Distributed cognition in the context of virtual collaborative learning.

In this article, the authors address the distribution of cognition in small virtual group learning. In particular, they examine how the distribution of cognition is manifested in the context of student-student interaction and how it can be studied in a virtual learning environment. For this reason, they define a conceptual framework of collaborative learning situations for modelling and understanding how learning evolves and how knowledge is constructed among students that cooperate on a task. One important issue to consider is the types and levels of interaction and knowledge, which is manifested, in a collaborative learning situation. Their approach aims at identifying the various types of interaction produced and examining how an interaction type is related to the learning that results from it. This framework allows us to study how distribution of cognition is transformed and becomes common to all group members.

**********

In this article an experimental collaborative problem-solving situation that takes place in a virtual learning environment is described. In particular, an approach to analysing and modelling asynchronous collaborative interactions which are realised by several groups of students that cooperate from a distance to carry out a project, is presented.

The analysis of the data performed was based on an integrated approach to modelling educational interactions and, in particular, on an integration of several models and methods: the Negotiation Linguistic Exchange Model (Martin, 1992); a model of Discourse Contributions (Clark & Schaefer, 1989); the types of learning activities underlying a participant turn (Self, 1994); and the Rhetorical Structure Theory (Mann & Thompson, 1988). Experimentation with this approach resulted in interesting insights that concern both the social and the material distribution of cognition at all the three general phases that comprise the realisation of the project: group formation, project development, and project evaluation.

Regarding the social distribution of cognition, the authors examined how the distribution of cognition is manifested in the context of student-student interaction and how it can be studied in a virtual learning environment. This involves the definition of appropriate collaborative learning situations and the distinction of two levels of student interaction, the discourse and the action level. At the discourse level, the essential element is the interaction among peers (participants need to interact with each other to plan an activity, distribute tasks, explain, clarify, give information and opinions, elicit information, evaluate, and contribute to the resolution of problematic issues, etc.). At the action level, task objects (e.g., documents, graphics) are created and manipulated.

The approach focused more on the analysis of the discourse level by seeing discourse as a medium and means through which the distribution of cognition is effected. More specifically, the distribution of cognition in discourse is studied and analysed through a dialogue model of asynchronous discourse which is defined in terms of types and structure of student-student interaction.

Regardind the material distribution of cognition, first the types of tools used to realise a real problem-solving situation cooperatively in distance are examined. Then, the relationship between the nature of the task, the tool, and the tool user is considered. And finally, the cognitive factors that affect both the learning process and collaborative dialogue features (such as roles and group dynamics) that support and promote more effective learning, especially distance collaborative learning is addressed.

AN APPROACH TO MODELLING A REAL COOPERATIVE PROBLEM-SOLVING SITUATION

In this section, the technological support provided to the students to carry out the project successfully is presented, then a number of constraints and conditions, which had to be taken into account and, which emerge both from the tool used and the nature of the experience itself are pointed out, and finally the approach to modelling a real cooperative problem-solving situation, where 30 students participated divided into 8 working groups is explained.

Technological Support

As mentioned before, cooperation is mainly achieved through asynchronous interaction and occasionally through a synchronous one. The asynchronous communication was performed using the BSCW tool (http:// www.gmd.gd/bscw), whereby the synchronous interaction was mainly achieved through the ICQ (http://www.mirabilis.com) chat tool.

BSCW is a tool for cooperative work that offers a range of possibilities such as: group management, asynchronous discussions, document management, and an automated daily workspace activity report. It also provides a help that explains how to use the tool (Bentley, Appelt, Busbach, Hinrichs, Kerr, Sikkel, Trevor, & Woetzel, 1997; Bentley, Horstmann, & Trevor, 1997; Horstmann & Bentley, 1997). In addition, we also provided the students a brief help document with the most important functionalities of BSCW so that to facilitate a faster and easier interaction and involvement with the tool.

Constraints and Working Conditions

Tool: BSCW:

* The tool uses the same reference for the notions of communicative situations and folders. This results in mixing the notions of folder (where information is organised) and communicative situation (where communication and asynchronous discussions take place). This is a conceptual constraint that had to be taken into account when designing the methodological approach to collaborative learning.

The nature of the experience

* Time: about three months were available to carry out the whole experience.

* The educational interaction considered is task-oriented: That is, it is assumed a group of students involved in a cooperative activity will reach a shared objective.

