Copyright 2003 IEEE. Published in the Proceedings of the Hawai'i International Conference On System Sciences, January 5-8, 2003, Hawaii

This work is supported by an NSF, KDI grant (award 9980182), and has benefited from discussion during a working group on scientific collaboration supported by the National Center for Ecological Analysis and Synthesis. The findings do not necessarily reflect the views of these agencies. Thanks go to members of our Distributed Knowledge Research Collaborative (DKRC) for discussion and input during the progress of this work (www.dkrc.org).

In 2001 we began interviewing members of a group - referred to here by the pseudonym ECAT - working on the creation of a database of culturally and linguistically appropriate materials for practitioners in their field. Interviews with 23 participants revealed that a key part of the learning entailed in getting their work done was about how to approach the issue of "appropriateness" on the way to deciding what to include in the database. The problem of exchanging knowledge around this emotionally-charged topic seemed to go beyond taking the tacit knowledge 'in their heads,' or their practices and experience, and making it explicit for others. Instead, solving this problem involved exploring and evaluating notions of appropriateness, making a definition and application that would be constantly open to thought and interpretation.

In coming to understand the operations of this group, we found the view presented by Cook and Brown [5] that knowledge is both possessed by the individual yet also, inextricably, present and expressed in practice, resonated with the kind of interaction and problem solving seen in the ECAT group. In this paper, we adopt Cook and Brown's perspective of the generative dance to interpret the experiences of members of the ECAT group. We also appropriate their terminology when we refer to generative knowledge, a term we use to name the kind of epistemically productive, constantly emerging knowledge we found created by ECAT as they came to describe an approach to the difficult issue of appropriateness, and to create a framework within which the judgment of appropriateness could be constantly held open for examination.

The knowledge management literature gives us many terms for talking about how to bring forth tacit knowledge - i.e., that which is embedded in practice and in know-how, and possessed 'in the heads' of individuals - to make it adoptable and repeatable as common practice and codified information, i.e., as explicit knowledge (see [13], [14]). Such tacit knowledge is often treated as static, something held by the individual which can be released by finding the right key, teaching method, writing style, or interview technique, and then transferred to others. However, while important for technology transfer, and the creation of routine, repeatable performance, as Cook and Brown [5] note, this view does not explain the creation of new knowledge, and in particular creative production by teams or other groups rather than by individuals. In the work of Cook & Brown, we find a more nuanced view of such tacit and explicit knowledge, and one that includes knowledge in groups and in practice.

Key to Cook and Brown's view is identification of the difference between knowledge that is possessed by its owner and knowing that is demonstrated and enacted as part of action. 'Knowledge' is seen as "a tool at the service of knowing" (p. 388), something brought into the action to be used during practice - e.g., as a psychologist might bring theories of analysis into a therapy session - while 'knowing' "is about interaction between the knower(s) and the world" (p.388) - e.g., as the psychologists modifies technique and direction during interaction with a patient.

While 'knowledge' and 'knowing' may be mapped onto the categories of explicit and tacit knowledge, in Cook and Brown's interpretation 'knowing' is considered inseparable from practice and interaction with the world. They see explicit and tacit knowledge as two distinct forms of knowledge: while we may try to articulate what we do in practice, this does not separate our knowing from our practice, nor can such an articulation make a person proficient at practice. While "each form of knowledge can often be used as an aid in acquiring the other," e.g., in receiving instructions on how to ride a bicycle, "neither tacit nor explicit knowledge can be used by itself to acquire the other: one must also, at the very least, get on a bicycle" (p.385).

Cook and Brown consider knowledge 'in the head' to belong to an epistemology of possession, which they see as insufficient on its own to explain how new knowledge is generated. They argue for adding a parallel epistemology of practice so we can "speak about the epistemic work done by human action itself - that is, about what is part of practice as well as what is possessed in the head" (p.382).

Important for the work presented here is that their view presents a framework for understanding the generation of new knowledge, and particularly knowledge in groups: "For human groups, the source of new knowledge and knowing lies in the use of knowledge as a tool of knowing within situated interaction with the social and physical world" (p.383), i.e., we must bring the knowledge in our heads into play in actual situations in order to go forward in what we know. Through conversation with others in the social world, and/or interaction with objects, devices, etc. in the physical world, the knowledge we possess is used in practice and "dynamically affords a generative dance within which the creation of new knowledge and new ways of knowing is possible" (p.393).

