within a drop of water, a world that is never seen while crossing the ocean by boat. We try to uncover general structures of the interaction that would be applicable to other cases and that thereby contribute to a theoretical understanding of collaboration. The conversational structures of small group collaboration are different from those of dialog commonly analyzed by discourse analysts, and this has implications for the theory of collaborative learning and of Computer Supported Collaborative Learning (CSCL) (Stahl, 2000, 2002b).

This approach to studying collaboration differs radically from both traditional educational research and from quantitative studies in CSCL, both of which can produce useful complementary findings. Experiments in the Thorndikian tradition focus on pre- and post-test behaviors, inferring from changes what kinds of learning took place in between. Such a methodology is the direct consequence of taking learning for an internal individual mental process that cannot directly be observed (Koschmann, 2002). However, if we postulate learning to be a social process, then the conditions are very different. In fact, it is not only necessary for the participants in a collaboration to make their evolving understandings visible to each other, this is the very essence of collaborative interaction. As we will see in a moment, when the evolving learning of the group is not displayed in a coherent manner everyone’s efforts become directed to producing an evident and mutually understood presentation of shared knowledge. That is, in the breakdown case the structures that are normally invisible suddenly appear as matters of the utmost concern to the participants, who then make explicit and visible to one another the meaning that their utterances have for them. As researchers who share a cultural literacy with the participants, we can take advantage of such displays to formulate and support our analyses.

Quantitative studies of collaboration are indispensable for uncovering, exploring and documenting communication structures. However, they cannot tell the whole story. Although measures of utterances and their sequences – such as frequency graphs of notes and thread lengths in discussion forums – do study the processes in which collaborative learning is constructed and displayed, they sacrifice the meaningful content of the discussion in favor of its objective form (Stahl, 2002a). This not only reifies and reduces the complex interactions to one or two of their simplest dimensions, it even eliminates most of the evidence for the studied structural relationships among the utterances. For instance, the content might indicate that two formally distinct threads are actually closely related in terms of their ideas, actors or approach. Coding utterances along these characteristics can help in a limited way, but is still reductive of the richness of the data. Similarly, social network analysis (Scott, 1991; Wasserman & Faust, 1992) can indicate who is talking to whom and who is interacting in a central or a peripheral way within a network of subgroups, but it also necessarily ignores much of the available data – namely the meaningful content – that may be relevant to the very issues that the analysis explores. We will look at a set of utterances that would be impossible to code or to analyze statistically; the structural roles of the individual utterances and even the way they create subgroup allegiances only become clear after considerable interpretive effort.

The other way in which both traditional experimental method and narrow discourse analysis tend to underestimate their subject matter is to exclude consideration of the social and material context. Some approaches methodically remove such factors by conducting controlled experiments in the laboratory (as though this were not itself a social setting) or basing their findings strictly on a delimited verbal transcript. Fortunately, countervailing trends are emphasizing the importance of in situ studies and the roles of physical factors, including both participant bodily gestures and mediating artifacts. Increasingly, the field is recognizing the importance of looking at knowledge distributed among people and artifacts, of studying the group or social unit of analysis and of taking into account historical and cultural influences. In our data it is impossible to separate the words from the artifact that they reference and interpret; we will see that artifacts are just as much in need of interpretation (by the participants and by the researchers) as are the utterances, which cannot be understood in isolation from physical and verbal artifacts.

The study of collaborative learning must be a highly interdisciplinary business. It involves issues of pedagogy, software design, technical implementation, cognitive theories, social theories, experimental method, working with teachers and students, and the practicalities of recording and analyzing classroom data. Methodologically, it at least needs its own unique intertwining of quantitative and qualitative methods. For instance, the results of a thread frequency study or a social network analysis might suggest a mini-analysis of the discourse during a certain interaction or among certain actors. Interpretive themes from this might in turn call for a controlled experiment with statistical analysis to explore alternative causal explanations. In this paper we present an attempt to uncover in empirical data the sort of meaning relationships that other methods ignore, but that might enrich their analysis.

What’s in a Sentence Fragment?

We naively assume that to say something is to express a complete thought. However, if we look closely at what passes for normal speech we see that what is said is never the complete thing. Conversation analysts are well aware of this, and that is a major reason why they insist on carefully transcribing what is said, not forcing it into whole sentences that look like written language. The transcript of our moment is striking in that most of the utterances (or conversational turns) consist of only one to four words.