* Tutor: the tutor's involvement was basically limited to control aspects related to the subject-matter contents and to a lesser extent methodological aspects concerning the group work. The tutor has not participated in the asynchronous discussions held by the students.

* Students lack the experience to deal with virtual group work: students had to combine learning about how to develop a project with how to resolve it working cooperatively. Due to the project size, they had to concentrate more on project issues rather than on methodological aspects of cooperative work.

* Differences in work pace: Studying from a distance implies that most students inevitably have different work paces. This sometimes created anxiety or disappointments and mainly affected the good functioning and the effective realisation of the asynchronous discussions.

Structure of Student-Student Interactions: Rhetorical Structure, Exchange Structure and Contributions to Discourse

In this section, it is shown how the approach is used to model the asynchronous cooperative interactions produced and thus provide a detailed account of how cognition is distributed. It is considered that the structure of a long interaction is constructed cooperatively by using the exchange as the basic unit for communicating knowledge. In that way, the effectiveness of the overall interaction can be determined through the examination of the degree of effectiveness of the exchanges which take place within the interaction.

Following, the Negotiation Exchange Model of Martin (1992), three general exchange structure categories are considered: (a) give-information exchange, (b) elicit-information exchange and (c) ascertain-information exchange. Assuming no interruptions, the number of moves that characterises these exchanges depends on the group size. For instance, for a group size of four students, the basic elicit- and ascertain-information exchanges may contain up to five and seven moves respectively (Tables 2 and 3).

Each of these categories describes the generic discourse goal an exchange is associated with. Thus, the goal of the actor who initiates the give-information exchange is to inform his/her partners about a certain situation with the purpose of changing the partners' mental states. Informing may include explaining, giving an opinion, describing, or reminding a situation in different ways. The actor's goal of the second exchange is to elicit the partners' state of mind (knowledge, beliefs, attitude, desires, abilities) of a situation, which the actor is not aware of or certain about. The actor's goal of the third exchange is to find out the state of mind (knowledge, beliefs, etc.) of the partners by normally assigning a problem or question to be resolved by the group. Typical examples of the three basic general exchange types (no interruptions are included), where groups of four students are collaborating, are presented in Tables 1, 2, and 3. The labels on the right-hand side are explained in the next section.

According to Martin, there is a move that constitutes the "obligatory move" of the exchange, since it either carries or indicates completion of the discourse goal for which the exchange is initiated. The obligatory move of each of the exchanges is: the first move of the give-information exchange, the second move of the elicit-information exchange and the third move of the ascertain-information exchange. The symbols (f) and (Ra) mean a follow-up move and a replay answer, which reformulates or slightly extends a previous answer, respectively.

According to Clark and Schaefer (1989), each move is seen as a contribution to discourse. This means that in a cooperative conversation, contributions are regarded as collective acts performed by the participants working together, resulting in units of conversation--typically turns (moves)--that aim to make a success of the discourse they compose.

Applying this theory to the types of exchanges considered previously, it was found that not all moves contribute in the same way toward the successful completion of the exchange. In particular, some moves do have a pure contributing function toward the realisation of the obligatory move of the exchange. This is the case of the first move of the elicit-information exchange, as well as of the first and the second moves of the ascertain-information exchange. In fact, without the presence of those moves, the obligatory move cannot be realised; thus, those moves really contribute toward the realisation of the obligatory move. Consequently, successful realisation of the obligatory move conveys evidence of (initial) success of the exchange.

In contrast, the other moves have a rather supporting function (provide evidence of support) toward the definite completion of the obligatory move and consequently of the exchange. This is the case of the follow-up moves of the three exchanges. Supporting moves are optional, so they may not be realised. In such a case, they convey an implicit support toward the obligatory move, that is, toward the definitive completion of the exchange.

LEARNING ACTIVITIES AND THE RHETORICAL EXCHANGE STRUCTURE IN COOPERATIVE EDUCATIONAL DIALOGUES

The notions of the Rhetorical Structure Theory (RST) of Mann and Thompson (1988), were the basis used to model the exchange and represent it's structure. In particular, three generic rhetorical relations: consent, elicitation, and ascertainment, which correspond to the previous three exchange categories respectively are provided. The name of each relation denotes the type (e.g., "ascertain-information") and the function of the exchange in the interaction (i.e., the particular discourse goal that is to be accomplished by the exchange). Moreover, the group partners have to be involved in a process of realising a number of learning activities which lead to the completion of the exchange goal. The consideration of the participants learning activities is based on the work of Self (1994) and a more recent work of Pilkington (1999).

