Dissertation Proposal

From Jsarmi

Bridging mechanisms in team-based online problem solving: Continuity in collaborative knowledge building

Dissertation Proposal by Johann W. Sarmiento

Doctoral Program in Information Science and Technology

College of Information Science and Technology, Drexel University

Dissertation Proposal Committee:

Gerry Stahl (Chair)	College of Information Science and Technology, Drexel University
Susan Gasson	College of Information Science and Technology, Drexel University
Robert B. Allen	College of Information Science and Technology, Drexel University
Wesley Shumar	Department of Culture and Communication at Drexel University
Steve Weimar	The Math Forum - Drexel University School of Education

Overview

Emergent theories and designs for collaborative knowledge building in fields such as social computing and information science and technology in general, continue to point to the pressing need to better understand how the power of virtual distributed teams and online communities can be harnessed. This kind of research and design work aims to realize the potential of such new forms of interaction to generate and advance learning and knowledge in organizations, communities of interest, academic disciplines, societies, and many other types of collectivity.

One particular aspect of the interdisciplinary study of computer-supported collaboration which is not well understood yet, is how interactions which are dispersed over time (e.g., long-term projects, multi-session problem solving engagements, etc.) and which cut across different collectivities (e.g., sub-teams, teams, communities, etc.) can more successfully be interlinked or "bridged" so that they lead to more constructive and sustained collaborative knowledge building. In fact, the study of this type of interactional bridging activity may provide a unique viewpoint into the interrelationship between the local small-group unit of analysis and other more wide-ranging groups (e.g., online communities, multi-team collectivities, etc.) as well as between the individual and the small-group. In addition to the need to understand the interactional nature of bridging, there is also a crucial need to learn more about which aspects of the computer tools provided to support collaborative knowledge building attend to these types of activities, and how such designs might be enacted in particular contexts.

This dissertation will investigate, from an interactional perspective, the ways in which bridging mechanisms contribute to sustaining continuity in the collaborative knowledge building of small groups in online learning communities. More specifically, it will pursue the following three goals by analyzing bridging phenomena within a particular online learning community appropriately structured for design-based research:

• (a) Defining how three different aspects of the online interactions of virtual collaborative learning teams are bridged: episodes, collectivities, and perspectives.

• (b) Exploring the effect of such bridging activity on the sustained knowledge work of virtual teams.

• (c) Investigating how bridging activity can be supported by designed artifacts.

Problem Statement

The Math Forum at Drexel University, an online community active since 1992, promotes interactions among teachers of mathematics, students, mathematicians, hobbyists, staff members and other interested parties interested in learning, teaching, and doing mathematics. As the Math Forum continues to evolve, support for synchronous interactions becomes increasingly essential for sustaining and enriching the mechanisms of community participation available. A step in this direction, the Virtual Math Teams (VMT) project investigates the innovative use of online collaborative environments to support effective secondary mathematics learning at the Math Forum. The research proposed here is aimed at understanding how teams of participants in VMT online community "bridge" the apparent discontinuity of their interactions (e.g., multiple collaborative sessions, teams and tasks) and exploring the role that such bridging activity plays in their knowledge building over time. In addition, this research aims at informing the design of appropriate computational supports for long-term collaborative knowledge building.

In order to motivate our proposed research, this section presents some pilot data from the VMT project and raises a series of questions about several bridging phenomena that can be identified in these data.

Recommencing problem solving work and managing participation

After the initial greetings and a discussion on what to do in this session, the following exchange takes place via the chat interface available in the virtual meeting environment:

This excerpt illustrates how the participants of this interaction chose to start a current collaborative task. Elements of their collaborative discourse signal to us that they are also engaged in using prior interactions as relevant resources in initiating their current work. Understandably, when teams sustain their collaborative work over multiple individual sessions, this task of recommencing knowledge-building activity becomes an issue that participants have to address. Furthermore, what participants do to "bridge" the discontinuity of their interactional episodes might have an impact on the ways collectivities evolve and the methods used to advance the development of group ideas. A close examination of this passage—by attending to the ways that the participants demonstrably orient to the interaction moment-by-moment—can help us construct ideas about how some aspects of such "bridiging" activity get done and guide us in the process of articulating relevant questions about collaborative knowledge building.

One way in which we can approach our analysis of this sequence of postings is to consider the characteristic question of ethnomethodology-oriented studies: "why this now?" In other words, what purpose do these particular textual postings serve at this point of the interaction of this virtual team? How is it that the ways in which Drago designed and presented his postings (e.g., the way they were segmented in the multiple postings, the way they were positioned after line 302, etc.) indicate their interactional role and their effects? We can see that Drago’s posting in line 303 ("last time, me and estrickm came up") stands as an uptake of the proposal for collective action put forward by Gdo in line 302 ("now lets work on our prob"). Furthermore, the distinctive juxtaposition between these two postings indicates to us the beginnings of the group’s orientation to the problem solving task. By contrasting "last time" with Gdo’s "now", Drago attempts to establish a particular kind of episodic continuity or "relevant history" of the team, categorizes Gdo and Mathwiz as newcomers and orients to them as such. Continuing our inquiry as this interactional sequence unfolds, we could pursue these observations and investigate how Drago and the rest of the small group of co-participants orient toward this reported history and the "spokesperson-addressees" framework and how this orientation evolves. In addition, we might be able to inquire about how these and other elements of their interaction influence and are influenced by the way the collaborative knowledge-building task is being approached by this team.

Turning now to the unfolding of the sequence which begins with Drago’s posting in line 306 ("you always have to move a certain amount to the left/right and a certain amount to the up/down") we can explore further the dynamics of bridging work. This posting in particularly seems central to our understanding how the bridging work that is currently being attempted is unfolding. The reply posted in line 307 by Gdo ("what?") and the subsequent elaboration attempted by Drago suggest that the posting in 306 was taken as a problematical response to the proposal to initiate the problem solving work. Perhaps additional work was necessary for line 306 to be fully sensible for the team; in other words, for Drago to successfully bridge prior work into the present. In the subsequent lines we can see the beginnings of an instance of the kind of interactional work that seems to be necessary for the team to engage with the reported past that Drago is presenting. Even without a thorough understanding of the mathematical task at stake, one can see that Drago elaborates on his initial posting by providing additional problem information (308, "for the shortest path") and adding further references to elements of the past problem solving activity (310-312, "since the problem last time stated that you couldn’t…"). In this way, Drago continues to use past resources to organize a potential present for the problem solving task of the team and in doing so, attempts to project that past history on the current team members. We will come back to further inquire about the nature of posting in line 306 after exploring in more detail its uptake.