Utterances are radically situated. In our analysis we will characterize spoken utterances as indexical, elliptical and projective. As we will see, they rely for their meaning on the context in which they are said, for they make implicit reference to elements of the present situation. We will refer to this as indexicality. In addition, an individual utterance rarely stands on its own; it is part of an on-going history. The current utterance does not repeat references that were already expressed in the past, for that would be unnecessarily redundant and spoken language is highly efficient. We say that the utterance is elliptical because it seems to be missing pieces that are, however, given by its past. In addition, what is said is motivated by an orientation toward a desired future state. We say that it is projective because it projects the discussion in the direction of some future which it thereby projects for the participants in the discussion. Thus, an utterance is never complete in isolation. This is true in principle. To utter a single word is to imply a whole language – and a whole history of lived experience on which it is grounded (Merleau-Ponty, 1945/2002). The meaning of the word depends on its relationships to all the words (in the current context and in the lived language) with which it has co-occurred – including, recursively, the relationships of those words to all the words with which they co-occurred. We will see the importance of co-occurrences for determining meaning within a discourse.

In analyzing the episode that we refer to as “the collaborative moment” in this paper, we make no distinction between “conversation analysis,” “discourse analysis” or “micro-ethnography” as distinct research traditions, but adopt what might best be called “human interaction analysis” (Jordan & Henderson, 1995). This methodology builds on a convergence of conversation analysis (Sacks, 1992), ethnomethodology (Garfinkel, 1967), nonverbal communication (Birdwhistell, 1970), and context analysis (Kendon, 1990). An integration of these methods has only recently become feasible with the availability of videotaping and digitization that records human interactions and facilitates their detailed analysis. It involves close attention to the role that various micro-behaviors – such as turn-taking, participation structures, gaze, posture, gestures, and manipulation of artifacts – play in the tacit organization of interpersonal interactions. Utterances made in interaction are analyzed as to how they shape and are shaped by the mutually intelligible encounter itself – rather than being taken as expressions of individuals’ psychological intentions or of external social rules (Streeck, 1983). In particular, many of the utterances we analyze are little more than verbal gestures on their way to becoming symbolic action; they are understood as not only representing or expressing, but as constituting socially shared knowledge (LeBaron & Streeck, 2000).

We worked for over a year (2000/2001) to analyze a video tape of students learning to use a computer simulation (on March 10, 1988). I say “we” because I could never have interpreted this on my own even if I had already known all that I learned from my collaborators in this process. The effort involved faculty and graduate students in computer science, communication, education, philosophy and cognitive science as well as various audiences to which we presented our data and thoughts at the University of Colorado at Boulder. It included a collaborative seminar on digital cognitive artifacts; we hypothesized that this video might show a group learning the meaning of a computer-based artifact collaboratively and hence potentially visibly.[1]

We logged the three hours of video, digitized interesting passages, conducted several data sessions with diverse audiences and struggled to understand what the participants were up to. Despite much progress with the rest of the learning session, one brief moment stubbornly resisted explanation. The closer we looked, the more questions loomed. In the following, we pursue a limited inquiry into the structure of that single moment. We try to understand what people meant by individual words and sentence fragments that they spoke.

The Complexity of Small Group Collaboration

Conversation analysis has largely focused on dyads of people talking (Sacks, 1992). It has found that people tend to take turns speaking, although they overlap each other in significant ways. Turn-taking is a well-practiced art; it provides the major structure of a conversation. The talk is often best analyzed into conversation pairs, such as question/answer, where one person says the initial part of a pair and the other responds with the standard complement to that kind of speech act. These pairs can be interrupted (recursively) with other genres of speech, including other conversation pairs that play a role within the primary pair (Duranti, 1998).

In much of the three-hour SimRocket tape from which our moment is excerpted, talk takes place between the teacher posing questions and one of the students proposing a response. The teacher indicates satisfaction or dissatisfaction with the response and then proceeds to another conversation pair. This is, of course, a typical classroom pattern (Lemke, 1990). In the collaborative moment, something very different takes place. Collaboration is a many-to-many interaction in which meaning occurs at the group level; it departs from the teacher-centric dialog and teacher-interpreted meanings. It somehow overcomes the sequentiality of directed turn-taking – the promise of CSCL is to provide media that allow people to handle the increased complexity of collaborative interaction and temporality.

