Collaborative concept mapping in a web-based learning environment: a pedagogic experience in architectural education.

This article describes a pedagogic work, carried out within a School of Architecture, using a web-based learning environment to support collaborative understanding of texts on architectural theory. Five architectural manifestoes of the European avant-garde were selected to be analyzed. First, each student interpreted one text, explaining the content with a multimedia presentation and summarizing the main ideas with three concepts. The individual work was followed by the collaborative work, which consisted of the construction of a critical vocabulary and a concept map. The collaborative work was carried out using a web-based system specifically created for the course. As students submitted their concept definitions to the system, they collaboratively created a critical vocabulary, summarizing the main ideas of the five texts. Then, students established relationships between pairs of concepts, which were visualized with a concept map. Subsequent exploration of the semantic spaces embodied in the map became a so urce of knowledge for both students and teachers. In this context, the concept map became an abstract machine, an artificial construct that allowed learners to produce meaning by interaction with it.

INTERTWINING PEDAGOGY AND TECHNOLOGY

This pedagogic experience took place during the academic year 2000/2001, within the course Sistemas de Representacion SDR, in the third year class of the Escola Tecnica i Superior d'Arquitectura La Salle, in Barcelona, Spain.

The course SDR is part of an innovative pedagogic project whose aim is to integrate--in pedagogically meaningful manner--information and communication technologies into the education of Architecture. The concept of Representation constitutes the subject matter of this course, which is structured in six themes: (a) text, (b) shape, (c) object, (d) image, (e) space, and (f) light (Figure 1). The theoretical content is made up of relationships between 'theory bits' taken from different disciplines: fundamentally architecture, but also art, graphic design, communication, aesthetics, psychology, philosophy, and computing. The theoretical base of the course, therefore, conforms to the nature of the Web, where relationships between items become more important than the items themselves.

Also in accordance with the spirit of the Web, exercises are carried out in an individual and collaborative manner, using the web-based environment SDR:NETWORKING (http://www.salleurl.edu/sdr), which has been specifically created for the course. The environment is divided into six modules, suited to the requirements of each one of the six themes. Thus, with SDR:NETWORKING [TEXT], students can analyze texts collaboratively by means of a concept map. With SDR:NETWORKING [OBJECT], they can carry out a process of form generation, collaboratively developing a three-dimensional object in successive stages. Also with SDR:NETWORKING [SPACE], they can establish associations among spatial units, which then give rise to cyber-narratives (http://www.salleurI.edu/sdr/info).

SDR:TEXT

The theoretical construct of the theme TEXT, is built up from different subject matters: linguistics (syntax and semantics, semantic triangle), cognition (semantic networks, concept maps), graphic design (posters, advertisement, media), interactive media (interface design, interactivity) and architectural theory (aesthetic principles of the Modem Art and Architecture). The exercise consists of the analysis and interpretation of five texts from the European artistic avant-garde of the early twentieth century: (a) Henry van de Velde, Credo, 1907; (b) Walter Gropius, Bruno Taut and Adolf Behue, New Ideas on Architecture, 1919; (c) Theo van Doesburg and Cornelius van Eesteren, Towards Collective Building. Commentary on Manifesto V, 1923; (d) Ludwig Mies van der Rohe, Industrialized Building, 1924; and (e) Le Corbusier and Pierre Jeanneret, Five Points Towards a New Architecture, 1926. The text analysis is first carried out individually, and then in collaboration using the web-based environment SDR:NETWORKING [TEX T]. Altogether, the time dedicated to the theme is 6 weeks. (1)

Individual Work: Interpreting and Summarizing an Architectural Manifesto

In the first part of the exercise--the individual work--each student chooses one of the five texts, analyzes its content, and explains it in public with a multimedia presentation. Students are expected to make personal and imaginative use of digital media to communicate, in the most effective way, what they have understood from the text. For example, Figures 2 and 3 show two screens of n web presentation explaining the Five Points Towards a New Architecture, from Le Corbusier. Content and expression are intelligently intertwined in this work: the five points (the basic principles of Modern Architecture) are depicted as the base on which the architectural theoretical body is built up. As the user interacts with the interface, the names of the five points (pilotis, roof garden, free plan, horizontal window, free facade) appear as vertical columns conveying the idea that architectural theory is constructed step-by-step, through practice, as Le Corbusier contends. At the end, the image of the Villa Savoye--the ic on of the Modem Architecture advocated by Le Corbusier-emerges: theory and practice become one.

