Dialogue in context, towards a referential approach in collective learning

AI & Soc (2006) 20: 314–330 DOI 10.1007/s00146-005-0023-4 ORIGINAL PAPER

Marie-Laure Betbeder Æ Philippe Cottier Colin Schmidt Æ Pierre Tchounikine

Dialogue in context, towards a referential approach in collective learning Received: 24 February 2004 / Accepted: 10 August 2005 / Published online: 17 December 2005 Ó Springer-Verlag London Limited 2005

Abstract In this article, we present research in the making of a collective work environment within the framework of a distance education course. We base our theoretical and methodological standpoints on examples of dialogical discourses recorded within the framework of this CSCL system called Symba. In fact, the results of previous research lead us to rethink our vision of the study of collaborative moments between participants in a computer-supported human learning environment that proposes several communication tools. Redefining the methodological process aiming at finding and understanding these rich learning moments is also necessary. We intend to describe ‘‘socio-technical’’ instances during which these collaboration phases appear. More generally speaking, our aim is to draw up both new theoretical and methodological perspectives that would be reusable in CSCL environments; in view of the nature of these two perspectives, and the diversity of the domain knowledge (sociology, cognitivism, linguistics, philosophy, statistics, etc.) brought to bear in the study of the environment in question, our approach constitutes a trans-disciplinary reassessment of the uses of the communication tools—and the study thereof—proposed. Keywords Context Æ Communication Æ CSCL Æ Dialogism Æ Pragmatics

Introduction Communication is the most important notion in collective1 learning activities. In what follows, we detail just one of the analytical phases of our current work on 1

In the CSCL and CSCW literature, it is usual to distinguish the collaborative approach (the different participants involved are working altogether on each issue of the work) and the cooperative approach (division of the work between the different people involved, each of them being in charge of one part of the activity) (Roschelle and Teasley 1995). The ‘‘collective’’ term will be used in this paper to denote a more general term. M.-L. Betbeder (&) Æ P. Cottier Æ C. Schmidt Æ P. Tchounikine LIUM, Laboratoire d’informatique de l’universite´ du Maine, Le Mans University, FRE 2730 du CNRS, Laval, France E-mail: [email protected]

315

Symba2 that will contribute to understanding how group learning works and how this understanding may help improve CSCL systems. But before getting down into empirical side of our work, we suggest that soundly grounding the context of discussion on CSCL issues in logic would be highly beneficial to the final explanatory process. As there seems to be some confusion as to just what constitutes communication, perhaps due to the coming of age of new technologies, developing a precise model of communication is extremely important, even if only to avoid calling certain activities communication when they are not. Turning points in CSCL design history are possibly at stake here. We can anchor the theoretical side of this enterprise in the progress made in the actual communication modelling process. Given that the main explanations for linguistic sequences are ex post facto in nature, they logically lead to monological models. This reminds us of the dawning of ‘‘serious’’ communication theories in human communication of which the first merely addressed the issue of inert information transfer, but called it The Mathematical Theory of Communication (Shannon and Weaver 1949). Over the years Francis Jacques has used rigour that is sound in our view and subtle enough to build a truly dialogical model of human communication that can be useful to CSCL. First of all monological models—which are still currently applied in the study of CSCL systems—crumble under the weight of his question: ‘‘what meaning should be assigned to the term [no tag]message’ when it involves a sequence that is addressed by no-one to no-one, but circulates between functional poles?’’ (cf. Jacques 1985, p. 190). This is in fact a good question to those that model interaction in any system because, if the poles in the exchange are human, no information is sent from one pole to another, there are only indications (voice sound wavelengths, light from written figures on paper, etc.) that the two are getting to know about some piece of information. As we shall point out further on, from time to time there are no indications at all of the collective activity of communicating as it is not always necessary for the ‘‘current’’ to pass, hence our putting the information being referred to, and the physical manifestation acting

S1

S1= human interlocutor S2= human interlocutor

S2

dialogue

Referent

Propositional Content

as an indicator of it, outside the realm of the communicating poles in the diagram below (cf. Jacques 1985, p. 213). The propositional content (only) indicates the state of the current communication process conjointly produced by the speakers S1 and S2. The word ‘‘dialogue’’ here shows the position of utterances—spoken or written—in our 2

Symba is a Web-based framework designed to support collective activities in a learning context.