Let us take a first look at the episode of the transcribed moment.[2] The group of 11-year-old boys is discussing a list describing eight different rockets that can be used in a rocket launch simulation (see Figure 1). They are trying to come up with a pair of rockets that can be used experimentally to determine whether a rounded or a pointed nose cone will perform better. The moment is concerned with the students noticing that rockets 1 and 2 have the identical engine, fins and body, but different nose cones, while rockets 3 and 4 differ only in number of fins.

About an hour and a half into the classroom session the teacher poses a question. For the past few minutes, the teacher has been dialoging primarily with Chuck, who has gone off describing some imaginary rockets he would like to design for the simulation to solve the problem of the nose cone. The teacher’s question, accompanied by his emphatic gesture at the computer, succeeds in reorienting the group to the list on the screen. After a significant pause during which Chuck does not respond to this question that interrupted his train of thought, Steven and Jamie utter responses as though talking to themselves and then simultaneously repeat, as if to emphasize that they have taken the floor. But their response was to disagree with the teacher, something not so common in a classroom. So the teacher restates his question, clarifying what it would take to justify an answer. Chuck responds in a confusing way, not directly answering the question, but attempting to apply the criteria the teacher has put forward.

1:21:53	Teacher	And (0.1) you don’t have anything like that there?
1:21:54		(2.0)
1:21:56	Steven	I don’t think so
1:21:57	Jamie	Not with the same engine
1:21:58	Steven	┌ No
	Jamie	└ Not with the same
1:21:59	Teacher	With the same engine … but with a different (0.1) … nose cone?=
1:22:01	Chuck	┌ =the same=
	Jamie	└ =Yeah,
1:22:02	Chuck	These are both (0.8) the same thing
1:22:03		(1.0)
1:22:04	Teacher	Aw┌ right
1:22:05	Brent	└ This one’s different

The teacher pauses at 1:22:03, encouraging student discussion, and Brent jumps in, cutting the teacher off, lurching forward and pointing at a specific part of the list artifact, while responding to the teacher’s quest for something “different” (see Figure 2). For the next 16 turns, the teacher is silent and the students rapidly interact, interjecting very short, excited utterances in a complex pattern of agreements and disagreements. From the conversational structure, one sees that the standard, highly controlled and teacher-centric dialog has been momentarily broken and a more complex, collaborative interaction has sprung forth. Normally reticent, Brent has excitedly rocked forward off his chair, pushed through a line of students, filled a void left by the teacher and directed attention pointedly at the artifact.

Dramatically transforming the stage within which talk takes place, Brent has signaled an urgent need to resolve some disturbing confusion. We can see the importance of this move in the bodily behavior of Kelly, a student who says nothing during the entire episode. Kelly had been slouched back in his seat, with his head rolling around distractedly up to this point in the transcript. As Brent leaned forward, Kelly suddenly perked up and leaned forward to pay attention to what was transpiring.

At 1:21:53 the teacher opened a conversation pair with a question. It was taken as a rhetorical question, that is as one that expected the conversation partner to see that there was something “like that there” and to answer in the affirmative, signaling that he had seen what the teacher was pointing out. We can see that it was taken as a rhetorical question because the negative answers supplied by the students were not accepted. The three students who tried to answer in the negative – first Steven and Jamie simultaneously, and later Chuck – repeated their answers, as if to re-assert answers that were not called for. Rather than accepting these answers, the teacher rephrased the question and paused for an affirmative answer.

Brent responded to the conflict between the expectation given by the rhetorical question and the attempts by the other students to give a negative answer. The following can be seen as an attempt by the group to resolve this conflict and provide the sought affirmative answer to the teacher’s question, finally completing the interrupted conversational pair.

The Problem

Brent interrupts the teacher with, “This one’s different.” The word “different” goes back to the teacher’s last statement. The teacher’s full question, elaborated in response to Steven and Jamie’s disagreement was: “And (0.1) you don’t have anything like that there? . . . With the same engine but with a different (0.1) nose cone?” In the meantime, Steven and Jamie had both picked up on the teacher’s term “same,” as had Jamie.

