WORKSHOP Scripted vs. Free CS collaboration

WORKSHOP Scripted vs. Free CS collaboration: Alternatives and paths for adaptable and flexible CS scripted collaboration”

Pre-Workshop Proceedings

Workshop Website http://mlab.csd.auth.gr/cscl2009/sfc-workshop.htm

SFC Workshop – CSCL 2009


2

ORGANIZERS Stavros Demetriadis, Aristotle University of Thessaloniki, Greece Yannis Dimitriadis, University of Valladolid, Spain Frank Fischer, University of Munich, Germany

PROGRAMME COMMITTEE Rafael Calvo, University of Sydney, Australia Stavros Demetriadis, Aristotle University of Thessaloniki, Greece, Pierre Dillenbourg, University of Geneva, Switzerland Yannis Dimitriadis, University of Valladolid, Spain Frank Fischer, University of Munich, Germany Maria Grigoriadou, Univerity of Athens, Greece Andreas Harrer, University of Duisburg, Germany Simeon Retalis, University of Piraeus, Greece Pierre Tchounikine, University of Grenoble, France Thrasyvoulos Tsiatsos, Aristotle University of Thessaloniki, Greece Armin Weinberger, University of Twente, Holland

THEME & GOALS This workshop focused on a decisive topic in collaborative learning, namely, how to provide constructive yet flexible support to collaborating students, so that any possible drawbacks emerging from unproductive instruction-imposed limitations are minimized or eliminated. Research has consistently emphasized that free collaboration conditions may fail to engage all team members in productive learning interactions (e.g., Hewitt, 2005; Bell, 2004; Liu & Tsai, 2008). Teacher’s consistent guidance and scaffolding is seen as a necessary prerequisite to achieve the desired learning outcomes, however, it is very difficult – if not impossible – for the teacher to consider all interacting parameters in order to provide valuable learning experiences to collaborating individuals (Dillenbourg et al., 1995). Scripted collaboration has been proposed as a remedy to detrimental peer interaction. Scripted collaboration – taking its origin in the scripted cooperation approach (O’Donnell & Danserau, 1992) – is the idea that collaboration can be orchestrated by didactic scenarios, aiming to engage students in fruitful learning interactions. Collaboration scripts structure and guide the collaboration process by prescribing the phases of the activity, assigning student roles and triggering peer interactions so that all learners are engaged in a meaningful learning situation (e.g., Weinberger, Stegmann, Fischer, & Mandl, 2007). The considerable interest that the scripting approach has gained in the CSCL community has motivated efforts for the formalization of collaboration scripts (Kobbe, et al. 2007), the operationalization of scripts in CSCL settings (Tchounikine, 2008) and the development of computer-based tools that allow instructors to design and parameterise the computer-based representations of collaboration scripts, which consequently guide the group activity (e.g. Bote-Lorenzo et al., 2007; Turani & Calvo, 2007). The CSCL community is currently systematically exploring the issue of collaboration scripting through various events and projects (see for example, Kaleidoscope ERT MOSILE and COSSICLE projects at http://cossicle.noekaleidoscope.org/events.html, Dimitriadis (2008) keynote speech, Fischer (2007) plenary talk, the Alpine RendezVous workshop (2007), etc.). However, there are also voices calling for attention on the issue of sacrificing the “fun and creativity of free collaboration” to attain effectiveness. Computer-supported scripting has been criticized for its loss of flexibility (i.e. difficulty of modifying a script in real time according to the needs of the instructional situation) (Dillenbourg & Jermann, 2007) and also the danger of “over-scripting” the collaborative activity (i.e. the pitfall of overemphasizing script imposed interactions and constraining natural collaboration) (Dillenbourg, 2002). The need has also been emphasized to differentiate between flexibility loss that is due to pedagogical design and undesired constraints of computer-based scripting techniques (Dillenbourg & Tchounikine, 2007). At the same time, suggestions for more adaptive and intelligent forms of CSCL scripting are already on the table (for example, Harrer, Malzahn & Wichmann, 2008; Tsovaltzi, et al. 2008; Demetriadis & Karakostas, 2008). Against this background, this workshop has argued that a significant issue for the CSCL community is to productively reflect on existing research regarding the pros and cons of scripted vs. free collaboration, with the focus being on methods and tools for the flexible representation, design and implementation of CS scripted collaboration activities.



3

PARTICIPATION The workshop attracted the attention of eight different research groups worldwide and the respective accepted and presented papers are: (1) Towards a Classroom Ecology of Devices: Interfaces for Collaborative Scripts Jochen Rick Department of Computing, The Open University, Milton Keynes, MK7 6AA, UK [email protected]

(2) Where Is the Script? Support for Computer-Based Small Group Collaboration and Plenary Activities by Means of Small-Group and Classroom Scripts Christof Wecker, Ingo Kollar, Sybille Langer, Frank Fischer University of Munich, Department of Psychology, Leopoldstr. 13, 80802 Munich, [email protected], [email protected], [email protected], [email protected]

(3) Structured activities in CSCL: a case study Francesca Pozzi, Donatella Persico Istituto Tecnologie Didattiche – CNR, Via De Marini 6, Genoa [email protected], [email protected]

(4) Emotional Pedagogical Agents for Collaborative Learning: an Application for Children with Learning Difficulties Konstantina Chatzara, Charalampos Karagiannidis Department of Special Education, University of Thessaly, Volos, Greece [email protected], [email protected]

(5) Supporting Collaboration at Different Levels in Computer Supported Education Manuel Caeiro, Luis Anido, Martín Llamas, Jorge Fontenla, Roberto Pérez University of Vigo, Spain, {Manuel.Caeiro; Luis.Anido; Martin.Llamas; JFontenla; RPerez}@det.uvigo.es

(6) CSCL Scenarios := A Cocktail of CLFPs Georgia Lazakidou1, Symeon Retalis1, Petros Georgiakakis1, Stamos Karamouzis2, 1 University of Pireaus, Department of Technology Education and Digital Systems, 80 Karaoli & Dimitriou, 18534 – Greece 2 Regis University, School of Computer & Information Sciences, USA [email protected], [email protected], [email protected], [email protected]

(7) Interrelating Assessment and Flexibility in IMS-LD CSCL Scripts E.D. Villasclaras-Fernández1, D. Hernández-Leo2, J.I. Asensio-Pérez1, Y. Dimitriadis1, L. de la Fuente-Valentín3 1 GSIC/EMIC, University of Valladolid, Camino del Cementerio, s/n, 47011, Valladolid, Spain 2 Pompeu Fabra University, Estació de França, Passeig de Circumval·lació 808003, Barcelona, Spain 3 Telematics Engineering Department, University Carlos III of Madrid, Spain. [email protected], [email protected], {juaase, yannis}@tel.uva.es, [email protected]

(8) Adaptive Patterns in Systems for Collaborative Learning and the Role of the Learning Design Specification Stavros Demetriadis, Ioannis Magnisalis, Anastasios Karakostas Aristotle University of Thessaloniki, POBOX 114, 54124, Thessaloniki, Greece [email protected]; [email protected]; [email protected]

It seems that these eight contributions fall nicely into two major thematic areas: (GROUP A) “Collaboration Scripts in the Classroom” Papers from 1 to 4 present specific cases and explore various perspectives on the impact of scripted collaboration in the classroom, with the support of technology. More specifically, Rick explores the potential of tabletops technology to support various aspects of scripted collaborative activity in the classroom. He presents three examples of how different interfaces can support collaboration and argues that understanding the impact of these interfaces eventually facilitates the efficient instantiation of collaboration scripts. Then, Wecker, Kollar, Langer and Fischer present the outcomes of two empirical studies on integrating small group collaboration in classroom instruction. They investigate a faded vs. unfaded scripting condition and also different degrees of structured classroom script. Employing a similar perspective, Pozzi and Persico analyze three online activities and reflect on the potential of each activity (with different levels of structuredness) to foster various dimensions



4

of the learning process. Finally, Chatzara and Karagiannidis bring into focus the role that emotional pedagogical agents may play in collaboration when used to support children with learning difficulties. They emphasize the importance of considering the emotional state of the user and the impact of emotional agents with collaborating functionalities. (GROUP B) “Modelling & Adapting the Scripted Activity” Papers from 5 to 8 focus on technologies and methods for modelling the collaboration and expressing design aspects that offer greater flexibility to teachers and learners especially when unpredictable events occur during collaboration. Within this perspective Caeiro, Anido, Llamas, Fontenla, and Pérez analyze the capacity of the existing proposals to effectively support the computational modelling of educational units and propose a new educational modelling language (PoEML) to cope with these requirements. Then, Lazakidou, Retalis, Georgiakakis, and Karamouzis, propose that to make a satisfactory cocktail of CLFPs an extra element named “variations” is needed that includes all proposals related to the suggested solution. They exemplify their approach in the case of a complex CSCL strategy. Next, Villasclaras-Fernández, Hernández-Leo, AsensioPérez, Dimitriadis, and de la Fuente-Valentín discuss the interrelation between script flexibility and the assessment plan. They argue that assessment can be a driving force to detect unpredictable situations or results and react accordingly by modifying the script to tackle them. Finally, Demetriadis, Magnisalis and Karakostas present their framework for adaptation patterns to increase flexibility in designs of scripted collaboration and they proceed to discuss issues regarding the limitations of Learning Design (LD) specification as a tool to express the adaptive features of collaboration scripts. In conclusion, the above contributions suggest that the workshop key-question can be approached from two complementary perspectives and courses of action. First, empirical research in the classroom is needed to provide insights regarding the role of various factors on achieving the right balance between necessary structuredness and desired flexibility. These factors may include innovative technologies (such as tabletops), fade/unfade script condition, script structuredness and its impact on the dimensions of the learning process, and also the presence of agents with collaborating functionalities. Second, efforts on modelling the collaborative learning activity should provide efficient ways of expressing the desired adaptive features of the activity. This may be accomplished by using more appropriately built educational modelling language, integrating new elements to increase the expressivity of modelling tools, employing the assessment plan as a reference point to increase script flexibility or introducing the notion of adaptation pattern as a framework for designing adaptive versions of collaboration scripts.

WORKSHOP SCHEDULE and ACTIVITIES The workshop will be conducted as a full-day workshop on Tuesday, 9 June (Karpathos room). Morning Sessions There will be two morning sessions and two afternoon sessions. During the morning sessions participants will present their contributions according to the following program. 09:00 – 09:10 09:10 – 09:35 09:35 – 10:00

10:00 – 10:25 10:25 – 10:50

10:50 – 11:00 11:00 – 11:30 11:30 – 11:55 11:55 – 12:20 12:20 – 12:45

Welcome – Overview Organizing Committee Towards a Classroom Ecology of Devices: Interfaces for Collaborative Scripts Jochen Rick Where Is the Script? Support for Computer-Based Small Group Collaboration and Plenary Activities by Means of Small-Group and Classroom Scripts Christof Wecker, Ingo Kollar, Sybille Langer, Frank Fischer Structured activities in CSCL: a case study Francesca Pozzi, Donatella Persico Emotional Pedagogical Agents for Collaborative Learning: an Application for Children with Learning Difficulties Konstantina Chatzara, Charalampos Karagiannidis Questions - Discussion Coffee break Supporting Collaboration at Different Levels in Computer Supported Education Manuel Caeiro, Luis Anido, Martín Llamas, Jorge Fontenla, Roberto Pérez CSCL Scenarios := A Cocktail of CLFPs Georgia Lazakidou, Symeon Retalis, Petros Georgiakakis, Stamos Karamouzis Interrelating Assessment and Flexibility in IMS-LD CSCL Scripts E.D. Villasclaras-Fernández, D. Hernández-Leo, J.I. Asensio-Pérez, Y. Dimitriadis, L. de la Fuente-Valentín


12:45 – 13:10

13:10 – 13:30 13:30 – 15:00


5

Adaptive Patterns in Systems for Collaborative Learning and the Role of the Learning Design Specification Stavros Demetriadis, Ioannis Magnisalis, Anastasios Karakostas Questions - Discussion Lunch break

The time slots are - to a certain extent - indicative. Presenters will have a 25 minutes slot to present (20 minutes for presentation and 5 minutes to answer questions). At the end of each morning session there will be time for posting questions to all session presenters and also for final comments/discussion. Afternoon Sessions During the afternoon sessions participants will form 2 groups, Group A and Group B. Group A includes contributions falling under the general thematic axis “Scripts in Classroom”, and Group B under the axis “Modelling and Adapting the Scripted Activity” (as already classified and commented above). In the first afternoon session (15:00 – 16:30) both groups will be given certain key-issues to explore collaboratively. These are: Issues for Group A (A-1) A teacher is interested in implementing scripted collaboration in the classroom. He/she would like to have an operational synthesis of your various conclusions and suggestions (by “operational” we mean something practical to guide teacher’s everyday practice). How would you advice the teacher? Please provide a synthesis of your perspectives emphasizing also possible differences and disagreements. (A-2) Research teams in group B have suggested various modes for modelling and adapting the collaborative activity. What interesting links do you detect between group A and group B work? You may comment on whatever seems important to your group (for example, a method suggested by group B might become a helpful tool for the type of script implementation that you envisage). Issues for Group B (B-1) A teacher is interested in understanding how your suggested methods can be integrated to CSCL systems and increase capacity for modelling and adaptation. What do you answer to the teacher? Please provide a synthesis of the practical implications of your proposals emphasizing also possible differences and disagreements. (B-2) Research teams in group A provide empirical evidence from applying scripted collaboration in the classroom. What interesting links do you detect between group A and group B work? You may comment on whatever seems important to your group (for example, some conclusions from group A might substantiate the need for the modelling or adaptive method that you suggest). In the second afternoon session (17:00 – 18:00) group moderators will present the deliverables of their groups to the above issues. The deliverables may have any format that the group wishes (for example, a doc file with a list of items, a concept map, etc.). An open discussion will follow finalizing the workshop conclusions on the above issues. So, the programme for the afternoon sessions will be as follows: 15:00 – 16:30 16:30 – 17:00 17:00 – 18:00

Group discussion and preparation of deliverables Coffee break Presentation of group deliverables (15 minutes for each group) Open discussion and finalizing the workshop conclusions

So, the workshop participants are kindly requested to read all the contributions and perhaps do some preparation for the above activity. Closing this short introduction the workshop Organizing Committee would like to thank all the participants who with their quality contributions provide valuable insights to the CSCL community. We would also like to thank the members of the Programme Committee for their constructive support during the review process. The SFC-Workshop Organizing Committee Stavros Demetriadis Yannis Dimitriadis Frank Fischer



6

REFERENCES Alpine Rendez-Vous workshop (2007). Report available at http://www.ell.aau.dk/The-CSCL-Alpine-RendezVous.384.0.html2007. Bell, P. (2004). Promoting students' argument construction and collaborative debate in the classroom. In M. C. Linn, E. A. Davis & P. Bell (Eds.), Internet environments for science education (pp. 114-144). Mahwah, NJ: Erlbaum. Bote-Lorenzo, M.L., Gómez-Sánchez, E., Vega-Gorgojo, G., Dimitriadis, Y.A., Asensio-Pérez, J.I. and I.M. Jorrín-Abellán (2007). Gridcole: A tailorable grid service based system that supports scripted collaborative learning, Computers & Education. Retrieved Nov. 20th, 2007, from http://dx.doi.org/10.1016/ j.compedu.2007.05.004 Demetriadis, S., & Karakostas, A. (2008). Adaptive collaboration scripting: A conceptual framework and a design case study. Proceedings of CISIS 2008: 2nd International Conference on Complex, Intelligent and Software Intensive Systems, Barcelona, Spain. Dillenbourg, P. (2002). Over-scripting CSCL: the risks of blending collaborative learning with instructional design, in P.A. Kirschner, (ed.), Three Worlds of CSCL. Can We Support CSCL?, Open Universiteit Nederland, Heerlen, 61–91. Dillenbourg, P. and Jermann, P. (2007). Designing interactive scripts. In F. Fischer, I. Kollar, H. Mandl, and J. Haake, (eds.), Scripting computer-supported collaborative learning: Cognitive, computational and educational perspectives, Springer, New York, 276-301. Dillenbourg, P. and Tchounikine, P. (2007). Flexibility in macro-scripts for computer-supported collaborative learning. Journal of Computer Assisted Learning, 23(1), 1-13. Dillenbourg, P., Baker, M., Blaye, A., and O’Malley, C. (1995). The evolution of research on collaborative learning. In E. Spada, and P. Reiman, (eds.), Learning Human and Machine: Towards an interdisciplinary learning science, Elsevier, Oxford, 189-211. Dimitriadis, Y. (2008). From design to evaluation of scripted networked collaborative learning environments. Keynote Speech. Proceedings of Networked Learning Conference, Sani, Greece. Available at http://www.networkedlearningconference.org.uk/people/Dimitriadis.htm Fischer, F. (2007). Learning through Scripted Discussion. Plenary talk at Kaleidoscope NoE 2007 Symposium, Abstract available at http://www.noe-kaleidoscope.org/group/symposium/programme/ Harrer, A., Malzahn, N., & Wichmann, A. (2008). The remote control approach - An architecture for adaptive scripting across collaborative learning environments. Journal of Universal Computer Science, 14(1), 148173. Hewitt, J. (2005). Toward an understanding of how threads die in asynchronous computer conferences. The Journal of the Learning Sciences, 7(4), 567-589. Kobbe, L., Weinberger, A., Dillenbourg, P., Harrer, A., Hämäläinen, R., and F. Fischer (2007). Specifying computer-supported collaboration scripts, International Journal of Computer-Supported Collaborative Learning, 2(2-3), 211-224. Liu, C., & Tsai, C. (2008). An analysis of peer interaction patterns as discoursed by on-line small group problem-solving activity. Computers & Education, 50, 627–639. O'Donnell, A.M. and Dansereau, D.F. (1992). Scripted cooperation in student dyads: A method for analyzing and enhancing academic learning and performance. In R. Hertz-Lazarowitz, and N. Miller, (eds.), Interaction in cooperative groups: The theoretical anatomy of group learning, Cambridge University Press, London, 120-141. Tchounikine, P. (2008). Operationalizing macro-scripts in CSCL technological settings. Computer-Supported Collaborative Learning, 3:193–233. Tsovaltzi, D., McLaren, B. M., Rummel, N., Scheuer, O., Harrer, A., Pinkwart, N., & Braun, I. (2008). Using an Adaptive Collaboration Script to Promote Conceptual Chemistry Learning. Paper presented at the 9th International Conference on Intelligent Tutoring Systems (ITS-08). Turani, A. and Calvo, R. (2007). The Potential Use of Collaboration Scripts in Synchronous Collaborative Learning. Proceedings of IMCL2007 Conference, Amman, Jordan, April 18-20. Weinberger, A., Stegmann, K., Fischer, F. and Mandl, H. (2007). “Scripting argumentative knowledge construction in computer-supported learning environments”, in: F. Fischer, I. Kollar, H. Mandl, and J. Haake, (eds.), Scripting computer-supported collaborative learning: Cognitive, computational and educational perspectives, Springer, New York, pp. 191-211.