316

model, outside the open pragmatic system of language usage rules; the referent is of course the theoretical object the Self (S1) is drawing to the attention of the Other (S2). This corresponds to the exchanges between students collected from the Symba CSCL system’s chat database (cf. excerpts below). We intend to analyse a good number of these excerpts with an original methodological approach in order to improve collaborative communication in using the system.

The Symba environment General presentation and key objectives Symba is a Web-based framework designed to support collective activities in a learning context. Through the design of Symba we are particularly concerned with the problem of providing students with a framework that leads them to work out organizational features, e.g. address problems such as decomposition of the work to be done into different individual and collective tasks, task scheduling or identification of the resources and tools to be used to perform these tasks. As an example, for the last experiment, the pedagogical activity we proposed concerned the ergonomics criteria and the interface criticism. Students had to research individually information about ergonomics criteria for the design of Web sites. Then they had to discuss and confront theirs opinions to draw up a collective document, proposing a set of ergonomics criteria to be taken into account for the design of a Web site. During a second cycle, students had to research and evaluate individually a distance learning Web site with the help of their collective list. Finally, they had to discuss about their individual document. Symba has been designed with a double objective: (1) to make students explicitly work out their organization and (2) provide tailorability features allowing students to decide which tools and resources they want to be accessible in order to achieve the tasks they have defined. Symba dissociates an ‘‘organizational level’’ and an ‘‘activity level’’. The organization level allows students to organize their activity as a set of phases, each phase being structured as a plan (sequence of tasks). When expliciting a task, students can define the tools and resources that should be available at the activity level to achieve it. This allows combining the two objectives: students can benefit from the activity level tailored according to their wishes by explicitly defining their organization. The literature and our own preliminary experiments (Betbeder and Tchounikine 2003) lead us to specify the key features that we have studied through this research, which correspond to the general specifications of Symba: 1. The framework must dissociate an organization level and an activity level. 2. The organization must be reified. 3. When achieving a task, the students must be offered tools that meet their wills (i.e. the activity environment must be tailorable).We detail these three points in the following sub-sections.

317

Dissociation of an organization level and an activity level Within RSC3 activities, students have to confront their individual productions and realize a collective work in a limited time. We observed that although a large amount of communication is concerned with coordination features (what tasks to achieve, who will achieve them, etc.), students experience difficulties in structuring their organization. Providing some groups with a proposed organization (they could follow or adapt) whilst others had to organize themselves from scratch highlighted that groups which are not supported more or less fail in their goal of coordinating their work, leading them facing problems when achieving their collective work (Betbeder 2003). Collaboration requires negotiating at different levels from which, in particular, problem-solving strategies (Dillenbourg and Baker 1996). Within our context it appears very clearly that students do not explicit their problem-solving strategy. Although one of the principal goals of the activity is to initiate them to collaborative work (and this is explicitly stated), they attach more importance to the result to be produced than to the effectiveness of their collaboration, and do not understand that this lack of explicitness has a negative impact on the final production. It therefore appears necessary to help students in clarifying their organisation and encourage an analysis on the articulation of the collective work by making this task explicit and supporting it. Many potential solutions can be envisaged for this problem. As computer scientists, the solution we propose is to study how a computer tool can lead students to explicate their strategies. The framework therefore makes them work out these features (the objective is not to automatically propose the best organisation and tools to the students, but to bring them to work out these questions). In order to emphasize the importance of working out organization features, Symba dissociates an organization and an activity level. The organization environment aims at supporting learners in being conscious of the specific task of organizing their collective work by tackling explicitly and collectively an abstract high-level task: organization. It enables them to define the tasks to be realized and to specify the tools and resources that are required to achieve them. The activity environment is the environment where the activity is realized. For a given task, it provides the tools and resources that have been asked for at the organization level. Symba also provides an awareness level. In this level, different tools make students aware of the activities of the other group members (see Betbeder and Tchounikine 2004 for more details). Figure 1 proposes the general structure of Symba: on the left, students can select in the browser bar different descriptions (general presentation, description of the phases, the tools) or access to the different levels (organization, activity or awareness), on the right top, the frame provides information selected by the browser bar (in Fig. 1, the organization level is selected with the general diagram of the activity), on the right below, the frame is dedicated to the communication tools.