How does knowledge building advance across collectivities?

Mathwiz posting in line 314 ("would you want to keep as close to the hypotenuse as possible? or does it actually work against you in this case?) engages with the bridging activity opened up by drago in a particular way. Mathwiz seems to suggest a particular way of clarifying Drago’s presentation of how the grid-world works while at the same time doing the interesting work of positioning drago as the one who is to assess this suggestion (i.e. testing whether this case "works against you"). In this way, Mathwiz’ posting seems to actively situate Drago’s narrative of the past by ratifying its relevance for their current problem solving and assessing its practical comprehensibility. This short sequence signals only the beginnings of the type of interactional work necessary to fully bridge prior knowledge-work into present joint-activity and yet it is sufficient to provide us with significant evidence of the nuanced aspects of this type of activity.

One further aspect of this interactional sequence is worth exploring. The sequence below follows the excerpt presented initially from session two, and allows us to observe how this collectivity dealt with some of the challenges in the uptake of Drago’s presentation of the past finding about short distances in the grid-world.

Despite the fact that Drago is orientating to the recommencement of the prior work performed by him and Estrickm (and to a narrator-explainer framework of participation), Gdo departs partially from that orientation in line 325. Gdo does this by making a solution proposal ("square root of 45") for the shortest path between the points that they are currently examining. Consequently, the sequence of postings from 325 to 330 seems to indicate a local engagement with the problem as the present matter and no longer as a bridging move to re-use prior findings. However, in the sequence starting at line 331 Drago uses, once again, elements of prior interactions ("the problem before said so but you weren’t here") to address what appears to be a problem in Gdo’s understanding (i.e. that you can’t go on the diagonal). This alternation between present and bridged resources for problem solving indicates a dynamic engagement by the group with its distributed history and its current problem solving activity.

Also particularly interesting in this sequence are Mathwiz’ postings in lines 339 and 342: "the hypotenuse is fine, but for the problem, you have to go on the grid lines…it’s like, you can’t walk in water, and the lines are dry lines." These postings seem to do the interesting work of ratifying Gdo’s use of the hypotenuse —as well as his participation in the task— while at the same time offering him a new "rule" to manipulate the grid. It seems to us that this interactional move bears resemblance with the types of activities we have analyzed before and deserves the label of bridging work. By offering this new perspective on how to imagine and manipulate the grid Mathwiz identifies a boundary between different perspectives or understandings (using diagonals or not) and goes beyond simply refuting it to offer a link between the two. This bridging of perspectives can naturally occur between different problem solving episodes or collectivities but here we see it happening within the flow of a team’s interactions. With the dynamic changes in team membership that characterizes naturalistic environments and the diversity of points of view among individuals and teams typical of online communities, sustained problem solving work seem to require that co-participants also engage in this type of bridging of perspectives. Here we may ask what kinds of resources are produced by individuals and groups to overcome the boundaries that emerge as relevant during their interactions and which might aid with challenges of continuity, coordination or affiliation. It is possible that the methods and processes of doing this kind of boundary work could characterize effective collaborative learning interactions, but this remains a notion to be investigated further.

So far, we have explored a few instances of bridging activity in the trajectory of a particular virtual math team. In doing so, we have offered a preliminary analysis on how the collective engagement with past work is constituted across different interactions or episodes, how changes in participation signal various aspects of the sustained knowledge-work of the teams, and how problem solving perspectives are subject to bridging as well. This preliminary analysis demonstrates that, in interactional contexts where there are continual sequences of discrete problem solving episodes and where the membership of a team might change over time, sustaining continuity of the team’s knowledge work becomes a particular challenge for which teams need to develop particular interactional strategies. Furthermore, our analysis seems to suggest that these attempts to establish continuity in collaborative problem solving involve the recognition and use of discontinuities or boundaries as resources for interaction (e.g., temporal or episodic discontinuity), changes in the participants’ relative alignment toward each other as a collectivity (e.g., narrators and interactive audience), and also the use of particular orientations towards specific knowledge resources (e.g., the problem statement, prior findings, what someone professes to know or remember, etc). Whether these are the central structural elements that characterize the establishment continuity in collaborative knowledge building and how such elements are used to sustain it, are questions that require further analysis. In fact, the analysis of other instances of this type of activity could lead us to uncover a range of bridging mechanisms used as part of this online learning community while at the same time helping us strengthen the methods utilized to conceptualize and analyze these phenomena.

How does knowledge building develop over time in small groups?

In addition to following the sequential unfolding of the particular instance of a team’s recommencement that we have pursued, we could also investigate retrospectively Drago’s "bridging" posting in line 306 ("you always have to move a certain amount to the left/right and a certain amount to the up/down"). We could do this by analyzing his prior work with Estrickm a few days back and trace the genesis of the reported finding. This approach would allow us to stretch the scope of our analysis not only from one problem solving episode into another but also from one particular collectivity (Drago- Estrickm-Gdo-Mathwiz) into a different one (the dyad Drago-Estrickm). In brief, let us just say that line 306 in session two appears as a re-statement of something that Drago and Estrickm discovered in the first session that they held as a team. The following excerpt illustrates how this idea was articulated then:

The relationship between line 173 in this excerpt of session one and line 306 of session two appears significant. On the one hand, the use of the adverb "always" in both postings seems to suggest a rule-like statement (or a conjecture) aimed at capturing a constructed understanding about the way the grid-world has been perceived to work. The fact that the creator of this text changes from Estrickm in session one to Drago in session two could be taken as an indicator that this rule is a collaborative understanding achieved by the dyad which is, later on, projected into a new collectivity and a "bridged" problem solving context. Based on this observation, we could construe the re-statement of prior findings and the change in authorship as indicators of sustainability in the co-construction of knowledge as the history of multiple teams in an online community evolves. Although not entirely conclusive, these two conditions certainly seem to point in that direction despite the fact that in small-group interactions the notion of authorship needs to be analyzed critically. For instance, if line173 is a response to line 172 and proceeds from the flow of the interaction, isn’t it really the dyad who should be credited with having produced the original rule about the grid-world? It is possible that we need to redefine certain notions such as authorship and construction of knowledge as we navigate individuals, small-groups and larger collectivities.