This generative dance is directed and interpreted. It is directed because it is performed as part of a "productive inquiry" ([5], citing Dewey), i.e., an active engagement in a problem, in pursuit of an answer to a question or inquiry. Individual and joint activities associated with this dance are intentional, not incidental, as is also the learning achieved by individuals and the group, and interactions associated with the inquiry (see [16] re intentionality and knowledge-building communities; [11] re design and participation in collaboratories).

It is interpreted because its form and outcomes differ according to the local knowledge, inquiries, and practices associated with the community in which the dance is performed ([2], [3], [6], [10]). It is also interpreted as individuals incorporate the new knowledge into their own world view, readjusting their thinking according to evidence, arguments, or directives stemming from this new knowledge. Moreover, as the new knowledge becomes a way of knowing, it is exhibited outside the head. It is embodied in new practice, becoming new tacit knowledge associated with and expressed through local, often physical, conditions and actions such as new procedures, routines, physical skills, communities, or technologies.

The generative dance is an important phenomenon to examine, particularly where the interest is in new knowledge and knowing, or "knowledgeability" as it is expressed by Orlikowski [15]. Current initiatives in interdisciplinary and multidisciplinary work, as well as in alliances and joint ventures, start with a premise that the combination of elements will create something new. Yet how does this proceed? How is the generative dance enacted?

This remainder of this paper explores the generative dance for the ECAT group whose productive inquiry focused on the creation of a database of culturally and linguistically diverse, yet appropriate, materials. In examining how this group managed this task, many elements of the generative dance became evident, including incorporation of new values into the practices and the product of the group (i.e., the database, its contents, and decision making procedures for evaluating potential content). We find that this incorporation involves more layers of practice and product than might be envisioned by a singular view of where new knowledge will be found. In particular, as the group dealt with discussion around the sensitive topic of cultural and linguistic appropriateness, they learned how to discuss sensitive issues and how to establish and run meetings that allowed for exchange of views. This learning runs parallel to the more 'concrete' learning of what constitutes an 'appropriate' material, or even the process now used to select materials. In what follows we explore not only how the generative dance proceeded, but also the layers of learning and new knowledge it entailed for these group members.

While details of the field in which ECAT was involved will not be given in depth here in order to help preserve the anonymity of the group, a few characteristics are important. The project was funded in 1996 for five years by U.S. government grants, and participants include university faculty, graduate students, advisory board members, field practitioners, and database developers. Participating organizations include two universities, a government-supported organization maintaining information systems for education-related literature, and an international professional organization for practitioners in the field. The ECAT project addressed the creation of a repository of educational materials to help practitioners providing services in this field to different races, languages, and cultures in the U.S. As such, there was an emphasis on evaluating materials not only for their scientific validity, but also for sensitivity and correctness in portraying different classwebs and addressing their day-to-day circumstances. At the time of writing, the project has officially passed its five year lifespan; the database is in place and considered a great success with over 4000 materials included, well over the forecast 500. Principal investigators are seeking further funding to maintain the database as an ongoing resource.

Data were collected in a series of semi-structured interviews conducted by the authors in 2001 and 2002. Interviews were taking place during the last six months of the five year project, and up to six months after. None of the authors are associated with the ECAT project, nor do any of the authors work in the same discipline or department as members of the ECAT project. This project came to the attention of the authors in connection with work looking at distributed knowledge processes and collaboration in research teams as an example of a successful collaboration that was nearing the end of its mandate.

Interviews were conducted with 23 members of the project, including faculty, graduate students, project coordinators, board members and database designers. All have been interviewed at least once, and all but a few twice. All interviews were tape-recorded and transcribed; where possible they were conducted face-to-face, otherwise by telephone. Each interview took from 1 to 11/2 hours; transcripts range in length from 3,000 to 15,000 words (9 to 20 pages) per interview.

