A case study analysis of a constructionist knowledge building community with activity theory

Chee S. Anga*, Panayiotis Zaphirisb and Stephanie Wilsonc

aSchool of Engineering and Digital Arts, University of Kent, UK; bDepartment of Multimedia and Graphic Arts of the CyprusUniversity of Technology, Cyprus; cCentre for Human Computer Interaction Design, City University, London, UK

(Received 25 March 2009; final version received 28 April 2010)

This article investigates how activity theory can help research a constructionist community. We present aconstructionist activity model called CONstructionism Through ACtivity Theory (CONTACT) model and explainhow it can be used to analyse the constructionist activity in knowledge building communities. We then illustrate themodel through its application to analysing the Wiki-supported community associated with a computer game. Ouranalysis focuses mainly on two perspectives: individual and collective actions, as well as individual and collectivemediations. Experiences and challenges from the analysis are reported to demonstrate how CONTACT is helpful inanalysing such communities.

Keywords: activity theory; constructionism; game community; knowledge building; Wiki

1. Introduction

Paper has an even more potent role than as astorehouse of knowledge. It achieves remarkablepower when it is a blank sheet, inviting studentcreativity . . . It is taking decades for educators torecognise that the most potent use of videotapeshappens when teachers o!er blank ones to students.

(Shneiderman 2002)

For decades, scholars have been relying oncognitivist or constructivist approaches for research-ing technology-aided learning and teaching. Myriadmodels of analysis, methods and frameworks havebeen developed to help design and evaluate educa-tional technologies (Lebow 1993, Oliver 2000).Although they are helpful, the advancement oflearning theories from behaviourism to social con-structionism; and technologies from standalone tonetworked computers has called for the developmentof more comprehensive theoretical or methodologicalframeworks. This has shifted the focus of instruc-tional design to constructional design (Resnick et al.1998). Constructional design advocates that learnersshould be treated more like experts constructing newknowledge than novices receiving existing knowledge.Knowledge has become a more personal propertythan a general entity; it is also a more concreteartefact than an abstract concept. Further, knowl-edge is considered to be stored not only within aperson’s head but also situated around a socialcultural context (Lave and Wenger 1991).

Some researchers have started shifting the focus tothe social cultural perspective (Barab et al. 2002) oflearning. This is an important paradigm shift asconstructionists base their arguments on individualdevelopment which is enhanced by shared constructiveactivity in the social cultural setting (Shaw 1995).

It is also crucial to realise that the advent of newmedia such as computer games not only changes howpeople perceive and understand new forms of multi-media messages but also how people use media toexpress themselves. Educators should come to under-stand that presenting knowledge in a media-rich, social-oriented technology is not the most desirable way ofteaching and learning. What is more important is thatstudents should use these media for expressing theircreativity (Shneiderman 2002).

We believe that constructionism as defined by Papert(Papert and Harel 1991) provides powerful theoreticaltools to explain learning in a social technologicalcontext. However, constructionism remains largely atheoretical model, and without operationalisation of itstheoretical constructs, it may be hard to use it directly toanalyse such learning communities. Therefore, in thisarticle we propose using activity theory (Engestrom2001) to operationalise constructionism (see Sections 2.1and 2.3 for detailed descriptions of constructionism andactivity theory).

Through this, we aim to develop a unified frame-work that incorporates models of knowledge building,perspectives and artefacts, and is grounded in empiri-cal analysis of collaborative interactions. Such a

*Corresponding author. Email: csa8@kent.ac.uk

Behaviour & Information TechnologyVol. 30, No. 5, September–October 2011, 537–554

ISSN 0144-929X print/ISSN 1362-3001 online! 2011 Taylor & FrancisDOI: 10.1080/0144929X.2010.490921http://www.informaworld.com

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

framework can guide the design of computer-basedartefacts as support for constructionist learningsystems with unified conceptualisations and theoreticalconstructs.

Specifically, the aims of this study are:

(1) to develop a model for analysing construction-ist learning based on Engestrom’s (2001)activity theory triangle and Papert’s (Papertand Harel 1991) constructionism

(2) to apply to model to analyse a Wiki-supportedknowledge-building game community in orderto give insights on how it evolves and develops.

This article is structured as follows: Section 2reviews the theoretical background which includesconstructionist theory of learning, computer-sup-ported collaborative learning (CSCL) and activitytheory. Section 3 presents the CONstructionismThrough ACtivity Theory (CONTACT) model whichis an adaptation of Engestrom’s (2001) triangle activitysystem by incorporating the concept of construction-ism. Section 4 is the major part of this article and itpresents the example application of the model. In thissection, Wiki technologies are introduced and theirpotential to support knowledge building communities(KBCs) is illustrated. Then, the data collection anddata analysis method is explained. The findings of thestudy as well the discussion of the results and theimplications are also presented in detail. Section 5discusses the future direction in which we are headingand Section 6 concludes the article.

2. Theoretical background

2.1. Constructivism and constructionism

Although some educators are still expounding thebehaviourist method of knowledge transfer fromexperts to novices, psychologists such as Piaget (Piaget1929), Vygotsky (1930) and others proclaim thatlearning is less about filling learners’ head withabstracted facts, than it is about constructing realityinternally through collaboration.

Constructivism as initiated by Piaget (1929) has hada significant epistemological and pedagogical impact.According to this theory, learners do not receiveknowledge from the external world. Instead they interactphysically and socially with the environment andconstantly construct and update their knowledge intern-ally. In other words, constructivists put learners into theexperts’ shoes, contending that everyone should con-struct their own versions of reality by extracting abstractand formal knowledge from the context.

Our work is based on Papert’s view of constructi-vism, which he terms constructionism. Papert’s (Papert

and Harel 1991) constructionism goes beyond what isconstructed inside learners’ heads. Being Piaget’sstudent, Papert’s constructionism is largely influencedby constructivist theory. But, although construction-ism embraces and builds upon Piaget’s constructivism,Papert eventually came to see some drawbacks inPiaget’s stage of cognitive development theory (Piagetand Inhelder 1969). He disagrees with Piaget’s stagetheory in which adults abandon concrete thinking infavour of formal thinking. To Papert, concretethinking and formal thinking are two di!erentcognitive styles which persist in adulthood. Thisepistemology pluralism states that people do not giveup concrete thinking in order to move forward to ahigher level of thinking. Both are equally important inhuman learning.

In terms of pedagogy, Papert claims that even foradults, learning remains essentially bound to context,in which knowledge is shaped by the use of externalsupports. Papert’s approach helps us understand howlearning is actualised when individual learners con-struct their own favourite artefacts or object-to-think-with (Papert 1980). In his own words, Papert definesconstructionism by stating that:

We understand ‘constructioNism’ as including, butgoing beyond, what Piaget would call ‘constructiVism.’The word with the V expresses the theory thatknowledge is built by the learner, not supplied by theteacher. The word with the N expresses the further ideathat this happens especially felicitously when thelearner is engaged in the construction of somethingexternal or at least shareable . . . a sandcastle, amachine, a computer program, a book. This leads us toa model using a cycle of internalisation of what isoutside, then externalisation of what is inside and soon.

(Papert and Harel 1991)

We attempt to visualise this concept with Figure 1in order to illustrate the fundamental idea of construc-tionist learning. Learning is thus understood as acyclical process of constructions in which the learnersexternalise their initial state of knowledge throughbuilding an object which helps them update their oldknowledge as well as interpret and construct newknowledge. As knowledge is being constructed ortransformed through the manipulation of objects, this

Figure 1. The fundamental concept of constructionism.

538 C.S. Ang et al.

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

new knowledge is internalised by the learner and thecycle continues.