7

Contents Towards a Classroom Ecology of Devices: Interfaces for Collaborative Scripts ..........8 Where Is the Script? Support for Computer-Based Small Group Collaboration and Plenary Activities by Means of Small-Group and Classroom Scripts..........................13 Structured activities in CSCL: a case study..................................................................18 Emotional Pedagogical Agents for Collaborative Learning: an Application for Children with Learning Difficulties..............................................................................23 Supporting Collaboration at Different Levels in Computer Supported Education ......28 CSCL scenarios := A cocktail of CLFPs ......................................................................33 Interrelating assessment and flexibility in IMS-LD CSCL scripts...............................38 Adaptation Patterns in Systems for Collaborative Learning and the Role of the Learning Design Specification......................................................................................43



8

Towards a Classroom Ecology of Devices: Interfaces for Collaborative Scripts Jochen Rick, Department of Computing, The Open University Milton Keynes, MK7 6AA, UK, [email protected] Abstract: The ShareIT Project has been investigating how interactive tabletops can support co-located collaborative learning. As interactive tabletops are still in their infancy, basic questions of how these interfaces support collaboration are still unanswered. To address these, we have been conducting conventional psychology experiments, where a small intervention is run with a significant number of groups and statistical comparisons can be made across conditions. While that is a useful research paradigm for understanding the potential of interactive tabletops by themselves, a different perspective will be needed to integrate them into a working classroom. This position paper is about bridging that gap by envisioning interactive tabletops as part of a larger classroom ecology of devices. It examines the potential of different devices and different interfaces on those devices to support different parts of a collaborative script. Based on our work on interactive tabletops, I present three examples of how different interfaces can enable, encourage, and enforce collaboration. A better understanding of these interfaces ultimately leads to better tools to instantiate collaborative scripts.

An Ecology of Devices The field of ubiquitous computing (ubicomp) has envisioned a future where multiple devices interact seamlessly to provide a fluid user experience (Greenfield, 2006; Weiser, 1999). We are approaching a time when that future can be realized in the classroom. Different technologies can work in tandem to form an ecology of devices that allows each to be used for purposes that best suits its interface and affordances. Handhelds are portable and can accompany the students. Interactive tabletops encourage small group work (Rick, Rogers, Haig, & Yuill, 2009). Electronic whiteboards best support whole class discussion (Moss et al., 2007). Different devices can become part of a coordinated system to support collaborative activities (e.g., Roschelle et al., 2007).

A Vision of the Future Students go on a field trip where they use handheld computers (e.g., PDAs, iPhone, etc.) and attachable probes to gather environmental data. Students are paired up for this task, so that one can operate the handheld while the other operates the probe. Students are encouraged to switch roles. Students take pictures of flora and fauna and an interactive application helps learners identify them. When back in the classroom, students add their data to the classroom-computing infrastructure. The next day, the teacher uses her electronic whiteboard to review the data with the whole class. Several research topics are identified as worth pursuing (e.g., how does soil pH and moisture affect plant life, how does plant life affect which bugs are where). The class is split into equal-sized groups to research these questions given the available data. Groups use interactive tabletops to analyze the data. These lend themselves particularly well to small group work as every group member has a good view of the data and can interact with it concurrently. Each group is also given a paper worksheet of questions to guide their work. Using the analysis tools, the groups examine the data to come up with some conclusion. They are then asked to create a presentation of their work. Analysis and creating a presentation takes several days. Many students take the presentations that they have home and work on them from their home PC. On presentation day, each group presents their findings to the whole class on the interactive whiteboard with the teacher running a whole-class discussion and feedback section. Based on that feedback, the groups revise their original presentation and submit it to the teacher. The following day, students are split into different groups based on a jigsaw pattern. Their assignment is to use the interactive tabletop to create a mind map of the different concepts. Since each learner brings individual insight from having been in a different group, the students must work together to build a mind map that encompasses all the concepts.

Interfaces for Collaborative Scripts A collaborative script organizes learning activities and classroom interaction along a predetermined trajectory to better support learning. The scripting can be done at either the micro level (how students engage a specific task step-by-step) or macro level (how learners are organized for the different phases of the learning activity). In the above example, the learning activity follows a macro script, from individual data gathering to whole class discussion to final small-group reflection. Macro scripts are useful as different configurations are better suited



9

for different phases. For instance, brainstorming has been shown to be most effective when done individually (Dugosh & Paulus, 2005). Often, group work benefits from an inequity of knowledge or divergent positions of the group members (Dillenbourg & Hong, 2008). At other times, no inequity is needed for collaboration to benefit learning (Roschelle, 1996). Creating, fine tuning, and executing educationally effective macro scripts is a difficult design challenge (Anastopoulou et al., 2008). While that is the ultimate problem, this paper addresses only a step towards a solution: How do we unlock the potential of each device to support collaborative learning? In the example, each step in the macro script is carried out with a computational interface that best supports it. Due to its large vertical display, the electronic whiteboard lends itself to whole class discussion. Due to its portability, the handheld computer can be taken into the field. Particular attention in this example has been paid to the potential of interactive tabletops to support group work.

Interactive Tabletops for Group Work From handhelds to desktops, the dominant paradigm for computing has been a personal one—one user per device. Interactive tabletops offer the potential to break that restrictive mapping. Already, much of the research on interactive tabletops examines how they can support collaborative activities. As tabletops become commercially available, they have the potential to support group work in a way no other popular classroom technology has been able to do (Rick, Rogers, et al., 2009). While tabletops are starting to be available for the classroom market (e.g., SMART Technologies is selling a multi-touch tabletop aimed at younger children), many of the interaction paradigms and technical features are still in flux. One particular challenge is understanding how tabletop interfaces can support collaborative tasks, such as found in a collaborative script. Both a change in task and a change in interface can change the nature of collaboration (Nacenta, Pinelle, Stuckel, & Gutwin, 2007). In our recent research, we are trying to understand how the task and the interface can enable, encourage, and enforce collaboration (cf. Benford et al., 2000). What follows are three examples of this work.

Figure 1. Three users position tables and students in their classroom using OurSpace.

OurSpace OurSpace is a desk positioning and seating allocation application designed for three concurrent users (Rick, Harris, et al., 2009). A bird’s eye virtual floor plan of the participants’ classroom is placed in the center of the tabletop so that all participants, irrespective of their position, have good access to it (Figure 1). Participants use their fingers to drag icons of students and desks onto the classroom plan. When a student is dragged over an available desk seat, the seat is highlighted and the student is oriented toward that seat position; when dropped, the student icon snaps to that seat. Once a student is in a desk seat, he or she moves along with the desk;



10

students can also be dragged out of their seat to relocate them. To rotate desks, users drop them on rotation areas at the bottom left and right of the screen. The research on OurSpace compared two interaction conditions: in the single touch condition, only one user’s touch input registers at a time; in the multiple touch condition, all three users can interact concurrently. While turn-taking dialog was significantly higher in the single touch condition, no differences were found in either physical or verbal equity (Harris et al., 2009). This result contrasts with other research where multiple touch systems were found to be more equitable. Our findings suggest that minimal transition time and effort is critical to avoiding inequity with a turn-taking interface (Rick, Harris, et al., 2009). Turn-taking interfaces have the additional benefit of slowing down the interaction and making participants more aware of each other’s contributions, which is useful for collaborative learning. The desk positioning and seating allocation task that the children were assigned was designed to be complex (i.e., with no simple solution). Thus, the collaboration was useful in creating a better solution as team members often approached the task from different perspectives. The tabletop interface enabled that collaboration. On the other hand, much of the work was done without much coordination. The multiple touch interface enabled the collaboration, but did nothing to encourage or enforce it. Eventually, as the design began to mature, the nature of the task required the participants to coordinate their actions. In contrast, the single touch interface slightly encouraged the groups to coordinate their actions from the beginning. This demonstrates how a slightly different interface can change the nature of the collaboration for the same task.

DigiTile DigiTile is an adaptation of DigiQuilt (Lamberty, 2007) for the DiamondTouch table (Rick & Rogers, 2008). Like DigiQuilt, it is a construction kit for learning about math and art by designing colorful mosaic tiles. In addition to being aesthetically pleasing, these tiles lend themselves to mathematical analysis. The designs embody fraction concepts and are often symmetric. The application provides feedback on the mathematical concepts; for instance, the fraction of the entire tile that is a certain color is displayed on the button for selecting that color. While DigiQuilt uses a conventional PC to support a single user, DigiTile is intended to be used by two concurrent users arranged side-by-side (Figure 2). The learners are given increasingly difficult challenges to accomplish, such as creating a design that is half red or creating a design that is horizontally symmetric.

Figure 2. Two users collaborating on a DigiTile challenge of half red and half black. Based on a thirty-minute session with DigiTile, learners achieved statistically significant increases in content understanding (Rick, Rogers, et al., 2009). While the interface allows both users to act independently, study participants frequently chose to coordinate their actions. They learned quickly that both participants adding to the tile independently causes confusion. One participant’s contributions interfere with the partner’s strategy. When participants discovered that uncoordinated contributions failed, they began to work together on one strategy at a time, often coordinating on strategies verbally. One participant might add pieces in a systematic



11

manner while the partner watched the updating fractions and told the other when they had reached the goal. DigiTile’s interface thus strongly encouraged collaboration. As with OurSpace, the task of completing a mathematical challenge benefits from collaboration. The learners benefit from articulating their understanding and providing feedback to their partners. While the task is not inherently collaborative, DigiTile’s interface encouraged collaboration by making it nearly impossible to complete the task without it.

WordCat WordCat is a word categorization game for two users positioned side-by-side (Figure 3). It is based on the SCOSS (Separate Control Of Shared Space) model of interaction, where learners work on the same task individually but must agree on a solution before they can proceed (Kerawalla, Pearce, Yuill, Luckin, & Harris, 2008). In WordCat, the challenge is to sort twelve words into the four central bins (three words per bin), so that each of the columns and each of the rows form categories related to the meaning or the shape of the word. In the example in Figure 3, the solution categories will be colors versus animals and words that begin with g versus words that begin with b. The words appear one at a time; the same word appears in the left yellow bin and in the right blue bin. Users then drag the words from the bin to the position in the central table where they believe it belongs. The left user can only move yellow words, while the right user can only move blue words; this rule is enforced by the tabletop interface. If both the yellow and blue version of a word have been moved to the same central bin, one green version of the word with bold text shows up. Once users are in agreement on all given words, a new word appears in the left and right bins.

Figure 3. Two users placing word tiles in WordCat. Unlike in OurSpace and DigiTile, one user cannot complete the task by themselves. Both users must move pieces into the same positions before a new word is added. Thus, WordCat’s SCOSS-based interface strongly enforces collaboration. Inherently, the word categorization game does not require collaboration; however, it does benefit from collaboration as players help each other out. Along the way, they must convincingly articulate their hypotheses to try to convince their partners to go along. While WordCat’s interface is quite strict, adjustments can be made to the interface to lessen the strength of the enforcement. For instance, new words could appear even if agreement had not been reached. Alternatively, all twelve words could appear in the respective bins at the start of the task. While the collaboration would still be enforced eventually (i.e., both partners must agree in the end), it would be so to a lesser extent (since the users could work largely independently until the end).

Conclusion As these three applications demonstrate, the interface affects how the task is carried out. It affects the nature of the collaboration and thereby the learning that occurs. To effectively utilize interactive tabletops in a classroom



12

ecology of devices, it is important to understand these interface differences and how to design for them. Yet, that is only a partial challenge to integrating interactive tabletops into a classroom. First, novel technologies tend to be expensive and it is unlikely that one classroom could afford more than one or two interactive tabletops in the near future. Thus, the challenge will be to script learning activities that are not unduly limited in effectiveness by the limited number of devices. Second, creating application suites that utilize multiple different devices is a challenge. Since they have different hardware and different interfaces, different devices often do not support the same applications. Third, sharing data seamlessly between different devices in the classroom and outside the classroom (e.g., children taking presentations home to work on them there) is both an implementation and usability challenge. Fourth, the problem only becomes more challenging as different devices (e.g., public displays, digital desks that allow document capture, tangible bits) are added to the classroom ecology.

References Anastopoulou, S., Sharples, M., Wright, M., Martin, H., Ainsworth, S., Benford, S., et al. (2008). Learning 21st century science in context with mobile technologies. In J. Traxler, B. Riordan, & C. Dennett (Eds.), Proceedings of the mLearn 2008 (pp. 12–19). Benford, S., Bederson, B. B., Akesson, K.-P., Bayon, V., Druin, A., Hansson, P., et al. (2000). Designing storytelling technologies to encouraging collaboration between young children. In Proceedings of CHI ’00 (pp. 556–563). New York: ACM Press. Dillenbourg, P., & Hong, F. (2008). The mechanics of CSCL macro scripts. International Journal of ComputerSupported Collaborative Learning, 3(1), 5–23. Dugosh, K. L., & Paulus, P. B. (2005). Cognitive and social comparison processes in brainstorming. Journal of Experimental Social Psychology, 41(3), 313–320. Greenfield, A. (2006). Everyware: The dawning age of ubiquitous computing. Berkeley, CA: New Riders. Harris, A., Rick, J., Bonnett, V., Yuill, N., Fleck, R., Marshall, P., et al. (2009). Around the table: Are multipletouch surfaces better than single-touch for children’s collaborative interactions? In Proceedings of CSCL ’09. Mahwah, NJ: Lawrence Erlbaum Associates. Kerawalla, L., Pearce, D., Yuill, N., Luckin, R., & Harris, A. (2008). “I’m keeping those there, are you?” The role of a new user interface paradigm—separate control of shared space (SCOSS)—in the collaborative decision-making process. Computers & Education, 50(1), 193–206. Lamberty, K. K. (2007). Getting and keeping children engaged with a constructionist design tool for craft and math. Unpublished doctoral dissertation, Georgia Institute of Technology, Atlanta, GA. Moss, G., Jewitt, C., Levačić, R., Armstrong, V., Cardini, A., & Castle, F. (2007). The interactive whiteboards, pedagogy and pupil performance evaluation: An evaluation of the schools whiteboard expansion (SWE) project: London challenge (Research Report No. 816). London: Department for Education and Skills, Institute of Education. Nacenta, M. A., Pinelle, D., Stuckel, D., & Gutwin, C. (2007). The effects of interaction technique on coordination in tabletop groupware. In GI ’07: Proceedings of Graphics Interface 2007 (pp. 191–198). New York: ACM Press. Rick, J., Harris, A., Marshall, P., Fleck, R., Yuill, N., & Rogers, Y. (2009). Children designing together on a multi-touch tabletop: An analysis of spatial orientation and user interactions. In Proceedings of IDC ’09. New York: ACM Press. Rick, J., & Rogers, Y. (2008). From DigiQuilt to DigiTile: Adapting educational technology to a multi-touch table. In Proceedings of TABLETOP ’08 (pp. 79–86). Los Alamitos, CA: IEEE. Rick, J., Rogers, Y., Haig, C., & Yuill, N. (2009). Learning by doing with shareable interfaces. Children, Youth & Environments, 19(1). Roschelle, J. (1996). Learning by collaborating: Convergent conceptual change. In T. Koschman (Ed.), CSCL: Theory and practice of an emerging paradigm (pp. 209–248). Mahwah, NJ: Lawrence Erlbaum Associates. Roschelle, J., Tatar, D., Chaudhury, S. R., Dimitriadis, Y., Patton, C., & DiGiano, C. (2007). Ink, improvisation, and interactive engagement: Learning with tablets. IEEE Computer, 40(9), 38–44. Weiser, M. (1999). The computer for the 21st century. SIGMOBILE Mobile Computing and Communications Revue, 3(3), 3–11.

Acknowledgements This work is part of the ShareIT project funded by the EPSRC, grant number EP/F017324/1. I would like to thank my collaborators in that project: (alphabetically) Victoria Bonnett, Kimberly Bryant, Sheep Dalton, William Farr, Rowanne Fleck, Caroline Haig, Amanda Harris, Samantha Holt, Eva Hornecker, Paul Marshall, Shems Marzouq, Richard Morris, Nadia Pantidi, Yvonne Rogers, and Nicola Yuill. We thank MERL for loaning us the DiamondTouch.



13

Where Is the Script? Support for Computer-Based Small Group Collaboration and Plenary Activities by Means of Small-Group and Classroom Scripts Christof Wecker, Ingo Kollar, Sybille Langer, Frank Fischer, University of Munich, Department of Psychology, Leopoldstr. 13, 80802 Munich, [email protected], [email protected], [email protected], [email protected] Abstract: This paper presents preliminary results from two empirical studies from an ongoing project on ways to integrate small group collaboration in classroom instruction. The goal of the project was to find ways to develop learners’ internal scripts to guide their collaborative activities. In study 1, the effects of an unfaded and a faded small group collaboration scripts on domain-specific knowledge were investigated. Learners in the condition with the faded script significantly outperformed learners from a control condition in the domain-specific knowledge test. This could be interpreted as an indication that learners built up an internal script which they used after the fading occured. In study 2, we compared a more structured classroom script involving phases of teacher modeling and reflection to a less structured classroom script. Preliminary evidence indicates that the learners in the more structured classroom script condition benefit more and build up appropriate internal scripts. A current issue in CSCL research is the question how to find the right balance between necessary guidance for collaborative activities by means of collaboration scripts and flexibility in these activities. In our contribution we build on the assumption that any collaborative activity is guided by scripts, whether external or internal (Schank & Abelson, 1977; Kollar, Fischer & Slotta, 2007). Against this background the question of scripted versus free collaboration shifts: Do the scripts that guide collaborative activities lead to fruitful collaboration, and how can learners be supported in acquiring appropriate internal scripts that are the prerequisites for collaboration “freed” from external scripting? We present two studies that investigate ways to build up these prerequisites: In study 1, the effects of an unfaded and a faded external collaboration script on learning outcomes were studied in a school classroom context over a period of several weeks. In study 2, we investigated ways to integrate computer-based small-group collaboration into regular classroom instruction by means of classroom scripts, thereby taking advantage of the teachers’ competence for developing learners’ internal scripts, e. g. by phases of modeling and reflection at the plenary level.

Study 1: Effects of unfaded and faded small group collaboration scripts for collaborative online search In study 1 we investigated the effects of these small-group collaboration scripts and of the fading of these scripts (cf. Wecker & Fischer, 2007) on the students’ scientific literacy. 111 ninth-graders from three urban high schools participated in an inquiry learning curriculum which spanned seven regular biology lessons conducted by their regular biology teacher. The topic of the instructional unit covered during this study was genetics and gene technology. The inquiry-oriented curriculum was centered around an ongoing discussion about advantages and dangers of so-called “green” gene technology, i. e. the genetic modification of plants for agricultural purposes. During the whole study, each student had the same laptop computer available. The unit started with an introductory lesson in which students acquired background knowledge about inheritance from an online “library” comprising fundamental information about the topics of genetics and gene technology. After this lesson, there were three inquiry and discussion cycles that lasted for two lessons each. In each cycle, student dyads were first asked to browse through an online library with information about genetics and genetic engineering that was developed in close collaboration with practitioners and pedagogical content knowledge experts. In a second phase, they collaboratively sketched an initial argument (e.g., that eating genetically modified food is dangerous for health) and searched the web to elaborate or change their initial argument. This phase was supported by a software tool which allowed for collaborative web browsing during their online search, i. e. both learning partners of a dyad always saw the same web sites on their laptop screens, no matter who of them navigated. During this phase, different conditions with respect to support by collaboration scripts were implemented. The final phase of each cycle was a classroom discussion in which the students presented and discussed their arguments and eventually experienced the need to further substantiate their views in subsequent cycles. During the second phase of each inquiry cycle, the manipulation of collaboration support by means of computer-based collaboration scripts took place. In both the condition with unfaded and the condition with faded small group collaboration script, there were two roles (A and B) in each dyad that switched after returning



14

to the search engine from any other web page encountered during the search activities. The two versions of the collaboration script were implemented as complementary text prompts in browser panels on the left hand side of the computer screens of both group members (see figure 1). At the beginning of the collaborative online search phase in each of the three cycles the students received prompts in the scaffolding area that were triggered by clicking on “search” on the project web page that was set as the default start page in each browser. Learner A was required to come up with an argument and present it to B, describe the argument chosen together with B in a note field, discuss the formulation with B and provide feedback about B’s tasks. Learner B also was given the task to come up with an argument and present it to A, write a sketch of the information necessary to support this argument (for instance, a study showing that the effects claimed have actually been observed, or a genetic explanation how the claimed effects could be possible), comment on A’s argument formulation and discuss his sketch of necessary information with A.