Beyond the changes in author, it is interesting to note how Drago’'s text in line 306 of session two is different than the original posting from session one. In Estrickm’s original posting there was only mention of moving "down and to the right" but in Drago’'s restatement one has to move a certain amount "to the left/right ''and a certain amount to the up/down.''" Why has Drago modified the original rule by adding the "/up" and "left/" elements? In order to investigate this question, we should point out that the environment

only included two points arranged in a similar way to the points A and B included in the diagram from session two. The arrangement of points used in session one matches Estrickm’s original rule that "you will always have to go down and to the right." As can be seen on Figure 2.2, there are two arrangement of points being considered in session two: The one including points A and B where the shortest path would be achieved by going down and to the right and another in which the movement would be up and to the right (linking the points labeled with circles). As a result, one can read Drago’s modification to the original rule as indication that he has adapted it to make it applicable to the two arrangements of points being considered by the team in session two. It is possible that Drago realized this generalization via further individual work in between team sessions, or that the position on the grid of the points that the team has selected in session two provided the need for the generalization to happen. Whatever the actual motive, Drago is presenting in session two a modified version of the finding constructed in session one suited to the current circumstances. Beyond simply citing prior findings, Drago has in fact bridged two problem solving contexts in an attempt to construct continuity. To further qualify this observation, we can contrast Drago’s tentative reasoning for why the rule works presented in line177 of session one (well...I guess its because whatever path you take, you will make tiriangles) with the sense of confidence that his presentation conveys in session two. This subtle change could illustrate a change in the strength of his understanding of the grid-world. Observations like these, although requiring further verification through triangulation and further analyses, start to point to critical interactional aspects of how knowledge work is sustained over time and hint towards longitudinal aspects of collaborative learning interactions. Furthermore, they reveal the need to understand how bridging interactions span across the individual and the different collectivities involved in an online community.

How do designed artifacts support bridging work?

Finally, we would like to consider a few questions regarding the collaboration supports used by the participants while engaged in the activities that we have presented. For example, it is reasonable to think that Drago is relying on his memory to produce, in session two, the account of the prior findings since the transcripts of prior chats were not publicly available. The arrangement of points and the graphical reasoning that we have presented in the snapshot captured in Figure 2.2. bear great resemblance with the work that Drago and Estrickm conducted in session one, but there was not a mechanisms available for them to reuse them once a session was over. The virtual room that the team was using was not available either to other teams who were working on the same problem despite the potential usefulness of these cross-team interactions. This motivates us to wonder about the forms in which records of prior interactions could be made accessible to virtual teams and how they would be used in cases where sustained problem solving is being conducted. Would access to such records aid teams in sustaining the processes of collaborative learning and problem solving? If, as in our current case, multiple teams are working on the same collaborative learning tasks, would access to other teams’ ideas promote a sense of community? Questions like these point towards the necessary translation between our understanding of sustained collaborative work and the design principles that underlie the support environments made available to participants.

In Summary

As we indicated earlier, the excerpts presented here originated from a pilot study of small-group collaborative learning conducted as part of the Virtual Math Teams (VMT) project of the Math Forum. Numerous relevant questions can be pursued about the data presented here and about other excerpts in this dataset. Our intent, however, has not been to present a full analysis of these data but to uncover some of the complexities that need to be investigated about these kinds of interactions and begin to outline elements of the analytical methods needed to investigate the processes related to the establishment of continuity and discontinuity of this type of collaborative knowledge work.

In summary, sustained collaborative learning in small virtual groups and online communities requires that co-participants "bridge" multiple elements of their interactions continuously as they interact over time— a non trivial and possibly very consequential undertaking. One way in which we can cluster the kinds of questions that we have been asking about the data presented could be to take "bridging" as the central target of our inquiry. In the sense we have used it, the term "bridging" defines interactional phenomena that cross over the boundaries of time, activities, collectivities, or perspectives and defines a set of methods through which participants deal with the discontinuities relevant to their joint activity. Bridging thereby might tie events at the local small-group unit of analysis to interactions at larger units of analysis (e.g., online communities, multi-team collectivities, etc.) as well as between the individual and small-group levels. Studying bridging may reveal linkages among group meaning-making efforts across collectivities or interactional episodes over time.

Taking "bridging" as the central interactional phenomena of interest of this research, our central aim is to characterize the ways in which bridging contributes to the establishment of continuity and discontinuity in the knowledge-building experience of online collaborative learning teams in the VMT online community.

In particular, we are interested in pursuing three specific goals:

(a) Defining how the bridging of episodes, collectivities, and perspectives in the online interactions of virtual collaborative learning teams is achieved;

(b) Exploring the effect of such bridging on the sustained activity of these virtual teams and the online community in which they are situated; and

(c) Investigating how this bridging activity can be supported by designed artifacts.

In addition, by synthesizing the observations that we have posed based on the data presented in this section, we define the following three central research questions which guide our proposed research:

Before presenting the details of the proposed methods and the experimental design selected to answer these three central questions, we first review the current state of the relevant literature in order to better ground our choice of problem in the larger research context, explore to what extent similar phenomena have been investigated in relevant research fields, and identify potential contributions of our work.

Literature Review

Given our choice of research problem and the context in which we propose to carry on our investigation, we find our work to be highly situated in the multidisciplinary field of Computer-supported Collaborative Learning (CSCL). CSCL has been defined as being concerned with understanding "the practices of meaning making in the context of joint activity, and the ways in which these practices are mediated through designed artifacts" (Koschmann, 2002). CSCL has also taken up the research program outlined by the sociohistorical psychologist Lev Vygotsky and his formulation of the "genetic law of cultural development" in which higher psychological functions in humans are demonstrated to originate at the social level (between people), and only later, through a "long series of developmental events", these functions are internalized by the individual. (Vygotsky, 1930/1978). As a branch of the learning sciences, CSCL is particularly concerned with the practices of meaning making that help us understand "how people can learn together with the help of computers" (Stahl et al., 2006). In its first ten years, CSCL studies have uncovered some of the unique ways in which groups create, represent and use knowledge (Roschelle, 1996; Schwartz, 1995), the particularities of how technology affordances support group collaboration (Schwartz, 1995; Stahl, 2004a; Suthers & Hundhausen, 2003), and the fact that, in many instances, the collaborative process needs to be explicitly supported or "scaffolded" for it to succeed (Jermann et al., 2004; Pea, 2004; Rummel & Spada, 2005). As illustrated in Firgure 2.1, CSCL encompasses fields as diverse as educational psychology, small-group research, groupware design, and other research areas that inform CSCL with theoretical models, analytical methods or contexts of applications.