Papert’s theory bears a lot of similarities toVygotsky’s (1930) social cultural theory which stressessocio-cultural aspects in learning, claiming that learn-ing is mediated by cultural tools in which socio-cultural knowledge amasses. Learners do not interactwith the environment directly; instead the interaction issocially mediated with artefacts, be it signs (language)or tools. Cognitive development occurs at two levels:knowledge is constructed socially (inter-psychologicallevel) before it is internalised into each individual(intra-psychological level).

Constructionism, however, goes beyond social cul-tural theorists’ claims that learning is mediated by thecultural semiotic system. Constructionism suggests thatlearning involves the e!ort to create external symbols tomove formal symbols constructed internally and locatethem in the environment. In other words, Papert’sconstructionists emphasise that learning is more e!ectivewhen learners are engaged in designing or constructingsomething tangible. For example, Papert himself hasspent more than two decades developing the LOGOprogramming language that enables students to graspcomplicated mathematical concepts by creating softwareartefacts (Papert 1980).

2.2. Computer-supported collaborative learning(CSCL)

The advancement of learning theories towards socialparadigms has resulted in the emergence of a new area ofstudy in e-learning, known as CSCL that emphasises thecollaborative aspects of learning as well as the socialinteraction among learners in knowledge construction. Itis a shift in the use of technologies from treatingcomputers as a personal tutor to a communicationmedium that promotes authentic group learning.

Although it is clear that the design, analysis,implementation and evaluation of CSCL is largelyinformed by social cultural theories (e.g. socialconstructivism), Koschmann (1996) notes that one ofthe problems with CSCL is the lack of connectionbetween constructivist views of learning and the studyof learning as social practice. He attempts to resolvethis question by concentrating on providing anadequate description of learning as a process ofmeaning-making rather than a focus on learningoutcomes (Koschmann 2001).

Learning is thus treated as the process of meaning-making through social interaction between learnersand their peers. Vygotsky’s (1930) conception that allhigher mental processes take place between peoplebefore they are internalised is congruent with con-structivist views of learning. The social aspect of

learning is understood as a process of social negotia-tion and joint knowledge construction. In the contextof collaborative learning, negotiation is viewed as aprocess by which learners attempt to attain agreementon aspects of the learning task and on certain aspectsof the interaction itself (Dillenbourg et al. 1996).

This is in line with Du!y’s claim that learning isinherently a social, dialogical process (Du!y et al.1993). That is, given a problem or task, peoplenaturally seek out opinions and ideas from others.Technologies can support this conversational processby connecting learners across the world. When learnersbecome part of KBCs, they learn that there aremultiple ways of viewing the world and multiplesolutions to life’s problems. As such, CSCL is alwaysmentioned in tandem with KBC which is a group oflearners committed to advancing the group’s knowl-edge of some shared problems through collaboration(Hewitt 2002).

Therefore, CSCL is thought to be more ambitiousthan previous approaches to e-learning, making itmore di"cult to evaluate the e!ectiveness and e"-ciency of such learning activities. Significant researchwork needs to be carried out to provide systematicevaluation, theoretical framework development andthe new CSCL system implementation.

For many years, theories of collaborative learningtended to focus on how individuals function in agroup. In CSCL, however, the focus has shifted so thatthe group itself has become the unit of analysis(Dillenbourg et al. 1996). Hence, some of the focusesof research would be looking at CSCL include:

. collaborative knowledge building

. group and personal perspectives

. mediation by artefacts

. micro-analysis of conversation provides a rich,multi-dimensional starting point for conceptua-lising and studying CSCL (Stahl 2002).

The notion of collaborative knowledge buildingdefines a useful paradigm for conceptualising learningas social practice in which shared knowledge isconstructed can be analysed as the result of inter-related group and personal perspectives. Thus, what isneeded is a theoretical framework incorporatingmodels of constructionist knowledge building toanalyse such learning communities.

2.3. Activity theory

Although Papert’s theory provides a solid frameworkfor understanding children’s and even adults’ ways oflearning by designing, it does not o!er a systematicframework for analysing the construction activities

Behaviour & Information Technology 539

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

within a learning community. The most importantanalysis includes the collective learning within acommunity as well as the development of an indivi-dual. We are also interested in finding out how toolssuch as computers help learners construct artefacts andknowledge. Hence, we would like to draw from theVygotskian naturalist approach which emphasiseshuman activity systems. Vygotsky (1930) formulateda theoretical concept which is very di!erent from theprevailing understanding of psychology dominated bybehaviourism at that time. This new orientation was amodel of tool-mediation and object-orientedness. Heproposes the classic triangle model to demonstrate theidea of mediation.

In Figure 2, the subject is the individual engaged inthe mediated action, the mediating artefact or toolcould include physical artefacts and/or prior knowl-edge of the subject. The object is the objective of theactivity. Figure 2 shows explicitly that the relationshipbetween the subject and the object is no longerstraightforward as in Figure 1. Instead it is mediatedby external and internal tools. For example, whenbuilding a website, the subject is working towards anobjective (e.g. to add a table to the webpage) using notonly the computer (external tools) but also her internalunderstanding of how websites and computers work(internal tools).

Leont’ev (1978) extends this notion of activity todi!erentiate between an individual action and acollective activity by proposing a hierarchy of activity(refer to Table 1). Collective activity is connected tothe object of the whole community, of which eachindividual subject is often not consciously aware. Anindividual action is connected to a conscious goal.Below the collective activity and individual actionthere is the level of operations that are dependent onthe conditions in which the action is performed. Thus,an activity system can be analysed at three levels: the

activity level which is oriented towards the object/objective and carried out by the whole community; theaction level which is directed at the individual goal, aswell as the operation level which is elicited byconditions and is performed unconsciously.

This hierarchy is crucial in explaining the learningprocess in an activity system. We would like toillustrate an example of this hierarchy in learning aforeign language (refer again to Table 1). The overallobjective is to be able to engage in a meaningfulconversation. In the beginning, the learner has to workon the grammar and the choice of words at a consciouslevel. When the learner has reached a higher profi-ciency level, these actions are transformed intooperations. The learner no longer needs to selectappropriate words and check grammar rules deliber-ately as these have been learned thoroughly and arenow operating unconsciously. The consciousness of thelearner is now focused on expressing himself properlydepending on the objective of the conversation.Grammatical rules become invisible to the learnerand he is only selecting appropriate goals to beachieved. Therefore, it can be inferred that activitytheory treats learning as the shift from the higher levelto the lower level in the hierarchy (e.g. from actions tooperations). Nevertheless, upon encountering newconditions, a learned operation might be shifted backto the action level to be reflected on at the consciouslevel (this will be further explained in Section 3).

Drawing on work by Vygotsky (1930) and Leont’ev(1978), Engestrom (2001) views all human activities ascontextualised within an interdependent activity sys-tem. Engestrom adds collective mediation to Vygots-ky’s tool mediation and presents the triangle model ofactivity system (Figure 3).

In Figure 3, the subject is the individual or a groupwho is selected as the point of view of the analysis. Theobject refers to the raw material or the problem spaceat which the activity is directed and which istransformed into outcomes with the help of externaland internal tools. Tools are the concepts, physicaltools, artefacts or resources that mediate a subject’sinteractions with an object (or objective). The com-munity refers to those with whom the subject sharesthe same general object (or objective). The division oflabour (DOL) is the classification of tasks among themembers of the community while the rules are the

Table 1. Hierarchy of activity.

Unit of analysis Stimulus Subject Language learning example

Activity Object Community Engage in a meaningful conversationAction Goal Individual conscious Sentence constructionOperation Conditions Individual unconscious Word selections, grammar rules

Figure 2. Vygotsky’s mediation.

540 C.S. Ang et al.

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

regulations, norms and conventions within the activitysystem.