Figure 1: Implementation of the small group collaboration scripts in S-COL. When the learners jointly moved to the search query form of a search engine, A was prompted to come up with a set of search terms and present them to and discuss them with B, while B had the task to first recall the information they had decided to look for, and comment on A’s suggestions for the search terms with respect to their likelihood of yielding suitable as well as inappropriate hits. At the results page, the scaffolding area asked learner A to scan through the list of results and evaluate them with respect to relevance, credibility, scientific support and impartiality on the basis of the title, the text excerpt and the URL provided, and to suggest the page to visit. Learner B was prompted again to recall the information they were looking for and to comment on the pages learner A suggested with respect to their relevance, credibility, scientific support and impartiality. When the group navigated to one of the web sites found by means of the search engine, learner A received prompts in the scaffolding area on how to localize the required information on the web site (e. g. by using the search function of the browser or the one on the page), to present the information in his or her own words to learner B and to discuss with B how to proceed. Learner B had the task to suggest to A to return to earlier steps of the search if he or she had the impression that the current page was not promising and to comment on the information presented by A with respect to its relevance, credibility, scientific support and impartiality. B’s task in this phase also comprised the documentation of the information retrieved (including the URL as a reference) and the discussion of the next step to take. When the dyad agreed to finalize their online search, any of the two members could click on a button that invoked support for the formulation of an argumentation to be used in the subsequent plenary discussion. The prompts asked B to summarize all the information collected during the previous online search in his or her own words and compose a written summary of this argumentation based on A’s comments. Learner A received prompts to comment on B’s spoken summary with respect to its persuasiveness and possible counterarguments



15

and to provide suggestions for improvement for the written version of their argumentation. If they had the impression that they needed further information, they could return to the online search. Before and after the seven lessons, tests for several aspects of scientific literacy that were developed on the basis of official educational standards were administered. A MANOVA with knowledge about concepts, factual knowledge and understanding as dependent variables revealed a significant main effect for condition (F(6,206) = 2.28; p < .05; Eta² = .06), especially with respect to knowledge about concepts (F(2,105) = 3.83; p < .05; Eta² = .07): Descriptively, the faded script condition was superior to the constantly scripted condition, which was superior to the unstructured control condition. However, there was also substantial between-class variance. These results provide evidence that the effects of small group collaboration scripts observed in the laboratory (e. g. Weinberger, Ertl, Fischer & Mandl, 2005; Kollar et al., 2007; Stegmann, Wecker, Weinberger & Fischer, 2007) also hold in applied settings involving real-world classrooms and extended time frames, at least if the collaboration script is faded as soon as it is no longer needed. They could be interpreted as an indication that learners built up an internal script for their collaborative activities and employed this internal script in a functional way after the fading occurred.

Study 2: Classroom scripts for the integration of small group collaboration and plenary activities The results of study 1 had shown that small group collaboration scripts may have positive effects on the acquisition of knowledge even under regular instructional conditions in classrooms. We wondered, however, how these effects could be even increased by a more systematic integration of small group and plenary activities that takes advantage of the particular affordances of a classroom situation, e. g. the presence of the teacher as well as other collaborating small groups. Dillenbourg and Jerman (2007), for example, have suggested combining small group collaboration with other classroom activities. In their opinion there should also be phases in which the students have the opportunity to work individually or phases in which the whole class interacts in a plenary setting. In her “synergistic scaffolding” approach, Tabak (2004) postulated that an appropriate combination of different scaffolds may have particularly benefical effects on the learning outcomes. She argued that different scaffolds may augment each other and interact in way that support the implementation of the task more than a single scaffold because many tasks consist of “a mélange of knowledge, skills, and values” (p. 318). Our approach to the integration of support on different social levels in the classroom consists in an extension of the framework for the description of small group collaboration scripts developed by Kollar, Fischer and Hesse (2006). According to this framework, a small group collaboration script can be characterized (1) by the instructional goals it is supposed to serve, (2) by the learning activities it specifies, (3) by the roles it assigns these activities to, (4) by the sequencing of these activities and (5) by the way the whole script is represented. In the larger social context of a classroom, activities can take place at different social levels. The separation among these levels is based on the “intended audience” of the activities in question: Whether a phase of a classroom script takes place at the individual, small group or plenary level depends on whether the activities belonging to this phase (a) are activities of communication directed at an audience (plenary and small group level) or not (individual level) and (b) whether the intended audience, i. e. the group of persons who are supposed to perceive and react to the activity in question, includes either all members of the class (plenary level) or just a subset of them (small group level). Accordingly, a classroom script can be characterized by a matrix of social levels, phases and roles that allocates each activity expected within the script to a particular social level, a particular phase and a particular role. Within this framework, the learning settings employed in our own prior research as well as many other studies can be described as a classroom script with little, if any, activity at the plenary level (cf. figure 2a). In study 2 we investigated whether learners can benefit from a more structured classroom script that to a larger extent allocates activities to the plenary level, e. g. the modeling of collaborative activities or phases of joint reflection on prior experiences. For this purpose we developed a more structured classroom script (cf. figure 2b). According to this classroom script, he teacher demonstrates the first steps of an expert online-search strategy in the first learning cycle (economic aspects of genetic engineering) before the first search phase. At the beginning, this modeling is conducted jointly with a student. It takes place at the front of the classroom, and the screens of both laptops are projected to the wall via a projector. Thus the whole class can follow the activities of the teacher and the student collaborating on a joint online search task. Later, the teacher withdraws from the joint modeling activities and is substituted by another student (cf. Palincsar & Brown, 1984). The last step of the online-search, i. e. the refinement of the argumentation, is modelled during the online-search in the small-groups in lesson 4. Each step of the collaborative online search strategy was modeled a second time during the project. In addition to these modeling phases, phases of reflection were inserted after the plenary discussions in each cycle. In these phases, students were ask to reflect which arguments had been successful in the discussion and which failed, what the features of these two sets of arguments were and what the could conclude from these insights for there subsequent learning activities in the following cycles.



16

a) less structured classroom script plenary plenary level …

small group level

t lesson 3

lesson 5

lesson 4

b) more structured classroom script

plenary level …

…

small group level

t lesson 3

lesson 4

introduction

sketch of argument

online library

selection of keywords

discussion

evaluation of results

reflection

localization of information on a single web page

lesson 5

refinement of argument Figure 2: Two differently structured classroom scripts. This excerpt shows the first of three learning cycles. The horizontal lines represent the social level at which the activity takes place (plenary or small group level). The different shades symbolize the type of activity. In study 2, we investigate the effects of classroom scripts of different degrees of structure. We expect that a more structured classroom script with activities more evenly distributed over social levels leads to higher gains in the learning outcomes in comparison to a less structured classroom script with activities distributed less evenly over social levels. In the design of this study we compare two conditions: One with the classroom script from study 1 (figure 2a) and one with the more structured classroom script just described (figure 2b). No small group collaboration scripts were implemented in study 2. As in study 1, the learners in each dyad were connected by the tool described above during their collaborative online searches. Learning outcomes are measured the same way as in study 1. Data collection is still taking place. Preliminary evidence indicates a higher degree of commitment of the students to their tasks as well as a higher level of collaborative activity.



17

Conclusions In this contribution we presented two studies of ways to structure learners’ collaborative activities without providing external scripting throughout their collaboration. Both the fading of a small group collaboration script and a more structured classroom script involving phases of teacher modelling and joint reflection on a plenary level can be regarded as ways to build up internal scripts that can guide students’ activities in phases in which they collaborate “freely”. With respect to faded small group collaboration scripts, study 1 indicated that this approach can be effective in obtaining positive effects on learning outcomes in school classroom contexts over extended periods of time. With respect to classroom scripts, we presented an extension of our earlier descriptive framework for small group collaboration scripts that can capture the peculiarities of collaboration on different social levels. We argue that even in such contexts, the activities of the people collaborating in a classroom, i. e. students and teacher, are guided by internal scripts that can be more or less functional. We characterized one version of a more structured classroom script by means of this extended framework and reported some promising preliminary findings concerning its effectiveness for integrating small group collaboration and plenary activities in the context of regular classroom instruction in order to foster scientific literacy. We suggest to further explore the integration of “free” and “scripted” computer-supported small group collaboration into whole class activities from the perspective of classroom scripts that guide the participants in their activities, whether they receive external computer-based scripting or not.

References Dillenbourg, P., & Jerman, P. (2007). Designing integrative scripts. In F. Fischer, I. Kollar, H. Mandl, & J.M. Haake (Eds.), Scripting computer-supported collaborative learning: Cognitive, computational and educatinal perspectives (pp. 275-301). Dordrecht: Springer. Kollar, I., Fischer, F., & Hesse, F. W. (2006). Computer-supported collaboration scripts - a conceptual analysis. Educational Psychology Review, 18(2), 159-185. Kollar, I., Fischer, F., & Slotta, J. D. (2007). Internal and external scripts in computer-supported collaborative inquiry learning. Learning and Instruction, 17(6), 708-721. Palincsar, A. S., & Brown, A. L. (1984). Reciprocal Teaching of comprehension-fostering and comprehensionmonitoring activities. Cognition and Instruction, 1(2), 117-175. Schank, R. C. & Abelson, R. P. (1977). Scripts, plans, goals and understanding: An inquiry into human knowledge structures. Hillsdale, NJ: Erlbaum. Stegmann, K., Wecker, C., Weinberger, A., & Fischer, F. (2007). Collaborative argumentation and cognitive processing - An empirical study in a computer-supported collaborative learning environment. In C. Chinn, G. Erkens & S. Puntambekar (Hrsg.), Mice, minds, and society: CSCL 2007 (pp. 661-670). ISLS. Tabak, I. (2004). Synergy: A complement to emerging patterns of distributed scaffolding, The Journal of the Learning Sciences, 13, 305-335. Wecker, C. & Fischer, F. (2007). Fading scripts in computer-supported collaborative learning: The role of distributed monitoring. In C. Chinn, G. Erkens & S. Puntambekar (eds.), Mice, minds, and society: CSCL 2007 (pp. 763-771). ISLS. Weinberger, A., Ertl, B., Fischer, F., & Mandl, H. (2005). Epistemic and social scripts in computer-supported collaborative learning. Instructional Science, 33(1), 1-30.



18

Structured activities in CSCL: a case study Pozzi F., Persico D., Istituto Tecnologie Didattiche – CNR, Via De Marini 6, Genoa Email: [email protected], [email protected] Abstract: This paper is rooted in the CSCL research field and aims to contribute to the open debate on how much an instructional designer should scaffold a learning activity with prescriptions, rules and procedures to be followed by students, as opposed to set up environments for “free collaboration”. By analyzing three online activities with different levels of “structuredness” (namely a Discussion, a Role Play and a Jigsaw) as they had been proposed within two online “twin” courses, it is possible to make some interesting reflections as far as the potential of each activity to foster the participative, the social, the cognitive and the teaching dimensions of the learning process.

Introduction In his well known paper concerning over-scripting CSCL, Dillembourg (2002) posed a number of questions concerning the way the research community goes about research concerning scripting in CSCL. In a nutshell, he puts two major issues on the table. The first is whether the whole idea of using scripts (or simply strategies and techniques) to facilitate interaction doesn’t impose a structure that hinders true collaboration rather than fostering it. The second is whether it makes sense to pursue what he calls “the golden script”, and whether it wouldn’t be better to strive to understand why some scripts are effective. The main standpoints of this paper derive quite directly from our considerations on these two questions. It may be true that sometimes scripting is too directive and prevents learners to really find their own way around a certain subject. This is probably the case for professional development and, more generally, in the context of informal learning and communities of practice. However, there still is a need for formal learning, that is, institutions in charge of organizing, setting up and running learning events of some kind. Designing these events and the learning environments where they will take place, entails making some decisions on behalf of the learners, mostly concerning the objectives to be achieved and the contents to be covered. This is something that is specifically required in formal learning, especially at the initial stages, when the learners are still quite unaware of new subjects. The role of scripts in these contexts is that of proposing certain objectives rather than others, and suggesting ways to achieve them that are (hopefully) more effective than others. This brings us to the second issue. It seems to us that there should be no quest for the “golden script”, but rather, that there is a need to understand better the qualities of some kinds of tasks and the pros and cons of certain types of scripts in order to support decision making in instructional design. The point is not that some scripts are effective and others aren’t, but rather that some must suit best some situations and others might be more fit for other contexts. This is to say that instructional design of learning events should be based on a “fit to purpose” principle, and this entails taking into consideration the learning objectives, the features of the target population, the nature of the content to be learnt. In our view, even the choice of not using any script is a possible, and sometimes suitable, instructional design decision, which is likely to work, for example, with fully fledged learning communities where mature, self-regulated learners are able to set their own individual and social objectives and pursue them quite autonomously. However, even self-regulation entails a set of competences that can and need to be fostered and nurtured, and this is particularly true for self-regulation in online environments, that include a set of abilities that aren’t always owned by today’s students, in spite of their ever increasing digital skills (Delfino et al, 2008). Starting from these general considerations, this study was aimed at investigating what kind of online learning processes are produced by three activities based on strategies which display different degrees of structuredness (namely a Discussion, a Jigsaw and a Role Play). The analysis of these activities in two real online contexts, whose results are presented in the following, allows to shed light on the two above mentioned research questions and provides interesting inputs for further research.

Context and method of the study The present study has been carried out in the context of “SSIS”, the Italian system for teacher training. Within this context, in 2007 two twin courses were run respectively in Liguria and Veneto on the issue “Educational Technology” (hereafter called “TD-SSIS Liguria” and “TD-SSIS Veneto”). The objectives and contents addressed within the two courses were exactly the same, so the only variable differentiating the two contexts was the community of students. In both cases the large size of the learning communities required the creation of smaller “classes” working in parallel (20/25 persons each), so to allow collaboration. In this study we



19

concentrate on two of the classes: the class of TD-SSIS Liguria, composed of 21 students, and the class of TDSSIS Veneto, which consisted of 24 students; the two classes were tutored by the same tutor. In order to scaffold collaboration within the two courses different activities were proposed, ranging from “highly structured” activities (i.e. a Jigsaw), to “medium structured” activities (a Role Play), to “low structured” activities (a Discussion). In particular, the first module, introduced by a “Socialization” phase aimed at making students familiarize with the other members of the community and with the CMC system used (First Class for TD-SSIS Liguria and Moodle for TD-SSIS Veneto), was devoted to the study of three e-learning models and was based on a Jigsaw. In this activity, the students were divided into three groups (7/8 persons each) and then each group was asked to analyze in detail one of the three models. Afterwards, students were rearranged in new, heterogeneous groups and then asked to collaboratively solve a problem of e-learning design, so that each student was triggered to put forward the competences acquired in the previous phase to help the group solving the problem. The Jigsaw lasted 3 weeks. During the second online module, students remained aggregated as in the last part of the Jigsaw. This activity (3 weeks) was based on a Role Play and dealt with the concept of “webquests”. In particular, learners were asked to pretend to be a group of teachers, whose school principal had asked them to analyze and evaluate a certain number of webquests. Since the Role Play imposed the analysis of the webquests to be carried out from pre-defined perspectives, i.e. by playing specific roles, at the beginning of the activity each student/teacher chose a role from a list of characters, including the “school principal”, the “rapporteur”, the “techno-sceptical”, the “bureaucrat”, the “defeatist”, etc. During the activity, the webquests were discussed and a common evaluation was negotiated by the students/teachers, who argued their position according to their role. At the end of the activity, the students/teachers produced a shared document containing the analysis, which took into account the different viewpoints played. Finally, in the third online module, called “Using blogs in educational settings”, students (who maintained the same groups of the previous module) were asked to develop a personal blog and then to collaboratively design an educational blog through a Discussion (3 weeks). In order to investigate the nature of the interactions occurred while performing the proposed online activities, an evaluation model was used, which had been previously developed by the authors as an adaptation of Garrison et al.’s model of Communities of Inquiry (1999) and had been extensively used to assess similar online experiences (Pozzi et al. 2007; Persico et al., 2009). The model considers four dimensions as those characterizing a learning process in CSCL contexts, namely the participative, the cognitive, the social and the teaching dimensions. The main advantage of using this model lies in the fact that it allows us to identify the potential of different activities (and their related scripts) with respect to the four components and therefore allows to relate this potential to the objectives and the abilities that one intends to develop while designing online learning. So, for example, if an activity turned out to be more promising in terms of social presence, while another worked better from the point of view of sustaining the cognitive presence, then instructional designers could draw their conclusions about what strategies best suit the objective of socialization, rather than that of focusing on contents. In the model, each dimension is defined by a set of relevant indicators, of both quantitative and qualitative nature. Those indicators are meant to express the actual manifestations of the four dimensions in the learning community. In particular, indicators of the participative dimension include the number of “active actions” by members of the learning community (in terms of sent messages, uploaded documents, etc.), the number of “reactive actions” (e.g. reading messages, downloading documents, etc.), as well as the level of continuity in participation across time. Indicators of the social dimension include clues of affection (which is typically revealed by expressions of emotion or intimacy, humour or irony, presentations of personal anecdotes) and group cohesion (such as vocatives, expressions revealing group-self efficacy, use of inclusive pronouns to refer to the group, phatics, salutations). As far as the cognitive dimension is concerned, the model makes a distinction between clues of individual and group knowledge building, by assuming that a collaborative activity in CSCL contexts typically requires a first stage entailing a personal re-elaboration of contents and the expression of individual points of view, and a second stage devoted to discussion and negotiation for collaboratively constructing shared meanings and common interpretations of reality. Moreover, according to the model, the cognitive dimension also encompasses meta-reflection, that is to say that students reflections on, and evaluations of, the learning process itself are considered an important component of the cognitive process. Lastly, indicators of the teaching dimension include taking care for organizational aspects, facilitating discourse and providing direct instruction. The following table summarizes the main indicators of the four dimensions: Participative

Social

P1 - Active participation P2 - Passive participation P3- Continuity S1 - Affection


Cognitive

Teaching


20

S2 - Cohesion C1 - Individual knowledge building C2 - Group knowledge building C3 - Meta-reflection T1 - Dealing with organizational matters T2 - Facilitating discourse T3 - Providing direct instruction Table 1 – Main indicators of the evaluation model

To gauge these indicators, an analysis of all the messages exchanged by the students during the activities was carried out. In particular, the indicators concerning the participative dimension have been gathered directly from the data tracked by Moodle, whereas the analysis of the cognitive, the social and the teaching dimensions is based on a “manual” content analysis. The manual coding procedure consisted of two coders reading each message and systematically identifying the frequency of given keywords or patterns or even expressions that are believed to reveal a feature of the communication act, and finally classifying each of them as belonging to a certain indicator category (Persico et al., 2009). The unit of analysis chosen by the two coders was the “unit of meaning” i.e. each message was split in parts defined on the basis of “consistent themes or ideas” (Henri, 1992), so that each unit could be classified as belonging to a certain indicator. In this study, the corpus of the coded messages was of 1164 (total number of messages exchanged by students during the three activities of the two courses). The inter-rater reliability was calculated on a sample of 110 messages and resulted in 0.87 (Holsti coefficient) and 0.83 (percent agreement).

