MUPEMURE: Towards a Model of Computer-Supported Collaborative Learning with Multiple Representations

Daniel Bodemer, Manu Kapur, Gaelle Molinari, Nikol Rummel, Armin Weinberger

Manu Kapur

Students need to understand and apply different representational formats (tables, graphs, algebra) (Ainsworth 2009)

• benefits and difficultires of learning with multiple representations (Cuoco & Curcio 2001)
• collaborative learning with multiple represnetaitons (Rummel & Braun 2009)

Self directed learning (Leuders, Hussman, Barzel & Prediger 2011)

Collaborative problem solving to yet unknown concepts

• most likely results in incomplete solution
• productive failure? (Kapur & Kinzer, 2009Kapur, M., & Kinzer, C. K. (2009). Productive failure in CSCL groups. International Journal of Computer-Supported Collaborative Learning, 4(1), 21--46. Springer.) evidence for collaborative learning from multiple problems

phase 1

• collaborative problem-solving to a yet unknown concept without content-related support or instruction
• generation and exploration
  • →
• teacher-led compare & contrast discussion leading to canonical solution

the more students generate in first phase, the more they learn (Kapur & Bielaczyk). We know it is productive, but don't know why.

first study

• two productive failure conditions
  • learning phase, collaborative problem-solving phase, groups of 3, tablet PCs, video/audio recording
    • what kind of support do students need to generate multiple representations
      • cognitive prompts
        • you are doing a good job together (condition PF)
      • motivational prompts
        • maybe there are situations where this doesn't work, look at this counter-example (condition PF+)
      • direct instruction
        • standard, based on correct representations (DI)
        • instruction based on typical student-generated representations (DI-S)
  • instruction based on typical student-generated representations in phase 2

results

• procedural
  • DI was slightly better, they were more effective
• conceptual
  • DI-S much better than DI, PF only slightly better than DI-S.

No effect for cognitive support. Persistence in collaborative problem-solving more important than specific cognitive activities and representations.

Important to include students' pre-conceptions in instruction, but you need to know what their preconceptions are, which means PF is useful.

Students in Singapore and Germany generated similar representations, pre-conceptions universal?

Quality of representation influence learning outcome?

• another project split high and low quality representations, which made the data more meaningful
• number of representations moderate learning?
• how do students collaboratively generate representations?

Daniel Bodemer, U Tubingen

Support learners in learning with multiple representations.

• translating between representations
• processing large amounts of interconnected information
• using varied information in systematic way

supporting translation between MERs,

• pre-integrated or linked representations (Chandler & Sweller, 1991; Kozma 1996)
• encouraging active integration (Bodemer)

work well, if you compare to pre-integrated formats - leave learners in passive situations. In these tasks, learners have to actively integrate representations mentally and externally.

• Requires external and mental translation processes
• Reduces complexity step by step
• Directing learners' attention to “unsolved” components

• Establishing references between external content and collaboration content (Dillenbourg and Betrancourt 2006)
• Constructing mutual understanding and common ground (Clark & Brennan 1991)
• Interacting in a structured and goal-oriented way (Bromme et al 2005)

Facilitating group awareness, collaborative integration tool

in the middle, statistics visualization, algebraic components, learners have to drag and drop to visualization

group awareness: visualization of integration of learning partner

• constrains content-related communication space
• provides cognitive group awareness information
• guides discussions on the basis of four cases of knowledge distribution

Experimental study

• spatially separated dyads
• individual learning phase
  • visual and algebraic on stats
• collaborative phase
  • text based chat and integrationt ool
  • displaying learner's and learning partner's integration
  • displaying learner's integration only
• better learning and task performance with awareness suport

Adaption of rainbow-scheme (Baker et al 2007)

6 categories

• outside activity
• social relation
• task management
• MER grounding → asking for, telling references
• content
• broaden & deepen → interactive elaboration

Broaden & deepen

• in control condition
  • almost no elaboration on conflicting perspectives (what makes collaboration productive)

Tacit guidance → efficient and germane communication

• reduced extraneous grounding activities
• discussion of conflicting knowledge / controversial perspectices
• collaborative elaboration

Representational guidance / informational guidance

• empty drop areas (salience)
• controversial content

Gaelle Molinari

external representation - concept map, can be self-generated by learners to visually represent how they organize individual knowledge in memory. Can be powerful meta-cognitive tool to help learner externalize, reorganize and foster communal knowledge.

Can be created by groups

When used collaboratively, can anchor learners to engage in knowledge co-construction.

Multiple representations and multiple perspectives.