2.3.1. Shortcomings of Engestrom triangle activitysystem diagram

Activity theory is a broad concept of human activitywhich can be customised to meet the need of eachresearcher in di!erent studies. We find activity theoryuseful to study KBCs from the perspective ofconstructionism. Even though it is possible to oper-ationalise activity theory principles within the con-structionist context, there still exists a need todemonstrate the mapping between activity theory andconstructionism. Due to its open-endedness, we find itdi"cult to use the traditional model of activityproposed by Engestrom for our purpose to study aconstructionist learning community.

First and most apparently, the hierarchy ofactivity, which implies the learning process, is notincorporated into the traditional model. Therefore, weattempt to solve it by explaining the relationship insubject-tool-object and subject-community-object withdi!erent levels of activity expanded from Leont’ev’smodel (Table 1), including individual and collectiveactions and operations.

By adopting this concept, we need to present thedynamics of each element in the activity system. As theactions shift to operations, the element in the activitysystem is changed too. The static representation ofactivity theory represents only a snapshot of aparticular time, thus making it hard to analyse theactivity across time. Although the concept of transfor-mation is a significant part of activity theory, it doesnot provide a standard way to operationalise thistransformation of the activity: especially the transfor-mation of each component and the relationship.

Third, whilst attempting to produce an activitysystem for the learning community, it is important thatthe process of modelling an activity system requiresbasic understanding or prior knowledge about the

situation being examined. The activity triangle modeldoes not provide this kind of insight as it does notcapture historical data prior to the current activitysystem. Fourth, the heart of the constructionistlearning concept, internalisation and externalisation,are not visualised. The construction of knowledgewithin an individual and within the community is notexplicated explicitly.

3. The CONTACT model

In this section, we present an adaptation ofEngestrom’s triangle activity system that can enableus to analyse constructionist learning. AlthoughEngestrom (2001) proposes expansive learning (thatlearning is the development through resolving contra-dictions for collective development of activity) as thelearning theory for activity theory, this proposition isnot directly reflected in his triangle model. We try toarticulate, within a modified triangle model, thelearning process through the hierarchy of activity inorder to analyse constructionist learning. Figure 4shows a graphical representation of the activity systemfrom the perspective of constructionist learning. Wecall it the CONTACT model. We attempt to incorpo-rate the idea of constructionist learning into activitytheory, in order to produce a unified model to explainconstructionist CSCL.

The most important adaptation is that weexplicate the externalisation process through media-tion. As the process of externalisation throughtangible objects is central to constructionist learning,there is a need to explicate this process within thediagram. This is marked by the shaded area and thenamed relationship of subject-tool-object and sub-ject-community-object. We also add an extra arrowthat links the outcome to the tool in order to showthe internalisation process. Then, we delimit themeaning of object and refer to it as the tangibleobject in order to be in line with the theory ofconstructionism.

Like the original triangle model, CONTACT ismade up of seven components. The subject in this caserepresents the learner, while the tool is usuallycomputational tools which include hardware andsoftware. Like Engestrom’s model, it also refers toinitial knowledge of the learner. The communityconsists of learners of di!erent backgrounds workingtogether to construct the object. The object is thetangible artefact which is being worked on by thelearners. It will then be transformed into the outcomewhich in this context is the knowledge and thecomplete sharable object. Last but not least, theinteraction between the community and the learner ismediated by rules and the DOL.

Figure 3. The triangle activity system diagram.

Behaviour & Information Technology 541

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

Although an object in a general activity systemcould be conceptual or abstract, constructionistactivities emphasise tangible objects which are createdby the learners to express their internal ideas. Mediatedby tools, usually computational ones, learners are ableto project their initial understanding of a concept andtransform it into knowledge which is situated in bothexternal objects and the learner’s mind. The concept ofobject should not be confused with the tool. Objectsare something incomplete that are currently beingconstructed by the learner. In a constructionistcontext, they are tangible artefacts. Tools are some-thing which are used by the learner to construct theobject and they can be physical (such as pencils) orconceptual (such as grammar rules). It should also benoted that a completed object (the outcome) caneventually become a tool in the next generation of anactivity system. This process is known as internalisa-tion. Note that unlike Figure 1, where internalisation isdirected to the learner, this process is directed to thetool in CONTACT, as activity theory treats tools asextensions to the subject.

We also attempt to visualise the concept ofexternalisation through the hierarchy of activity bydefining the relationship between subject, tool, com-munity and object (the shaded area in Figure 4). Basedon Leont’ev, the subject operates the tool uncon-sciously to act on the object at a conscious level. Inaddition to this, we propose that actions and opera-tions also take place at the collective level. The subjectcollectively operates with the community to collectivelyact on the object.

One example of such collective actions andcollective operations in the context of constructionismis agreement and negotiation: an agreement on thedeveloping artefact takes place through the process of

negotiation. Thus, the subject agrees with the commu-nity to negotiate on the object. We also suggest thatindividual operations and agreements happen uncon-sciously. The di!erence between the two concepts isthat individual operations are taken individuallywhereas agreements are taken collectively. In otherwords, an operation is conditioned towards individualunconscious, while an agreement is conditionedtowards what Carl Jung (as cited in Boeree 1997)refers to as collective unconscious. Table 2 presents ourproposed hierarchy of collective actions and collectiveoperations.

Mediated by the tool and the community, thelearner externalises her initial stage of knowledgethrough object construction. The individual externali-sation (mediated by the tool) can be broken down intoactions and operations. Actions are directed towards apersonal goal and are carried out with carefuldeliberation. For example, in order to write a bookwith a word processing program, the author (thesubject) needs to construct sentences carefully in orderto express herself correctly. On the other hand,operations are autonomous responses to a conditionand are carried out unconsciously without delibera-tion. For example, for a regular computer user, movingthe mouse cursor around and clicking the cross buttonto close a software program are operations. Some-times, an action might become an operation if it ispractised many times. We call this a shift of operationsand actions and we demonstrate this through theexample of a book writing activity (Figure 5).

In a book writing activity, the author (an expertword processor user) will operate (e.g. typing) the wordprocessor at the unconscious level and consciously acton the book (to select appropriate words, constructmeaningful sentences and paragraphs) she is writing.

Figure 4. The CONTACT model.

542 C.S. Ang et al.

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

At a certain point, the author encounters a newcondition with the word processor which she is notfamiliar with: say to insert a table into the book. Underthis new condition, a breakdown is said to havehappened. The conscious e!ort of the author is nolonger placed on the book itself but instead is nowplaced on the word processor (e.g. to achieve theaction: insert tables, the author performs the opera-tion: read help files). Once the author has thoroughlylearned about the table insert, she can again act on thebook consciously and development is said to havehappened.

In a similar way, the collective externalisation(mediated by the community, rules and DOL) can bebroken down into negotiations and agreements.Negotiations are brought about by the conflict in acommunity and are carried out consciously withcareful deliberation. For example, in order to write abook collaboratively, every author needs to negotiateand achieve an agreement about the format of eachchapter. For that reason, we say that agreements aredriven towards a set of established rules among acommunity. Once an agreement is achieved, eachmember in the community no longer needs todeliberately negotiate about it. Thus it can be saidthat final agreements are reached unconsciously with-out any deliberation. For example, once the format ofeach chapter is decided, each author will refer to the‘formatting rules’ without having to negotiate again.

Figure 6 shows the shift of agreement (collectiveoperation) and negotiation (collective action). In acollaborative book writing activity, the editor willagree with other authors and negotiate the contents ofthe book (e.g. for them to collaboratively achieve theaction: write the book, they will collaborativelyperform the operation: agree topics). At a certainpoint, the author encounters a new condition whichhas yet to be agreed by the community of authors: sayan author raises an issue of text formatting. Abreakdown is said to have happened as the consciouse!ort of the authors is now placed on the formattingrules, not the book itself (e.g. now for them tocollaboratively achieve the action: decide the textformat, they will have to collaboratively perform theoperation: discuss with the community). Once theauthors have agreed the text format, the rules areupdated and they can again act on the bookconsciously and the development is said to havehappened (operation: agree with the topic and textformat, action: negotiate the contents and collectivelywrite the book).