Figure 2.1. Composition of the multidisciplinary field of Computer-supported Collaborative Learning

Significant progress has been made within CSCL in advancing our understanding of the nature of learning in small-groups and how to support it with designed artifacts (Koschmann et al., 2002; Koschmann et al., 2005; Stahl, 2002; Wasson et al., 2003)  but much remains to be discovered, especially about the longitudinal patterns of meaning-making in naturalistic small-groups and larger collectivities (e.g., online communities). Next, we review some of the theoretical frameworks and areas of research relevant to the proposed research questions.

What interactional mechanisms explain collaborative learning over time?

Successful collaborative learning in small groups is, in fact, a complex phenomenon that involves numerous factors (e.g., nature of the task, number of participants, time, pooled expertise, communication patterns, etc.). These contextual factors interact in non trivial ways (Dillenbourg et al., 1995) to shape the dynamics and outcomes of collaboration. Given such complexity, current research attempts to understand the interactional mechanisms that underlie different collaborative interactions and the ways that they could explain particular collaborative outcomes. A number of these interactional mechanisms have been studied in the CSCL literature and also in related fields. For instance, researchers have argued that collaborative knowledge-building advances thanks to the power of peer explanations and the sharing of knowledge among peers (Dansereau, 1988; Webb, 1985, 1991, 1992) or through argumentation derived from cognitive conflict (M. J. Baker, 1991). An attempt to characterize the dynamics of bridging activity, as presented in the previous section, as peer knowledge sharing or peer explanations seems to miss important parts of the dynamics of such interactions.

For example, some of the limitations of the "peer sharing" perspective are made evident when we revisit the data we have analyzed before in our trace of the interactions of team 5. Even in the brief excerpts we have analyzed, we clearly see how explanations are not simply furnished directly from an "explainer" to a "tutee", nor are summaries of prior activity necessarily presented only by a "recaller" to a "listener." The type of activity that we have identified as "bridging" clearly goes beyond simple peer-explanations or "recalling" of findings, primarily, because of the active and multi-dimensional engagement that characterizes what team participants seem to orient to. In addition, the boundaries imposed by multiple interactional episodes and changing collectivities seem to inflict significant constraints on how the traditional interactional mechanisms, discovered mostly from studying single-episode dyadic interactions, are enacted. Furthermore, recent research on dyadic explanation and argumentation has recognized that these interactional mechanisms, when successful, are co-constructed jointly by both peers through collaborative interactions (Hausmann et al., 2004) Co-construction, however is seen as synchronous and local phenomena and theorized to work through continuous, moment-by-moment engagement in attending and monitoring shared issues of understanding (Chi et al., 2004), exploring and transforming the joint problem-space (Roschelle, 1996), and engaging with each other’s proposals (Stahl, 2006c). Our challenge is to investigate how the apparent discontinuity of the interactions affect these processes when virtual teams collaborate over time as part of an online community, and what new processes of co-construction emerge in these contexts.

Does the "contribution model of grounding" from communication theory explain bridging activity?

One of the communication theories most commonly used in CSCL, CSCW, and other fields to understand the process of co-constructing shared understanding in small-group collaboration, is Clark and Schaefer’s "contribution model of grounding" (Clark & Brennan, 1991; Clark & Schaefer, 1989). In this theory, rational parties in a conversation not only produce and receive messages but also monitor their mutual understanding by seeking and providing feedback that the message has been understood or by "specifying some content and grounding it" (Ibid, p. 124). The mechanics of this process, the structure of presentation and acceptance phases, and the underlying concept of "common ground", outlines how conversational parties are seen to reach identical or closely aligned mental contents. Several CSCL and CSCW studies have used grounding as a central theoretical concept to explain interactional aspects of how knowledge is established through conversation (M. Baker et al., 1999; McCarthy et al., 1991). Recently, significant criticism has been expressed about the limitations and inadequacies of this model to truly capture the interactional aspect of collaborative meaning-making and its applicability in the design of collaboration support systems (Koschmann & LeBaron, 2003; Stahl, 2006b). On the one hand, it is not clear how the systematics of the model scale up to interactions that span larger collectivities and time scales from the short dyadic exchanges studied by Clark and Brennan. In addition, the concept of common ground as a psycholinguistic object might not be sufficient to explain how complex shared understandings, routines, and community norms, are created and maintained in sustained collaborative interactions. In fact, in the data excerpts presented in Section II, we can already see that the notion of common ground can only explain a small portion of how we see these sequences of interactions unfolding. Contributions to the process of constructing shared meaning in the groups’ discourse can bee seen to come from multiple sources and be more contingent than what the theory of common ground may suggest. We can see that bridging implies contributions from the individuals’ knowledge of the task, their resources, skills, memories, etc., but equally important in this process is the role of the teams’ distributed knowledge of their local history (e.g., what the team has accomplished up to a point in time), the cultural knowledge of the community in which bridging is localized.

Is bridging the same as "transactive memory" as described by the social-psychological study of teams:?

The construct of "transactive memory" recently adapted from the study of interpersonal relationships to research in team and organizational learning (Moreland et al., 1996) appears also as highly relevant to the study of longitudinal, knowledge-building interactions. Transactive memory is theorized as a distributed memory system through which a collectivity stores and recalls information and is used as the basis to explain how a team becomes a "knowledge-acquiring, knowledge holding, and knowledge-using system" (Wegner et al., 1985, p. 256). The theory argues that a collectivity composed of individuals develops a memory system by constantly updating a "directory" of expertise (knowing who knows what), communicating to allocate information, and communicating to retrieve information. These processes are highly dependent on the establishment of a shared conception of the topics that the individual members know, something that the theory of transactive memory predicts is achieved through the grounding processes described by Clark and Brennan and discussed before. The interdependence developed in this way, is theorized to create a holistic system of people, knowledge, and tasks responsible for the performance benefits usually attributed to teams. In this sense, the value of teams is related to an enhanced memory system that supports unique distributed operations.