1:22:05

Brent

└ This one’s different ((gestures with pen at computer 1 screen))

The teacher had used the terms, “same” and “different” to clarify what he meant by “like”. In rhetorically asking, “Don’t you have anything like that there?” The teacher was suggesting that the list of rockets (“there” where he was directing their attention) included a rocket whose description was “like” the rocket they needed, namely one that had the same engine but a different nose cone from the one that they would compare it with.

The teacher’s original statement at 1:21:53 was elliptical in its use of the term “like”. It assumed that the audience could infer from the context of the discussion in what ways something (“anything” “there”) would have to be like the thing under discussion (“that”). After two students responded that they could not see anything like that there, the teacher tried to explicate what “like” meant here. He did this by picking up on Jamie’s “Not with the same engine” and defining “like” to mean “with the same engine, but with a different nose cone.” Scientific talk tries to avoid the elliptical ways of normal conversation. Throughout the session, the teacher models for the students this explicit way of talking, often taking what a student has stated elliptically and repeating it in a more fully stated way. Now the teacher is doing just that. Sometimes one of the students will pick up on this and start to talk more explicitly. Here Brent has picked up on the term “different” as a key criterion for determining likeness.

Of course, the problem for us as researchers is that Brent’s exclamation, “This one’s different,” is itself elliptical. In what way is “this one” different?

The Confusion

There is also the interpretive problem of reference or indexicality. Brent is pointing at the list of rocket descriptions, but it is impossible to tell from the video data which description he is pointing to. Even if we knew which one Brent was pointing to, his utterance does not make clear which other rocket he is comparing with the one to which he is pointing. We have to deduce the answers to both these questions from the ensuing discussion, to see how the participants themselves took the references.

Jamie’s immediate follow-on utterance begins with “Yeah, but” indicating a response that is partially supportive. Since we know that Jamie is responding to Brent, we know that Jamie’s use of “it” refers to Brent’s “this one.” Chuck in turn builds on Jamie, reclaiming the floor by interrupting and completing Jamie’s incomplete utterance of the term “nose cone.” So Chuck’s subsequent utterance – which he ties to the preceding with “but” uses “it’s” to refer to Brent’s “this one” as well.

1:22:06	Jamie	Yeah, but it has same no…
1:22:07		(1.0)
1:22:08	Chuck	Pointy nose cone=
1:22:09	Steven	=Oh, yeah=
1:22:10	Chuck	=But it’s not the same engine

Here we see the conflict begin to be stated. Chuck’s “but” suggests a disagreement with Brent and possibly with Jamie also. In the next second both Jamie and Brent come back with “yes it is,” showing that they took Chuck’s comment to be a clear disagreement with what they were saying.

Kelly’s non-verbal behavior again indicates that something unusual is going on. Now he rocks forward onto his elbows to follow events more closely. He stays in this position for the rest of the moment.

At this point in our interpretation, we have several shifting factions of opinion. At first, all the students seemed to be disagreeing with the teacher. Following Brent’s bold gesture, some of the students seem to be disagreeing with others. We have not yet worked out the basis of this disagreement because of the elliptical and indexical nature of the utterances that form our data.

We have actually overcome the problem of the elliptical – but not the indexical – character of the utterances by looking closely at how the individual utterances build off of each other, repeating the use of the same words or using conjunctions like “but” or “yeah” to signal continuity of topic. However, it is harder to know, for instance, which rockets are indexed by pronouns like “it”. It seems likely that Jamie and Chuck are, in fact, indexing different rocket descriptions with their use of the pronoun “it.” This would certainly cause confusion in the discussion because the repeated use of the same word should signify commonality of reference. To determine which rockets they are each indexing in their utterances, we will have to continue our interpretive effort.

The Repair

In the next couple of seconds, Jamie and Brent state virtually the same thing simultaneously. This indicates that the state of the group discourse – from the perspective in which Jamie and Brent are viewing it – must be very clear. That is to say, the network of indexical references as interpreted in Jamie and Brent’s utterances is univocal. Within this set of references, Chuck’s claim that “it’s not the same engine” is clearly wrong. Jamie and Brent insist that “it” is the same engine.