It is also important to note that although we dividethe triangle into two levels, i.e. individual andcollective externalisation, these two levels are in factinter-related. The interaction between two learnersmight result in the development of each individualtool. For example, two book authors might discuss theuse of the word processor and through this eventuallydevelop their individual skill on using the wordprocessor.

4. An application of CONTACT to a Wiki community

In this section, we demonstrate how the CONTACTmodel can be applied to analyse a Wiki-supportedknowledge building game community. Before this, it isimportant to acquaint the readers with Wikitechnologies.

Table 2. A hierarchy of collective actions and operations.

Unit of analysis Stimulus Subject

Individual action Goal Individual consciousCollective action Collective consciousIndividual operation Conditions Individual unconsciousCollective operation Collective unconscious

Figure 5. The transformation of individual action-operation.

Figure 6. The transformation of collective action-operation.

Behaviour & Information Technology 543

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

4.1. Wiki technologies

The ‘Wiki’, named from the Hawaiian word ‘quick’, isa new technology that allows everyone to be theauthor. It is a freely expandable collection ofhypertexts which can be easily edited by any userwith knowledge of a very simple mark-up language. Itdoes not require any specialised tools; all you need is aform-capable web browser client. This simple editingmethod gives the freedom to everyone reading the pageto amend or correct it. Wikis explicitly supportcollaboration as they decentralise the e!ort of creatinga website from the hands of the few and distribute it toa huge community of internet users. In a Wikienvironment, users are not only editing, they are alsoencouraged to create their own content and their ownpages. A link to existing pages can be made easily, anda new page can be created by making a new link. Thus,apart from contents, the users also co-design thestructure of a Wiki site.

The goal of Wiki sites is to become a sharedrepository of knowledge, with the knowledge basegrowing over time. Unlike chatrooms, Wiki content isexpected to have some degree of seriousness andpermanence. In a Wiki, it is the users who create thecontent in collaboration and over time. Like Weblogs,Wikis have been around for some time and are popularamong the technology community. However, Weblogscan be highly personal while Wikis are intenselycollaborative.

Recently there has been an increasing interest inusing Wikis for learning (Jones 2003, Wang et al.2005). Although any knowledge building applicationthat demands the absolute and immutable integrity ofthe content is not really suitable for a Wiki, it is usefulin situations where communities of people are devel-oping shared ideas, values or resources. A teacher ofbiology, for example, could start a Wiki site by postingsome material creating a tentative structure for thesubject, and uploading some media files. Whenstudents visit, they could expand the contents bymodifying them, by posting more material, or bymaking links to new pages, thus enriching the learningresources. Through shared construction cycles, thestudents feel closer to the learning system as theycontribute to its development instead of being pas-sively presented with the information.

A project has already been undertaken to re-build aweb-based learning site for spectroscopy using Wikitechnologies with the goal of making the resource morerelevant and content-rich, so as to attract authors fromdi!erent backgrounds to provide content in multiplelanguages to support international users (Mader 2004).Perhaps the most famous educational Wiki in existenceis Wikipedia (Wikipedia 2005). It is a free-content

encyclopaedia operated by the Wikimedia foundationthat anyone can edit. It is the result of a vastcollaboration and currently contains over a millionentries in multiple languages.

The Wikimedia foundation has also created theWikibook project which is a collaborative bookwriting implementation aimed at building communitiesthat contribute to the development of hypertext booksavailable free of charge to the public. The Wikipediaand Wikibook projects show that the Wiki conceptmight work, and that groups of people can collabora-tively create shared knowledge artefacts.

4.2. Method

The development of the CONTACT model was basedon the grounded theory method, in which thetheoretical model emerges through the iterative cycleof qualitative data analysis.

We started our analysis with the original ATtriangle model as proposed by Engestrom, andthroughout the research, we constantly revised andmodified it as new concepts and relations emergedfrom the data analysis (Glaser and Strauss 1967).

For each of the Wiki pages, we started by browsingand reading the information to get a general overview/impression of what kinds of information was created.After that, the contents were read carefully to extractimportant information that would reveal construc-tionist learning. The goal was to find out all possibleprocesses observed in the Wiki community. Emphasiswas put on issues that occurred frequently or that weredeemed of fundamental importance.

We categorised these processes according to thetraditional activity theory triangle model. Processesnot visualised in the original triangle were added to thediagram as we revised and developed CONTACT. Forinstance, the process of collective externalisation wasobserved frequently in the community in variouscontexts, in which a number of users constantlymodified a piece of text until an agreement hadarrived. This important process was then added tothe diagram. Please refer to Table 2 for newexternalisation processed identified through theanalysis.

As we expanded and modified the CONTACTmodel, we found that the model was increasingly ableto explain more activities/process in the Wiki commu-nity. We went through a series of iterative cycles ofdata analysis for this study and stopped only whentheoretical saturation has been reached, e.g. when themodel can explain all the processes we observed in thecommunity. The validation of the final model wasachieved through additional cycles of analysis whichdemonstrated that the examination of new data

544 C.S. Ang et al.

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

revealed no new information regarding the theoreticalconstructs and their relations. Therefore, our proposedmodel is both data driven in its construction andvalidation and can be inductively generalised to thepopulation represented by the selected sample (Glaserand Strauss 1967, Strauss and Corbin 1990).

Then, by using the CONTACT model, we examinehow the community of a Wikibook evolves and whatare the characteristics that lead to the growth of such aKBC. More specifically, in order to demonstrate theusefulness of the CONTACT model presented in theprevious sections, we apply it to describe actual datacollected from a Wikibook of a computer gamecommunity.

4.2.1. The Wikibook

We examined the ‘Grand Theft Auto: San Andreas’game Wikibook that provides game-related informa-tion to the player. ‘Grand Theft Auto: San Andreas’Wikibook (Figure 7) was selected for this studybecause of our familiarity with this computer gameand a preliminary scrutiny on the site revealed that thedevelopment of this community is rather active, thus itprovided su"cient data for analysis.

Each page in the Wikibook has its own historypage that documents its evolution by recording all thechanges/edits made by every participant. It containsthe following information: who made the change, whatchanges were made, when they were made and anoption to let the participant explain why a change wasmade. Figure 8 shows the history page for the main

page of this Wikibook. It displays a list of all theversions of the page from its creation to its currentversion. To identify the di!erences between twoversions of the page, the user must select and comparethem from the history list (Figure 9).

4.2.2. Data collection method

We collected data at from 11 pages of the Wikibook.These pages were the work of 65 participants (17registered, 48 unregistered) who contributed a total of421 edits. The collected data spanned across 226 days.

We also collected the discussion among the partici-pants on the discussion page (if any). Participants canstart a new discussion for each page they are working onby clicking the discussion tab (Figure 8). Apart fromthese, we also examined the title and description of eachedit (Figure 8). Some participants described the changes/edits they made, and why a change was made; suchinformation gave some insights into the goal of theiractions, thus helping us to carry out the categorisation.Meta-book pages which provide information regardingthe Wikibook project were examined as well. Empiricaldata presented in this study are also comprised of fieldnotes kept during the study. These field notes consistedof personal reflections on the method of analysing thedata with activity theory.

4.2.3. Analytical method

For each of the pages, we recorded and examined everysingle change/edit as recorded from its history page

Figure 7. The main page of the ‘Grand Theft Auto: San Andreas’ Wikibook (Grand Theft Auto: San Andreas 2005).

Behaviour & Information Technology 545

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

and then categorised the change. For example: was it achange of game contents or was it a deletion or acorrection of grammar? The categorisation for eachchange of the page was recorded and the same processwas repeated for other pages until no new categorycould be found. Using card sorting technique, wegroup these categories together until high levelcategories emerged.