The generic rhetorical relations provide a construct that contains a detailed specification of the multi-turn surface-level structure of the exchange, which includes a description of the learning activity functions that takes place in the three basic exchange types. In fact, the details of this specification are captured by specific interlocutory rhetorical relations (Tables 1, 2, and 3) that hold between the exchange moves and describe the move relationships, the activities realisation and goal completion.

Another important type of information captured by the above relations is the role of the participant move in the exchange. According to RST, two basic roles are distinguished: the Nuclear role and the Satellite role. Taking into account the definition and significance of the obligatory move in the exchange, it is considered that this move plays the role of the nucleus (N) in the exchange structure. In regards to the rest of the moves, depending on its position and its function in the exchange, a move may initially adopt a temporal satellite role and then change it to a local nucleus and vice-versa. This fact captures the dynamic change of the move function in the exchange and reflects both the satisfaction of the current intentions of a participant and his/her setting of expectations of how the other participants have to proceed cooperatively to achieve a common objective. This approach allows us to capture the way the distribution of cognition is manifested in the context of student-student interaction and how learning is evolved.

To simplify the presentation of the model of exchange structure, simple examples of the three exchange categories which do not include interruptions are discussed. Yet they are sufficient to show the workings of the exchange model and how this is used to model asynchronous collaborative interactions, and thus give an account for the social distribution of cognition. In these exchanges, the interlocutory rhetorical relations that hold between the moves are identified, as well as the role(s) that these moves play in the exchange. Each relation captures and controls the evolution of the learning activity performed by a participant move by setting the expectations of the type of learning activity, which has to be realised next by the other participants so that the current local goal set by the initial move can be accomplished. The exact specification of the learning activities is influenced by the contributing or supporting function that the move may have in the exchange.

More specifically, in the give-information exchange, the following particular types of support relations, which help achieve the learning goal associated with the exchange have been identified; no contribution moves apply to this exchange.

First, the main learning activity of the initial student- 1 move is to present the other group members a piece of information. Then, student- 1 expects the others' positive evidence of understanding of the knowledge conveyed to consider that the exchange goal has been successfully achieved. That is, the next learning activity (realised by student-2) is to give an assent to the student- 1 expectations of such a piece of evidence. This is done by uttering "OK." Thus, the relation that links the student- 1 informing move N and the student-2 follow-up move S is that of Assent, a particular type of the interpersonal Evidence relation. The interpersonal meaning of this relation clearly connects the student-2 state of mind with the propositional content of the student-1 previous move.

At the same time, the other students of the group may wish to reinforce the fact that the goal of the exchange has been completed successfully. This possible new activity indicates the multiple function of the initiating move in the exchange: the expectation of "consensus" by all group members. This is captured by reassigning the initiating move the nuclear role N of a consensus relation. Parentheses around the relation name are used to denote the opening of a cycle where the relation applies, whereby brackets indicate the completion of the relation cycle.

As soon as student-1 intentions are initially accomplished (the expected assent satellite is realised), student-2 sets his own expectations about what move may follow, that is, what learning activity is to take place. Under the current context (created by the semantics of the assent relation), the new expected activity is nothing but "reaching to the conclusion" of the exchange. This dynamic change of the function of the student-2 move is captured by promoting its satellite role (of the Assent relation) to a local nuclear role (N1) of a Conclusion relation (symbol a). This is explicitly expressed by the utterance "As regards me, I agree." The learning activity of reaching to a conclusion involves the expression of agreement of the rest of the participants as regards the accomplishment of the exchange goal. This occurs when student-3 utters "I see it too." In this context, the student-4 last move serves both as a satellite (S2) of the Conclusion relation and as a satellite (S) of the Consensus relation, since it expresses the mutual beliefs of all participants and it definitely concludes the exchange. The detailed description of the rhetorical exchange structure and the learning activities involved in the other two exchange types is done in a similar manner (Daradoumis, 1997).

Building a Larger Cooperative Educational Dialogue

The approach considers that exchanges are related to one another through high-level specific rhetorical relations to form a larger coherent dialogue (Daradoumis, 1995). Exchanges themselves are regarded as dialogue units that play a nucleus or satellite role according to the way they contribute to the completion of the overall dialogue goal for which they were generated.