Experimental research in social psychology and the recently emerging area of "team cognition" appears to have confirmed the existence of the transactive memory construct (Hollingshead, 1998a, 1998b) and quantified its positive impact on team performance when, for instance, promoted through collective training programs (Moreland, 1999; Moreland & Myaskovsky, 2000). Interestingly, some of this research has documented that group performance decreases with the turnover in group membership but that giving newcomers access to information about the knowledge of other group members (and vice versa) has positive effects.

Research on transactive memory extends the concept of common ground in ways that seem to address some of our observed limitations but, at the same time, does not seem to provide a completely adequate perspective to investigate "how" teams develop and advance their knowledge building over time. On the one hand, its use of grounding as the central mechanism through which transactive memory is established seems to limit its applicability to our context for the reasons discussed before. Furthermore, the interactional mechanisms through which the distributed system of knowledge and memory is achieved as theorized by this line of research are highly underspecified as "communication" processes. In this respect, our goal is to extend our preliminary analyses of data from virtual teams engaged in sustained problem solving to review the adequacy of constructs such as common ground and transactive memory to explain the mechanisms by which joint understanding is achieved in our longitudinal contexts.

In fact, in the interdisciplinary field of small group research, lead researchers have recently pointed out that new theoretical frameworks are necessary to truly address and understand the complexity of small group interactions (Arrow et al., 2000). Key group processes such as those related to group formation, development, and adaptation have only been superficially understood by the laboratory experiments that had dominated this field. Furthermore, researchers have expressed their concern that the topic of group development has faded away as a result of the preference for laboratory experiments with ad hoc groups that have "no past and no anticipated future" (Arrow et al., 2005). This has resulted in the recent resurgence of studies that pay more attention to temporal analyses and to what Arrow and colleagues call "groups’ traces, trajectories and timings." Some of this work which views of groups as complex systems has attracted particular attention in the field of CSCW (Fitzpatrick, 2003; McGrath & Arrow, 1995) and other areas of socio-technical research concerned with the dynamics of knowledge management and organizational learning. We will review more of the relevant work from the field of CSCW later in this section especially as it relates to groupware supports for continuity and knowledge management.

Progressive problem solving and the theory of knowledge building

In general, even in CSCL only a handful of studies treat the small group as the unit of analysis while instead treating the collectivity as a context in which individuals interact. Our proposed focus on bridging and its relationship to collaborative learning over time, attempts to contribute to the emergent preoccupation for treating collectivities as holistic units. One unique line of research in CSCL contrasts with the shortage of studies examining the continuous nature of meaning-making mechanisms underlying longitudinal sequences of small-group interactions: the theory of knowledge building. In this theory developed by Carl Bereiter and Marlene Scardamalia (Scardamalia & Bereiter, 1996) (as well as in the computer supports designed to support it), it is argued that successful collective learning results from the intentional engagement in a progressive process of idea refinement and communal discourse as part of a shared enterprise. Progressive problem solving is seen as the identifying characteristic of both individuals becoming experts and also experts working "at the edges of their competence" both strongly situated in a sociocultural setting. Ten years of research in this theory has documented the viability of decentralized, open knowledge building and the development of collective knowledge as well as the necessary pedagogy and technological supports for it to flourish (Scardamalia & Bereiter, 2006). A series of principles have been advanced to characterize successful knowledge building including idea diversity, collective responsibility, epistemic agency, and symmetric knowledge advancement (Scardamalia, 2002). Many of these principles operate at the macro or community level and their interactional achievement at the small-group and individual levels is still a matter of research. We see our proposed research as expanding this line of inquiry by illuminating the interactional aspects of how progressive knowledge building is actually achieved by small-groups situated in an online community and, specifically, how bridging of interactional episodes, collectivities, and perspectives contributes to the sustainability of knowledge building.

Research about virtual communities, a field closely related to CSCL, has also advocated a more encompassing approach to investigation of how knowledge is built over time in large online interaction spaces. Unfortunately, most of the online communities investigated are typically based on asynchronous mechanisms of participation (e.g., online forums and discussion boards) and offer views highly anchored by this factor. Despite this limitation, the foundation established by this research is highly relevant to our research interests. For example, Renninger and Shumar (Renninger & Shumar, 2002) in their introduction to the first collection of research on how learning and change can be fostered by online communities, argued that "the connection (that participants develop in) virtual communities is supported by affordances that invoke imagination about and identification with a site, such as autonomy, time, space, choice, opportunity, support, and depth of content. Furthermore, the learning that is undertaken as participants work with a site has an agency and opportunity for changed understanding of self" (Ibid, p. 7). It is our goal to examine through our analysis of virtual teams the similarities with these observations. The authors also state that "the availability of stored resources and information, coupled with the flexibility in the time and space of usage, may well account for the attributions of utopian possibilities for community via the Internet" (Ibid, p. 11). Our interest in the longitudinal interactions of virtual teams and in the ways that designed artifacts can support them is aligned with this observation and will potentially expand it’s applicability to other forms of interaction. Other authors (Stahl, 2004b) also argue that small groups represent the central mediating force between individual learning and community learning, and that "community participation takes place primarily within small group activities. The proposed plan of research offers an opportunity to test this conjecture empirically.

Another framework commonly used in researching large-scale collaborative learning environments originates form the study of social networks and the theory of social capital. Robert Putnam in particular, distinguishes two type of social capital: bonding and bridging. Bonding social capital is produced through social networks between homogeneous groups of people, while bridging emerges thanks to the linkages between socially heterogeneous groups (Putnam, 2002). Bridging social capital is theorized to produce the highest benefit for communities, societies, and individuals. The dynamics of "bridging," the type of social capital responsible for ties across dissimilar groups, however, remain vaguely understood. What is that "bridgers" do that generates the theorized benefits for the collectivities? In the particular context of knowledge-building communities, what processes characterize bridging? These unanswered questions require that the mechanisms through which bridging is achieved and their role in the continuity of knowledge-building collectivities be investigated.