As a conclusion of this part of the exercise, students must propose three concepts that summarize the text they have studied, which are submitted to the web-based learning environment. Then, it starts the collaborative part of the exercise.

Collaborative Work: Vocabulary and Concept Mapping

The collaborative work is divided into two major tasks: (a) building a vocabulary, and (b) establishing relations between pairs of concepts.

TASK 1: Building a critical vocabulary. Students submit the three words (2) and their corresponding definitions to the web-based system SDR:NETWORKING [TEXT] (Figure 4). These words do not have to appear explicitly on the text. Preferably, students should come up with their own terms to express the main ideas discovered in the text. Similarly, rather than looking up in the dictionary the meanings of the concepts they are proposing, students must use their own words to define them in relationship to the analyzed text.

As students submit their concept definitions, a critical vocabulary is collectively created in the web space. (3) A simple user-interface allows students to have access to the vocabulary (Figure 5). As words are introduced, they appear in the list on the right side. The size of the font indicates the number of definitions of a word: the larger the font, the bigger the number of definitions. Obviously, the fact that a word has got more definitions than others does not make it more relevant, from a conceptual point of view. It only gives us an indication of the areas of words on which students are concentrating meaning. After selecting a word in the list, its definitions are displayed in the main window. For example, by clicking on the word MATERIAL one can see the different meanings that students have assigned to it: "What makes form possible;" "What sets bounds to form." It is also possible to see concepts in other ways. For example, selecting one of the authors (e.g., Gropius, Le Corbusier, etc.) will give a list of all the words associated to one text (Figure 6).

At this stage of the exercise, the vocabulary has become the depository of a common knowledge, out of which more knowledge can be elicited. As students browse through the vocabulary, they can begin to figure out relationships between the texts by comparing definitions of the different concepts. They can realize, for example, that MATERIAL was a common preoccupation in the texts of both Van de Velde and Mies, although it has different connotations for each one. This way, the individual text is no longer the object of study, as it was at the beginning of the exercise. Rather, the goal now is to understand--in collaboration with the other students--the ideas that underlie the five texts and the links between them.

This simple web-based workspace, which allows students to construct a vocabulary collaboratively, has effectively contributed to promoting discussion and debate in and outside the class. Rather than using the web system as surrogate of the classroom-as in distance learning--it has been integrated it into the dynamics of a group. In this hybrid model, the role of the educator as interface between students and system is fundamental for the success of the whole pedagogic experience. For instance, as the vocabulary develops, the educator must help to unveil the meanings of the concepts and relations that are being built up in the system. (4) Without the educator acting as a bridge between the students and the system, there is a high risk that a web-based learning environment becomes a pure depository of information, as opposed to a distinct source of knowledge.

TASK 2: Relating pairs of concepts. Once the collective vocabulary has been created and discussed in the classroom, students proceed to create relationships between pairs of concepts; more precisely, between pairs of concept definitions. This task is performed on an interface where the vocabulary is listed on left-hand and right-hand sides (Figure 7). After comparing the definitions, students can set up relationships between concepts, describing them with a short text.

Discovering a relation between two terms, related to different texts and defined by other students, is a demanding task which tests the language skills of a student, as well as his or her knowledge of the underlying theoretical issues. For example, these two definitions of MATERIAL ("What makes form possible"), and FORM ("The expression of function"), were related by a student on these grounds: "Material gives us the form. The form is in relation with the function of the design." The relation the student proposed suggests that there are two apparently conflicting requisites to which FORM must conform: the MATERIAL and the FUNCTION.

All sorts of possibilities can occur in the construction of a triad concept-relation-concept. For example, it can happen that the same student has been the author of the two concept definitions and the relation. The most frequent case, however, is that a third student establishes a relation between two concepts that have been defined by other peers. In this collaborative scenario of knowledge construction, authorship is no longer an issue. Ideas and arguments emerging from this process cannot be attributed to a single person: they are the result of a collaborative effort.

The dynamic process that takes place as students establish relationships between concepts is captured by the web system. As they log in, students receive information regarding the "activity" of their concepts. For example, the system reports that "student 1" has related one of his or her concepts to another concept defined by "student 2." This way, the web system contributes to promoting the debate among students (within the system, but also in the classroom) and the collaborative development of ideas.