ECAT members include four Principal Investigators (PIs) working at two universities (three interviewed), and two management staff involved in the running of ECAT (both interviewed). There are 16 collaborators on the database construction and management, who worked at the participating organizations and in various university departments (eight interviewed), and two web designers (both interviewed). Early after its inception, a national advisory board was created to increase the diversity of cultures, regions, and institutions with a voice in the work of ECAT. Board members came from across the U.S., representing regional and national cultural groups, and educational groups (five interviewed). Finally, there were many graduate students involved in ECAT, coming and going over the years of the project (four were interviewed).

Interviews and analysis proceeded using a grounded theory approach [17]. Discussions of results from the first set of interviews were used to guide questions for the following interviews. Analysis consists of coding the data looking for themes relating to the category of main interest, which for this paper has been the generative dance around sensitive issues, comparing across individuals for commonalities and differences and analyzing the characteristics of the themes that emerged.

The first interview collected data on the role of the individual in the ECAT project, e.g., their position and involvement in the work and discussions, as well as their general impressions of the work process. A series of questions asked about the development of the database and the discussions around that work, their role in those discussions and that work, their characterization of the process (we asked for a metaphor to describe working on the database and the project). We explored the chronology of the project, their contributions and how these changed over time, changes in assumptions over time, and challenges and problems during the project. We also asked how their knowledge and expertise were brought into the project, and how they represented the database to others. The second interview followed up relevant threads from the first interview, particularly exploring their take on the meaning of ECAT, whether (and how) perceptions of what it meant changed over time, and what they perceived as its impact at what was then the end of its five year lifespan. Questions also asked about knowledge exchanges and interactions by asking extensively about what they learned from others and what others learned from them.

While there are many avenues of exploration in the rich data from the 23 interviewees, we concentrate here on the topic that emerged as a cornerstone to the group's operations: the enactment of the generative dance around difficult and sensitive matters. The dance is most evident in how ECAT members came to share understanding of how to approach cultural and linguistic appropriateness, and how to understand and represent multiple voices in their final product. We term this generative knowledge because what ECAT members were creating was not an answer to a problem of 'what is appropriate' but rather a way to approach the question, a framework within which appropriateness could be examined. In this way such knowledge is epistemically productive as it opens the conditions for the further creation of knowledge. Key to their work has been uninhibited discussion of the "assumptions" about cultural and linguistic appropriateness that underpinned the selection procedures for materials for the database. The discussions led to incorporation of values into individual's practice as they came to understand in a new way what it meant to support cultural diversity. They also were instantiated in group practice in the guidelines for reviewing materials for appropriateness, and the instructions to users on how to approach and use materials in the database.

Interviews reveal that the success of these discussions (which built, in turn, the success of ECAT) depended on sharing "voices" representing different views, different cultures, and the creation of an environment in which such sharing could take place despite differences in outlooks. However, sharing alone would not have been enough. Success and trust in the process also depended on the visible and evident integration of what was voiced in those discussions into both other individual's world views and the structures of the project. Participants needed to see that others associated with ECAT had taken their opinions and incorporated them into their own values and approaches, as well as into review guidelines for materials, and into the database.

In keeping with the generative dance, the group's focus on its intended product - the database of culturally and linguistically appropriate materials - acted as the constant reference point for their "productive inquiry." Discussion of its contents, the proper way to evaluate materials for it, and how to train reviewers to do that evaluation, served as an embodiment of their goals, a way of presenting explicitly the tacit knowledge they were accumulating through group interaction. It also acted as a way of showing how changes in those views were updated, revised and made visible, e.g., through changes in database content and guidelines for that content.

Moreover, the development of guidelines for evaluation of materials, rather than instructions for how to accept or reject of materials, reveal the epistemologically productive aspect of their work. Guidelines, as well as training sessions for material reviewers, were designed to create active engagement by non-ECAT members in decisions about appropriateness, examination of their own assumptions about appropriateness, and where and how to use materials found in the database or elsewhere.

In exploring further how the generative dance was enacted in ECAT, our discussion highlights these aspects of the group's inquiry and outcomes:

• how the generative dance unfolded around the emotionally-charged "hot" topic of cultural and linguistic appropriateness
• how the visible instantiation of discussion outcomes into individual and group practice built trust among group members
• how group members learned to engage in the generative dance, and in particular how to engage in the generative dance around a sensitive topic
• and the layers of learning and knowledge their dance entailed.