Thus, depending on the overall dialogue goal, a nuclear exchange may be preceded and/or followed by a determined number of satellite exchanges that contribute or support the nuclear exchange and aim at the successful completion of the overall dialogue goal. The rhetorical dialogue structure is then a mirroring of the rhetorical exchange structure, at a higher level of abstraction.

Relations, at dialogue level, capture and control the evolution of the learning activity that is associated and performed by each exchange and in turn provide a way to model how cognition is distributed and propagated in the course of the group interaction. Thus, failure to complete a nuclear or satellite exchange indicates failure in the communication process and the need to analyse and revise the collaborative and cognitive processes that took place.

DISTRIBUTED COGNITION IN COOPERATIVE EDUCATIONAL DISCOURSE

This section aims to explain and illustrate the way discourse can be used as a medium and means through which the distribution of cognition is effected. As seen in the previous section, distribution of cognition is manifested when a group of students is engaged in a cooperative dialogue to solve a task, at two levels of description. At the exchange level, where exchanges constitute the basic unit of a cooperative dialogue, distribution is manifested through the realisation of the three exchange types and their respective moves. At this level both the quantity and the quality of each exchange type performed is examined. Both these aspects of an exchange are measured by the collaborative effort of the members involved to achieve the discourse goal for which the exchange was initiated. The term collaborative effort means both the number of contributing and supporting moves issued by a participant (which indicates an active or passive participation) and the type and effectiveness of these moves (which indicates the way a participant contributes toward the achievement of the shared discourse goal, as regards knowledge possession and transfer, reasoning capability and positive attitude). As explained in the previous section, several move types are distinguished, whereby their effectiveness is measured by the role (nucleus or satellite) they play in the exchange and the degree of success of that role.

According to the exchange model, completion of an exchange expresses the mutual beliefs of all participants about the accomplishment of its discourse goal. This in turn implies that distribution of the cognition that this exchanges carries, is effected among group members to a degree which is at least equal to the degree, of contribution and completion of each member moves.

To give a more concrete example of how cognition is distributed through operationalisation of the exchange, the give-information exchange type is examined. It's successful completion implies that the knowledge possessed by the initiator of the exchange is assumed to be transferred and distributed successfully, at informative level, to all other members of the group. To examine then, at a comprehension level, whether this piece of knowledge has been assimilated by all group members, an ascertain-information exchange should be issued next. That is, the type of distributed cognition effected, the way distribution is manifested, and the level cognition is distributed depends on the types of exchanges issued, their degree of individual completion, the way each exchange type has been completed (which includes not only the move types described in this article but also the types of interruptive moves occurred and the way they have been handled), and the successful combination of these exchange types into a larger di alogue.

In general, the three types of exchanges suggest a certain type of distribution, giving, eliciting, and ascertaining information represent standard discourse structures for handling information. These discursive structures enable the participants to take turns, share information, exchange views, monitor the work done, and plan ahead. Most importantly, they provide a means to represent and operationalise the cognitive product at an individual level, that is, the way the reasoning process is distributed over the participants as it is shared in a collaborative discourse.

Distributed cognition can also be modelled at a higher level, by combining and relating exchanges through rhetorical relations to form a larger coherent dialogue. Analysis at that level is also important, since it gives us an account of how cognition is manifested and evolved beyond the individual level, that is, at the group level, by examining several parameters such as the combination of the different exchange types, the purpose and the role that each exchange plays with respect to the related exchanges. The result of the analysis at this level can yield very useful conclusions on aspects such as group working and dynamics, which allows the tutor to obtain a global account of the progress of the group work and to identify possible conflicts, which then can be analysed deeper at the exchange level.

In fact, the next section collects, classifies and summarises preliminary insights from the experimental study and the analysis carried out so far in terms of the cognitive factors that affect a number of collaborative issues, such as the learning process, group dynamics, roles, and the collaborative tool used.

PRELIMINARY INSIGHTS FROM THE EXPERIMENTAL STUDY

The analysis has been mainly based on the collaborative interactions that took place at the BSCW virtual workspace (asynchronous communication/discussions). In addition, further triangulation with measures outside the dialogue are provided, by gathering data that reflect the cooperative work carried out, from a variety of sources such as: group interviews, questionnaires, student self-assessment reports, tutor reports, student reflection about cooperative work, and teaching materials about collaborative learning.