3

Collective activity that alternates a Research (individual) phase, a Structuring (individual) phase and a Confrontation (collective) phase.

318

Fig. 1 Snapshot of Symba

Reification of the organization as a plan Symba proposes editors that enable students to describe their organization as a plan. A plan defines how the students intend to decompose the general tasks into subtasks, how these tasks will be scheduled, who (an individual or a subgroup) will be responsible for the task and what tools will be used to achieve it. Plans are used here as proposed in Bardram (1997), i.e. means to organize the work and reflect the responsibility of the actors involved. A misunderstanding would be that we want students to bury their activity in a predefined intangible plan. We argue with others (Bardram 1997) that expliciting a plan is not in contradiction with the fact that action is situated (Suchman 1987). The plans we are concerned with here are (1) a means to make students work out organization features (before, while and after achieving the work) and (2) resources to carry the work out. A plan does not intangibly determine the students’ activity nor does it constrain the tools they can use, it reifies how they perceive (a priori or a posteriori) their organization. First level: a plan as a sequence of tasks At a first level, learners can define each phase as a set of tasks through a specific interface. The right part of the interface (Fig. 2) proposes a set of predefined abstract tasks such as ‘‘search for information’’, ‘‘analysis of a production’’, ‘‘discussion’’, ‘‘vote’’, ‘‘search for a meeting time’’, etc. (this list can be customized for a given application). A plan is defined by selecting some of these tasks and ordering them (left part of the interface). Students can also define new

319

types of tasks if required (as an example, in our experiment they defined ‘‘elaborate a final document’’). The interface is shared, students can connect themselves synchronously and browse the plan while it is being defined: all of the students see the interface, one of them can modify it (alternatively, using a shared interface system). Another part of the interface proposes a chat to discuss the plan synchronously. Second level: definition of the tasks At a second level, learners can describe the different tasks of the plan. The notion of ‘‘task’’ in Symba is conceptualized according to the activity theory and, more precisely, the Engestro¨m triangle (Engestro¨m 1987) and the mediation relationships it highlights (subject, object, community; rules, tools, division of labour) (Lewis 2000). Symba task description interface (Fig. 3) is isomorphic to the Engestro¨m triangle (although it is not presented in this way).4 For every task, the students have to define the objective (in natural language), the nature of the task (individual or collective), the subject (an individual, a sub-group, all the group), the beginning and ending dates (rules), the tools (that will be accessible at the activity level), the resources (files names) and the production (files names). The system dynamically generates a natural language explicitation of the description in order to avoid possible misunderstandings of the interface. Similarly to the description of the plan, the task description interface is shared and coupled with a chat. A tailored activity level As stated before, the analysis of the previous experiments highlighted that every group adopts its own organization (including groups that are presented with a proposed plan). It is not possible to predict this organization and, consequently, the tools that will be used by the students. A collective activity must be tailorable enough to allow the learners to decide on their organization (phases, tasks) and on the tools and resources that should be accessible at the activity level to achieve these tasks. This tailoring feature is proposed in Symba by the fact that students can decide at the organizational level (in the task description) what tools they want to be provided with in order to achieve the task. Symba generates from each task description a Web page that proposes, in an integrated frame, an access (a hyperlink) to the different tools that have been asked for. Figure 4 presents a snapshot of the activity level generated for a task whose associated tools are a browser and a chat. This Web page corresponds to the frame within which the students will achieve the task. An important point to be noted is that this approach allows students to tailor their environment without having to deal with any programming notions. The principle we have adopted in Symba is not to embed the framework with 4

We state as a working hypothesis that, although the activity theory is a descriptive framework for analysis, the notions it provides can also be used in a more prescriptive way to design computer-based frameworks.

320

Fig. 2 Defining a plan as a set of tasks. Snapshot of a plan constructed by students (2003 experiment). The plan is analysis productions (the different individual propositions from the previous phase), discuss/confront, elaborate a proposition (collectively) and make it available

specific tools but to allow the access from the framework to external tools. Symba can therefore easily be adapted/upgraded: adding a tool to Symba (i.e. enabling the students to use it) only requires describing the tool in order to allow the students to understand its possible uses and programming a piece of software that generates an access to this tool. Component integration is a research domain in itself, and how heterogeneous software components can be integrated is the object of a large amount of computer science studies (Bourguin and Derycke 2001). The way we integrate tools in Symba is a simple approach, but, given our objectives, sufficient.