It was impossible to ignore from our interviews the key importance of the sensitive nature of the foundation to the group's activities - i.e., the nature and meaning of linguistic and cultural "appropriateness." This theme, and the "hot" discussions around it, appear centrally through the 46+ hours of the interviews, mentioned by every participant regardless of their role - whether a database developer, student or PI. As such it represents a distinguishing feature of this group that differentiates it from the kinds of work or project teams often considered when dealing with knowledge issues.

As our interviewees explained, the problem in addressing diversity is that as we strive to include one voice, we may exclude another; and if we cling to one view of appropriateness or view situations through only one lens, we exclude or fail to see the possibility of other views. Dealing with diversity is a difficult problem, and such difficulties should not be reduced to the simplicity of a lack of desire to solve the problem or an inability to comprehend it. The problem is complex, and conditions vary with context, yet we tend to approach them in a simplifying manner. For example, our discourses often favor similarity over difference (see [4] for a discussion of this problem for talking about "communities"; see also [7]). Interviewees pointed out how we simplify differences by language when there are multiple variants of culture to go with each language (e.g., the multiple U.S. and non-U.S. cultures that make up Spanish speakers); and we simplify diversity in culture by identifying classweb from one race as one group (e.g., the many nations that make up the Native American population). Simplifying categorizations such as these can put undue burden on classweb to speak for many cultures, and not necessarily ones with which they are familiar. They can unduly shape the way we view problems and thus what decisions we make about data collection and organization ([1], [18]).

There are parallels also in how we simplify to make knowledge mobile. We strive for the generic, the reproducible, and even the transdisciplinary [8] when we seek to take what we know in one modality and use it in another. In our efforts to make knowledge mobile, we move away from local conditions to find what holds across all environments [3]. Re-writers, practitioners, and users are left to translate and fit these generic materials to their local environment, classweb whose expertise is in knowledge and ways of knowing about local conditions. They re-embed into local context what has been disembedded for mobility (Giddens, cited in [3]).

The ECAT group set out to explore and integrate diversity into their work, and to bring together a set of materials that re-embedded diversity into materials for practitioners in the field. Their knowledge problem was mobility in reverse, yet it was not a static, one-off operation. They also had to keep the problem open and continuously interpreted. Unlike many knowledge mobilization situations, the goal was not to come to a common answer to a problem, but instead to a common approach to a problem. Recasting the problem in terms of creating a process rather than finding an answer or creating a product appears to have happened as a results of the "facilities and frustrations" ([5], p. 389) afforded by interaction with their social world - one that included those both experienced and inexperienced in handling cultural diversity.

In their first interactions, ECAT members found dramatically, in a hot encounter, that their first challenge was to understand what they meant by diversity, and by cultural and linguistic appropriateness. Cook and Brown [5] write of the difference between individual knowledge and knowledge held in common by the group. In their example, a doctor knows how to use his or her hands to make a diagnosis, but only medical groups know what constitutes acceptable practice for that diagnosis. While some ECAT members knew appropriateness for some cultures, they did not individually cover all cultures for whom they wanted to select materials; and while some members had ideas of what "appropriateness" meant, as a group they did not yet know what constituted acceptable practice. Through conversation and interaction with the world - in this case, social interaction among board members, PIs, staff, and students - they came to develop their own new knowledge about appropriateness, one instantiated in the heads of those who possessed it, as well as in the actions of the group as it established routines for reviewing and incorporating materials into the database.

The heat of this issue and ensuing discussion - particularly at the first board meeting - might have sunk the whole endeavor, but it appears that processes during the first meeting established quite a different long-term tone:

Such comments on the positive outcomes of the first meeting are echoed by others of the advisory board members as well as staff members, such as this interviewee who elaborates further on how trust developed around this work:

Yet, on its own, trust during meetings would not have been enough. To come back to the table and to contribute again, participants needed to see the knowledge and ways of knowing exhibited and shared at meetings become part of the practice of ECAT. They saw this both explicitly, in written documents and procedures:

and tacitly, in incorporation into others' practice, for example:

Trust emerges as a mediating factor in group processes in two ways. First, trust regulated the extent to which members were willing to share experiences and give input for the project. As the advisory board and others gained trust in the ECAT staff and its processes, e.g., as they saw staff incorporating their views into the ECAT project effort, and as they saw the opportunity to do something of lasting value, the more willing they were to contribute to discussions. Trust in the staff mediated the desire to add input to the whole project. Second, trust regulated the extent to which input was incorporated into the project. For example, ECAT staff gained trust in the process as their own efforts were accepted positively by board members. Trust in the relationship between board and staff increased the willingness and ability to incorporate input into the work.