Continuity and Computer-supported Cooperative Work (CSCW) research

Research in the field of CSCW has dedicated a significant amount of attention to issues of continuity of collaborative work and the designed environments aimed at supporting it. In fact, CSCW researchers have led the integration of design and investigation goals in their attempts to, both, understand collective action and support it through design artifacts. For instance, in their call to "take CSCW seriously" Schmidt and Bannon (Schmidt & Bannon, 1992) proposed "articulation work" —the coordination of work among team members — to be the central concern of studies of joint work and argued that CSCW needed to go beyond socio-technical studies of work in order to implement design research that is better suited to support group work. In general, the problem of coordination of work has been central to CSCW (Malone & Crowston, 1990; Montoya-Weiss et al., 2001). Coordination can also be seen as the interactional work necessary to overcome the "gaps" that characterize collective activity and which, in some cases, become magnified in computer-based environments(Ishii et al., 1993). Coordination work includes temporal continuity, coordination across team members and also across organizations. As we will see in the next paragraphs, coordination and bridging activity share a lot in common.

Initially, a great deal of CSCW research revolved around decision-support systems and electronic meeting environments (e.g., Nunamaker et al., 1991). Within these environments, it became evident that supporting continuity of interactions was both an opportunity (given the digital recordings available) but also a significant challenge. Diverse approaches to meeting synthesis and summarization emerged, some which attempted to build intermediate semantic representations of the structure and content of the artifacts available as guides for the creation of summaries. Few of these approaches have evolved into mature summarization systems given the complexity of such approach. However, it is interesting to note that these line of research has concluded that providing users with appropriate interfaces for them to manage their own issues of continuity might be a more effective strategy than attempting to create perfect summaries of interactions (Farrell et al., 2001; Waibel et al., 2001). More recently, Greenber and Roseman have argued that using designs based on the room metaphor is an effective way to overcome the numerous "gaps" identified in computer-based joint activity by almost 20 years of CSCW research (2003). In particular, the authors explore how room-based designs with persistent records can ameliorate four different types of gaps: the gap between individual and team work, the gap between synchronous and asynchronous interaction, the social awareness gap, and the gap that needs to be overcome in order to foster a sense of community among teams. Unfortunately, no experimental data is provided to validate these claims. In our case, we expect to provide and empirical analysis of how these boundaries or gaps are actually bridged and how the proposed designs are enacted in naturalistic interactions.

The last ten years of research in CSCW has led to the development of the area of "social computing" or "social systems," largely as a result of a commitment to better understand the realities of social interaction, one of the critical failures of initial CSCW research pointed out by Grudin (1990) and others. A crucial goal of this area of work lies in realization that mutual awareness of the histories and interrelationships among participant in a collectivity is critical to the collective outcome. To support this kind of activity, some researchers have proposed the use of "social proxies" (Erickson et al., 1999) and other strategies aimed at creating "socially translucent" environments. In addition, the design and use of systems that support "persistent conversations" (Erickson & Kellogg, 2001; Smith, 2002) has also emerged as a need to understand these new forms of interaction and their role in organizations and general culture. Persistent records of interactions, an apparent solution to problems of continuity, do not come without consequences and, as several researchers have pointed out, system designs have to go beyond "recording and reporting" (Bodker & Christiansen, 2006) and avoid the naïve view that "everywhere and forever" is always the best alternative (Grudin, 2002). Even in contexts where knowledge work is sustained over time, it is in the analysis of the practices that participants engage in that CSCW has been able to make progress in the understanding of processes such as "knowledge distillation" (M. Ackerman et al., 2003), "organizational memory" or the use of boundary objects (M. S. Ackerman & Halverson., 1999). In general, work in this area is the result of sustained design experimentation and analysis of users’ interactions and serves as the basis for the claim that CSCW, and human-computer interaction in general, need to dedicate more attention to understanding the "collaborative user experience." Our proposed work to study bridging in the context of virtual problem solving teams extends this orientation by considering the close relationship between collaboration and knowledge work.

Similarly, CSCW research has also evolved beyond single-team collaboration systems to consider larger arrangements of collective activity such as those in multiple team configurations, and those supporting larger organizational knowledge management. In fact, the issue of group-to-group collaboration in distributed settings has become extremely important recently. Some researchers, for instance, argue that a "new class of interaction problems" emerge when collective activity is analyzed in these contexts (e.g.,Mark et al., 2003p. 101). This interaction problems all stem from the need to overcome different terminology, perspectives, and work procedures across individual, sub-teams, teams and larger collectivites very much as we have described in our problem formulation. At the moment, it is clear that support mechanisms provided at the data level (e.g., offering access to records of interactions) or at the process level (e.g., controlling workflow) might be insufficient or too rigid (Miao & Haake, 1998) unless we understand how bridging activity works. Interestingly, research on group-to-group collaboration has highlighted the importance of studying the "space between" collectivities and understanding the connections, interdependencies and gaps across groups and organizations (Weick & Sutcliffe, 2005), a goal shared with research in organizational science and knowledge management.

Research in knowledge management has also made significant recent contributions to the understanding of the dynamics of sense-making (Carlile, 2002; Thomas et al., 2001; Weick, 1996). For example, the field has recently increased its attention towards studying the development of expertise in organizational contexts and to interdisciplinary teams, boundary objects, and boundary-spanning work (Gasson, 2005; Star, 1989). The unit of analysis that is suggested by the concept of "boundary objects" is of particular interest to our approach. The concept, proposed by Star based on historical case studies of scientific work involving both professional scientists and amateurs (Star, 1989), suggests that the participants: "(1) cooperate without having good models of each other’s work; (2) successfully work together while employing different units of analysis, methods of aggregating data, and different abstractions of data; and (3) cooperate while having different goals, time horizons, and audiences to satisfy." (p. 46). Star suggested that in the activity observed, it was the boundary objects that made cooperation possible. Boundary objects are "objects that are both plastic enough to adapt to local needs and constraints of the several parties employing them, yet robust enough to maintain a common identity across sites while sitting "in the middle of a group of actors with divergent viewpoints." Recent research has highlighted the intrinsically cross-functional nature of boundary objects (Carlile, 2002). From our preliminary analysis of trajectories of knowledge building in Virtual Math Teams, the cross-functional aspect of these interactions does not appear as salient as in other contexts and, yet, we have identified bridging artifacts co-developed within the interaction and used as a part of activity geared towards mediating or bridging contexts. For example, in the last excerpt presented in section II, we noticed how Mathwiz’ postin in line 342 ("it’s like, you can’t walk in water, and the lines are dry lines") seem to be a resource created to bridge a boundary of viewpoints and understandings and wondered whether the methods and processes of doing this kind of boundary work could characterize effective collaborative interactions. We expect to expand this analysis and pursue the investigation of the processes involved in the creation and use of these types of "bridging" objects in the Virtual Math Teams community. This type of analysis may provide a starting point to answer the question that Star has recently posed about the concept of boundary objects: "How are boundary objects established and maintained? Does the concept scale up? What is the role of the technical infrastructure?" (Bowker & Star, 2002)