Previous research and findings on the use of the e-learning application called Symba: highlights and limitations Symba is an experimental computer application used by M.Sc. level students at Le Mans University (France) for distance learning. The M.Sc. course is being taken both by distance learners and on-campus students and together they contribute to commonly proposed activities. Putting this application to use responds to several objectives the most important of which is to ‘‘teach students to learn in a collective manner [no tag]through practice’’’. The work theme changes every year and the learners’ activities are carried out according to a pre-established cycle, namely ‘‘Information Retrieval, categorising or sorting out of information and confrontation of individual works with a view to creating a

321

Fig. 3 Defining a task. The description in natural language generated by the system is every actor (of the group) must discuss and confront productions from the files ‘‘listes_criteres.doc’’ and elaborate a collective document put in the file ‘‘liste_coll.doc’’. The tools that will be used for this purpose is a mail service, a forum and a chat

collective production’’, the first two phases being individual exercises, and the latter of course, collective in nature. Practically speaking, the Symba environment allows participants to select the communication tools they wish to work with and organise their activities as they see fit. It thus has a good level of adaptability for the diversity of today’s student uses and changing habits. The actual state of the environment is the result of several iterative design stages (starting in 2001). It is within this design and research framework that we have conducted analyses on how the environment is being used in aim to advance knowledge on two fronts: to improve the nature device and to isolate some key moments of collaboration by students in its usage, moments in which several participants seem to carry out a truly collective activity rather than an individual one. This implies two stances. Firstly, our designers investigate into both the relevance of the pedagogical structure and organisational issues as well as the technical configuration of the environment with a concern for computer-supported human learning research and re-engineering (Schmidt and Cottier 2003). Secondly, what we have started to do more recently is to turn the spotlight on user behaviour in regards to the Symba environment whilst focalising more particularly on questions communicational in nature; our aim is to pinpoint those sequences of interaction which show that the students have internalised the inter-relation so predisposed to learning together. From our point of view as producers of a new learning technology, this latter must logically serve as torchbearer for the former. We shall now speak about two distinct approaches for analysing behaviour. Through the use of quantitative analysis grids which help in revealing and classifying interactions going on within chats, forums and e-mails, the first

322

approach allows us to show that ‘‘each group acts, interacts and organises itself differently [...]. From their organisation straight through to the means of communication they use, nothing is predictable [...]. The groups have a lot of difficulty getting organised [...]. The participants do not have required information on the activities of the other members of their group [...] an awareness function must be added’’ (Betbeder and Tchounikine 2004). The second approach essentially consists of semi-directive interviews with the learners and targets understanding the sociological and interactive context of the users and their practices. These interviews also show that the strong distinctions between learners (present on-campus and distant learners) and the dissonance they experience with respect to time frames and workspaces (domestic settings or at school) as well as the diversity in their career planning generate specific uses of the environment. We could mention a few of them, namely, the emergence of communication topics having nothing to do with the proposed system, the use of synchronous or asynchronous communication tools depending on the percentages in the mixture of students, the driving role of institutionally imposed due dates, the participants desire to take part in the experiment, which participants achieved maximal usage of tools and who accepted to test new ones. From the point of view of research on the uses of these such applications, both of these approaches have their limits: the first one is centred on the objectives of the system, essentially teleological in nature, and does not reveal ‘‘deviant’’ practices or the misuses of the system by the students. And on another note, the quantitative apprehension of the interactions, above and beyond the incontestable results it supply, does not allow for getting down to meanings the users attribute to their activities and to their interaction in the absence of communication. In casting an eye upon the technical object in an off-centred manner, revealing rather statuses, roles of actors and situations and aiming at rendering behaviours explicit, the second approach fails to determine the ‘‘socio-technical’’ configurations and the key moments of the activity with regards to learning together.

Fig. 4 Snapshot of a generated activity-level environment

323

Since this application involves some human and some technical material and wishes to create the conditions of a common activity between actors, the identification and description of collaboration’s key moments proves to be fundamental here.