Meetings with the advisory board were only one of two major examples of the way the ECAT group members had to confront the challenging issue of talking about sensitive issues. The other example relates to the graduate students, who again represented a greater range of cultural and linguistic diversity than the principal investigators and management staff. Knowledge inherent in the experiences of the graduate students had to be recognized, and then incorporated into ECAT practice.

Again, the report suggests that the evident action on the proposal regarding the graduate students, and its incorporation into management's values, was important, at least to this board member. It is not as clear what trust was built between ECAT staff and students. Many students stayed through to the end of the project, but some moved on to other endeavors. However, reports from some interviewees suggest that a mid-course correction improved the way student experiences and opinions were incorporated into ECAT.

Generative knowledge, as we have termed it here, is evident not only in the way interaction among ECAT members approached the problem, but also in how they then presented that problem to others. They do not present a solution, but rather a way of approaching the problem. Here one of the staff members explains how they present the ideas to those who come to the ECAT project from outside the group. This quote also describes how they had learnt to accept the "tension" of the generative dance, and accepted that as part of the process of dealing with this sensitive issue.

The kind of knowledge produced by this group is itself generative, i.e., composed of components of both knowledge and knowing continuously enacted through interaction with members of the social world and artifacts from the physical world (e.g., materials for the database). While much work on knowledge in organizations works on existing or ongoing knowledge in a relatively stable social structure (i.e., the structure of the organization is stable, even as classweb may come and go within it), ECAT was breaking new ground, breaking in new participants, and breaking in new processes. ECAT members came to know how to enact a generative dance around a sensitive subject. They - as individuals and as a whole - learned not just how to assess, evaluate, talk about, and think about cultural and linguistic appropriateness (a learning of practice), but also how to effect discussion and productive inquiry around such a theme (a learning of meta-processes around practice). Just as Orlikowski [15] found that a key to success in the multi-national Kappa organization was "a collective competence in knowing how to deliver innovative yet complex products in a timely fashion,"(p. 256), a key to ECAT's success was that it developed a collective competence in knowing how to conduct productive inquiry predicated on inclusive representation of diverse sensibilities. By the end of its five year lifespan, individuals and the ECAT group as a whole had learned how to create an environment in which such a generative dance could take place.

Key to the generative aspect of this knowledge is that is it continuously open for examination. This openness corresponds to Latour's [9] concept of negative modality. He suggests that in scientific writing a positive modality is found in statements that lead away from the conditions of production. By contrast, negative modalities lead toward the conditions of production; these are "opened up and many specific conditions of production will be added" ([9], p.43). One of the questions asked in interviews was whether there were topics with a lot of discussion, and whether there were topics around which there was 'no' or 'little' discussion. A typical response to the latter was "I don't know if there was little discussion on anything!" suggesting everything was open for examination of the 'conditions of production.' This same person continues:

Thus, we have the interesting situation of having a productive inquiry toward a database, open to continuous discussion of the conditions of production. A common situation for new work, new knowledge, and new groups, yet one that in its generative state represents a markedly different challenge than the management of problems with more well-defined conditions of production.

Members of ecat shared their knowledge and experiences as members of their culture, professionals providing service to members of different cultures, educators teaching future professionals, administrators and consultants for regional programs, and experts in subfields associated with the project. While sharing knowledge about diversity in practice, they constructed something new for the group - the "assumptions" which became outlined in a written document. As well, individuals learnt from and about others. They incorporated into their own views new ways of knowing:

There is learning not just about other classweb's experiences, but also in how to relate to others who have these experiences. The learning includes both knowledge about what it is like to be outside the mainstream culture of the U.S., but also knowledge about how to talk about it, and eventually how to create a working group around the discussion of it.