1:22:11	Jamie	Yeah, it is, =
1:22:12	Brent	=Yes it is,
1:22:13	Jamie	┌ Compare two n one
	Brent	└ Number two

Here Jamie and Brent support their counter-claim precisely by clarifying the references: they are talking about similarities and differences between rocket number two and rocket number one on the list in the simulation artifact.

Jamie’s imperative, “compare two and one,” is first of all an instruction to Chuck to look at the descriptions of rockets two and one on the list. At the same time, it is a reminder that the purpose of the whole discourse is to conduct a comparison of rockets in order to determine the best nose cone shape. Jamie’s utterance serves both to propose an explicit set of indexical references for the problematic discussion and to re-orient the discussion to the larger goal of solving a specific scientific task. His utterance thus serves to state both the indexical and the projective basis of the discourse. He is saying that the group should be indexing rockets one and two in the list comparison so that they can then conduct a comparison of one and two in the datasheet artifact as their projected future task.

Jamie and Brent have now solved our task of interpreting the indexical references. Of course, we might still want to try to reconstruct the networks of references that different participants had at different points in the discourse. We would thereby be retrospectively reconstructing the process of construction that the discourse originally went through to reach this point. We would be “deconstructing” the discourse.

If we go back to the minute of discussion between the teacher and Chuck that preceded our transcript, we indeed find the source of the confusing references. Chuck had switched the discussion from nose cones to fins and had in fact solved the problem of how to determine the best rocket fin configuration. He said to compare rockets 3 and 4, which are identical except that rocket 3 has 3 fins and rocket 4 has 4 fins. Then Chuck wanted to return to the problem of nose cones. He proposed making the simulation software modifiable by users so that he could either change the nose cone of rocket 3 or 4, or else change the engine of rocket 2 to match the engine of 3 and 4 so he would have a pair with the same engine as his baseline rocket (3 or 4) but different nose cones. So Chuck was actually following the right theoretical principle already. However, his description of the changes he would make got quite confusing – plus it made unrealistic assumptions about the software.

So the teacher’s opening remark, directing Chuck and the others back to the list on the screen can now be seen as a projective attempt to have Chuck recognize that rockets 1 and 2 could be compared as is without changing one of them to be comparable to 3 or 4. In other words, the list had this built-in structure – that Chuck was not seeing and taking advantage of – that it had been organized to solve the problem of rocket comparisons. Unfortunately, because the discussion had been focused on rockets 3 and 4 as the basis for comparison, none of the students could see at first that 1 and 2 met the criteria. As Jamie said, there was no rocket with a pointed nose cone, “not with the same engine,” where we can see that “same” referred to same as the engine in rockets 3 and 4.

When Brent points to what must be rocket 2 and says, “This one’s different,” his utterance refers to the fact that rocket 2 has a pointy nose cone, which is different from all the other rockets. At this point, Brent’s and Jamie’s utterances must be taken as comparing 2 to rocket 1. Because when Chuck keeps insisting that “it’s not the same engine” (meaning 2’s engine is not the same as 3 and 4’s), Brent and Jamie retort “yes it is” and explicitly refer then to 1 and 2. As they repeat that they are looking at descriptions of rocket 2 and another rocket with the “same” engine, even Chuck gradually aligns with the reference to rockets 1 and 2. With this look back at the situation prior to our moment, we can reconstruct how our moment developed out of its past and we can determine a consistent and meaningful interpretation of the references of the utterances, as understood from the perspectives of the different participants’ utterances.

The Resolution

In the final segment of our transcript, Chuck responds to Jamie’s clarification. When Jamie says “compare two and one,” Chuck actually turns to the computer screen and studies it. With gradually increasing alignment to what Jamie is saying, Chuck says tentatively, “I know.” This is the first time during this episode that his utterances are agreements. Jamie goes on to instruct on how to make the comparison of rockets one and two: note how they “are the same.” Chuck’s “Oh” response indicates a change in interpretation of things. Brent makes even more explicit how Jamie’s “are the same” is to be taken, namely that both rockets have the same kind of engine.

1:22:14	Chuck	(0.2) I know.
1:22:15	Jamie	(0.2) Are the same=
1:22:16	Chuck	=Oh
1:22:17	Brent	It’s the same engine.
1:22:18	Jamie	So if you ┌ compare two n one,
1:22:19