To apprehend the object differently: situated action and dialogical communication Within the framework of such an environment (for collective work in a pedagogical setting), besides finding out how the tools are actually used and learning about new emerging practices, we aim to lead the students to developing stages of work that are collaborative because we presuppose that certain cognitive processes can operate through them. In this way, we think that the collective activity is apt to generate a process of ‘‘construction and maintenance of a shared vision of a problem’’ (Dillenbourg et al. 1996). This idea of shared or situated cognition (Suchman 1987) supposes that in interaction and through language, conceiving a problem becomes shared. What we mean to bring forth here is the dialogical character of such situations of communication, namely those described by Jacques (1979, 1990) and his notion of reference. Every dialogue implies a convergent evolution of beliefs, desires, intentions, knowledge, etc. or to put it shortly, inter-subjectivity. Identifying these dialogical phases in the interaction is not an end in itself. Above and beyond pointing them out, what is at stake in our work is to understand the role played by the variables of the socio-technical system at the time of these specific moments of interaction. Even though it would be inappropriate to trace any causality there may be here—our model of communication is a far cry from being linear—this socio-technical environment is an element that has to be challenged in order to envisage the complexity of the intricacies between the social, technical and cognitive aspects of e-learning. To express this differently, the experience we have accumulated on the activity in question to date allows us to state that such dialogical phases exist, even if we can only base this on empirical apprehension of the phenomena. We consider that these dialogical instances are, as well as being the product of interactions between individuals, more generally the fruit of a socio-technical context which of course includes the ‘‘interactants’’, but also logically accounts for social, professional, economical and institutional elements—that is our working hypothesis. First of all, working by this hypothesis means to isolate the exchanges concerning the content and the common definition of concepts (i.e. a discussion in a chat between students in which they try to agree on the notion of ergonomics means leading them to establish a common definition together). We call the initial interactions ‘‘meaning negotiation phases’’. This discursive phase is identified and seems to show a concomitant evolution of how things are conceived of. It would then be called upon to be corroborated with other ulterior phase of re-usage of the concepts defined (i.e. ‘‘ergonomics’’), phases we call ‘‘re-enforcement’’. This then allows us to determine

324

that, at least in appearance, this is indeed a dialogical process having favoured the emergence of a socio-rational plan of a common conception of things.

Methodological approach It would then be necessary to put the specific interactions into perspective with several variables. The next (big) task then is to identify the types of relations that are at work between (1) the emergence of these dialogical phases (negotiation and re-enforcement), (2) the ‘‘environment’’ including people (cultural, psychological and social aspects), (3) the tools used (for communication, work, agents, etc.) and (4) the ways of teaching plus organisational methods, all at that specifically special moment of the ‘‘meeting of the minds’’ that we wish to study. Several factors must then be considered: nature of the learners—their socio-professional characteristics, capital cultural, individual contexts; group configuration—the emergence of a leader, any possible mixture in the types of learners; the approach to working—organisation (or auto-organisation) of task handling, the divvying up of the work, etc.; the ways of communicating—through and without the technical environment; pedagogical issues—objectives adopted, the carrying out of student help, etc.; technical configurations—the technical resources used during phases of negotiation, agents, communication tools; the institutional contingencies—deadlines, course validation, etc. The central question to our work may be formulated as follows: within the framework of the Symba project, the collective distance activities grouping together the on-campus and distance students, therefore computer supported, allows collaborative activity phases to develop through interactions that are dialogical in nature. What are the configurations of the environment at that time and are there processes that favour these configurations? In order to respond to this question, several steps would seem to be necessary. We develop them here based on an initial empirical work with a goal of finding the analytical tools necessary. These empirical findings will allow for the comparison systematic of the Symba environment with other environments or work sessions. Pinpointing meaning negotiation phases This work is carried out in two steps starting with the identification of negotiation phase markers which is then followed by the identification of the object of the exchange. In a third step, the actual application of the markers is carried out on another corpus in order to retain the validity and transferability of the indicators. Finding meaning negotiation markers is made possible through lexical strings that are composed of relevant terms. If we take, for instance, the example we give hereafter, finding meaning negotiation markers means pointing out words such as definition, to define, concept, notion, etc. (the words in grey below), words that we consider to be traces or detectors of an

325

activity having a dialogical character. In order to identify the object of exchange, the referential elements of dialogue, two types of indicators are used: redundancy of one or several occurrences throughout the totality of the corpus studied and their proximity to the lexical string chosen beforehand. In the following example, ‘‘FOAD’’ or ‘‘Formation Ouverte et a` Distance’’, which means Open and Distance Courses in French, constitutes the object of discussion which the speakers attempt to define together. These terms are framed in the excerpt below. [Speaker 1] : Alright, shall we start off by agreeing on the definition of a foad? [Speaker 2] :

The autonomy of the learner and providing him with help are all notions that have to do with FOADs as well as that of the Engineering underlying it.