In order to validate the category, we ran a focusgroup with three PhD students who have experiencesin similar qualitative research. We aimed to reach apoint of convergence among the participants throughthe focus group in order to come up with a consistentcategory set. This exercise was carried out in a focusgroup room which was equipped with an interactivewhiteboard. We started by explaining the aim of theresearch and a brief of the study. Then, the categoryscheme was described, supported by examples. Eachparticipant was required to apply the scheme indivi-dually to check if it was able to explain the actions in anew Wiki page. After that, each individual result wasdiscussed in a group setting until an agreement hadbeen reached. In principle, all participants agreed withthe scheme, however they felt that some definitions ofprocess and element in the model were ambiguous.Therefore, definitions were further explained andexamples were given to clarify these issues. All findingsand results presented in Section 4.3 went through thisanalytical process to ensure reliability.

Guided by CONTACT, we begin our analysis withthe most basic aspect of the constructionist model byexamining the relationship between the subjects and

the object. Then, we analyse Vygotsky’s mediationmodel of activity system consisting of individualactions and tools. The analysis is then extended tothe whole community of this system to includeemerging rules as well as the DOL that mediate thecommunity. The focus is mainly on the constructionistconcept of externalising the internal meanings onto asharable artefact through mediation. More specifically,we look into:

. Subject and object: What are the constructionistactions that act on the object and transform theobject into outcome?

. Action and operation: How do actions shift tooperations and vice versa? How do tools mediateindividual actions and operations? What is thenature of the mediating tools? How do theysupport knowledge building?

. Negotiation and agreement: What is the negotia-tion that transforms the object to outcome? Howdo negotiations shift to agreements and viceversa?

. Rules and DOL: What is the nature of implicitand explicit rules that mediate collective actions(negotiation)? How is DOL manifested in thecommunity? How do rules and DOL supportknowledge building?

Note that the analysis is not only limited to theabove aspects. There are various ways of looking at thedata using CONTACT, but for the purpose of clarityand due to the limitation of space, we chose to focus on

Figure 8. The history page.

546 C.S. Ang et al.

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

four aspects which we trust are the main strength of themodel.

4.3. Findings

4.3.1. Individual actions: subject and object

An individual action on the object is the focal point ofa KBC. It is the foundation that all other collectiveactions build upon since a single contribution from anindividual subject will invite more actions and inter-actions among members in the activity system.

Therefore we first study individual actions from thehistory entries of each page. Our initial analysis iscentred on the nature of actions by excluding theinteraction among the participants. As this stage, westudy an individual engaging in goal-oriented actions toexpand the Wiki page. By reflecting on the goal of eachaction, we identified six categories as shown in Table 3.

Figure 10 shows the number of occurrences of eachaction category.

Please note that these actions might becomeautonomous operations through practice. This will bediscussed in more detail in Section 4.3.2.

The first and of course the most important categoryis the book content which is the targeted knowledge ofthe community. This category involves adding, editingand deleting content. Apart from game information inthe form of texts and pictures, this also consists ofmeta-book content, the information about writing thebook such as FAQ (Frequently Asked Question) andproject meta pages. It can be said that the goal of this

category of actions contributes directly to the outcomeof this activity system.

Like its antecedent Wikipedia.org (Emigh andHerring 2005), the game Wikibook aims for a certainlevel of formality and standardisation of language use,such as barring informal abbreviations and slang.Thus, the second category (writing style) of actions isdirected towards providing a homogenous style ofwriting and avoiding ambiguous meanings. An exam-ple of such actions is presented below:

64.175.238.38 (09:51, 8 May, page_4)

Changed from

Also starting in the small towns like Angel Pine willmake it alot easier and faster to beat, since the peoplespawn about 10 secs away from the hospital.

To

Starting in the small towns like Angel Pine will makethis mission a lot easier, since the patients spawn about10 seconds away from the hospital.

The third category (English language) involveslanguage-related actions such as correcting spellings,grammatical errors and punctuations. The goal of thiscategory of actions is to produce error free and correctsentences.

The fourth category is structure and can begenerally divided into two sub-categories: internalstructure and external structure. Internal structurerefers to the modification of organisation and layout ofthe content (including texts, images, etc.) in a page. It

Figure 9. The version compare function.

Behaviour & Information Technology 547

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

includes moving text to di!erent locations, groupingtext in a new section, etc. External structure refers tothe creation and editing of links among pages. It isobserved that the book starts with no clear externalstructure; as information grows substantially, morepages are created and linked to each other in order toprovide an accessible structure to the reader. From ourobservations, a huge amount of e!ort is put intomodifying the external structure such as creating

anchors, changing page names for linking, indexing,creating navigational links, etc.

The fifth category (format) is concerned with thetypography and the format of the book including thefont type, the font size, lists, tables, etc. The goal is tomake the presentation of information more attractiveand clearer. The sixth category (Wiki mark-up) is theuse of the Wiki mark-up syntax. Since Wiki provideslimited WYSIWYG features, the participant has torely on the mark-up syntax. While English languageand writing style are important for constructingcontent, the mark-up syntax is important for structur-ing and formatting the information.

4.3.2. Individual externalisation: action and operation

In this sub-section, we attempt to observe if some ofthe actions described in the previous section mightbecome operations which, as explained earlier, takeplace unconsciously.

It is important to point out that on encounteringchanged conditions, we may have to reflect on theoperation consciously again, and thus shift uncon-scious operations back into conscious actions. Refer-ring to the action categorisation (Table 4), we canspeculate that due to their nature (i.e. their proficiencyis dependent on practice) English language, writingstyle and Wiki mark-up might become operations as

Table 3. Individual action categorisation.

Action categories Goals Examples of actions

Book content Directly related to the collective object, which is tobuild the game knowledge repository

Adding, editing or deleting the following:. Textual game information. Pictorial game information. Meta-book contents

Writing style To present the information more clearly RewordingRephrasingFormality of writing

English language To present the information correctly Correcting the following:. Spelling. Grammar. Punctuation

Structure To put the contents in a proper hierarchy and thusmore easily accessible

Sectioning textsCreating new pagesIndexing and table of contentsAdding, editing or deleting:. Links among pages. External links. Navigational links

Format To make the contents tidier and more readable Adding, editing or deleting:. Table. Font. Heading. List

Wiki mark-up To structure and format the contents correctly Correcting Syntax

Figure 10. The number of occurrences of each actioncategory.

548 C.S. Ang et al.

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

they mediate other actions such as book contents,structure and format. A possible instance of thehierarchy of activity for an individual could be thefollowing:

Activity: to build a game guideActions: game information, structure, formatOperation: mark up syntax, English language

The following shows an attempt of a participant,NSRegentPark, to learn the Wiki mark-up languagesyntax for creating links:

NSRegentPark (13:04, 23 Dec, page_7): attempt to adda link, but the link contains syntax errors ([‘‘‘BigSmoke:’’’])NSRegentPark (13:04, 23 Dec, page_7): try to fix thelink ([[‘‘‘Big Smoke:’’’]])NSRegentPark (13:04, 23 Dec, page_7): try to fix thelink ([[‘‘‘#Big SmokejBig Smoke:’’’]])NSRegentPark (13:08, 23 Dec, page_7): link fixed([[#Big SmokejBig Smoke]])

It is observed that this participant was trying tolearn the mark-up syntax to add a link through a seriesof individual actions. It seems that the author tried toproject his current internal understanding of Wikimark-up syntax. The externalised object (the particularWiki page) became a tool that let the participant seeand update his/her understanding of the mark-upsyntax. For this type of case, those who are fluent inthe syntax will perform these edits through uncon-scious operations.