Results In the following, data are reported concerning the participative, the social, the cognitive and the teaching dimensions, as they developed during the execution of the Jigsaw, the Role Play and the Discussion respectively in TD-SSIS Liguria and in TD-SSIS Veneto. In TD-SSIS Liguria, as far as the participative dimension is concerned, the students sent the highest number of messages during the Discussion, while in the Jigsaw and the Role Play the number of sent messages is lower and quite similar to each other. Besides, the mean number of messages per student is quite high in all of the three activities and the percentage of read messages is satisfactory too. In TD-SSIS Veneto the overall number of messages and the mean number of messages per student is lower than in TD-SSIS Liguria, but the overall trend is quite similar, since again the Discussion resulted to be the most participated activity, followed by the Jigsaw and then the Role Play. Unfortunately, due to the features of the interaction platform used, it was not possible to collect data concerning passive participation in TD-SSIS Veneto. Going further, we looked at more qualitative data: Figure 1 and 2 contain data concerning the social, the cognitive and the teaching dimensions obtained by the three activities in the two courses. In particular, by looking at Figure 1 (TD-SSIS Liguria), it is interesting to note that the indicators distribution seems to be similar regardless of the activity proposed, namely: S1 of the social dimension (affection) is always quite low and especially the value of S1 in the Jigsaw and that in the Role Play resulted very close; in contrast S2 (cohesion) reached the highest values in all the three activities and again values of the Jigsaw and the Role Play are very similar. As far as the cognitive dimension is concerned, C1 (individual knowledge building) is always lower than group knowledge building (C2), whereas C3 (meta-reflection) is almost absent in all of the three activities. Values of C1 for the three activities are again quite close and the same applies to values for C2 and C3. Finally, the three indicators of the teaching dimension: T1 (Dealing with organizational matters), T2 (Facilitating discourse) and T3 (Providing direct instruction) are more or less all at the same level with the only exceptions of T1 in the Discussion and T2 in the Role Play, which both reached higher levels.



21

Indicators of the three activities (TD-SSIS Liguria 2007) 350

326

300 250

228

Role Play

205

200

Jigsaw

179 163 142

150

Discussion 121

117 97

107

100

87 75

72 56

71

6364

50

121 77 67 52

2829 6

0 S1

S2

C1

C2

C3

T1

T2

T3

Figure 1. Social, cognitive and teaching dimensions in TD-SSIS Liguria As already mentioned, TD-SSIS Veneto (Figure 2) registered an overall lower number of messages (in spite of the number of students being slightly higher). Still, again here there is a common trend in all the three activities. In particular, a bias is registered in the social dimension between affection (S1), which is quite low, and cohesion (S2), which is very high (with the only exception of the Role Play, whose S2, even if higher than S1, is sensibly lower than S2 of the other two activities). In all the three activities C1 is lower than C2, with the Jigsaw developing the highest cognitive dimension, followed by the Discussion and then by the Role Play. Again here meta-reflection (C3) was not particularly developed by none of the proposed activities. As far as the teaching dimension is concerned, again here differences among T1, T2 and T3 are not so evident across the three activities, with the exception of T2 during the Discussion and T3 in the Role Play, which were sensibly lower than those of the other two activities.

Indicators of the three activities (TD-SSIS Veneto 2007) 180

167 156

160

148

140 121 113

120

Role play

86

81 76

80 61

60

102

99

100

68

62 60

59

56

69

47

44

Jigsaw Discussion

44

36

40 20

5

11

18

0 S1

S2

C1

C2

C3

T1

T2

T3

Figure 2. Social, cognitive and teaching dimensions in TD-SSIS Veneto

Discussion and Conclusions In the following we draw some considerations based on the cross-analysis of the results just described.



22

First of all, it is worthwhile stressing the difference between the levels of participation in Liguria and in Veneto: it is evident that in the former course we were in presence of a pro-active group, while in the latter the group was much less keen on writing. Nonetheless, it should be noted that in both the courses, despite some differences in the values assumed by the indicators, these seem to follow the same trend independently of the structure of the activity used. In particular, the group cohesion always shows high values, while affection tends to be much lower; at the same time, it seems that individual knowledge building is on average quite low during this kind of activities, while group knowledge construction is usually high (and this is reasonable as we are in a collaborative learning context), whereas meta-reflection indicators are quite scarce in all of the three proposed activities. Besides, it seems that all the activities have supported adequate levels of teaching dimension with no particular predisposition for one or the other aspect of it. It should also be noted that individual knowledge building and meta-reflection are latent variables and therefore, as De Wever et al (2006) pointed out, their low levels might not necessarily mean that they did not take place but that, simply, they were not made explicit in the student messages. Together with such a general “common trend”, one should also consider that each activity in our study revealed a specific ability as for supporting one or another dimension, namely: the Discussion resulted more participated by both the groups and the one which mostly fostered the social dimension; the Role Play always obtained the lowest levels for C1, as well as for C2 and C3 while it seems to be quite good as for the teaching dimension (especially for the aspects of discourse facilitation which concern taking responsibility of the group learning process); the Jigsaw obtained in both the courses the highest level of group knowledge building. This leads us to think that, if on the one hand there is no activity that - in principle - is better than others, on the other hand, the task and the strategy used may have a different impact on the different dimensions, i.e. a low structure seems to foster more the social dimension (as people feel more free to express their own impressions and feelings), whereas a higher degree of structuredness seems to have more positive effects on the cognitive dimension. By taking into account these final remarks, it is also worthwhile noting that some of the data in our study may have been even affected by the order in which the activities were proposed; for example the fact that the Discussion was proposed as the last activity of the course, when groups have reasonably already consolidated a certain sense of community and people have already become familiar with the technology and the collaborative working methods, may have (at least partially) affected the results. It would therefore be interesting to carry out further investigations to ascertain whether there are significant changes in the distribution of the indicators when the same kinds of activity are used at different stages of the course. Thus, overall, this study seems to suggest that choosing too strictly between structured and nonstructured activities within a course may not be particularly useful; rather, a more effective decision making criteria could be that of alternating structured with non-structured activities, depending on which dimension needs to be fostered and on the phase of the course (at the beginning, mid-way through or at the end). For example, at the beginning of a course, when it is usually important to build a sense of community, free discussion or another weakly structured activity is preferable, while at stages where it is important to encourage the group to take responsibility for the learning process, the Role Play may be more suitable. Similarly, the choice could fall on the Jigsaw when knowledge building is the primary focus.

References Dillenbourg, P. (2002). Over-scripting CSCL: The risks of blending collaborative learning with instructional design. In P. A. Kirschner (Ed.), Three worlds of CSCL. Can we support CSCL (pp. 61-91). Heerlen, NL: Open Universiteit Nederland. Delfino, M., Dettori, G., Persico, D. (2008). Self-Regulated learning in virtual communities, Technology, Pedagogy and Education, 17(3), 195-205. Garrison, D. R., Anderson, T. & Archer, W. (1999). Critical inquiry in a text-based environment: computer conferencing in higher education. The Internet and Higher Education, 2(2-3), 87-105. Henri, F. (1992). Computer conferencing and content analysis. In A. R. Kaye (Ed.), Collaborative Learning Through Computer Conferencing (pp. 115-136). The Najaden Papers, New York, Springer. Pozzi, F., Manca, S., Persico, D. & Sarti, L. (2007). A general framework for tracking and analyzing learning processes in CSCL environments, Innovations in Education and Teaching International, 44(2), 169-180. Persico, D., Pozzi, F. & Sarti, L. (2009). A model for monitoring and evaluating CSCL. In A.A. Juan, T. Daradoumis, F. Xhafa, S. Caballe & J. Faulin (Eds.), Monitoring and Assessment in Online Collaborative Environments: Emergent Computational Technologies for E-learning Support. IGI Global. De Wever, B., Shellens, T., Valcke, M., & Van Keer H. (2006). Content analysis schemes to analyze transcripts of online asynchronous discussion groups: A review. Computers and Education, 46, 6-28.



23

Emotional Pedagogical Agents for Collaborative Learning: an Application for Children with Learning Difficulties Konstantina Chatzara, Charalampos Karagiannidis Department of Special Education, University of Thessaly, Volos, Greece [email protected], [email protected] Abstract. This paper presents our work in exploring emotional pedagogical agents in elearning environments for children with learning difficulties (LD). We argue that the learning environment may become much more natural and realistic through these agents, and it can offer a more efficient and effective communication between the user (learner) and the machine (learning environment). Most of the existing adaptive learning environments do not take into account the emotional state of the users, as well as the relations between the learners that use the system. Children with LD are especially sensitive regarding their self esteem in the school environment. Emotional pedagogical agents might help to overcome this problem through their ability to adjust their behavior to the individual‘s behavior and emotional state.

Introduction Pedagogical agents are attracting considerable attention during the past few years, since they have the potential to facilitate a more natural communication between users and computers. Agents are being used in many learning applications and services, promising to open up exciting new possibilities for learning (e.g. agents that help users escape from errors, comfort them when they are disappointed of their competence level, etc). The empathy that is communicated through the agents can result in learning environments which can engage and motivate learners more effectively (especially children, who seem to develop an immediate and deep affinity to such characters) (Forbus and Feltovich, 2003). Some of these agents are capable of demonstrating emotional reactions, and several models have been designed in order to represent this emotional process (Sehaba, 2007). Animated lifelike agents can facilitate rich learning interactions (in many cases, similar to face-to-face ones). Overall, the combination of animated agents that show empathy for the users have the potential to improve the quality of education (especially for children). Agents also have the potential to also cater for the requirements of different (categories of) users and have been therefore used in many adaptive (learning) applications (Ganzha at al, 2007). This paper discusses the use of animated, emotional pedagogical agents for children with learning difficulties. Learning disability refers to a disorder in one or more of the basic psychological processes involved in understanding or in using language, spoken or written which may manifest itself in an imperfect ability to listen, think, speak, read, write, spell or to do mathematical calculations (Mercer et al, 1996). Learning disabilities are specific, not global impairments. As such, they are distinct from intellectual disabilities and include a wide range of disorders that are associated with learning. They are related with oral language, reading, written language and mathematics skills. Users are heterogeneous, come from different backgrounds and have different level of knowledge in the use of information systems. This is more obvious when we are referring to children with LD. Each pupil faces different difficulties in the learning procedure and applications should be able to accommodate all users in order to avoid negative emotions that might leave the users disorientated, feeling stressed out and disappointed, end up giving up the learning procedure to avoid further annoyance. In order to cater to different user needs, information systems can be tailored manually by the user or automatically by the system. This paper describes our work in emotional pedagogical agentσ and how they can affect humancomputer communication in order to help children with LD. We develop an agent which adapts its emotions based on the student’s emotions. Our aim is to investigate whether and to what extent it can help these students to improve their self esteem that is related with the learning procedure and help children learn in a more productive and efficient way. A pilot study has already been conducted, and a large scale experiment is planned to test the effectiveness of the agent.

Related Work A number of models have been developed to simulate emotions in artificial intelligence (Li & MacDonnell, 2008). Several researchers have also demonstrated that the presence of empathic emotions in embodied computer agents can have significant positive effects on users’ opinions of that agent (Brave et al, 2005; Prendinger & Ishizuka, 2005; Kim, 2005). Some examples of the previous research are Bates’ OZ project (Reilly & Bates, 1992; Bates et al, 1992a; Bates et al, 1992b), Cathexis Model, Elliot’s Affective Reasoner,



24

Rousseau’s CyberCafe, TAPA’s project. In most of these attempts many interesting aspects of emotions were addressed; however the underlying models still have some limitations. Even though most of the experiments employed Ortony’s emotional synthesis process (Ortony et al., 1988), which emphasized the importance of expectation values, the models did not attempt to simulate the dynamic nature of expectations. The Cathexis Model uses a multi-agent architecture and it simulates emotional responses but again did not incorporate adaptation in modeling emotions. Another multi-agent model, called Affective Reasoner, was developed by C. Elliot (Elliot, 1992). In the Affective Reasoner project, agents portray twenty-four different emotions, including joy, disappointment, resentment and boredom and can generate about 1200 different emotional expressions that correspond to different emotional states. Even though Elliot’s model presents an interesting simulation describing emotion generation, emotional expressions, and their role in interactions, the model still faces some problems. The model does not examine some important factors, such as conflicting emotion, the impact of learning on emotions or expectations, filtering emotions, and their relation to motivational states (El-Nasr, 2000). Blumberg and Galyean are developing believable agents that try to simulate real characters. These agents were designed to simulate different behaviors and emotional states. A learning model was developed but they did not link the learning algorithms they used with the emotional process (Blumberg & Galyean, 1995). The TAPA project is the one that is more relevant to our research objectives. It is the only one that is addressing children with learning difficulties (Mohamad et al, 2004). They used Interface agents to complement the emotional system to convey emotional expressions to the user and they used affective communication through touch-screen, speech recognition, mouse, keyboard, skin conductivity sensors, diverse virtual input options via GUI buttons and questions through the interface agents (generally from the user interface). In 2005 in China State Key Lab of CAD&CG, Zhigeng Pan and his team proposed believable brain architecture in order to allow synthetics characters to achieve high level of autonomy (Pan et al, 2005). In 2008 in the same institute Yang and his team proposes a comprehensive computational model of emotions that can be incorporated into the physiological and social components of the emotions (Yang et al, 2008). Only a few of the previous efforts include collaboration between students, and only the TAPA’s project addresses children with learning difficulties. Nevertheless the results regarding the use of emotions in computer systems are promising; in most experiments the interaction between the users and the application was more efficient, productive and enjoyable.

Current Progress The scope of this paper is to present a model for intelligent agents for learning environments that accommodates the emotional process in order to create believable agents that will act as a medium between the user and the machine in a naturalistic manner. Memory, experience, previous knowledge, transformative learning, adaptive behavior and appropriation are the key points that are taken into account in the realization of the model. Our work in exploring emotional pedagogical agents for collaborative e-learning is an area that is not very much addressed in previous research. We propose a model for the representation of emotions and personality dynamics in interactive e-learning environments. By introducing emotional reactions, the learning system becomes much more natural and realistic and offers a more efficient and effective communication between the user and the machine. Children are especially sensitive regarding their self esteem in the school environment. Especially children that face learning difficulties (LD) are much more vulnerable due to the particularity of their character which has been modified by the specific LD. Relations in classrooms for these children are crucial. Usually children tent to belong to groups that share same personality characteristics and leave out the ones that differ. These children are craving for acceptance by others, teachers and classmates. Their emotional state is extremely fragile and affects the learning procedure dramatically (Bender & Wall, 1994). We propose the use of an emotional caring agent that acts as a learning companion as well as an academic advisor. This pedagogical agent interacts with students via different ways such as facial expressions, voice, gestures and text. We use the OCC model (Paiva et al, 2005) of emotions inspired by appraisal theory by Scherer. Research that took place in MIT (Aist et al, 2002) addresses the emotions that are possibly relevant in learning: Anxiety-Confidence, Boredom-Fascination, Frustration-Euphoria, Dispirited-Encouraged, TerrorEnchantment. Table 3: Emotions that are communicated through the agent Anxiety-confidence Ennui-Faschination Frustration-Euphoria Dispirited-Enthusiasm Terror-Exitement

Anxiety Ennui Frustration Dispirited Terror

Worry Boredom Puzzlement Disappointed Dread

Discomfort Indifference Confusion Dissatisfied Apprehension

Comford Interest Insight Satisfied Calm

Hopefulness Curiosity Enlightment Thrilled Anticipatory

Confidence Faschination Euphoria Enthusiasm Excitement


Humiliated-Proud

Humiliated


Embarrassed

Self-conscious

Pleased

Satisfied

25

Proud

Embodied pedagogical agents which are highly expressive may increase students’ perceptions of trust, and these agents are perceived as believable, helpful, and concerned (Lee et al, 2007). The fact that the agent not only adapts its own behavior but adapts the whole interface of the application itself in order to suit to the individual’ s with LD, needs and preferences, creates a personalized version of the application which caters for the specific student. This happens as an active dialogue (graphical and verbal) between the agent and the user so he/she is always aware of the changes that are taking place in the application, therefore he/ she feels in control of the educational procedure. The e-learning system can be used by several students that have to provide a few personal details in order to use the system. The system uses the given information by the students in order to alter the user profiles that are feed in the system. The emotional pedagogical agent (EPA) collects this data and reacts accordingly regulating its behavior and the academic level for each pupil. The EPA uses facial expressions, hand gestures and body movements in order to communicate with the user. The agent is “talking” to the user through verbal and written language and provides empathic feedback to help him/her recover from negative emotions as well as encourages learners to overcome academic problems. (Lee, 2007).

Figure 4: Architecture model The Pedagogical Agent collects information about child’s emotion, intensity of motion, and student’ skills from the students in the form of written text that the user sends to the Analysis module. The Analysis Module collects information from users’ operations of the application by recording his/her actions e.g. puts in different variables users actions and by using an algorithm that corresponds to a five grades scale assign the emotional state of the user. The users performance is been timed so assumptions about users emotional state may be made e.g. if the user spends more time than expected in a given task the agent takes action and communicates with the user to ask him/her if there is a problem and gives him/her clues for the given task. The pedagogical agent uses an encouraging voice message that embeds emotional tones (empathic feedback) to prompt the child to express him/her self by writing messages to the agent. The agent gets the appropriate behaviours from the emotion expression database and voice database to express motion with encouraging voice in order for the user to feel that the agent empathise with him/her. All data, from recorded user’s operation and direct user’s input text is been collected in the Analysis Module and categorized. In Decision module the appropriate emotional reactions of the agent is decided and via the response module the emotional state of the agent is portrayed. Relations between students in a learning environment are very important. E-learning applications often can not accommodate interaction between students. Especially when we are referring to education for children with LD, interaction between them and other pupils is often problematic. Children often do not have the correct behaviour towards this special group of children; many times they underestimate them and look down at them. This adds one more problem to the already difficult learning procedure and might causes the resentment from both parts. Agents can be programmed to “show” the correct social behaviour and through them a channel of communication is opened to serve for better interaction among students in order to increase student’s self esteem. Each student that uses the system can choose the EPA they prefer to accompany them. In the end of each task in the learning procedure each user may leave messages regarding the specific application (problems he/she had to overcome, points not well understood, give hints and tips for better results). These messages in conjunction with the EPA of their choice are there for every student to see. The collaboration of students is



26

achieved through the exchange of users’ feedback through their agents. The data that has been imported from students to be used from other pupils is stored and categorized in collaboration module in order to be available on request from the next user.