[Speaker 1] :

Just a minute, we have to define foad first

[Speaker 3] :

The notions of distance and openess come first and autonomy comes of these, doesn’t it?

[Speaker 1] :

Wait! Before attacking the concepts, we must come up with a clear definition of a FOAD.

[Speaker 2] :

For learners, autonomy is more so a necessary condition for having success with a FOAD application

[Speaker 4] :

Don’t you thing we should start off with some sort of an outline? for our work

[Speaker 2] :

“A Formation Ouverte et A Distance is an organised finished product recognised as such by the actors. It takes into account each person’s singularity in their individual and collective dimensions and is based on complimentary and mixed learning situations in terms of

[Speaker 1] :

to put it short, you mean the Chasseneuil definition????

[Speaker 1] :

why not use our own definition

[Speaker 2] :

in terms of time, place, pedagogical, human and technological mediations as well as resources”.

[Speaker 2] :

what would your definition be?

[Speaker 1] : A Formation Ouverte et A distance ( FOAD ) is an organised application based on the learning situations of a group of learners that uses human, pedagogical and human resources.?????????????

Next we validate the chosen markers in applying them to another dialogue. In our situated way of going about this, this dialogue is produced by another group of learners taking part in the same course at the same time; their collective

326

activity has the same end. The lexical analysis applied to this new corpus points out a significant usage of terms such as definition, concept and notion (112 uses of the term concept, 65 times for definition and 6 uses of notion). It also shows the importance of the terms isolated in our previous work: 70 uses of FOAD and formation ouverte et a` distance. Some of the phases of meaning negotiation could be identified. They include our negotiation and referential markers. Let us now look at one of them: [Speaker 1] [Speaker 2]

[Speaker 4] [Speaker 3]

[Speaker 4] [Speaker 2] [Speaker 2] [Speaker 3] [Speaker 1] [Speaker 4] [Speaker 3] [Speaker 6] [Speaker 1] [Speaker 1] [Speaker 6] [Speaker 2]

[Speaker 6]

What do you think of our definition of a foad ? I was just in the process of answering. I find that your informed input says just about everything! To me it seems to be a definition that is allencompassing and operational… It seems quite complete to me… We must try to obey the turn-taking procedure and not talk too much so that everyone gets a chance to participate We have a list of all the types of requirements The current definition allows for the inclusion of various styles of distance learning systems It is in fact very complete. We tried to come up with an all-purpose definition What do you think “Speaker 6”? Does everyone agree on this def.? Just a minute, we are waiting for a response from “Speaker 6”. I agree with Speaker 1’s file for the all-purpose definition definition unanimously adopted!! Have we got an exhaustive list of concepts according to our analyses? I think we have almost finished with the concepts Maybe we could be more specific: do we have a sufficient amount of concepts in order to define the application, the people and the situations? or does this question still seem premature at this time? But the conceptual relations between the concepts are not varied I don’t think, are they? It is not premature? Because if it is, we would not totally agree. There are quite a few things to discuss here, aren’t there?

We must however put these first results into perspective. F. Jacques points out that a dialogue can have an object that is formally absent from its syntax. In our example, the term FOAD could even not be present and constitute the object of the discussion. If we limit our analysis to what is said, we will miss out on part of the meaning. It is thus necessary to take into consideration other factors: ‘‘the history of the communication process, i.e. the list of all the phrases pronounced in the order of their emission, before the saying in question is said’’ (Jacques 1979, p. 144). So far our approach allows us to identify dialogical phases (like the exchange here about the FOAD notion) but it does not allow for a truly