To summarize the activities and work of this group, we present five themes (or activities in Orlikowski's [15] terminology) we identified from our interviews that relate to the achievement of knowledge creation and exchange in ECAT (outlined in Table 1). These themes emerge in relation to the type of knowledge problem faced by this group. Their task is not technology transfer, nor is it a master-apprentice inculcation of practice. It is instead a coming to understanding of process - i.e., how to maintain a continuously open approach to a difficult problem. The topic is hot and sensitive. There is no one right answer but rather a process to be learned. It is a constantly open problem that doesn't reach closure, and has no explicit definition or codification except in how to unfreeze, unbundle, recombine, question foundations, and open premises for examination.

Developing trust is a necessary starting condition for exchange of any sort, and one much discussed in the knowledge management literature. Face-to-face meetings brought together ECAT participants for discussion. Action based on these discussions, as well as the obvious incorporation of discussion points into products and practices acted as catalysts for more trust. Along the way, participants came to know each other and how they could work together on this problem in this context.

Engaging in productive inquiry, Maintaining openness, and Doing the generative dance are each part of the active process of moving the project forward. Engaging in productive inquiry entailed maintaining a focus on the topic of inquiry, its relation to the database product, its eventual users and preferred design characteristics. This focus gave purpose to discussions, and a goal to keep in mind. While the goal of a collection of materials was known from the beginning, the discussion allowed refinement of how the goal would be met in practice.

Maintaining openness is a key factor in dealing with this kind of knowledge problem - a topic that must constantly be kept open for examination, where the knowledge that can be gained is not an answer to a specific question but instead a way of doing and thinking about things. As one interviewee described, there were cultures and languages for which materials were being selected when no one on the board or staff came from that background or knew about it. What had to be put in place was a process that could deal with that, not an answer for any specific combination of culture, language, race, etc.

Doing the generative dance entails change in outlooks, incorporation of new viewpoints, synthesis into one's own world view, accepting the creative tension. Its acts both inwardly, changing the views of the recipient, and outwardly, effecting change in another's viewpoint. Doing the dance requires contribution, receptiveness to change, and focus on the chosen inquiry.

Creating productive process is another theme that is important for the kind of knowledge dealt with here. Accepting open evaluation of diversity entails instantiating this principle into how work gets done. It's not a surface level adherence to procedure, nor the acquisition of master level expertise that can be applied to production of another database. It is instead a meta-level skill, one that operates like transdisciplinarity, at a generic, cross-situational level. Individuals, and the group, are learning how to deal with doing the generative dance around difficult practices. Their knowledge is at the level of how to do this kind of inquiry again, in another context, and on another difficult practice.

Note: This table borrows its form from Orlikowski [15].

We have describe how interactions among participants in ECAT, enacted in service of productive inquiry directed to the creation of the database of culturally and linguistically appropriate materials, progressed as a generative dance about how to approach the issue of 'appropriateness' as they made decisions about what to include in the database. While the database content exists as a visible, explicit result from the generative dance, equally important has been the tacit incorporation of values about appropriateness both by individuals and the group as a whole. Members of this group not only do the dance, but learn that it is necessary, on what it is focused, and how it is performed with these others in this context. Moreover, the group gained in knowledgeability about process; they learned how to discuss and move forward a project that is based in the discussion of 'hotly contested' areas, in the face of 'passionately held' beliefs.

In engaging in the generative dance, the database co-constructed by participants in the ECAT project has become something much more than an instrumental repository of resources. In purposefully negotiating with each other in a productive inquiry into what constitutes appropriateness, how materials should be evaluated, and how to get these data out to others, they created a narrative beyond the basics of the data (see [12] re databases and narratives). The definition of the database emerges as an instantiation of the meaning of the group, i.e., as the explicit representation of the multiple exchanges and interactions regarding tacit group members' knowledge. Not only does this database remain as a technological artifact embodying the generative dance so far, it also stands as a base for future work in this area. As such, its definition and existence represents an important stage in a longer generative dance, one that spans 'generations' of funding cycles and project definition.

As organizations increasingly include diverse members who span not just nations and national cultures, but also local cultural boundaries - whether based on experiences, traditions, or disciplines - it becomes important for knowledge management to learn how to facilitate discussions in such environments. It is also important to recognize that not all problem discussion leads to a definitive, codified answer. Instead, difficult, sensitive topics need to be continuously open, and this is something that is learned in observing and doing the generative dance around generative knowledge.