Figure 5: Architecture model with collaboration

Conclusions and Future Work We expect that the presence of the emotional pedagogical agent (EPA) empathic responses as well as the emotion of collective work that might be achieved through the collaboration of EPAs of the users will make students obtain higher interest in the tasks. Our current work includes an experiment between a group of children that have LD and use EPA and a group of children that have LD and use EPA with collaborating functionalities. The experiment will be made in a group of 80 children in primary school for children with special needs. A small scale, pilot experiment has shown a positive perception of the agent and children seem to accept advices from other children’ agents. They were particularly interested in the messages the other children send them through their agents Although the sample size was not large enough to draw conclusions, most subjects indicated they preferred to interact with the pedagogical agent that expresses emotions rather than the one that was just instructing them in the learning procedure. Most subjects seem to like the collaborating functionalities. The preliminary results revealed that the caring agent may not only increase the interaction between students and the learning system, but also have positive effects on students’ emotions to make them more engaged in learning. Children with autism will be excluded in the first experiment. Previous work (Sansosti & Powell-Smith, 2008) showed that they do not correspond well to emotional reactions due to their difficulty to recognize emotions. This kind of communication may be perceived negatively for this group of children. The results will lead to conclusions regarding emotional agents and non emotional agents for students with LD, with and without collaborative functionalities.

References Aist, G., Kort, B., Reilly, R., Mostow, J., & Picard, R. W. (2002). Experimentally Augmenting an Intelligent Tutoring System with Human-Supplied Capabilities: Adding Human-Provided Emotional Scaffolding to an Automated Reading Tutor that Listens. In Proc. International Conference on Multimodal Interfaces (Pittsburgh, PA, USA, October 14-16, 2002), pp. 483-490 Bates, J., Bryan Loyall, A., & Scott Reilly, W. (1992a). An architecture for action, emotion, and social behaviour. Technical Report CMU-CS-92-144, School of Computer Science, Carnegie-Mellon University, Pittsburgh, PA, USA. Bates, J., Bryan Loyall, A., and Scott Reilly, W. (1992b). Integrating reactivity, goals, and emotion in a broad agent. Technical Report CMU-CS-92-142, School of Computer Science, Carnegie-Mellon University, Pittsburgh, PA, USA. Bender, W., N., & Wall, M., E. (1994). Social-Emotional Development of Students with Learning Disabilities. Learning Disability Quarterly, 17, 323. Blumberg, B., & Galyean, T. (1995). Multi-level direction of autonomous creatures for real-time virtual environments computer graphics. In Proc. SIGGRAPH ‘95 Conference (Los Angeles, CA, USA, August 6-11, 1995), pp. 47-54.



27

Brave, S., Nass, C., & Hutchinson, K. (2005). Computers that care: investigating the effects of orientation of emotion exhibited by an embodied computer agent. International Journal of Human-Computer Studies, 62, 161-178. Elliot, C. D. (1992). The Affective Reasoner: a process model of emotions in a multi-agent system. Northwestern University doctoral thesis. Evanston, IL, USA. Forbus, K. D., & Feltovich, P. J. (2001). Smart Machines in Education: The Coming Revolution in Educational Technology. Menlo Park, CA: AAAI Press/MIT Press, pp.269. Ganzha, M., Paprzycki, M., Popescu, E., Badica, C., & Gawinecki, M. (2007). Agent-Based Adaptive Learning Provisioning in a Virtual Organization. In Wegrzyn-Wolska, K. M., & Szczepaniak, P. (eds.), Advances in Intelligent Web Mastering (pp. 112–117), Springer, Berlin. Kim, Y. (2005). Empathetic virtual peers enhanced learner interest and self-efficacy. In. Proc. Workshop on Motivation and Affect in Educational Software (12th International Conference on Artificial Intelligence in Education, AIED 2005, Amsterdam, The Netherlands, July 18-22, 2005). Lee, T. Y., Chang, C. W., & Chen, G. D. (2007). Building an Interactive Caring Agent for Students in Computer-based Learning Environments. In Proc. International Conference on Advanced Learning Technologies (Niigata, Japan, July 18-20, 2007), pp.300-304. Li, L., & MacDonnell, S. (2008). An Emotion-based Adaptive Behavioural Model for Simulated Virtual Agents. The International Journal of Virtual Reality, 7, 59-64. Mercer, C. D., Jordan, L., Allsopp, D. H., & Mercer, A. R. (1996). Learning disabilities definitions and criteria used by state education departments. Learning Disability Quarterly, 19, 217-232. Mohamad, Y., Velasco, C. A., Damm, S., & Tebarth, H. (2004). Cognitive Training with Animated Pedagogical Agents (TAPA) in Children with Learning Disabilities. In Proc. Computers Helping People with Special Needs Conference (Paris, France, July 7-9, 2004), pp. 187-193. Ortony, A., Clore, G. L., & Collins, A. (1988). The Cognitive Structure of Emotions. Cambridge University Press. Paiva, A., Dias, J., Sobral, D., Aylett, R., Woods, S., Hall, L. E., & Zoll, C. (2005). Learning By Feeling: Evoking Empathy With Synthetic Characters. Applied Artificial Intelligence, 19, 235-266. Pan, Z., Xu, B., & Zhang, M. (2005). Building a Believable Character for Real-Time Virtual Environments. In Proc. 1st International Conference on Affective Computing & Intelligent Interaction (Beijing, China, October 22-24, 2005), pp. 683-690. Prendinger, H., & Ishizuka, M. (2005). The empathic companion: a character-based interface that addresses users’ affective states. Applied Artificial Intelligence, 19, 267-285. Reilly, W. S., & Bates, J. (1992). Building emotional agents. Technical Report CMU-CS-92-143, School of Computer Science. Sansosti, F. J., & Powell-Smith, K. A. (2008). Using Computer-Presented Social Stories and Video Models to Increase the Social Communication Skills of Children With High-Functioning Autism Spectrum Disorders. Journal of Positive Behavior Interventions, 10, 162-178. Sehaba, K., Sabouret, N., & Corruble, V. (2007). An Emotional Model for Synthetic Characters with nd Personality. In Proc. of the 2 International Conference on Affective Computing and Intelligent Interaction (Lisbon, Portugal, September 12-14, 2007), pp. 749-750. Seif El-Nasr M., Yen J., & Ioerger. T. (2000). FLAME-Fuzzy Logic Adaptive Model of Emotions, Autonomous Agents and Multi-Agent Systems, 3, 219-257. Yang, H., Zhigeng, P., Mingmin, Z., & Chunhua, J. (2008). The Knowledge Engineering Review, 23.



28

Supporting Collaboration at Different Levels in Computer Supported Education Manuel Caeiro, Luis Anido, Martín Llamas, Jorge Fontenla, Roberto Pérez University of Vigo, Spain, {Manuel.Caeiro; Luis.Anido; Martin.Llamas; JFontenla; RPerez}@det.uvigo.es Abstract: CSCL (Computer Supported Collaborative Learning) still lacks of a clear integration of the different collaboration modes, namely: synchronous and asynchronous, structured and unstructured or predefined and ad-hoc ways of communication, co-operation and coordination. Educational Modelling Languages (EMLs) have been proposed to enable the computational modelling of educational units, including also the variety of collaboration modes. Nevertheless, current EMLs have not achieved a comprehensive solution. This paper analyzes the requirements to support such modes and the capacity of the existing proposals. Grounding on this analysis, a new EML is proposed to satisfy the identified requirements: PoEML (Perspective-oriented Educational Modelling Language). Separation of concerns, the key-underlying model behind PoEML is discussed in this paper.

Introduction Separation of concerns is a long standing idea that simply means a large problem is easier to manage if it can be broken down into pieces; particularly if the solutions to the sub-problems can be combined into a solution to the whole problem. This is an important design principle in many areas, such as architecture and software design. The goal is to design systems so that functions can be modelled independently of other functions, meaning that failure of one function does not cause other functions to fail, and in general to make it easier to understand, design and manage complex interdependent systems. This paper introduces an Educational Modelling Language (EML) based on the separation of concerns principle. EMLs (Rawlings et al. 2002, Koper 2001), as computational languages, are intended to support the modelling of educational units (e.g., a simple lesson, a lab practice, a collaborative seminar, an e-learning course) and the eventual execution of such models. This EML proposal is specially focused on supporting the modelling of collaborative learning educational units. Collaborative learning comprises a broad range of educational practices that take advantage of human interaction in order to achieve more effective and efficient learning (Dillenbourg 1999). In practice, collaborative learning educational units may be arranged in accordance with different structures (e.g., a discussion based lecture, a group-based workshop, a brainstorm, a cooperative project) and using different collaboration schemes (e.g., synchronous or asynchronous, strict-coordination or free collaboration, face-to-face or distance). The goal is to support the computational modelling of collaborative learning educational units taken into account this variety of structures and schemes. To do it, this proposal follows a separation of concerns approach. The key idea is to decompose the modelling of educational units in several concerns. For example, participants’ awareness and authorizations can be modelled as different concerns. The modelling of awareness is concerned with the way in which events produced during education have to be processed and notified to appropriate participants (e.g., learners' actions have to be notified to a teacher). The modelling of authorization is concerned with the assignment of permissions to participants (e.g., learners and teachers have different permissions to use the functionalities of a simulator). The concerns identified in this proposal are named as perspectives and the proposed language is named as Perspectiveoriented EML (PoEML). Next section introduces the requirements identified to support the modelling of collaborative learning educational units. These requirements have been mainly obtained from the domains of workflow and groupware in relation with their support for human collaboration. Then, the PoEML separation of concerns proposal is introduced. The paper ends with some conclusions.

Collaboration in Educational Units Broadly speaking, collaborative learning is a situation in which two or more people learn or attempt to learn something together (Dillenbourg 1999). Depending on the application, collaboration may be arranged in different ways: involving two or several participants, freely or in a constrained way, synchronous or asynchronously, it may involve communication among participants or the performance of an activity in conjunction, etc. An EML should support the modelling of educational units involving all these alternatives. In order to provide a comprehensive analysis of the needs this section introduces a classification of the collaboration modes that should be supported. Please, note that these collaboration modes are not exclusive of educational units, but they can be applied in other areas where collaboration is required (e.g., groupware). Anyway, this paper is centred in the collaborative learning context as it is the area where it has been developed. Eventually, the results of this



29

research are of interest both in Computer-Supported Collaborative Learning (CSCL) and in ComputerSupported Cooperative Work (CSCW). Collaborative educational units involve people collaborating. To collaborate, people need to exchange information (i.e. to communicate), to act together in shared spaces (i.e. to co-operate), and to organize themselves during the shared activity (i.e. to coordinate) (Ellis & Wainer 1999).

Communication It encompasses the process of transfer and exchange of information between participants (Ellis & Wainer 1999) (Raposo et al. 2004). Typical communication tools are: e-mail, desktop conference systems, chat, whiteboard, etc. The provision of communication functionalities usually involves mechanisms to control the participants’ access and use of these functionalities. These control mechanisms are usually specified in conference and conversational models: • The conference model is concerned with the control of communication resources and participants: how the communication is initiated and finished, how new participants can join the communication, etc. Usually, these issues are arranged in a session part (about the management of applications) and a membership part (about the management of participants). Many e-learning systems adopt conference models; an example is the SimulNet virtual laboratory that enables to control the involvement of participants in communications (Llamas et al. 2001). • The conversation model is concerned with the interchange of messages during the communication: the conversational moves allowed in the communication, what the appropriated conversational replies to the moves are, etc. There are some CSCL developments involving specific conversation models (Guzdial & Turns 2000; Baker & Lund 1997) where possible reply messages are suggested to participants.

Co-operation Sometimes the collaboration among a group of participants is centred on the access and change of a shared artefact (Ellis & Wainer 1999; Raposo et al. 2004). In these situations, the goal of the collaboration is the construction of this shared artefact. Examples of systems that provide these functionalities are shared editors, virtual whiteboards, shared repositories, etc. These functionalities also require some control mechanisms: • Access rights. It concerns with the assignment of permissions to participants. In many educational units learners and teachers do not have the same rights to access artefacts or to perform operations. • Floor control. Some tools use a mechanism of floor control to avoid simultaneous access to artefacts. For example in a blackboard only one user has the right to change the artefact at each time. • Version and info. In some applications it is important to store stable situations of an artefact and to allow the artefact to be restored to such stable situations. It is possible to consider the storage of versioning information, time stamp and author information. Using this information, some tools allow just to view the data of interest.

Coordination Coordination is considered as the process of managing dependencies among tasks (or activities) (Ellis &Wainer 1999; Malone & Crowston 1990). There are different ways in which coordination can be supported. Traditionally, two extreme approaches are distinguished that have been the centre of a heated discussion (Raposo et al. 2004) (Schmidt & Simone 2000) (Bernstein 2000). On the one hand, there are normative models that try to structure collaboration by restricting the tasks to be performed. On the other hand, there are those advocating that collaborative systems should take flexibility to the extreme, leaving the coordination to the users and simple mediating in the interaction (i.e. supporting communication and co-operation). There have been several attempts to conciliate and integrate these two extreme positions, arguing that both kinds of activities are “seamlessly meshed and blended in the course of real world”. (Bernstein 2000; Schmidt & Simone 2000). As an example, Bernstein (Bernstein 2000) identifies four categories of approaches to support coordination from highly unstructured to highly structured: • Context. It involves the more flexible and simple coordination solution as the system only supports communication and co-operation functionalities. Participants are responsible for the coordination. • Awareness. It involves the notification to certain participants of what is happening in the collaboration and what other participants are doing or have done (e.g., teachers are notified of events produced by learners). • Constraints. Participants are provided with constraints about how goals can be achieved. These constraints limit the participants' work, but they can work quite freely. For example, a lab practice needs teachers to provide the main deadlines to obtain a solution, but learners can define and arrange their own activities. • Directions. Participants are provided with strict instructions about what they have to do. There are several issues that can be arranged for the provision of directions: • Restricting the activities that must be performed. In addition, activities can be enabled and disabled in accordance with an established order.



30

• •

Restricting the persons that must perform the activities. Restricting the information that must be used to perform the activities. In practice, depending on the pedagogical approach, the different collaboration modes may be more or less important. For example, a discussion-based classroom requires a major focus on communication while a project-based laboratory demands a main role of co-operation or coordination. In a similar way the different coordination alternatives also introduce learning implications (Dillenbourg 2002). High-structured units provide clear guidance to learners' action. In this way it can be very easy for learners to follow the intended plan. Nevertheless, such solutions can disturb the natural learning capabilities and result in forced interactions. On the contrary, low-structured tasks offer more freedom, but they use to require more effort by participants. Therefore, there is not a unique solution that can be applied to all cases. The right level of collaboration support has to be a design decision that depends on the pedagogical approach.

PoEML Foundations This section introduces a separation of concerns proposal to support the modelling of collaborative learning educational units. This modelling involves many different requirements that need to be satisfied, specially the several collaboration modes. To do it, the whole modelling problem is decomposed into several parts (named as perspectives) grouping and separating the various modelling concerns. Then, each perspective may be modelled separately while abstracting from the concerns considered in other perspectives. Following this approach the new EML has been named as Perspective-oriented EML (PoEML).

Separation of Concerns Principle The key strategy for the development of PoEML has been the separation-of-concerns principle. Separation-ofconcerns is a long-standing idea that simply means that a large problem is easier to manage and solve if it can be broken down into parts and each part can be approached separately. It is an important design approach in many areas, such as software design (Parnas, 1979), used to facilitate the understanding, design and management of complex systems. In addition, UML (Fowler, 2003) is a modelling language for software engineering where different diagrams are proposed to model different issues: use cases, analysis, design, etc. A similar example in other domain is the architectural plans, which follow a separation of concerns for the development of buildings. In the learning domain there are also some proposals in which a certain kind or separation of concerns is provided, such as (Strijbos et al., 2004), where learning units are considered through several orthogonal axis. Anyway, as long as we know, PoEML is the first attempt that takes the separation of concerns as driven development principle.

The Activity Theory Another important foundation of PoEML is Activity Theory (Roth and Lee, 2007). This theory has been used to analyze the issues involved in learning units and to identify an appropriate separation of concerns (Caeiro et al., 2007). The Activity Theory is a meta-theory about activities and their con- stituent components (Engestrom et al., 1999). Considering that any learning unit can be described as a set of tasks (activities during the run-time), the Expanded Mediation Model conceived in this theory provides an interesting framework to identify a suitable separation of concerns (see Figure 1).

Figure 1. Perspectives in accordance with the Activity Theory Expanded Mediational Model. The core of this model is that any activity involves a subject, playing a role, acting on an object to achieve a certain goal. The environment and the community influence the connection between the subject and the object



31

where the activity is performed. In other words, the work on the activity depends on the environment and the community. The environment contains the tools and resources that can be used by the subject to act on the object. The community puts the emphasis on the social context where the subject operates, involving the influence of two new issues: rules and division of labour. The rules component highlights the fact that within a community, subjects are bound to rules and regulations that affect the way they interact in the activity, including also the interaction with the environment and its elements. The division of labour refers to the breaking down of the goal into sub-goals and the distribution of responsibilities among the available subjects. As a result new subsidiary activities (sub-activities) are produced. The Activity Theory has guided us in the separation-of-concerns in learning units towards the identification of 13 perspectives. Figure 1 shows how the several perspectives are located in relation to the Expanded Mediation Model.

Support of the Collaboration Modes The perspectives introduced are devoted to support the modelling of educational units, specially the collaboration modes identified in section 2. Table 1 shows the relationships among the issues considered in each perspective and the collaboration modes. There are some important points to notice: • The distinction between the Structural and Goals perspectives is original in PoEML. The Structural perspective is based on the activity concept. Educational units are conceived as a set of activities that group all the other elements (goals, actors, environments, etc.) in a modular and hierarchical way. The Goals perspective is about the goals that have to be satisfied in educational units. This separation between structure and goals provides a more flexible solution enabling to change each one of them without affecting the other. For example, it is possible to specify goal maps changing the activities that have to be performed without modifying the structure of the educational unit. • The first five perspectives (Structural, Goals, Participants, Environments and Data) are key to support the several coordination needs. For example, they have to be used to determine the mandatory or optional character of a goal (Functional perspective), the learners and teachers enrolled (Participants perspective), the environments and their contained resources (Environments perspective) and the data managed associated with goals, participants and environments (Data perspective). • The Tools perspective is very important to support the basic communication and co-operation functionalities. EMLs are not intended to support the modelling of all the functionalities that may be required in educational units. This would be an impossible goal as the number of required functionalities is infinite. Instead, EMLs promote the integration of external tools that provide such functionalities (e.g., chat systems, shared areas). In this proposal the integration of such tools is performed through the Tools perspective through their “abstract” featuring. This perspective is essential to support the modelling of communication and co-operation functionalities. • The Order perspective is used to determine the order in which activities have to be performed, such as sequence, parallel, synchronization points, etc. Therefore, it is directly related with the Directions coordination mode.

X

Communication

X X

Conference Conversation

X

Co-Operation

Version & Info

X X

X

X

X X

X X

X X

X X

X X

X X

Directions

X X

X

Awareness Contraints

X X

X X X

Floor Control

Context

X X

X

Access Rights

Coordination

X X

X

X

X X

Causal

Interaction

Awareness

Authorization

Temporal

Order

Organizational

Tools

Data

Environments

Participants

Particular

Goals

Mode

Structural

Table 1. Support of the collaboration modes with the proposed perspectives


•

•


32

The Temporal perspective involves temporal specifications and constraints. In this way it is useful to specify conference and conversation models in communications and floor-control and version & info models in co-operation. The Authorization, Awareness and Interaction perspectives are original contributions of this proposal to support the collaboration modes, particularly those related with the support of communication and cooperation models. Notice that the issues supported by these perspectives are not provided by current EMLs.