327

exhaustive search of all dialogical phases in its present state. This means that we must take into account more relevant pragmatic markers in our future work as well as put the sayings studied with their interlocutionary context into perspective, namely by conducting interviews of the learners of the group being studied. This is the future direction of our research. Finally, as we pointed out before, it would be useful to attest to the nature of these dialogical phases by the identification of what we called phases of reuse or re-enforcement of the discussed concepts. This work is based on a lexical analysis of the exchanges, but this is just the start. The activity studied here is effectively only the first step of the students’ course to be carried out over one whole year. Finding re-enforcing phases will require a more long-term analysis of electronic exchanges between learners, well after the computer-supported activity of Symba has been carried out. The contextual approach Analysing dialogue as we have therefore allows us to isolate dialogical phases and the phases of meaning negotiation, however, imperfect it may be. This step in our work does not, however, constitute an end in itself; the analysis of the emergence of context is necessary for understanding the complexity of the situation. Our theoretical perspective does not really take the technical object here as its central concern. There is in fact a dynamic pull between the technical offer and the actual use of anything (Perriault 1989) and it is useful to consider the system as including technical objects, individuals and institutions without any of these factors prevailing over another. The socio-technical relation is complex and the logic of the elements interact and get intertwined without giving us the possibility to really clearly point out the degree of influence of one over the other. Applied within the framework of our research, this stance thus implies that we are not only trying to identify the technical characteristics that could have favoured the aforementioned dialogical situations (which is an techno-centred attitude) but to envisage the logically complex game at work in such a system. We must recall that the logic of these parts of the system are particularly linked to the nature of the learners, the configuration of the work groups, the ways of working, communicating, teaching as well as the institutional policies on teaching and setting up technical material. Our work on the usage of Symba allows us to situate the contours of our exploratory framework. We are currently characterising what best relates this project, which of course lends a hand to envisaging new work hypotheses. Our initial analysis5 made it possible to reveal that two types of learners differ quite clearly within the framework of this partially distance course: those present on-campus and those off-campus. This has a multitude of consequences in their relation to the environment, the structure of their common activities and how they perceive the prescribed scenarios or those that seem to be prescribed by designers. The on-campus learners constitute a population of students living on their own in student accommodation for the majority of them. The off-campus ones 5

Carried out through interviews and observations.

328

almost all have a professional activity, a family to feed and consequently daily lives that allow for little free time. These are thus two very dissociated universes when it comes to capital time available and the management of that time. Much time is allotted to the activity by the first group which ‘‘consumes’’ it within the university walls whilst the second group generally works at home, sometimes at work and always in short time slots. We noticed just how much the motivations and the types of projects of the two groups were very different. The on-site learners have a typical academic approach to the work in which the value of acquiring knowledge lies in its validation by the diploma itself whereas the off-campus learners care more about whether what they are learning will help them in their daily professional tasks. We must mention that the interviews and the observations that we conducted allow us also to affirm that, outside of all the prescribed and expected communication to take place between these two types of actors via the chats, forums and e-mails provided within the Symba environment, other means of interacting did emerge: face-to-face or other tools outside the Symba environment (personal mail, telephone, etc.). Finally what unites these two types of learners is the desire to conform to the instructions or make up others if they do not seem to be enough. The learners develop the desire to play along, test ways of doing things and experiment with new ways of playing the game. We are now able to hypothetically envisage the fact that a strong relation of dependence exists between the emergence of collective work within Symba in the framework of the course studied and (1) the weight of university deadlines, (2) the structuring of activity into ‘‘tasks’’, (3) software or tutor group assistance, (4) the emergence of student leader to organise tasks, (5) the capacity of attraction the tools could have on learners trying to use them in a maximal way, and lastly, (6) the synchronous or asynchronous nature of the proposed communication tools. This first approach also teaches us about context and the necessity to historically situate the dialogical events we will finally search for. In order to do this, and because the Symba environment is not the only communication space used in the game, interviewing and observing with the speakers during and after the activity proofs to be indispensable for the reconstitution of the chain of facts.

Conclusion Over and beyond the first empirical results, the different work conducted on the Symba environment today allows us to envisage a specific approach for treating truly dialogical situations within the framework of collective distance learning activities. A methodological approach is springing forth and we aim at using it to find dialogical phases and specify the contexts that yield the emergence of these phases. We must sharpen our analytical skills and better get to know the process that leads to the emergence of dialogical situations. We wish to test our method of approach to the situation on new corpuses and improve our way of apprehending this type of situation. This will happen notably through work on defining more precise referential markers, taking into account the dialogical situations in which we can lexically isolate the object of exchange.