Since in many cases tools are internalised knowl-edge (thus operate at the unconscious level), we

identify at least two possible mediation tools in thisactivity: language and Wikis. In addition, the subjectof the activity system is the participant of the study andthe object is the Wiki page under construction.

A Wiki site as a web technology can be regardedas a physical tool that helps the participant inexternalising constructions. It is found that someparticipants tend to make a couple of changesconsecutively. One possible inference is that theyattempt to externalise their internal understanding soas to have a closer look at it. The change that isincorporated in the Wiki becomes the symbolic toolto mediate their next actions. In such cases, a Wikisite can also be considered a psychological tool, as itconsists of various signs such as English languageand images. It can be said that Wikis help theparticipant externalise their internal tools (e.g.internal understanding of grammar) and situatethem in the Wiki page as external tools (e.g. words,sentences on the page).

For new users who are not familiar with Wiki, theWiki itself might be initially the object of the activitymediated by simple keyboard typing operations. Afterbeing used for some time, the Wiki will become thetool to mediate the ‘real’ objective of this activitywhich is to create a Wikibook of game knowledge.Further, the tool will be updated when the userencounters a di!erent situation, say to add a table. Ifthe users have not yet learned the mark-up syntax forconstructing tables, they have to reflect on the toolconsciously again. Thus tools are dynamic and underconstant change.

Table 5. Negotiation in the community.

Negotiation Goal Examples

Game content To resolve the di!erent understanding/mastery of the game

Correct or clarify game information

English language To resolve the di!erent understanding/mastery on the language

Correct grammatical errors and typos in anotherparticipant’s writing

Structure To resolve the di!erent view on whata clear structure is like

Change the structure within a page or among pages

Format To resolve the di!erent preferenceson the format

Modify the format of the text

Table 4. Division of labour.

Typology of DOL Description Related actions

Authors Contribute new information Book infoReviewers Check and correct information Book infoProof readers Check and correct grammar and typos, writing style, etc English, writing styleEditors Format the type face, etc., structure the pages and links Formatting, structure, markupVandalism monitor Control vandalism –

Behaviour & Information Technology 549

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

4.3.3. Collective externalisation: negotiation andagreement

A community is not just made up of a sum of separateand unconnected individual actions. Every individualparticipant is related and interacting with every otherin a community to transform the object into theoutcome which is a Wikibook of game knowledge. Wethus extend the activity system to examine the socialdimension to investigate the emerging rules and DOLthat mediate the community. Collective actions takeplace when more than one user is trying to modify thesame piece of information be it a sentence, a paragraphor the structure of a page.

We have previously described the collective actionsas negotiation and collective operations as agreement.We also pointed out that agreements are usuallyunconsciously made, while negotiations require con-scious e!ort.

It is worth mentioning that the term ‘negotiation’ isnot used in a conventional sense in this context. Due tothe very nature of Wiki-technologies, which allow onlyasynchronous interaction, negotiation is thus referredto as the process of at least two participants engagingin a series of at least two related actions on the sameobject. For example:

Participant_A (time_1, page_n): modify abcd to ababParticipant_B (time_2, page_n): modify abab to abef(negotiation occurs)

At this point, we study the interaction of partici-pants engaged in object-oriented negotiation to expandthe game knowledge repository. By reflecting on thegoal of each negotiation, we classify them into fourcategories as shown in Table 5.

The first category is the negotiation on the gamecontents which are most directly related to theoutcome. Each participant who plays the computergame comes to the Wiki site with their own under-standing of the game (about how to win the game).Therefore, negotiation at this category is the mostimportant and it is focused around constructinginformation which is socially agreed by the wholecommunity.

The second category is the negotiation of theEnglish language such as grammar. Despite our beliefthat language is a tool, and thus should be mediated atthe unconscious (agreement) level, at the collectivelevel the language mediation does occur at theconscious (negotiation) level:

Dizzle (07:52, 29 May, page_6): typos/grammar, ‘duel-wield’ is changed to ‘deul-weld’. . . (other actions)69.196.128.202 (18:05, 20 Jun, page_6): typos/gram-mar, ‘duel-weld’ is changed to ‘duel-wielded’

The third category involves the internal andexternal structure as discussed previously. One of themost common examples in this category is thenegotiation on the name of a link. It is found thatexternal structuring is negotiated by a fixed group oftwo participants. Finally, the fourth category is thetypology and formatting of the texts such as the fonttypes:

Master Thief Garrett (04:17, 9 Jun, page_1): add titleof the page with di!erent fonts (Beckett, Diploma)Master Thief Garrett (11:20, 9 Jun, page_1): changefont face for the title Old English Text MT12.220.161.65 (12:34, 16 Jun, page_1): change font sizefrom 12 to 10Aya (17:28, 17 Jun, page_1): add font template

Like the relationship between operations andactions, negotiations can be transformed into agree-ments. For example, two di!erent participants mighthave di!erent opinions on the format of the Wikipage and keep editing each other’s format until bothof them are satisfied at certain stage. Once this stagehas been achieved, new rules will arise and anegotiation is said to have been transformed intoan agreement. However, when new conflicts arise,the community has to negotiate it consciously. Oncethe conflict is transformed into rules, agreements willbe achieved. A possible instance of the hierarchy ofactivity for a community could be the following:

Activity: to build a game guideNegotiation: English language, game contents, struc-ture, formatAgreement: mark up syntax, writing style

Not all actions need to be negotiated as someactions come to a conclusive agreement. For example,mark-up syntax does not involve negotiation as it ismore objective; if one makes a mistake, the mistakecan be corrected without conflict. In this case study,there is no disagreement on general writing style either;therefore for the specific Wikibook we are studying it issafe to assume that most participants have agreed touse a formal writing style.

4.3.4. Rules and division of labours

Like individual actions, collective actions are alsomediated actions. Collective actions such as negotia-tions result in transformation of collective mediationsuch as rules and DOL. When a new negotiationbecomes an agreement, rules or DOL will be updated.In addition, negotiations might also result in revisingthe tool. For example, negotiations on grammar willeventually lead to an agreement that updates theconceptual tool about English in each individual.Indeed, rules and DOL are also some kinds of

550 C.S. Ang et al.

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

conceptual tools. But instead of just meditatingindividual actions, these tools also regulate collectiveactions.

Like any other community, the community inheritsrules from the social cultural background of eachparticipant. Some participants are already part ofother Wiki communities before they join this gameguidebook Wikibook. However, since every commu-nity is to a certain extent unique, existing rules will beamended and new rules will be introduced throughnegotiations.

This study identified some cases where the ruleswere made explicit. These rules are written in aspecific Wiki page called meta-page and are sup-posed to be followed and controlled more strictly.These rules are about formatting which includes thetemplate for the format of pages and the linkingname convention. While some rules are explicitlywritten in a page and are followed and controlled,some rules are implicit, like the use of non-abusivelanguage and writing style. These rules are agreedwithout being stated explicitly and are looselycontrolled. Even though this game, Grand TheftAuto: San Andreas, itself contains explicit violenceand inappropriate language, it is observed that theWikibook community generally has agreed to avoidthe use of abusive language in writing the game’sWikibook:

206.149.4.17 (18:25, 22 Jan, page_7): add abusivelanguage. . . (other actions)69.242.43.174 (14:57, 10 Apr, page_7): the abusive infois removed

Rules make sure that an agreement can be achievedwithout having to consciously negotiate the actions.The project standard meta-page, once being under-stood and agreed by the community, becomes a rule tomediate the actions and negotiation on the object pagein progress (the object).

The DOL guides the collective actions by settingthe roles and responsibilities of the communitymembers. In fact, DOL is a subset of rules that controlthe division of tasks among the members. In this study,the DOL is not clearly marked in the Wikibook asmost participants play many di!erent roles. However,by examining the nature of actions we are able toderive five categories of DOL:

Authors are the game players who contribute newinformation about the game. Reviewers are usuallyalso authors and they check the validity of theinformation. As they read, they find errors and correctthem. Proofreaders are those who check the correct useof grammar, and writing styles. They might besomeone who visits the site to get information about

the game and makes corrections on any errors he orshe encounters.