Conclusions The paper introduces an EML based on the separation of concerns approach: PoEML. The proposal devotes a special attention to the support of the variety of collaboration modes that may be involved in collaborative learning. This is a complex design problem involving many different human interaction contexts and schemes. The separation of concerns principle has been applied to simplify this complexity trying to decompose the whole problem into several independent sub-problems. A main contribution is the set of perspectives considered to support the modelling of collaboration modes: Tools, Awareness, Interaction and Authorization. These perspectives have not been in previous works, but they are essential to support many of the CSCW requirements and in particular to enable the integration of both domains.

References Baker, M. & Lund, K. (1997), ‘Promoting reflective interactions in a CSCL environment’, Journal of Computer Assisted Learning (13), 175.193. Bernstein, A. (2000), How can cooperative work tools support dynamic group processes? Bridging the specificity frontier?, in 'Proceedings of ACM Conference on Computer Supported Cooperative Work'. Dillenbourg, P. (1999), What do you mean by collaborative learning?, in P. Dillenbourg, ed., ‘Collaborativelearning: Cognitive and Computational Approaches’, Elsevier, Oxford, chapter 1, pp. 1.19. Dillenbourg, P. (2002), Over-scripting CSCL: The risks of blending collaborative learning and instructional design, in P. P. A. Kirschner, ed., ‘Three Worlds of CSCL. Can We Support CSCL?’, Open Universiteit Nederland, Heerlen, The Netherlands, chapter 2, pp. 61.91. Ellis, J. & Wainer (1999), Groupware and computer supported cooperative work, in G. Weiss, ed., ‘Multiagent Systems: A Modern Approach to Distributed Artificial Intelligence’, MIT Press, pp. 425.457. Engestrom, Y., Miettinen, R., Punamaki, R.-L. (Eds.) (1999). Perspectives in Activity Theory. Cambridge University Press, New York, USA. Fowler, M. (2003). UML Distilled: A Brief Guide to the Standard Object Modeling Language, Third Edition. Addison-Wesley Professional. Guzdial, M. & Turns, J. (2000), ‘Effective discussion through a computer-mediated anchored forum’, Journal of Learning Sciences 4(9), 437.469. Koper, R. (2001). Modelling units of study from a pedagogical perspective. The pedagogical metamodel behind EML. Tech. rep., Open University of the Netherlands. Llamas, M., Anido, L. & Fernández, M. J. (2001), ‘Simulators over the network’, IEEE Transactions on Education 44(2). Malone, T. W. & Crowston, K. (1990), What is coordination theory and how can it help design cooperative work systems?, in ‘Proceedings of CSCW’, pp. 357.370. Parnas, D. L. (1979). On the criteria to be used in decomposing systems into modules. Yourdon Press, Upper Saddle River, NJ, USA, pp. 139–150. Raposo, A. B., Pimentel, M. G., Gerosa, M. A., Fuks, H. & Lucena, C. J. P. (2004), Prescribing e-learning activities using workflow technologies, in ‘CSAC’04’, pp. 71.80. Rawlings, A., van Rosmalen, P., Koper, R., Rodrguez-Artacho, M., Lefrere, P. (2002). Survey of educational modeling languages(EMLs). Tech. rep., CEN/ISSS WS/LT Learning Technology Workshop. Roth, W.-M., Lee, Y.-J. (2007). ”Vygotsky’s neglected legacy”: Cultural-historical activity theory. Review of Educational Research 77 (2), 186–232. Schmidt, K. & Simone, C. (2000), 'Coordination mechanisms: Towards a conceptual foundation of CSCW systems design', CSCW 4(2-3), 155.200. Strijbos, J. W., Martens, R. L., Jochems, W. M. G. (2004). Designing for interaction: six steps to designing computer-supported group-based learning. Computers and Education 42 (4), 403–424.

Acknowledgments This piece of research is supported by the eContentplus programme ECP 2007 EDU 417008 (www.aspectproject.org), a multiannual Community programme to make digital content in Europe more accessible, usable and exploitable. Additionally, it has been partially funded by the Spanish Ministerio de Educación y Ciencia under grant TIN2007-68125-C02-02.



33

CSCL scenarios := A cocktail of CLFPs Georgia Lazakidou, Symeon Retalis, Petros Georgiakakis, University of Pireaus, Department of Technology Education and Digital Systems, 80 Karaoli & Dimitriou, 18534 - Greece [email protected], [email protected], [email protected] Stamos Karamouzis, Regis University, School of Computer & Information Sciences, USA [email protected] Abstract: Collaborative Flow Design Patterns are used to describe CSCL scenarios and strategies known as best practices. However, very often there is a need to make effective blending of strategies. Here, we describe a proposal of adding an extra element called “variations” that includes all proposals related to the presented solution and we show our proposal by setting the example of a complex CSCL strategy.

Motivation One of the great challenges of learning design in computer supported collaborative learning (CSCL) is to design effective interactive learning scenarios consisting of task distribution, specification of roles, role playing rules, resources, tools, deliverables, etc. (Dillenbourg et al., 2009). Luckily, there are several collaborative learning strategies such as the Pyramid, the Jigsaw, the Think Pair Share, etc. which have been widely used in the CSCL design practice. It is quite a simple task for designers to create CSCL scenarios which are based on a specific strategy but it is more complicated for designers to mix and match strategies in order to create their scenarios. The starting point of these design cases has always been a primary issue in every strategy used. However, the designer wants to make a configuration of a strategy and makes use of issues from other strategies. The need for this mixture of strategies stems from contextual requirements for balancing a variety of organizational, administrative, instructional and technological components. In order to help designers in creating complex CSCL scenarios, strategies need to be described in a designer friendly way. Such a way is the use of collaborative flow design patterns (CLFP) (Hernandez-Leo et. al, 2009). The term “flow” is used to portray the coordination and the sequencing of tasks during the learning process. A CLFP defines the sequence of the tasks that the strategy dictates as well as other elements needed for the various tasks, such as the duration of a task, the use of a particular tool for a given task and so on (Turani & Calvo, 2006). A CLFP is an attempt to illustrate and disseminate the “best design practices” with respect to a problem or class of problems, to share the experience, to transfer knowledge from experts to novices. These resources are richer than guidelines or scripts because they contain well justifiable solutions and examples to design problems as well as the rationale behind these solutions (Goodyear at al, 2004). CLFP’s are all about reusability, which seems to be the keyword in achieving an economy of scale for developing affordable and effective CSCL learning scenarios (Garzotto & Retalis, 2008). Although there are some simple algorithms that can help a designer choose among which scenario will be created based on CLFP (see the recommendations integrated into the Collage tool (Hernandez-Leo et. al, 2009)), there are no such recommendations that will enable designers to make effective blending of strategies. An idea, which is presented in this paper, is to add an extra element to the CLFPs, called variations, which include ideas for variations of a proposed solution which is based on a strategy. CLFP writing is a team effort rather than an individual task. They are usually drafted by someone and then shared, analyzed, commented, assessed (evaluated) during action and refined through an extended process of collaboration. As a result, the various variations will also be stated into a CLFP only after they have been proven to be effective. As a matter of fact, when proposing the variations of a strategy, the infrastructure needed to support the actors’ activities might change. In the paper, we introduce a CFLP in connection with a new CSCL strategy, called eARMA, as well as its variations which have led to the development of a new CSCL tool, called SyCo. Like the preparation of a cocktail, where the cornerstone is an understanding of the relationships between strong and weak, sour and sweet, in CSCL scenario development, a designer should blend the phases, the resources, and the tools (i.e. the ingredients) together cautiously. Otherwise, this scenario will resemble a weak or watery cocktail with inappropriate flavour, texture and colour.



34

Configuring variations of CLFPs The format of e-learning design patterns differs. However, the structure of a CLFP design pattern includes elements such as a design of a problem’s description, the related context and a documented suggestion for a solution to the problem with concrete examples and references. Table 1 shows in detail the format of a CLFP which is similar to the suggestion of (Hernandez-Leo et. al, 2006). Table 1. The CLFP design pattern format Element

Explanation

Name

Name of the FDP

Context

Environment type in which the CLFP could be applied

Problem

Learning problem to be solved by the CLFP

Solution

Description of the proposal by the CLFP for solving the problem

Actors and Actions

Description of the actors involved in the collaborative activity described by the CLFP as well as their activities

Types of Tasks

Description of the types of groups of learners identified and how they are related, types of tasks, together with their sequence, performed by the actors involved in the activity.

Example

A real-world learning activity capable of being structured according to the CLFP

Related patterns

Names of other interrelated patterns

References

List of resources for further reading

As pointed out by Hernandez et al (2006) CLFPs provide software developers with information about the flow of learning activity types that are expected to occur during a collaborative learning scenario based on a CSLC strategy in which the pattern analyses. Using this information, software developers can identify the functionality of a tool needed in order to better support the sequence of activities that the pattern describes. In order to specify the variations of a strategy in a CLFP, a new element needs to be added, entitled “variations”. This element will specify what to change and under which conditions in the sequence of activities, the resources or the tools. All these changes need to be justified in order to illuminate the rationale behind these variations. In the Appendix we include a short version of a CLFP with its variations which refers to the eARMA CSCL strategy (Lazakidou & Retalis, 2009). eARMA can be supported with various synchronous collaborative concept mapping tools which allows users to share ideas in the form of concept map and chat. However, due to some special characteristics of the strategy, the functionality of a typical mind mapping tool should be enriched. This fact led to the development of the SyCo tool (Kefalidis et al, 2009). The SyCo tool (see Figure 1) differs for concept/mind mapping tool in some aspects such as: 1. It offers private thinking space apart from a shared space. It also allows the learners or the moderator to create more collaborative shared spaces for sub groups. 2. It supports the insertion of sticky notes which will have hyperlinks as well as links to resources 3. Its library of drawing objects can be extended with more shapes than the basic concept modeling shapes, such as flowchart shapes, entity relationship shapes, and images 4. It allows the learners to seek advice for better performance or execution of a task. For example, during the steps of the problem solving process, the learner can look at the questions that needs to be answered. For example in step 2 “definition of the problem”, questions like “are there any key points which could help me”, “how is the specific problem connected to what we have been taught”, etc will be shown to the learners thus offering scaffolds.



35

Figure 1. A screen shot of the SyCo tool

Conclusions The idea of mixing strategies or making variations of a strategy in order to make complex CSCL scenarios is not new. There is a need, however, to give advice to designers on how to make effective mixtures of strategies without harnessing the basic educational philosophy and added value. Moreover, variations of a strategy can lead to new functionality of a CSCL tool. This is why in a CLFP, the variations of a strategy needs to be documented. Based on the preliminary findings from a simple research study with involved school teachers who wanted guidance in designing CSCL scenarios, we can state that this proposal seems to be well appreciated. More experiments with teachers are needed along with evaluation sessions with developers from the educational software industry who want to develop innovative CSCL tools. Also, we are trying to apply the same ideas to the domain of collaborative creativity for innovative product development (within the idSpace FPT7 EU funded project - ref num: 2008-216199). Variations of well known creativity techniques will be proposed.

References Dillenbourg, P., Järvelä, S. & Fisher, F. (2009). The evolution of research on computer-supported collaborative learning: From design to orchestration (pp.3-19). In Technology-Enhanced Learning. Principles and products. Goodyear, P., Avgeriou, P., Baggetun, R., Bartoluzzi, S., Retalis, S., Ronteltap, F., & Rusman, E. (2004). Towards a pattern language for networked learning, Proceedings of the 2004b Networked Learning 2004, Lancaster, UK, 5-7 April 2004. Hernandez-Leo, D., Villasclaras-Fernandez, E. D., Asensio-Perez, J. I., Dimitriadis, Y., Jorrin-Abellan, I. M., Ruiz-Requies, I., & Rubia-Avi, B. (2006). COLLAGE: A collaborative Learning Design editor based on patterns, Educational Technology & Society, 9(1), 58-71. Hernández-Leo, D., Villasclaras-Fernández, E.D., Asensio-Pérez, J.I., & Dimitriadis, Y. (2009). Generating CSCL scripts: From a conceptual model of pattern languages to the design of real scripts, In Goodyear P. & Retalis, S.(Eds.) E-learning Design Patterns Book, Sense Publishers. Kefalidis C., Lazakidou, G., & Retalis, S. (2009). SyCo: A Collaborative Learning Tool for Private and Public Activation, Interaction Design for Children Conference, IDC2009, 3-5 June 2009, Como Italy. Kollar, I., Fischer, F., & Hesse, F.W. (2006). Collaboration scripts - a conceptual analysis. Educational Psychology Review, 18(2), 159-185. Lazakidou, G., & Retalis, S. (2009). Using computer supported collaborative learning strategies for helping learners acquire self-regulated problem solving skills in Mathematics. Computers and Education. Turani A. & Calvo R.A. (2006). Beehive: a software application for synchronous collaborative learning, Campus-Wide Information Systems, 23(3), 196-209



36

Appendix The example of a CLFP with variations – The case of eARMA The pattern that follows, concerns the eARMA CSCL strategy which consists of a sequence of learning activities for augmenting problem solving and self-regulation skills (Lazakidou & Retalis, 2009). Due to space limitations, a short version of the eARMA design pattern will be presented. A full version can be found at (http://cosy.ted.unipi.gr/patterns) . Element Explanation Name e-ARMA In primary and secondary schools the acquisition of problem solving skills is of great priority. Context The usual teaching approach which advocates the repetitive practice at problem solving can help learners gain routine expertise. They may also develop speed and accuracy at routine problem solving, but fail to develop the ability to reflect on what they have done or to adapt to solving new problems in a flexible manner. However, a process-based teaching approach can aid learners in developing problem solving skills as well as trigger learners’ awareness of their own thinking processes (Brown, 1987). Also, the organization of collaborative problem solving sessions in the spirit of proactive cooperation and shared effort can lead to better, more creative and effective solutions as well as shared knowledge construction. Learners’ interaction toward reaching a common goal often tends to regulate each other's actions resulting to the solution of difficult problems that learners might not be able solve when working independently. How can a teacher help learners acquire individual problem solving skills, but also increase their Problem collaborative behavior? Use eARMA which is a strategy for helping learners understand the problem solving process as Solution well as for fostering the ability to regulate their thinking when solving problems in collaborative settings. eARMA includes three phases : 1st Phase: Observation, 2nd Phase: Collaboration, 3rd Phase: Semi-Guidance. A solver after having performed the three phases of problem solving tasks the problem solver can reach to a level of autonomous problem solving. During each phase a problem solving model which consists of well defined steps is being utilized (e.g. problem definition, relevance to past problems, etc.). eARMA suggests the use of Sternberg’s problem solving model. The main idea behind the eARMA strategy is that it provides problem solvers with adequate devices (schemes) to help them learn how to solve a problem following a series of well specified steps and in a discourse with peers. The gradual removal of the collaborative learning tools occur after taking into account children's increasing mastery of the problem-solving strategy as well as their self-regulating skills. The phases of eARMA is shown graphically below: Actors and actions

During the first phase a learner observes a problem solving model [display solution] and how it is implemented during the problem solving process. For example, a teacher, who plays the role of an expert solver, can make her thinking tasks explicit when applying the Sternberg’s problem solving model (Sternberg, 2003) that includes seven steps: Identification of the problem, Definition of the problem, Constructing a strategy, Organizing information, Allocation of resources, Monitoring the solving process, Evaluating the solving process and outcome. During the second phase learners try to mimic the observed problem solving process in quite similar problems. Each learner is responsible for one step. Peer learners watch how a team member performs a step and they either accept the proposed actions or discuss other alternatives. The goal is that all group members share responsibility for the actions of each step irrespective if a specific learner had the duty to perform the specific actions of a step. In the collaboration of two, learners solve at least two similar problems alternating the roles of the active and the passive solver. As an active solver s/he solves the problem based on the observed problem solving process. As a passive solver s/he observes their partner to solve the problem maintaining the right to ask, object, disagree and propose different answers. The third phase includes an individual problem solving process where learners have to find solutions to problems by following the steps of the problem solving model in the way they have been taught.


Examples – Known uses

Variations

Related Patterns References


37

All collaborative learning actions can take place in a computer supported collaborative learning environment where problem solving space is either shared among the group members or isolated allowing some space for individual thinking [shared collaborative space]. Also, well chosen resources such as a video of a problem solving process, mathematical formulae, etc. need to be given to the learners in order to be utilized during the problem solving process. eARMA has been effectively applied to primary and secondary schools. A typical example is the following. In terms of mathematical problem solving in the 5th grade of primary school, a set of ten story-problems had been created (Lazakidou & Retalis, 2009). Every story included mathematical problems concerning the topic of diet (nutrition, malnutrition, genetically modified foods, eating disorders etc.). An example of a problem-story “Tom’s mother suffers from anorexia and she weights 42 k. She needs to take 800 calories every day to increase her weight up to 44 k. in two months. What combination of food could she make in order to cover the necessity of 800 calories every day?” Use of the Synergo tool, a synchronous computer supported collaborative concept mapping tool which also allows learners to chat while sharing, the various collaborative tasks of the eARMA strategy has been applied to help learners acquire self-regulated problem solving skills in Mathematics. Instead of the Sternberg’s model a teacher may select the IDEAL problem solving model (Bransford & Stein, 1984). It includes five steps according to which they must: a) Identify the problem, b) Define the problem through thinking about it and sorting out the relevant information, c) Explore solutions through looking at alternatives, brainstorming, and checking out different points of view, d) Act on the strategies and f) Look back and evaluate the effects of their activity. Otherwise, there are other problem solving models that a teacher may select one of them depending on the type/domain of problems that are needed to be solved. The collaborative phase of the e-ARMA strategy can be a mixture of phases from other collaborative strategies. For instance, instead of having each group working alone, one may select that learners, who share similar roles such as those of the identifier (of the problem space), organizer (of the given data) explorer (of the potential strategies), and evaluator (of the problem solving process), can collaborate like Jigsaw strategy suggests. This variation is needed when problems are quite complex, the solvers are not expert and groups are not in a competitive mode. A graphical illustration of this variation is shown in the figure below:

DISPLAY SOLUTION, WELL CHOSEN RESOURCES, FORM GROUPS, SHARED COLLABORATIVE SPACE

Bransford, J.D., & Stein, B.S. (1983). The IDEAL problem solver: A guide for improving thinking, learning and creativity. NY: W.H.Freeman. Brown, A. L. (1987). Metacognition, executive control, self-regulation, and other more mysterious mechanisms. In F. E. Weinert & R. H. Kluwe (Eds.), Metacognition, motivation, and understanding (pp. 65-116). Hillsdale, New Jersey: Lawrence Erlbaum Associates. Lazakidou, G., & Retalis, S. (2009). Using computer supported collaborative learning strategies for helping learners acquire self-regulated problem solving skills in Mathematics. Computers and Education. Sternberg, R. J. (2003). Cognitive psychology (3rd Ed.). Thomson, Wadsworth.

Acknowledgments This work has been partially supported by the ISTFP7 idSpace project: Tooling of and training for collaborative, distributed product innovation (ref num: 2008-216199).