329

In regards to those special situations in computer-mediated communication, we must question and then test some of the concepts and methods used in the fields of the pragmatic analysis of dialogue and the sociology of uses: lexical analysis, content analysis, interviews and observations. We should also programme for the transferability of the observation techniques elaborated with the framework of computer-supported human learning engineering and for following up on collective activities. This of course calls for interdisciplinary work between computer scientists and researchers in the humanities which we have already started as it seems to us that part of our approach can be automated and implemented with a view to developing awareness functions in the systems that could facilitate tutoring students. Furthermore, we intend to promote theories that are more useful for modelling student–tutor–system–teacher relations. In CSCL systems, the claim has often been that the sequential form of ‘‘dialogue’’ produced the relationship between interlocutors when in fact it is a pure product of it. This is why the notion of reference is central for coming to grips with the communication problems of CSCL systems as the referent is always outside of the communication system (cf. model above), and separate from the group of rules that control activity there.

References Bardram JE (1997) Plans as situated action: an activity theory approach to workflow systems. Paper presented at European conference on computer supported cooperative work, Lancaster, UK, pp 17–32 Betbeder M-L (2003) Symba : un environnement malle´able support d’activite´s collectives en contexte d’apprentissage. The`se de doctorat, Le Mans, Universite´ du Maine, France, 2003, 238 pp Betbeder M-L, Tchounikine P (2003) Symba: a framework to support collective activities in an educational context. In: International conference on computers in education (ICCE 2003), 2–5 de´cembre 2003, Hong Kong, China, pp 188–196 Betbeder M-L, Tchounikine P (2004) Mode´lisation et perception de l’activite´ dans l’environnement Symba . In: Actes de RFIA’04, 28–30 janvier 2004, Toulouse, France, pp 1217– 1225 Bourguin G, Derycke A (2001) Integrating the CSCL activities into virtual campuses: foundation of a new infrastructure for distributed collective activities. Paper presented at Euro-CSCL, Maastricht, The Netherlands, pp 123–130 Dillenbourg P, Baker M (1996) Negotiation spaces in human–computer collaborative learning. Paper presented at COOP’96, second international conference on design of cooperative systems, Juan-les-Pins, France, pp 187–206 Dillenbourg P, Baker MJ, Blaye A, O’Malley C (1996) The evolution of research on collaborative learning. In: Reimann DP, Spada H (eds) Learning in humans and machines: towards an interdisciplinary learning science. Pergamon, Oxford, pp 189–211 Engestro¨m Y (1987) Learning by expanding. An activity–theoretical approach to development research. Orienta-konsultit, Helsinki Jacques F (1979) Dialogiques: recherches logiques sur le dialogue. Presses Universitaires de France, Paris Jacques F (1985) L’Espace logique de l’interlocution. Presses Universitaires de France, Paris Jacques F (1990) De ‘‘On Denoting’’ de B. Russell a` ‘‘On Referring’’ de P.F. Strawson, l’avenir d’un paradigme, Herme`s VII: Bertrand Russell, de la logique a` la politique. Editions du CNRS, Paris, pp 91–100 Lewis R (2000) Human activity in learning societies, Invited paper presented at ICCE/ICCAI, Taipei, Taiwan, pp 36–45 Perriault J (1989) La logique de l’usage, Essai sur les machines a` communiquer, Flammarion

330 Roschelle J, Teasley SD (1995). Construction of shared knowledge in collaborative problem solving. In: O’Malley C, Shannon C, Weaver W (eds) Computer-supported collaborative learning. The mathematical theory of communication. University of Illinois Press/Springer, Urbana/Berlin Heidelberg New York Schmidt CT, Cottier P (2003) Relational issues in improving the management of collective activities in E-learning. In: Proceedings of E-Learn 2003 the 8th world conference on ELearning in corporate, government, healthcare and higher education, Association for the Advancement of Computing in Education, 7–11 November, Phoenix, Arizona, Etats-Unis, pp 1768–1771 Shannon C, Weaver W (1949) The mathematical theory of communication. University of Illinois Press, Urbana Suchman L (1987) Plans and situated actions: the problem of human-machine communication. Cambridge University Press, Cambridge