Editors keep the page tidy and make it look nicerby formatting the font types, styles, etc. They also helpto maintain the clarity of the page structure: e.g. thehierarchy of links. In our study only two participantswere actively maintaining the structure. They were alsothe two most active authors. It was also observed thatvandalism results in a new role, vandalism monitoring.For example, there were some participants like,Geocachernemesis, whose only contribution was tofix vandalism:

69.114.183.180 (20:19, 14 Jun, page_7): add nonsensetextsGeocachernemesis (2:01, 14 Jun, page_7): revert to aprevious version

Although a technology like this invites ‘vandalism’(this includes the massive deletion of the text, addingirrelevant contents, using abusive language, and anyother actions that are directed against the objective),which was identified in a number of occasions in ourstudy, it is basically controllable. The Wiki itself hasthe revert function that can reverse a page to its earlierversion. In addition, we also discovered that there is aself-regulating mechanism within the community thatprevents the website from being vandalised.

4.4. Discussion

Let us revisit our main aims of this study. First, wehave developed a model, known as CONTACT, foranalysing constructionist learning based onEngestrom’s (2001) activity theory triangle andPapert’s (Papert and Harel 1991) constructionism.We have presented the findings of an activityanalysis of a Wiki community. Based on CONTACTmodel which draws largely from Leont’ev’s hierarchyof activity and Engestrom’s triangle model of activitysystem, we successfully cast some light on theconstructionist activity in a Wiki-based community.Our findings demonstrate both individual actionsand collective actions (negotiation) which constituteactivity oriented towards the objective of building agame knowledge repository. We also demonstratedhow learning might occur individually through theaction-operation transformation, and collectivelythrough the negotiation-agreement transformation.

Apart from these, CONTACT also helps analyse thetools, capture the rules and the DOL which mediatethese actions. These must be further explained in orderto di!erentiate individual mediation and collectivemediation. Individual mediation places its emphasis on‘how a user uses the tool to write the game guide,without taking into account how other users act in the

Behaviour & Information Technology 551

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

community’. In other words, it is about the a!ordance ofthe tool to support what an individual can do.

Collective mediation is about the community,which consists of two major components: rules andDOL. Rules define what can be done and cannot bedone in a community. This should not be confusedwith the a!ordance of the tool. The tool might a!ordcertain actions such as writing in abusive language, butthe rules might want to ban this action. DOL is self-explanatory: how the work load is divided amongmany users in a community.

Based on our observation, individual actions helpsharpen the mediation tool, while negotiations bringabout new rules or refine existing rules that mediate thecollective action. In short, CONTACT is useful toanalyse the community in the following ways:

. It helps understand the individual mediationprocess: subject-tool-object

. It clearly presents the communal mediationprocess: subject-community-object

. It reveals the emerging rules and DOL in thecommunity

CONTACT appears to be a promising frameworkas it gives an analytical lens on analysing andinterpreting the data. It provides di!erent perspectivesof analysis, as it casts di!erent light on the data asresearchers can examine it from many perspectives byfocusing on di!erent sub-triangles of the activitysystem diagram. It also helps us examine the learningprocess: how learning occurs individually and collec-tively through the externalisation and transformationof hierarchy of activity from action to operation.Furthermore, both individual and collective aspectsare given equal importance. CONTACT informs thedevelopment of the whole community as well as theindividual development. It explains how individualdevelopment contributes to the community growth andvice versa.

Our second aim was to use CONTACT to getinsights on how this kind of communities evolve anddevelop. In a Wiki space, knowledge is sociallyconstructed; it is created individually with tools,negotiated and agreed within a community based onemerging rules and DOL. It starts as a single unit ofinformation (a page in the case of the Wikibook) andgrows organically and evolves into a complex and well-structured set of knowledge. From our findings, weconclude what contributes to the development of thecommunity is the fact that:

. the users share some historical backgrounds:they already share some of the tools/rules beforejoining the community, they also share an

interest on the same topic (a specific game in thiscase)

. users share the same object (goal) (in this casebeing to build a game guide book)

. a user’s individual action: this goal-orientedindividual action triggers negotiations that leadsto the growth of the space

. the community’s agreement on the object: notonly the community shares the same object, thecommunity must be able to negotiate and agreeon the object

. they share tools that support these actions andnegotiations

. they share emerging rules that coordinate theactivity

. they follow a DOL that divides theresponsibilities

Apparently, the evolution of a KBC needs morethan a group of devoted users who share the sameobject. It also involves negotiation and agreementamong the users on the object. Although every usertends to act towards their own goal, it takes thecompromise of the entire community to agree on theobject.

4.4.1. Practical uses of CONTACT

CONTACT incorporates Leont’ev’s conception of thehierarchy of activity within Engestrom’s trianglemodel. Apart from this, we have also proposed acollective dimension of the hierarchy, namely negotia-tion and agreement that explicate the externalisationprocess at the collective level. These add a newperspective to qualitative data analysis which webelieve is applicable for CSCL research. For instance,one of the focuses on CSCL is to examine learners’performance both as a group and as an individual.With CONTACT, we are able to visualise individualand collective externalisation in a community and howthese might eventually lead to development. This isparticularly interesting in the case of collectiveexternalisation as our analysis reveals that this kindof externalisation might lead to individual develop-ment or collective development. One example found inour study was that the negotiation among participantscould not only result in the development of the mutualunderstanding of formatting rules but also could causeeach individual participant to update his or herpersonal knowledge about the game.

While looking at a knowledge building or a CSCLcommunity, it is crucial to also look at the mediationtools, which in the case of CONTACT, are dividedinto individual tools and collective tools (i.e. rules andDOLs). We trust that this distinction is important,

552 C.S. Ang et al.

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

while trying to analyse from both individual and groupperspectives, as it helps us identify the development.For instance, when two learners are interacting andacting on a specific learning task, we would like toknow if this interaction will eventually update indivi-dual mediation tools (e.g. learning to use the interfaceof the CSCL system or learning certain concepts of thesubject matter) or collective mediation tools (e.g.agreeing upon the appropriate protocol of behavingor agreeing upon the role of each learner in thecommunity).

This points to the internalisation arrow added tothe original triangle model. In order to fully under-stand the social dynamic of a KBC, it is noted that atany particular instance, the outcome is channelledback to the activity system, modifying each element ofthe triangle, thus updating the activity as a whole.

To demonstrate the practical usefulness of CON-TACT, we present two possible cases of CSCL inwhich this model can be useful:

. Design: CONTACT can provide a holistic viewof CSCL design. For example, when designing aCSCL system, it should be noticed that technol-ogies (e.g. the computers and web systems) areonly one part of the design process. Focus shouldalso be placed upon the formation of commu-nities through supporting social dynamics. Pos-sible contradictions (both individual andcollective) should be included into parts of thedesign. Besides, while designing CSCL systems,mediation tools should not mean to be justtreated as individual but also collective as well.

. Evaluation: as mentioned before, in order toevaluate the learning process of the learners,both individual and social perspectives can betaken into consideration through the individual-collective distinction made by the CONTACTmodel on the hierarchy of activity which is beingvisualised explicitly in the triangle.

As such, we believe that the intricate relationshipbetween individuals and groups can be capturedthrough the use of CONTACT.