38

Interrelating assessment and flexibility in IMS-LD CSCL scripts 1

2

1

1

E.D. Villasclaras-Fernández , D. Hernández-Leo , J.I. Asensio-Pérez , Y. Dimitriadis , L. de la Fuente3 Valentín 1 GSIC/EMIC, University of Valladolid, Camino del Cementerio, s/n, 47011, Valladolid, Spain 2 Pompeu Fabra University, Estació de França, Passeig de Circumval·lació 808003, Barcelona, Spain 3 Telematics Engineering Department, University Carlos III of Madrid, Spain. E-mail: [email protected], [email protected], {juaase, yannis}@tel.uva.es, [email protected] Abstract. CSCL macro-scripts enable the design of the structure of collaborative scenarios before actually carrying out the learning activities. In order to cope with unexpected situations, scripts need to be complemented with the possibility of modifying the script as it is being enacted. On the other hand, this ability of introducing changes can also be considered in advance in order to conceive the pedagogical method of the script. This position paper discusses the interrelation between script flexibility and the assessment plan. The latter, being regarded as part of the script, can be a driving force to detect unpredictable situations or results and react accordingly by modifying the script to tackle them.

Introduction CSCL macro-scripts have been found to be useful for structuring the learning activities so that interactions can be more productive and therefore achieve the learning goals (Dillenbourg, 2002). However, over-scripting may impose excessive prescription to learners and teachers, and therefore fading-out mechanisms (Kollar et al., 2006) or flexible and adaptable scripts should be provided (Dillenbourg and Tchounikine, 2007). On the other hand, assessment plans and accordingly assessment activities may be interwoven with the classical learning activities (Black & Wiliam, 1998; Shepard, 2000) and therefore, assessment may be used as an integral part of the CSCL macro-script (Villasclaras-Fernández et al., 2009). Taking into account that regulation and scaffolding may be directly related to and derived from the assessment activities, it is clear that an effective use of assessment activities requires a certain degree of flexibility; at the same time, assessment offers valuable information to intervene and introduce changes in the learning activities as they are carried out (Stiggins, 2002). For instance, the assessment activity may indicate that a certain group should undertake a new activity independently from the rest of groups and therefore the teacher should be able to modify adequately the existing macro-script during enactment. Scripts not only describe sequences of activities, but they also reflect the pedagogical plan devised by the teacher or designer; this plan can include assessment activities as well as the possibility of introducing changes in the script as a way of improving the chances of learning. In spite of the need for flexibility in CSCL scripts, existing educational modelling languages and tools, mainly those related to the IMS-LD specification (IMS, 2003), lack an usable and standard mechanism to enable assessment activities as triggers for flexible scripts. This is especially evident with respect to teachers who are non-proficient in handling complex computational representations. On the other hand, current available IMS-LD compliant engines and associated players offer different non-standard ways of introducing modifications in the script during its enactment. For instance, the Grail IMS-LD compliant player (del Cid et al., 2007) includes a cockpit for this function (de la Fuente Valentín, 2008), and an extension to the CopperCore player (Martens et al., 2005), focused on IMS-LD script flexibility, have already been proposed (Zarraonandia et al., 2006). As a consequence, those engine-specific modification mechanisms are not reflected in design processes enforced by existing IMS-LD authoring tools. This separation between script language, design process and enactment systems has a negative impact in the flexibility of scripts. This paper addresses the problem of effectively employing assessment plans in CSCL macro-scripts and provides an approach that allows teachers to increase the flexibility of the scripts. The focus of this approach is to take advantage of IMS-LD script features that can enable script flexibility. Therefore, much emphasis will be put in the design phase, in which the script is created. This approach is studied through a case study in which CSCL macro-scripts have been adapted during enactment according to assessment results. This case study involved modelling the macro-script in the Web Collage authoring tool, as well as its enactment with current technologies, namely Grail and a Wiki website. These technologies allowed runtime modifications of the script through transparent mechanisms that hide the complex computational representations. Although current implementation is not fully automatic, the case study showed that such an approach may be effective and efficient, thus improving the usability of available IMS-LD players in real situations.



39

Position statement In order to cope with unexpected situations, it has been argued that flexibility mechanisms should be available for teachers and students during the enactment of the script (Dillenbourg and Tchounikine, 2007). On the other hand, flexibility may be also a mechanism that is actually specified within the pedagogical method of the script itself. In other words, the script may include its own adaptation scheme in order to regulate undesired situations. The responsible for performing such an adaptation would still be a human agent (the teacher or perhaps the students); however, the script itself can provide instructions to that human agent (a participant in the learning situation) concerning the moments in which these changes are appropriate, and how they are related to the pedagogical method of the script. For instance, let us consider the activity flow depicted in Figure 1, which may be part of a CSCL script. There are three interrelated activities: first, students complete a learning activity. Next, the teacher assesses their work. Based on this assessment, the teachers finally may introduce some modifications in the script: • The teacher may introduce new activities, if it is suggested by the assessment results. • These activities need to be configured with respect to time schedule, description and objectives. • Learning resources may be included in the activities.

Students perform activity.

Teachers assess student’s work, and activate/modify additional activities accordingly.

Students carry out additional activities if needed.

Figure 1. Example of a sequence of activities, which shows the relation between assessment and flexibility (white boxes represent students’ activities, gray boxes are teacher’s activities). Therefore, the script itself integrates flexibility as a mechanism to improve the chances of achieving the learning objectives. While this case is quite different from that of unexpected situations that may require changes in the script on the fly, actually the technology requirements are comparable: introducing the changes through a specific IMS-LD player may be done in the same way in both cases, independently of the motivation for the changes. However, this application of the flexibility concept has a fundamental difference: it is included in the script, and therefore it is foreseen during the design phase. The ability to actually predict the needed changes is, of course, rather limited. Only the moments and perhaps the type of modifications are indicated. The concrete content of the changes remains open. The fact that changes in the script may be anticipated allows the designer and authoring tool to prepare the script, in the design phase, to enable flexibility. This is especially relevant considering existing problems associated with flexible collaboration scripts and current technology. The problems identified here are not entirely caused by limitations of the IMS-LD specification, but also by two facts: (a) IMS-LD authoring tools and players remain unintuitive and complex, making it difficult to use all the features of this specification; and (b) changes in the script may need to be introduced independently for each group or individual students. With respect to this second issue, typically a collaborative learning session will involve several groups working in parallel. Such learning scenarios present two challenges: first, the technology should allow group-specific changes; second, the teacher has to independently manage each group. Using conditional capabilities of IMS-LD (such as properties or notifications) may facilitate the introduction of flexibility, especially when group-specific changes are desired (Koper et al., 2005), as discussed in the following section. However, these capabilities have shown to be complex and require a deep understanding of the specification. In addition, the script may define the usage of different non-standard flexibility mechanisms offered by IMS-LD players. Therefore, the role of the design phase in assessment-related flexibility is very relevant: some flexibility mechanisms may require the configuration of the script (for instance, by including IMS-LD properties or other components); and the script may need instructions (included in the teachers’ activities) explaining when and how to introduce changes in the player during enactment. These problems make it difficult for non-expert teachers to create and enact CSCL scripts combined with flexibility mechanisms. In order to alleviate this problem, this paper proposes: (A) flexibility mechanisms can be part of the pedagogical method of a CSCL script, and particularly in the case when they are coupled with the assessment plan; and (B) the configuration of learning and assessment activities in the script can lead to the automatic integration of flexibility mechanisms within CSCL scripts in the authoring tool, during the design phase; and the provision of intuitive and user-friendly interfaces to make certain changes in the script in the IMS-LD player during the enactment phase. While so far the paper has argued the adequacy of integrating flexibility mechanisms in the script itself (issue A), the possibility of automated configuration of flexibility mechanisms (issue B) is necessary to improve



40

the usability of the specification for non-expert teachers. Therefore, the following section will also deal with issue B, discussing a case study that shows the feasibility of automating the creation of flexibility mechanisms.

Case study The sequence of activities and the flexibility mechanisms mentioned in the previous section have been actually applied in an authentic collaborative learning situation, set up in the context of a doctoral course, and analyzed in the case study described here. This learning scenario comprises two face-to-face sessions and several distance activities between and around them, spanning a period of six weeks. The activities were planned using a CSCL macro-script, formalized with IMS-LD. The script was enacted in the Grail IMS-LD player. Five students and two teachers participated in the case study. The creation of the script required two steps. The first, which was completed with Web Collage, involved the design of the pedagogical method. The method of the script is based on the Jigsaw CLFP, enhanced by the integration of two assessment activities to be carried out by the teachers, consisting of the review of reports generated by the students. These assessment activities were integrated in the script by inserting them in the sequence of activities as shown in Figure 2, and with the objective of supporting the pedagogical method of the Jigsaw. Jigsaw Individual work

Expert phase

Jigsaw phase Assessment

Additional activities

Assessment

Additional activities

Figure 2. Complete script used in the case study. After the pedagogical design had been defined, a second step in the creation of the script was carried out. This step consisted in the integration of IMS-LD features that allowed the implementation of the required flexibility mechanisms. In more detail, this step involved: • The creation of activities whose default visibility is hidden. These hidden activities may be considered as additional activities, so they can be activated in case the assessment of the students’ reports indicates that they are necessary. • The integration of IMS-LD level B properties and conditions in the script. Thus, default visibility is automatically set in a different manner for each group. • The location of initially-empty resources for the additional activities. These resources were created in a Wiki website so the teacher is able to fulfill their content on demand. Wiki documents were attached to activities by a simple URL, and each group receives a different document. The possibility of modifying the resources during enactment was made possible due to the use of this Wiki system, though this mechanism is not supported by the IMS-LD specification. • The organization of the document flow: each group’s report is passed to the teachers; they have to create feedback on the report, which is delivered to the students in the additional activity, if this is activated. • The creation of detailed instructions for the teachers: instructions explain how to access each group’s report, write group-particularized feedback, as well as how to activate and configure the hidden activities if necessary. These instructions are delivered to teachers during enactment: they constitute the description of the activities (support-activities in IMS-LD) that teachers must carry out to complete the script. This configuration of the script is only one possibility to achieve flexibility; in this case study this configuration was chosen with Grail (the player employed in the case study) features in mind. For other players, other mechanisms or alternative implementations of flexibility may be available. This fact increases the complexity of configuring flexibility in the script, as the teacher would need to be aware of the operation of the player. Actually, the second step (adding IMS-LD features for flexibility) was performed manually by a staff member in the case study, without intervention from the teachers. The last item of the list, the detailed instructions, was necessary given the complexity of the usage of the flexibility mechanisms in the player. Moreover, the provision of instructions to make changes in the script reinforces the idea of flexibility being considered as part of the script; in addition, the teachers responsible for introducing changes are seen as participants in the script, rather than external staff. In the real sessions, students were organized in two groups in



41

both occasions when the assessment activities were carried out. Therefore, the second step of the design phase had to take into account this fact: properties and resources were created in order to support two groups, each possibly having different documents in the additional activities. Figure 3 shows the actual learning flow considering two groups. Pedagogical method

Activities in real session

Students’ work

Group A’s work

Group B’s work

Teachers’ assessment and activity modification

Assess Group A

Assess Group B

Additional activities (activated only if necessary)

Additional activities for Group A

Additional activities for Group B

Figure 3. Relationship between the pedagogical method (script design, left) and activities in real session (script instantiation, right). Thus we propose the adoption of this approach of separating the design phase in two steps; the first dedicated to the pedagogical method of the script (for which the teacher is responsible) and the second dedicated to the configuration of the technical details of the script. In this approach, the second step needs to be automatic in order to enable non-expert teachers to take advantage of IMS-LD flexibility affordances in current systems. This second step can be performed automatically by CSCL script authoring tools if the following information is available to be processed by software systems: the type of assessment activity (review of a student’s report); the reports to be assessed (in order to handle document flow); the concrete way in which the assessment results would be used by the teachers (e.g., give feedback to students and activate/modify a later learning activity); and, finally, the number and composition of groups (to handle group-particular additional activities and document flow). By creating the needed components of the script automatically (i.e., activities, properties, IMS-LD monitoring services), group particular resources (e.g., a different activity description for each group), and appropriate instructions to teachers (with information about how to use the LMS and Wiki website to activate and modify the activities), the script could be enacted by teachers without a great knowledge about IMS-LD or the technical operation of the LMS.

Discussion and future work This paper highlights script flexibility as a mechanism that can be embedded in the very same CSCL scripts that may result modified. Under this perspective, flexibility refers to possible changes in the script that are designed beforehand, and are considered as part of the pedagogical method of the script itself. This view is interesting since it creates the opportunity, during the design phase, to prepare the script for the introduction of changes. This is in contrast with flexibility as a reaction to unexpected situations arising during the enactment of learning activities. Assessment plays an important role in this view of flexibility: the assessment plan establishes a series of moments in which flexibility is anticipated to be useful or even necessary. Additionally, the assessment plan indicates what information will be available to decide what changes are to be introduced. One important aspect is that the agent that will introduce the changes, typically the teacher, may have limited knowledge and/or experience with the IMS-LD specification. With respect to this, the case study presented here adopted the approach of hiding the complexity of flexibility mechanisms (in the design, instantiation and enactment phases of the script), and producing a script that actually guides the teacher in the assessment and script modification activities. The operations to create scripts with these features can be automated to some extent. In our case, this was possible given certain information about the pedagogical method of the script, as indicated in the previous



42

section. A tool capable of interpreting this information would be able to create automatically the needed mechanisms and instructions to enable the teachers to introduce changes in the script after the assessment activities. In this way, this approach adds necessary functionality to IMS-LD scripts (and, therefore, to currently available IMS-LD players), while at the same time keeping the expertise demands on the teacher low. In other words, the authoring tool should be usable by teachers who do not know the details of the IMS-LD specification. This automatic script configuration, however, is highly dependent not only on the formal specification used, but also on the differing features of IMS-LD players. This might limit interoperability of the produced scripts, at least with respect to the flexibility mechanisms (the rest of the script components may remain IMSLD compliant). A greater limitation, however, appears due to the fact that only a very specific form of flexibility is employed in the case study: introducing an additional activity and modifying it (independently for each group). Other flexibility mechanisms would lead to completely different results. However, this can be tackled by the identification of typical practices in flexibility. In addition to introducing new activities, other potential flexibility mechanisms are modifying groups, changing the time schedule of the activities, or even larger modifications to the activity flow. Therefore, future work will deal with the identification of potential common practices in script flexibility, and their implementation in the authoring tool Web Collage. In this way, it is ecpe so that flexibility-enabled CSCL scripts may be produced by non-expert designers and focusing on their operationalization of these scripts in currently available systems.

References Black, P., & Wiliam, D. (1998). Assessment and classroom learning, Assessment in Education: Principles, Policy, and Practice, 5(1), 7–74. del Cid, J. E., de la Fuente-Valentín, L., Gutiérrez, S., Pardo, & A., Kloos, C. D. (2007). Implementation of a Learning Design Run-Time Environment for the .LRN Learning Management System, Journal of Interactive Media in Education, Special Issue on Adaptation and IMS Learning Design. Dillenbourg, P. (2002). Over-scripting CSCL: the risks of blending collaborative learning with instructional design. In P.A. Kirschner (ed.), Three worlds of CSCL. Can we support CSCL (pp. 61-91), Heerlen:Open Universiteit. Dillenbourg, P. and Tchounikine, P. (2007). Flexibility in macro CSCL scripts, Journal of Computer Assisted Learning, 1 (13), 1–13 de la Fuente Valentín, L., Pardo, A., & Delgado Kloos, C. (2008). Change is Good. Improving Learning Design Flexibility at Run-time, Crafting didactic materials based on IMS LD: from requirements to evaluation. Workshop on the International Conference on Advanced Learning Technologies, ICALT 2008. IMS Global Learning Consortium. (2003). IMS Learning Design specification. Last access November 2008 at http://www.imsglobal.org/learningdesign/ Kollar, I., Fischer, F., & Hesse, F. (2006). Computer-supported cooperation scripts—A conceptual analysis. Educational Psychology Review, 18(2), 159–185. Koper, R., & Burgos, D. (2005). Developing advanced units of learning using IMS Learning Design level B. International Journal on Advanced Technology for Learning, 2(4), 252-259. Martens, H., & Vogten, H. (2005). A reference implementation of a learning design engine. In R. Koper and C. Tattersall (ed.), Learning Design, a Handbook on Modelling and Delivering Networked Education and Training (pp. 91-108). Springer, Heidelberg. Shepard, L.A. (2000). The role of assessment in a learning culture. Educational Researcher, 29(7), 4-14. Stiggins, R.J. (2002). Assessment Crisis: The Absence of Assessment FOR Learning, Phi Delta Kappan, 83(10), 758-765. Villasclaras-Fernández, E.D., Hernández-Leo, D., Asensio-Pérez, J.I. & Dimitriadis, Y. (2009). Incorporating assessment in a pattern-based design process for CSCL scripts, Computers in Human Behavior, Special Issue on Design Patterns for Augmenting E-Learning Experiences, in press. Zarraonandia, T., Dodero, J. M., & Fernández, C. (2006). Crosscutting Runtime Adaptations of LD Execution. Educational Technology & Society, 9 (1), 123-137.

Acknowledgments This work has been partially funded by Sofocles project (Spanish Ministry of Science and Innovation project TIN2008-03023/TSI). The authors would also like to thank the rest of GSIC/EMIC Group at the University of Valladolid and the Gradient Lab at University Carlos III of Madrid for their support and ideas.



43

Adaptation Patterns in Systems for Collaborative Learning and the Role of the Learning Design Specification Stavros Demetriadis, Ioannis Magnisalis, Anastasios Karakostas Aristotle University of Thessaloniki, POBOX 114, 54124, Thessaloniki, Greece Email: [email protected]; [email protected]; [email protected] Abstract: Research on the impact of adaptive system behaviour as a method to support collaborative learning and improve its outcomes has provided encouraging results and various research teams cope with the pedagogical and technical issues raised in the field. The scope of this paper is to illustrate our theoretical framework for adaptation patterns as the basis for integrating capabilities of adaptive design into systems for collaborative learning. From this perspective we also discuss the role of LD (benefits, possibilities and limitations) as a tool for formalization and support of adaptivity and flexibility during the design and development of adaptive systems for collaborative learning.

Introduction Collaborative learning is important for students both for social and cognitive reasons (Dimitracopoulou & Petrou, 2003). However, the CSCL (Computer-Supported Collaborative Learning) community is well aware of unproductive peer interactions when unsupported groups of learners use the computer as communication medium and try to collectively work towards a learning objective (e.g. Liu & Tsai, 2008). As a remedy it has been proposed that teachers trigger and guide group interaction by implementing specific didactic scenarios (collaboration scripts). Scripts aim to engage students in fruitful learning interactions by providing explicit collective workflow description, guidance and support during the collaborative activity. The enactment of CSCL scripts (computer-based representations of collaboration scripts) has been reported to result in improved learning outcomes (e.g. Rummel & Spada, 2007). Nevertheless, CSCL scripting has been also criticized for its loss of flexibility (difficulty of modifying a script in run time according to the needs of the instructional situation) (Dillenbourg & Tchounikine, 2007), and the danger of “over-scripting” collaborative activity (the pitfall of overemphasizing script imposed interactions and constraining natural collaboration) (Dillenbourg, 2002). Although a teacher can be flexible enough and adjust various collaboration parameters during script run-time, CSCL systems for scripted collaboration are far from exhibiting a comparable level of flexibility. This opens the question of how to adaptively transform various script characteristics during ‘runtime’ and cater for the level of flexibility that real-time script enactment may demand. Consequently, the design and development of adaptive systems for collaborative learning (ASCL) emerges currently as a significant issue at the crossroad of adaptive educational hypermedia and CSCL research traditions. Against this background the scope of this paper is twofold. First, to illustrate our position on integrating adaptive characteristics in systems for collaborative learning based on the notion of adaptation patterns. Second, to discuss the role that the Learning Design specification (LD) is expected to play as a formalization tool in the design of such systems.