Bridging as we have tried to define it can also be seen as related to the longitudinal process of expertise development. Recent research in expertise development states that continued improvements in achievement are not automatic consequences of more experience but that, instead, successful "aspiring" experts seek out particular kinds of experience. This special experiences are characterized as "deliberate" practice and characterized by they types of activities designed, typically by another expert or a mentor, for the sole purpose of effectively improving specific aspects of an individual’s performance (Ericsson et al., 1993). Only these types of activity provide optimal opportunities for performance improvement through cycles of feedback and re-construction of knowledge and skills. The careful study of bridging as an interactional phenomenon may provide an entry into the nature of these cyclical processes at the small-group level. Interestingly, in a recent review of the literature on problem solving, Pertz, Napes and Stenberg (2003) urge researchers to devote more attention to the early phases of the problem solving cycle related to problem formulation. Although considerable empirical research has been conducted on the latter stages of problem solving, the authors point to the little that is known about "what makes a person more likely to engage him or herself in seeking out ill-defined problems and experimenting with various ways of representing them" (p. 27). Interestingly, there is a clear opportunity in investigating these phenomena in group-interactions going beyond theories of individual problem solving and exploring new constructs such as group and team cognition (Salas & Fiore, 2004; Stahl, 2005, 2006a). By studying open-ended tasks in collaborative contexts and attending to the moment-by-moment unfolding of the interaction we hope to inform precisely these areas of problem exploration, problem finding, and problem definition.

Methodological perspectives in the study of interaction

Our review of the literature provides not only theoretical guidance and validation for the significance of our research questions but also informs our choice of research method. We will discuss in depth our selected research method in the next section and will restrict our discusion here of a few alternative methods emerging from several different perspectives in the literature. For instance, the methodological framework originated from the field of Computer-mediated discourse (CMD) studies is certainly one that is closely related to the proposed research questions. CMD studies evolved from research in Computer-mediated communication and as such investigate a diverse array of interpersonal communication carried out on the Internet via e-mail, instant messaging systems, mailing lists, newsgroups, web discussion boards, and chat rooms (Herring, 2001). However, although CMD research often encompasses perspectives from the pragmatic, discourse-analytic, and sociolinguistic points of view, it still centers on the study of linguistic artifacts more than language-mediated interactions. In CMD studies the concentration is usually in the collection of linguistic artifacts, the analysis of their representativeness or typicality within a genre, and the processing of such artifacts to derive conclusions about discourse practices (Herring, 2004). In some cases, CSCL studies adopt a methodological orientation similar to that used in CMD studies. Different coding schemes have been developed and used in CSCL in order quantify aspects of observed collaborative processes (e.g., the DISCOUNT coding scheme of Kneser & Ploetzner, 2001; Pilkington & Parker-Jones, 1996). In these schemes, roles (e.g., information seeker, explainer, task performer and reflector), "moves" (e.g., statement, couter-proposal, elaboration, etc.), episodes (e.g., negotiation, explanation, etc.), and other theorized elements are labeled and their quantitative patterns analyzed as a way to understand, and sometimes assess, processes such as externalization of knowledge, elicitation of knowledge, or consensus building. In such cases, regardless of whether the methods are used in CMD or CSCL studies, the linguistic artifacts are seen as representing a message or meaning to be uncover by the analyst, a position which contrasts with the tenants of ethnomethodology (EM) which "is not in the business of interpreting signs. EM is not an interpretive enterprise. Enacted local practices are not texts which symbolize "meanings" or events. They are in detail identical with themselves. The witnessably recurrent details of ordinary everyday practices constitute their own reality. They are studied in their unmediated details not as signed enterprises." (Garfinkel, 2002). Only in a few cases, CMD studies do explore how individuals use linguistic resources to participate in every-day activities mediated by computer networks and adopt an interactional perspective that make them strongly aligned with our methodological choices.

Furthermore, our approach differs significantly from the traditional method of interaction analysis in which a system of codes and categories (e.g., giving information, questioning, harmonizing, dominating, etc.) are used to label and analyze quantitatively the different ways in which teams engage in joint activity (Bales, 1951; Jordan & Henderson, 1995). On the one hand, these categories are centered on what an observer (i.e. the analyst) perceives while we strive to uncover the perspective of the participants and how they orient to the moment-by-moment interaction. In addition, classical interaction and other coding approaches take the linguistic unit as their unit or analysis when assigning a code to a sentence or posting. In our case, our unit of analysis requires that we define activities and networks of activities that can span from a few seconds to a series of interactional episodes.

In summary, diverse research literatures have approached, directly or indirectly, the problem of sustaining continuity in collective knowledge building. However, the constructs and theories explored do not account for the entire range of phenomena that we have characterized as "bridging" and which we have illustrated through he preliminary data analysis provided in section II. Next we elaborate on how some of these concepts will be used to construct our choice of theoretical framework and present the details of our proposed method proposed to investigate the three three research questions outlined.

Plan of Inquiry

Having explored the relationship between our research questions and the current state of relevant knowledge, we outline now the particular process of inquiry devised to investigate the phenomena of "bridging" in a particular computer-supported collaborative learning context. Considering that our central aim is to characterize the ways in which bridging contributes to the establishment of continuity in the knowledge-building experience of virtual teams in online learning communities, our means of inquiry are fundamentally descriptive and grounded on naturalistic data resources collected longitudinally. In the following subsections we describe the theoretical framework guiding this plan and the research method selected, including the strategies and time estimates devised for data collection and analysis.