• Have to read text, and translate it into a personal concept map
• Comparing, transferring, transformative, constructive processes
• Creating group map

Conditions

• Read the same text
  • compare and confront OR
  • competence threat
• read different but complementary (20% overlapping)
  • more collaborative effort
  • dependent on the partnet

How learners use and coordinate visualizations of their respective prior knowledge (personal maps) to create a collaborative map

individual to group transfer: old concepts and links from personal maps to collaborative map group creativity: creation of new concepts and links

to what extent does sharing either identical or complementary information resources influence group transfer and creativity

use eyetracking to analyze building of collaborative map

you can see your own map, your partner's map, and your collaborative map

learners were more similar to partner with respect to outcome knowledge level, when they worked with the same information

even when text was the same, overlap between personal maps was quite low (similar to with different information).

There was little difference in amount of information transferred from individual to collective map.

The more learners spent time consulting their own map, the less they built together on common ground.

Gijlers, can Dijk, Armin Weinberger

How do learners create join representations of a science text? Learners share self-generated drawings and coordinate the creation of one joint presentation

young learners first steps towards knowledge representations and abstractions

externalization of knowledge that helps

• disambiguating learners' understanding
• identifying misconceptions
• working memory offloading
• self-explanation
• increased and shared task focus
• improved elaboration when drawing for others

how learners' elaborate better when they are communicating and drawing for others - learning effects in collaborative scenarios

technical problem

• sharing
• revising
• reusing

learning problem

• sharing external representations is not sufficient for learners to arrive at shared understanding

two kinds of instructional support

• awareness features
  • what can learners be aware of
    • presence / activities of others (Prinz 2001)
    • knowledge states of others (Engelman et al 2009)
    • workspace (Gutwin et al 1996)
    • group structure (Gutwin & Greenberg 2002, Gutwin et al 1996)
  • real-time tracking and representation of learners' process data
• scripts (has often been contrasted with awareness)
  • provide normative model (whereas awareness features - this is what you are doing, you can decide what to do about this meta-information)
  • scripts
    • specify
    • sequence
    • distribute
  • roles and activities in collaborative learning
  • families
    • jigsaw
    • epistemic
    • peer tutoring
    • peer review (all: Weinberger et al., 2005Weinberger, A., Reiserer, M., Ertl, B., Fischer, F., & Mandl, H. (2005). Facilitating collaborative knowledge construction in computer-mediated learning environments with cooperation scripts. Barriers and biases in computer-mediated knowledge communication, 15--37. Springer., Weinberger et al., 2007Weinberger, A., Ronen, M., Tchounikine, P., Harrer, A., Dillenbourg, P., HaakeJ\"{o}rg, Kali, Y., Fischer, F., & Kohen-Vacs, D. (2007). Languages and platforms for CSCL scripts. Proceedings of the 8th iternational conference on Computer supported collaborative learning, CSCL'07. International Society of the Learning Sciences. , Weinberger et al., 2007Weinberger, A., Clark, D., Dillenbourg, P., Diziol, D., Sampson, V., Stegmann, K., Rummel, N., Hong, F., Spada, H., McLaren, B., Brahm, T., & Fischer, F. (2007). Orchestrating learning activities on the social and the cognitive level to foster CSCL. Proceedings of the 8th iternational conference on Computer supported collaborative learning, CSCL'07. International Society of the Learning Sciences. )

Awareness and scripts

• macro-scripts
  • changing learners' awareness apriori, eg. about their roles and workflow, before they begin to collaborate
    • when a learner knows that his product will be reviewed by a peer, you will cognitive engage more in the process
  • adaptive micro-scripts
    • just as awareness widget, on automatic runtime analyses and inform learners about deficits including an instruction about what to do with it, for example “have you considered all relevant concepts?”

Experiment

• participants Dutch 11 year olds
• photosynthesis
• jointly produce shared representation of photosynthesis that has explanatory value to peer
• introductory texts

platform

• shared workspace
• logging learner actions and drawing
• annotations
• clustering and sketch-recognition algorithms
• supportive prompts

phases

• individual drawing phase
• review each other's drawing in split screen and integrate components of two different drawings
• agree on and finalize common drawing

Coding

• how learners drew
• open recall texts

result

• quality significantly higher of scripted students
• awareness was worse than control condition
• cued and open recall: both experimental groups significantly higher than control group

Discourse analysis

• script
  • more frequent use of annotations as a means to manage to conflict
  • using the learning partner as an additional resources
• awareness feature
  • more frequent looking up the learning material
    • might have improved individual learning, but disrupted joint drawing quality

Kristine Lund (ICAR, U Lyon)

Theory ↔ MUPEMURE model (predictive) ↔ experimental field (CSCL and MERs)

Ability-treatment interactions (Snow, 1994)

Abilities are affordances.

• people are tuned in (or not) to affordances
• affordances invite actions, but a person is capable of effectivities

affordances in academic tasks requires analysis of the affordance-effectivity pattern matches of different learners and different instructional tasks and treatments