This type of analysis explores the relationship between collaborative dialogue features (such as roles and group dynamics), learning process and tools. The focus is more on the project development phase, in which the analysis of data showed that several cognitive factors influence both the accomplishment of the project and learning through this practice. In particular, the cognitive factors that affect the learning process of the student are:

* The type and purpose of the dialogue episodes that take place in the interaction.

* The number of exchanges that involve knowledge transfer (give-information exchanges), problem-solving activities (ascertain-information exchanges), questioning and elicitation of knowledge (elicit-information exchanges), and the number of exchanges that each participant initiates.

* The number of contributing and supporting moves made by a participant, which indicates whether the participant has been active or passive in the conversation.

* The type and the effectiveness of the contributing or supporting moves made, which indicates the degree of collaboration that a participant provides toward the achievement of the shared objectives.

* The type of activities in which the participants have been involved: these include planning activities, knowledge contribution, reflection, and revision of the methodology of both the problem-solving situation and the way collaboration is performed, evaluation of the progress made, expressing opinions, beliefs, reasoning, attitude, needs, desires, capabilities, rights, and consciousness.

In fact, there is a need to distinguish between situations that involve discussions to carry out a learning goal, debates on a particular subject, or negotiations and collaborative argumentation, especially situations that involve knowledge building.

There are also cognitive factors that influence the group dynamics and roles, such as:

* mutual confidence;

* acquisition of the compromise and responsibility to pursue and accomplish the shared objectives. (the compromise acquired in-group reinforces the individual compromise);

* support and motivation of each other;

* work pace: a too rapid or too slow pace of a group member influences negatively the group work (very different work paces provoke anxiety and insecurity to the group members);

* well-defined member attitudes: a member's attitude may influence positively or negatively the work of the whole group, the other members' attitudes, efficiency and final performance;

* group organisation and task planning: a good organisation of the group itself as well as a good planning and partitioning of the tasks that have to be carried out is a necessary condition for the group success (this also includes an appropriate initial and continuing re-organisation of the working space); and

* group interaction: a symmetrical and equal-based interaction can provide a better guarantee for the achievement of the planned learning objectives. This interaction should be also well balanced throughout the project development both as regards the participation (active member, initiator) and the type of significant contributions made. The amount and the quality of discussion about a particular subject/objective also influences the objective completion.

Finally, the utility of the main tool (BSCW) to asynchronous collaborative learning, has played an influential role as a tool that facilitated the realisation of asynchronous discussions. In addition, it allowed the accessing, sharing, and organising of information, to establish face-to-face or virtual meetings, to leave notes or comments on posted documents, to get to know about the daily activities taking place (through the automated daily reports). In general, it is believed that the influence of the main tool used for collaborative work/learning depends on the possibilities, user-friendliness, and the strong or weak points of the tool. These issues may:

* transmit confidence or insecurity to the group;

* influence the group predisposition to work with the tool and to use it to interact with each other; and

* affect the successful completion of the shared learning objectives.

CONCLUSION

This article has shown a complex work that was carried out to develop a conceptual model of cooperative problem-solving in a virtual learning environment. The model is based both on an integrated approach to modelling educational dialogues and on initial pedagogic and learning principles. These issues guided the design of virtual communicative situations that allowed to perform asynchronous interactions aiming at accomplishing specific goals of a real cooperative problem-solving situation. Experiments with this problem have shown very interesting results regarding the way cognition distributed among people and artefacts. The conclusion is that collaborative study of a subject matter is a very important issue in distance education since it could prove very useful for the students in the sense that it can improve and motivate learning, and foment their social interaction.

Table 1

A Description of the Rhetorical Structure of the Basic
"Give-Information" Exchange by Means of Specialised Types of Support
Relations


Student 1:      You know, there are types of problems
                in which the output data are
                determined through the calculation
                of a function. So, we need to
                learn how to specify functions
                and use them at the right moment.

Student 2 (f):  OK... As regards me, I agree.

Student 3 (f):  I see it too.
Student 4 (f):  That's fine with me too.