Adaptation patterns in systems for collaborative learning Various studies exploring the impact of these adaptation types have reported encouraging evidence regarding their potential to support and enhance collaborative student learning. For example, Gweon et al. (2006) provide research-based evidence in favor of adaptive collaboration support through scripting, when learning in an online collaborative environment. The authors use the term “scripting” to refer to the provision of support to collaborative students in the form of prompts (not necessarily within the framework of a collaboration script). In their study, they show how students increase their contribution over time when they adaptively receive prompts indicating ways of improving their within group interactions. From a similar perspective Tsovaltzi et al. (2008) use the term “adaptive scripting” to describe the situation in which a system wizard, who observes the students as they collaborate, provides adaptive support via prompts sent to the students, to promote explanations, reflection, and help giving/receiving. An interesting issue is that these studies implement a “wizard-of-Oz” research method to explore the impact of adaptive design; that is, the adaptive system is not actually built but some human (a teacher) acts “behind the scenes” simulating the behaviour of the system. This means that building an adaptive system for collaborative learning (ASCL) is far from being trivial. From our point of view, we have emphasized so far the need for a generalized conceptual framework of adaptive scripting, relevant to all types of collaboration scripts, as a basis for formalizing the design of flexible adaptive interventions to support group learning (Demetriadis & Karakostas, 2008). This framework should not only consider the learner’s (or group) characteristics but also the specific characteristics of the implemented



44

script. We suggest that an adaptive system for collaborative learning should satisfy at least three criteria: (a) it is a CSCL system, that is, it somehow supports collaborating groups of students; (b) it includes a user model (learner’s cognitive characteristics and preferences), a group model (data relevant to the synthesis and the dynamics of the group) and a script model (computer-based script representation comprising information on specific script characteristics); (c) finally, it comprises an adaptation model; i.e. a set of rules to initiate some adaptation pattern based on available input. An adaptation pattern epitomizes when and how an adaptation strategy should be implemented to provide improved conditions in a collaborative learning situation. For example, organizing students in groups of mild heterogeneity is a practice generally expected to increase the possibility for fruitful student interaction. A relevant adaptation pattern (a “heterogeneity grouping” pattern) implemented in a script authoring tool (script editor) should offer to teachers the possibility of defining all relevant parameters for implementing heterogeneous grouping at a certain point of their script enactment. An adaptation pattern, therefore, is a process (or a set of processes) that takes into account the user, a group and/or script model (or other modelled entity) and adjusts certain aspects of the collaborative activity in order to maximize student engagement, satisfaction and, consequently, the learning outcomes. For an adaptation pattern at least three issues should be defined: (a) conditions of initiation, (b) aspects of script to be adapted, and (c) processes to be executed. In general, we envision a situation where teachers would be able to select and implement the type of adaptivity they deem necessary in any demanding situation during collaboration. Of course, this generalization leads to the question of how to define what a demanding situation is and how to develop accordingly the needed adaptation patterns. We have proposed and exemplified elsewhere (Karakostas & Demetriadis, 2008) a design methodology (DeACS) for identifying adaptation patterns to be embedded in adaptive scripting systems.

Figure 1. Left: Abstraction of adaptation patterns through applying the DeACS methodology. Right: Integration of adaptation patterns in script design process and enactment. The DeACS methodology proposes three major processes: (a) a top-down process: integration of identified adaptation patterns in the ideal script (the form of the script that the teacher initially wishes to put into practice) based on particular activation conditions, (b) a bottom-up process: identification of adaptation patterns that emerge from students’ needs for help, support, adjustments, etc. during script runtime, (c) an evaluation process aiming to assess the added value of the adaptation patterns in the previous two categories. If the evaluation of patterns reveals beneficial impact on student learning then these patterns can become part of the computerized script representation embedded to the collaboration support system (see Figure 1). Naturally, the important technical challenge is how to link the core non-adaptive pedagogical design of the script (currently supported by various non-standardized script editors) with adaptive design functionalities. Our position on this is that adaptation patterns can be built either as software add-ons or web services that are invoked by a script editor when available (i.e. the software extends its functionalities depending on the available add-ons library or list of web services). The teacher then could integrate the selected adaptation pattern at the appropriate point of the computerized script representation and parameterize the properties and methods of the pattern as desired (Figure 1). In this way the “adaptive logic” can reside at a separate software component



45

(outside IMS-LD manifest) and pedagogically effective UoLs are decoupled from the flexibility it is desired to have under certain circumstances. This way also we take advantage of modifying the adaptive strategy without touching the original pedagogy pattern expressed with LD. We also maintain that not any script feature can be candidate for adaptation. The script “intrinsic” constraints (Dillenbourg & Tchounikine, 2007), that is, the core features that give to the script its specific pedagogical character and value, should not be adapted in any way. Only “extrinsic” constraints can be adapted in order to enhance script flexibility. Extrinsic constraints can be considered as belonging to either of two categories: (a) “Non-pedagogical”, that is constraints without any pedagogical relevance. These can be altered by the teacher and/or the students simply to make the script to better accommodate the conditions of the specific implementation (for example, extending the duration of a phase because of a learner’s temporal inability to meet a deadline). (b) “Pedagogical” constraints that can (should) be adapted in order to provide a well suited learning experience (for example, increasing the level of support when diagnosing learners’ misconceptions). Computerized script representations should clearly define intrinsic and extrinsic (also pedagogical and nonpedagogical) script features, and adaptation patterns should affect only those features characterized as extrinsic. For example, a teacher may decide that working in dyads is an essential script condition. Consequently, he/she should be able to identify this attribute as an intrinsic aspect of the script not to be affected by any adaptation pattern. Having said the above, it is clear that a number of issues should be considered when different types of adaptation need to be supported with some formalization method, such as LD. In the following sections we discuss what the Learning Design (LD) standardization can offer to formally express the adaptive design of collaborative learning activities.

The basics of LD and adaptivity capabilities Learning Design (LD) is primarily a modeling tool which uses the metaphor of a theatrical play for describing a teaching-learning process (Halm et al., 2005). Its main components are: metadata, roles, acts, environment, rolepart (i.e. activities of actor, who does what, when and how), sequence of activities within a role-part, conditions and notifications (interactivity and control over a live learning design as a form of event driven messaging system within an LD player). Through LD tool we formally express a unit of learning (UoL), that is, a complete, self-contained unit of education or training, such as a course, a module, a lesson etc. For completeness we refer to the transformation of simple learning material/assets through Learning Objects to a UoL. Thus, we indicate that learning objects serve as resources within a UoL (or equivalently a LD). To be usable by computers, Learning Design has to be given a concrete syntax and semantics. Thus, we come to Learning Design specification (IMS LD, 2003). LD specification consists of three levels of implementation and compliance and each level is mapped to separate XML Schemas: (a) Learning Design Level A: contains all the core vocabulary needed to support pedagogical diversity. (b) Learning Design Level B: adds Properties and Conditions to level A, which enable personalization and more elaborate sequencing and interactions based on learner portfolios. (c) Learning Design Level C: adds Notification to level B, much like an event-driven messaging system, which provides more interactivity and control during CSCL script runtime. The approach taken in LD specification is therefore not to define a single large schema with a core of mandatory elements and numerous optional elements, but rather to define a complete core that is yet as simple as possible, and then to define two levels of extension that capture more sophisticated features and behaviours. Analyzing the LD structure Burgos et al. (2006) identify three levels of support that the specification can offer to various types of adaptation: (a) well supported (for learning flow, content, evaluation and interactive problem solving support), (b) partially supported (for user grouping, interface adaptation, adaptive evaluation and full modification of a course on-the-fly), and, finally, (c) no support (for dynamic modification of learning structure and method in run-time, and adaptive information filtering and retrieval).

LD limitations Here we criticize LD on the basis of what can be modeled by the specification. We also state some conclusions as to what could be done in order to overcome these limitations. At first, a framework should be established on how pure decentralized P2P interaction models can be incorporated inside LD. This should be done in contrast with centralized P2P models (Halm et al., 2005). Put more clearly, LD should permit to declare whether the LD player itself is web-based and called from a server or it can be on a client machine, runs independently and when in need for communication then it uses networking. What could run on an individual’s machine? For instance, could a machine host dynamically a synchronous VOIP meeting? Could all necessary interaction information of an individual be stored locally and then, when network permits, be transmitted on a server centrally? These and similar questions are interrelated with the idea of ubiquitous learning. That is everywhere –even in everyday workplace– a learning activity occurs; do learners have the chance to learn everywhere in a CSCL sense? Can ubiquitous learning have collaborative characteristics whether synchronous or asynchronous? LD specification at its current version does not



46

incorporate elements that could model such situations. This obviously suggests a possible LD specification enhancement. LD should facilitate representation of “loose” scripts where persons and groups have the opportunity to self-plan, just like a good actor/actress does in a movie. This requirement emerges for instance from the need to model/incorporate games within LD. An educational game is a medium for content rather than the content itself. As a quick answer one could say that here exists a possible role for a game meta-language variant of LD (possibly RDF or OWL/OWL-S), perhaps with extensions of the semantic web variety that will allow a wider range of both representations and operations on those representations (Richards, 2005). In literature we find criticism for the capabilities of LD to express adaptive behaviour. Paramythis A. (2008) concludes that LD offers: (1) No support for modelling groups, (2) no support for modelling artefacts (e.g. a vote, an argument, an answer etc), (3) poor support for dynamic features modelling, (4) poor support for modelling complicated control flow, (5) poor support for modelling social interaction, (5) no exchange of information across UoLs, (6) poor modelling of services and their characteristics (additional services maybe “name-spaced” into the LD specification), (7) acts within plays cannot be re-sequenced or structurally modified. Due to above limitations, LD cannot support alternative policies for role playing, for example assigning more than one roles to the same learner to be played during a specific script phase. Another issue is that we cannot use IMS-LD to maintain collaboration activity data and support the identification of group activity patterns in semantically meaningful way. More limitations are identified by Towle and Halm (2005) including: (1) difficulty of supporting multiple rule interactions (e.g. student profile with multiple characteristics); (2) lack of user/group driven activity ordering; LD is agnostic to the eventual user/group experience (e.g. users’ capability to perform selected activities based on their preferences is not supported); (3) manifest-centered vs. server-centered (LD is a manifest-based representation, so once delivered cannot be changed on the fly); (4) knowledge is embedded in manifest and can not be accessible through metadata for use in new arbitrary strategies.

LD research roadmap Lastly, a few words about some relevant promising research directions. One such direction, in the field of using LD to support CSCL, is to combine initially Web service standards and service oriented architectures (SOA) (Wilson, 2005) with LD. Another direction is the approach of Intelligent Tutoring Systems (Wegner, 1987) in order to build agents to support flexible personalized/“groupalized” learning (Miao & Hoppe, 2005) by using knowledge about the domain, the student and about teaching strategies. We can think of agents as actors of roles in LD. Other research efforts in the area of LD include mapping possibilities of IMS-LD and LMS elements (Burgos et al., 2007) or building systems for managing the creation of a LD by offering to the tutor options that guide the system to propose an appropriate LD. Such an attempt is Collage (Collage, 2009). Work can be done on this or a new tool in order to incorporate extensions of LD (e.g. Web services) and adaptive elements. Attempts are also made towards the direction of building knowledge bases for automatic generation of LDs by the use of rule-based systems (Pacurar et al., 2006). It is challenging to construct systems that go further, adapt a LD during runtime and “explain” the choices of adaptation according to rule bases.

Conclusions and future work In this paper we illustrated our position on the implementation of the DeACS methodology to identify and integrate adaptation patterns to systems for adaptive scripting. Furthermore, we discussed the advantages and limitations related to using LD as a modelling tool to formally express the adaptive design. We are currently working towards identifying possibilities to satisfy our “greed” for LD compliant adaptability and develop system prototypes for adaptive scripting based on the adaptation pattern framework. Another line of our research focuses on exploiting the adaptation capabilities in broadly available systems for collaborative learning such as LAMS (http://wiki.lamsfoundation.org). We conducted a first experiment with LAMS where we implemented two adaptation patterns; one relevant to supporting processing of domain knowledge (DK pattern) and one relevant to enhancing peer interaction (PI pattern). The two patterns were enacted during chat-based communication of student dyads. The DK pattern prescribed that the system adaptively prompts students, reminding them of domain key-words that the dyads had missed in their chat-based discussion. Respectively, the PI pattern prescribed also that the system adaptively prompts the students on how to improve their interaction when needed (for example, offer more explanatory feedback to their peers). Because of technical limitations (LAMS does not offer some means for detecting detrimental online peer interaction) this pattern was not really implemented as an adaptive process but only in the form of on-demand available student scaffolding (i.e. students at certain instances were prompted to see the advices if they felt that this might be useful for them). Despite this limitation, we believe that if the outcomes prove to be encouraging then teachers could seriously consider applying useful adaptation patterns in their collaborative learning designs implemented in LAMS.



47

References Burgos D., Tattersall C., Dougiamas M., Vogten H., & Koper R., (2007). A First Step Mapping IMS Learning Design and Moodle. Journal of Universal Computer Science, 13(7), 924-931. Burgos, D., Tattersall, T., & Koper, R. (2006). Representing adaptive eLearning strategies in IMS Learning Design. In: Proceedings of the 2006 TENCompetence Conference, Sofia, Bulgaria. Collage (2009), Collaborative learning design editor – Collage. Retrieved April 29, 2009 from http://ulises.tel.uva.es/collage/ Demetriadis, S., & Karakostas, A. (2008). Adaptive collaboration scripting: A conceptual framework and a design case study. In F. Xhafa & L. Barolli (Eds.), Proceedings of the CISIS 2008: 2nd International Conference on Complex, Intelligent and Software Intensive Systems (pp. 487-492). Los Alamitos, CA: IEEE Computer Society. Dillenbourg, P. (2002). Over-scripting CSCL: the risks of blending collaborative learning with instructional design. In P. A. Kirschner (Ed.), Three Worlds of CSCL. Can We Support CSCL? (pp. 61-91). Heerlen: Open Universiteit Nederland. Dillenbourg, P., & Tchounikine, P. (2007). Flexibility in macro-scripts for computer-supported collaborative learning. Journal of Computer Assisted Learning, 23(1), 1-13. Dimitracopoulou, A., & Petrou, A. (2003). Advanced Collaborative Distance Learning Systems for Young Students: Design Issues and Current Trends on New Cognitive and Meta-cognitive Tools. THEMES in Education International Journal. Gweon, G., Rosé, C. P., Carey, R., & Zaiss, Z. S. (2006). Providing Support for Adaptive Scripting in an OnLine Collaborative Learning Environment. Paper presented at the SIGCHI conference on Human Factors in computing systems. Halm, J., Olivier, B., Farooq, U., & Hoadley C., (2005). Collaboration in Learning Design Using Peer-to-peer technologies. In: R. Koper & C. Tattersall (Eds.), Learning Design: A Handbook on Modelling and Delivering Networked Education and Training (pp. 203-213). Berlin-Heidelberg: Springer Verlag IMS LD (2003). IMS Global Learning Consortium: Learning Design Specification. Retrieved April 29, 2009 from http://www.imsglobal.org/specifications.html. Karakostas, A., & Demetriadis, S. (2008). Systems for Adaptive Collaboration Scripting: Architecture and Design. In A. Paramythis, & S. Weibelzahl (Eds), Proceedings of Adaptive Collaboration Support Workshop, 5th International Conference on Adaptive Hypermedia and Adaptive Web-Based Systems (pp. 7-12). Retrieved February 1, 2009, from http://www.ah2008.org/index.php?section=62. Liu, C.-C., & Tsai, C.-C. (2008). An analysis of peer interaction patterns as discoursed by on-line small group problem-solving activity, Computers & Education 50, 627-639. Miao, Y., & Hoppe, U. (2005). Adapting Process-Oriented Learning Design to Group Characteristics. Proceedings of Artificial Intelligence in Education, Looi, C., McCalla, G., Bredeweg, B., Breuker, J. (eds.) IOS Press, 475-482 (2005). Pacurar, E. G., Trigano, P., & Alupoaie, S. (2006). Knowledge base for automatic generation of online IMS LD compliant course structures. Educational Technology & Society, 9 (1), 158-175. Paramythis, A. (2008). Adaptive Support for Collaborative Learning with IMS Learning Design: Are We There Yet? In A. Paramythis, & S. Weibelzahl (Eds), Proceedings of Adaptive Collaboration Support Workshop, 5th International Conference on Adaptive Hypermedia and Adaptive Web-Based Systems (pp. 17-29). Retrieved February 1, 2009, from http://www.ah2008.org/index.php?section=62. Richards, G. (2005). Designing Educational Games. In: Koper, R. & Tattersall, C., Learning Design: A Handbook on Modelling and Delivering Networked Education and Training (pp. 227-237). BerlinHeidelberg: Springer Verlag. Rummel, N., & Spada, H. (2007). Can People Learn Computer-Mediated Collaboration by Following A Script? In F. Fischer, I. Kollar, H. Mandl & J. Haake (Eds.), Scripting Computer-Supported Collaborative Learning (pp. 39-55). New York: Springer Verlag. Towle B., & Halm M. (2005). Designing Adaptive Learning Environments with Learning Design. Learning Design: A Handbook on Modelling and Delivering Networked Education and Training (pp. 216-226). Berlin-Heidelberg: Springer Verlag. Tsovaltzi, D., Rummel, N., Pinkwart, N., Scheuer, O., Harrer, A., Braun, I. & McLaren, B.M. (2008). CoChemEx: Supporting Conceptual Chemistry Learning via Computer-Mediated Collaboration Scripts. In P. Dillenbourg and M. Specht (Eds.): Proceedings of the Third European Conference on Technology Enhanced Learning (EC-TEL 2008), Lecture Notes in Computer Science 5192 (pp. 437448). Berlin: Springer. Wegner E. (1987). Artificial Intelligence and Tutoring Systems, Morgan Kaufman, San Francisco. Wilson S. (2005). Architectures to Support Authoring and Content Management with Learning Design. Learning Design: A Handbook on Modelling and Delivering Networked Education and Training (pp. 41-62). Berlin-Heidelberg: Springer Verlag.

WORKSHOP Scripted vs. Free CS collaboration

WORKSHOP Scripted vs. Free CS collaboration

Suggest Documents

Scripted collaboration in serious gaming for complex ... - CiteSeerX

Building Community Collaboration Workshop

Computer-Supported Collaboration in a Scripted 3 ... - Semantic Scholar

Communication vs. Computation - CS Technion

European-American Collaboration Workshop - Semantic Scholar

CHI 2008 Sensemaking Workshop paper - CS Stanford

Scripted Reading Instruction? - Pdkmembers.org

the mason institute - SCRIPTed

SCRIPTed book review template

Classicism vs. Connectionism Debate Revisited - CEUR Workshop

Free VDI Workshop

Android Free Workshop

Opt-In Dystopias - SCRIPTed

Elaine Kasket - SCRIPTed

Peer Surveys - SCRIPTed

The Crusades - Scripted Genius

Download PDF - SCRIPTed

Download PDF - SCRIPTed

Download PDF - SCRIPTed

Elaine Kasket - SCRIPTed

Download PDF - SCRIPTed

Data Protection - SCRIPTed

Opt-In Dystopias - SCRIPTed

Opt-In Dystopias - SCRIPTed