5. Conclusion and further research

In this study, we investigated how activity theory canhelp research a constructionist community. Althoughwe were able to analyse and interpret a KBC in light ofactivity theory, our analysis uncovered some weak-nesses of the triangle activity diagram. We showed thatactivity theory is useful because instead of focusingonly on the interaction between the user and thetechnology or the interaction among users, it includes

both dimensions of a community: the individual andthe collective aspects. However, the traditional modelof activity system does not fulfil our requirement inanalysing the dynamics of knowledge construction inthe activity, as well as the interaction of twointerrelated activity systems. Our aim is to operatio-nalise activity theory to study/research social construc-tionist learning activities in order to do that weproposed and presented in our CONTACT model.

Further research needs to be conducted in order todevelop useful analytical tools based on activity theorythat could be used to design, and evaluate a construc-tionist community. The new model should overcomethe shortcomings of the traditional triangle model asdiscussed in Section 2.3.1. We would also like toemphasise that the main purpose of this study was notto refine Engestrom’s model, but to adapt and apply itto analyse CSCL. Refining the model is such a bigundertaking that it is beyond the scope of this article.

We must also reiterate the fact that activity theoryitself is not limited to what is presented in the trianglediagram as proposed by Engestrom. Although hismodel is useful, it overlooks several significantconcepts of activity theory. It is understood thatEngestrom’s model is intended to be open so that it canbe used in various domains but this has proven to posea serious di"culty among the practitioners as someresearchers have started to operationalise it so that it ismore practical in day-to-day methodology (Korpelaet al. 2000, Barab et al. 2001, Mwanza 2002). Ourpurpose went in line with these scholars; we expandedactivity theory so that it can be used by educators, aswell as game designers in general and game communitydesigners in particular.

Most importantly, it should be able to not onlyanalyse human activities particularly constructionactivities but also the knowledge construction thatarises from such activities. Moreover, more empiricalstudies will also be conducted to further enhance andverify this model.

References

Barab, S.A., et al., 2002. Using activity theory to understandthe systemic tensions characterizing a technology-richintroductory astronomy course. Mind, Culture andActivity, 9 (2), 76–107.

Barab, S.A., Hay, K.E., and Yamagata-Lynch, L.C., 2001.Constructing networks of action-relevant episodes: an insitu research methodology. The Journal of LearningSciences, 10 (1 and 2), 63–112.

Boeree, C.G., 1997. Personality theories [online]. Availablefrom: http://www.social.psychology.de/do/pt-jung.pdf.

Dillenbourg, P., et al., 1996. The evolution of research oncollaborative learning. In: E. Spada and P. Reiman, eds.Learning in humans and machine: Towards an interdisci-plinary science. Oxford: Elsevier, 189–211.

Behaviour & Information Technology 553

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011

Du!y, T.M., Lowyck, J., and Jonassen, D.H. eds., 1993. Thedesign of constructivistic learning environments: Implica-tions for instructional design and the use of technology.Heidelburg, FRG: Springer-Verlag.

Emigh, W. and Herring, S.C., 2005. Collaborative authoringon the web: a genre analysis of online encyclopedias. In:Proceedings of the Thirty-Eighth Hawai’i InternationalConference on System Sciences (HICSS-38), LosAlamitos.

Engestrom, Y., 2001. Expansive learning at Work: toward anactivity theoretical reconceptualisation. Journal of Edu-cation and Work, , 14 (1), 133–156.

Glaser, B.G. and Strauss, A.L., 1967. The discovery ofgrounded theory. Chicago: Aldine.

Grand Theft Auto: San Andreas, 2005. Wikibook [online].http://en.wikibooks.org/wiki/Grand_Theft_Auto:_ Sa-n_Andreas [Accessed 21 November 2005].

Hewitt, J., 2002. From a focus on tasks to a focus onunderstanding: the cultural transformation of a Torontoclassroom. In: T. Koschmann, R. Hall, and N. Miyake,eds. Computer supported cooperative learning volume 2:carrying forward the conversation. Mahwah, New Jersey:Lawrence Erlbaum Associates, 11–41.

Jones, G.B., 2003. Blogs and wikis: environments for on-linecollaboration. Language Learning & Technology, 7 (2),12–16.

Korpela, M., Soriyan, H.A., and Olufokunbi, K.C., 2000.Activity analysis as a method for information systemsdevelopment. Scandinavian Journal of Information Sys-tems, 12, 191–210.

Koschmann, T. ed., 1996. CSCL: Theory and practice of anemerging paradigm. Mahwah, NJ: Lawrence ErlbaumAssociates.

Koschmann, T., 2001. Revisiting the paradigms of instruc-tional technology. In: G. Kennedy, et al., eds. Meeting atthe Crossroads. Proceedings of the 18th Annual Con-ference of the Australian Society for Computers inLearning in Tertiary Education. Melbourne: BiomedicalMultimedia Unit, The University of Melbourne, 15–22.Available from: http://www.ascilite.org.au/conferences/melbourne01/pdf/papers/koschmannt.pdf [Accessed 6November 2005].

Lave, J. and Wenger, E., 1991. Situated learning: legitimateperipheral participation. Cambridge University Press.

Lebow, D., 1993. Constructivist values for instructionalsystems design: five principles toward a new mindset.Educational Technology Research and Development, 41(3), 4–16.

Leont’ev, A.N., 1978. Activity, consciousness, and personality.Englewood Cli!s, NJ: Prentice-Hall.

Mader, S., 2004. OpenSpectrum: a wiki-based learning tool forspectroscopy that anyone can edit. Winter 2005CONGCHEM: trends and new ideas in chemical educa-tion, on-line conference [online]. Available from: http://www.chedccce.org/confchem/2005/a/index.html [Ac-cessed 21 November 2005]

Mwanza, D., 2002. Towards an activity-oriented method forhci research and practice. Thesis (PhD). Open University.

Oliver, K., 2000. Methods for developing constructivistlearning on the web. Educational Technology, 30 (6), 5–16.

Papert, S., 1980. Mindstorms: children, computers, andpowerful ideas. New York: Basic Books.

Papert, S. and Harel, I., 1991. Situating constructionism. In: IHarel and S. Papert, eds. Constructionism. Norwood, NJ:Ablex Publishing.

Piaget, J., 1929. The child’s conception of the world. London:Routledge and Kegan Paul.

Piaget, J. and Inhelder, B., 1969. The psychology of the child.New York: Basic Books.

Resnick, M., Rusk, N., and Cooke, S., 1998. The computerclubhouse: technological fluency in the inner city. Highertecnology and low-income communities [online]. Availablefrom: http://archnet.org/library/documents/one-document.tcl?document_id!8782 [Accessed 8 November 2005].

Shaw, A.C., 1995. Social constructionism and the inner city:designing environments for social development and urbanrenewal. Thesis (PhD). MIT.

Shneiderman, B., 2002. Leonardo’s laptop. Cambridge, MA:MIT Press.

Stahl, G., 2002. Contributions to a theoretical framework forCSCL. In: G. Stahl, ed. Computer support for collabora-tive learning: foundations for a CSCL community.Proceedings of the CSCL 2002 conference. Hillsdale,NJ: Erlbaum. Available from: http://newmedia.colora-do.edu/cscl/81.pdf [Accessed 21 November 2005]

Strauss, A.L. and Corbin, J., 1990. Basics of qualitativeresearch. Newbury Park, CA: Sage.

Vygotsky, L., 1930. Mind and society. Cambridge, MA:Harvard University Press.

Wang, H.C., et al., 2005. An empirical exploration of usingWiki in an English as a second language course. In:Proceedings of 5th IEEE International Conference onAdvanced Learning Technologies (ICALT 2005), Kaoh-siung, Taiwan, IEEE Computer Society Press, 155–157.

Wikipedia, 2005. Wikipedia [online]. Available from: http://en.wikipedia.org/wiki/Wikipedia#References [Accessed21 November 2005].

554 C.S. Ang et al.

Dow

nloa

ded

by [C

ypru

s Uni

vers

ity O

f Tec

hnol

ogy

] at 1

1:54

22

Aug

ust 2

011