Student 1:





                (N) (Assent) (Consensus)
Student 2 (f):  (S)[Assent] [right arrow]
                (N1) (Conclusion)
Student 3 (f):  (S1) (Conclusion]
Student 4 (f):  (S2) [Conclusion]
                (S) [Consensus]
Table 2

A Description of the Rhetorical Structure of the Basic
"Elicit-Information" Exchange by Means of Specialised Types of
Contribution and Support Relations


Student-1:      Do you understand how a              (S) (Inducement)
                function is defined?
Student-2:      I have no problem.                   (N) [Inducement]
                I think I understand it quite well.  (N) (Content)
                                                     (Consensus)
Student-3 (f):  I have no problem with this          (S) [Content]
                definition either.                   [right arrow]
                                                     (N1) (Conclusion)
Student-4 (f):  Neither do I have.                   (S1) [Conclusion]
Student-1 (f):  That's great, folks!                 (S2) [Conclusion]
                                                     (S) [Conclusion]
Table 3

A Description of the Rhetorical Structure of the Basic
"Ascertain-Information" Exchange by Means of Specialised Types of
Contribution and Support Relations


Student-1:       Let's first revise by doing
                 a simple specification
                 exercise, before we start with
                 the real problem. Specify the
                 following problem completely:

                 Given an integer number,
                 determine whether it is even
                 or not.

Student-3:       Well, here is my solution to
                 the problem:

                 INPUT: Let X "belong to" (INT)
                 OUTPUT: Write PAR "belonging
                 to" (BOOL) so that
                        PAR=(0=MOD(X,2))
                 What do you think about it?

Student-4:       As far as l know, it seems OK
                 to me.
                 I think your answer is right.



Student-2 (Ra):  In fact,

                 you need not put any
                 parentheses
                 around the types INT and BOOL.

Student-3 (f):   You're right.
                 I'll have it in mind.


Student-1 (f):   OK, I think this is the
                 definite solution
                 solution to the problem.

Student-2 (f):

                 Fine, I think we can move on
                 now.


Student-1:



                 (S) (Enablement-doing)



                 (S1) (Solutionhood)

Student-3:
                 (N1)[Solutionhood]
                 (Means)




                 (S) (Evaluation)

Student-4:
                 (N) [Enablement-doing]
                 (N) [Evaluation]
                 (N) (Adjust) (Positive-
                 implication) (Consensus)

Student-2 (Ra):  (S) [Adjust] [right arrow]
                 (N2)

                 (Confirmation)
                 (S1) [Means]

Student-3 (f):   (S) [Positive implication]
                 (S2) [Confirmation] [right
                 arrow] (N3) (Conclusion)

Student-1 (f):

                 (S3) [Conclusion]

Student-2 (f):   {... After some time with no
                 further contribution...}

                 (S) [Consensus]

References

Bentley, A., Appelt, W., Busbach, U., Hinrichs, E., Kerr, D., Sikkel, K., Trevor, J., & Woetzel, G. (1997). Basic support for cooperative work on the WWW. International Journal of Human-Computer Studies: Special Issue on Innovative Applications of the WWW. 46(6), 826-846.

Bentley, R., Horstmann, T., & Trevor, J. (1997). The World Wide Web as enabling technology for CSCW: The case of BSCW. Computer-Supported Cooperative Work: Special Issue on CSCW and the Web, 6(2/3), 111-134. Dordrecht, the Netherlands: Kluwer Academic Press.

Clark, H., & Schaefer, E. (1989). Contributing to discourse. Cognitive Science, 13, 259-294.

Daradoumis, T. (1995). Using rhetorical relations in building a coherent conversational teaching session. In R.J. Beun, M.J. Baker, & M. Reiner (Eds.), Dialogue and instruction, pp. 56-71. Heidelberg-Berlin: Springer-Verlag.

Daradoumis, T. (1997). An integrated approach to modelling tutorial dialogues under an overall rhetorical structure. Doctoral Dissertation, Politechnical University of Catalonia, Barcelona, Spain.

Horstmann, T., & Bentley, R. (1997). Distributed authoring on the Web with the BSCW shared workspace system. ACM Standards View, 5(1), 9-16.

Mann, W., & Thompson, S. (1988). Rhetorical structure theory: Towards a functional theory of text organization. TEXT, 8(3) 243-281.

Martin, J.R. (1992). English text: Systems and structure. Amsterdam: Benjamin Press.

Pilkington, R.M (1999). Analysing educational discourse: The DISCOUNT scheme (Tech. Rep. No. 99/2, Jan. 1999). The University of Leeds, UK.

Self, J.A. (1994). Dormobile: A vehicle for metacognition. In T.W. Chan & J.A. Self (Eds.), Emerging computer technologies in education, pp. 120. Charlottesville, VA: Association for the Advancement of Computing in Education.