Conceptual framework:

We define "bridging" as the interactional work performed by co-participants to establish continuity in a context in which virtual teams collaborate in learning tasks across multiple sessions and where multiple teams might form an online community. One of the central conjectures of our proposed research is that bridging is highly consequential for the nature of the overall knowledge-building experience. We seek to understand how bridging is achieved in interaction and to explore its role for the collaborative learning activities conducted by the teams we propose to study. In an attempt to identify the relationship between bridging and collaborative learning and specify a conceptual framework, we start by characterizing these interactions as strongly determined by the particulars of the participants, the activities that they orient to, and the resources at their disposal; all contextual factors which are made relevant in and through interaction. This foundational position emerges from our preliminary analysis of virtual teams and is also supported by the stance of situated cognition (Engestrom & Middleton, 1998; Suchman, 1987; Winograd & Flores, 1986) , also known as the "situative" perspective (Greeno, 2006).

The situative perspective argues that the organization of action is an emergent property of moment-by-moment interactions between actors, and between actors and the environments of their action. Furthermore, it argues that learning and knowledge are distributed across people and artifacts, and because of this, focuses on understanding activity and changes in the activity systems in which knowledge is co-constructed and used jointly. As Greeno has recently argued (2006) the situative approach encompasses other perspectives such as those underlying sociocultural psychology (Rogoff, 1995; Wertsch, 1991), activity theory (Engeström et al., 1999), and distributed cognition (Hutchins, 1995) given their fundamental commitment to analyze performance and transformations of activity systems. This can also be extended to apply to the recently proposed theory of group cognition (Stahl, 2006a) in which collaborative group discourse is examined as the central force behind the emergence of group knowledge, as we discussed in section III.

The initial conceptual model of bridging (which we expect to refine through the research proposed), theorizes bridging interactions as mediators of effective, sustained collaborative learning. We expect bridging activity to be closely related to the productive knowledge-building experience of teams. As indicated in Figure 4.1, bridging might occur at many social and temporal levels with the arrows indicating bridging across interactions among individuals, teams and communities and across time .

Figure 4.1 Conceptual model of timelines and trajectories of sustained collaboration

As individual teams engage in their episodes of knowledge building and collaborative-learning they generate ideas and co-construct resources that can be – and are designed by the participants to be – taken up by the same team, later on, or by other teams as part of a larger collectivity (e.g., an online community). In fact, in these particular contexts where longitudinal work is pervasive, the retrospective and projective relationship of moment-by-moment interactions is expected to signal active engagement with the knowledge work of the team itself and, possibly, its larger context. This process of uptake, the establishment of continuity and relevance that attempts to span across interactional boundaries, is expected to play a significant role in the sustainability of knowledge-building work across individual episodes and across multiple collectivities or teams. The particular dynamics of this bridging activities constitute the object of study of our proposed research and is expected to reveal how past episodes are made relevant as constitutive resources (e.g., contrasting, confirming, expanding present work) which co-participants use in particular ways. In addition, instances of bridging are expected to signal the interdependency between the local small-group unit of analysis and higher units of analysis (e.g., online communities, multi-team collectivities, etc.) as well as between the individual and small-group levels. It is important to note that we restrict our goals to the characterization of bridging and its role in online collaborative learning in settings like VMT.

An important contribution of this dissertation will be the development of a set of theoretical constructs for characterizing the dynamics of bridging and for analyzing bridging mechanisms. This framework builds on core concepts of situated theory, activity theory, distributed cognition and group cognition, but goes beyond them based on detailed analyses of examples in our data. Some phenomena and terms that we anticipate explicating and refining are: bridging, bridging artifact, projective bridging, bridging interaction, bridging methods and interactional continuity. Central to our conceptual framework is the view that, to understand bridging and its functions, the interactional perspective is essential. In addition to the elements of our conceptual model of bridging, the commitment to bridging as interactional phenomena is central to our theoretical framework. This commitment also guides our choice of naturalistic data and the use of micro-level methodologies as we will explore in subsequent sections.

Method

In the following sections we present our choice of research framework and the corresponding data analysis techniques proposed. In addition, we define our proposed unit of analysis and other central elements of these research plan.

Research Framework: Design-based Research

As discussed in section III, the research proposed here is situated in the context of the field of Computer-supported Collaborative Learning (CSCL) and, more generally, within the emerging field of the Learning Sciences. Research in the learning sciences attempts to "better understand the cognitive and social processes that result in the most effective learning, and to use this knowledge to redesign classrooms and other learning environments so that people learn more deeply and more effectively" (Sawyer, 2006) . Our research goal is highly rooted in this orientation for we strive to understand "bridging" as interactional phenomena related to knowledge building in online collaborative learning teams, and use our findings to advance the design of interaction supports for the Virtual Math Teams (VMT) online community. Consequently, we have adopted the framework of "design-based research" proposed within the Learning Sciences as the guiding structure of our method and of inquiry.

The concept of Design-Based Research (DBR) was introduced in the early nineties, in writings by learning scientists Ann Brown (Brown, 1992) and Allan Collins (Collins, 1992) who proposed the use of "design experiments" as a strategy to cope with the complexities of investigating how designed artifacts (e.g., curricula, computational tools, etc.) contributed to learning in real-life settings . Design experiments were conceived as a way to extend laboratory experiments, ethnographies, and large-scale studies by providing a framework for formative research which combines incremental design and the progressive development of theory (Cobb et al., 2003; diSessa, 1991). The typical design experiment is defined by Cobb et al. (2003) as "both engineering particular forms of learning and systematically studying those forms of learning within the context defined by the means of supporting them" (p.9). This iterative combination of applied design and systematic theoretical development is the central characterizing element of design experiments and the motivation behind its current widespread use (Barab & Kirshner, 2001; Design-based Research Collective, 2003; Edelson, 2001). It is precisely because of this iterative synergy between incremental design and systematic theory building that we propose to adopt the framework of design-based research. In our case, we aim to incrementally expand our understanding of specific interactional aspects of learning in online teams and virtual communities and utilize this knowledge to inform the design of computational supports.