Concept Maps in Education

As the class carries out the process of matching pairs of concepts, a semantic space is collaboratively created which will then be visualized as a concept map. Concept maps are visual representations of a semantic space, made up of nodes connected by arcs. Other names are also used to refer to this kind of representation, for example, cognitive map, semantic network, and knowledge maps. Even though a clear-cut distinction among the different terms would be difficult to establish, each has distinct connotations. According to Sherratt and Schlabach, (1990),

Concept mapping involves identifying concepts or ideas pertaining to a subject, and then describing the relationships that exist between these ideas in the form of a drawing sketch. The purpose of the map is to represent an individual's understanding of a body of knowledge, and to illustrate the relationships among ideas that are meaningful to him or her.

This representation of an individual's knowledge could be referred as a cognitive map, and it should be distinguished from a concept map. Thus, according to Novak and Gowin (1984), a cognitive map is "a representation of what we believe to be the organization of concepts and propositions in a given student's cognitive structure. Cognitive maps are idiosyncratic, whereas concept maps should represent an area of knowledge in a way that experts in that field agree is valid" (p. 138). A semantic network is, according to Sowa, "a graphic notation for representing knowledge in patterns of interconnected nodes and arcs....a declarative graphic representation that can be used either to represent knowledge or to support automated systems for reasoning about knowledge." In contradistinction to the "informal" nature of a cognitive map, a semantic network is based on a formal language. Finally, the term knowledge map has been used to refer to the application of concept maps as pedagogic tools that help a community of lea rners to derive knowledge from them in collaboration. (Lambiotte, Dansereau, Cross, & Reynolds, 1989)

Applications of conceptual maps as tools to support learning can be found in almost all levels of education. In the basic school level, Novak and Gowin (1984) developed a system of concept maps for the evaluation of students' learning; Scardamalia et al. (1992) created CSILE, a networked communal database designed to encourage group processes and progressive discourse; and Suthers, Erdosne, & Weiner, (1997) developed the Belvedere system, a shared workspace for coordinating and recording collaboration in scientific inquiry. In higher education, we can find examples of applications of conceptual maps in specialized fields like biology, physics, mathematics, management, physiology, nursing and many others. (5)

Concept maps are particularly well suited to support a constructivist model of education, whose aim is the active involvement of students and teachers in a collective process of knowledge construction. In a constructivist scenario, concept maps become representation tools with which knowledge--at the individual and collective levels--can be externalized. An example of this approach is the KMap developed by Gaines and Shaw (1995), a tool that allows students to develop "consensual maps," which become representations of a common knowledge acquired by negotiation. As McAleese (2000) contended: "[In the construction of conceptual maps] learners will come to negotiate and modify, what it is that they appear to know in terms of what is known." In this regard, a conceptual map is more than a visualization to facilitate communication among individuals: it is an abstraction with which knowledge can be discovered.

Constructing the concept map collaboratively. The concept map tool we have developed for our web environment is rather straightforward and simple. Basically, it visualizes the relationships between concepts pairs collaboratively created by a group of users. The concept map is constructed step-by-step, as the user explores a domain of the network of relationships (Figure 8). It is both a cognitive map and a knowledge map tool, considering the previous discussion about concept map definitions.

At the outset, a student selects a concept definition from the list of words that have been already related to other words. As a result, a node representing the selected concept definition is placed at the center of the graphic window. This concept becomes the focus of the map. A color-coding indicates to which of the five source texts the concept is related (Van de Velde: yellow, Gropius: green...). The "+" sign appended to the word tells that there are other concept definitions associated with it. Toggling on and off a node allows one to see or hide all the concepts related to it. After expanding the relations twice, it is necessary to replace the focus node with another one to continue exploring the map. Then, the user can keep expanding the network two more levels. (6)

Once the user has delimited a domain of words in the concept map, it is possible to switch to the mode "RELATIONS" to get further information about the concept relations (Figure 9). In this mode, the lines connecting the nodes can be selected. When a line is selected, the definitions of the triad concept-relation-concept are displayed at the lower part of the window.

While inquiring about the relationships, students are challenged to come up with argumentations that can be built up with the set of words that make a particular domain. For example, if we begin selecting the word MATERIAL ("What makes form possible"), clicking on the word shows two definitions of FORM that have related to it: "The expression of function;" "The external appearance of an object, derived from the constructive process." All three nodes (MATERIAL, FORM, FORM) have the color that is associated with the text from Van de Velde. Indeed, the question of the relation between form, material, and essence of an object is raised in that text. Two different students explain the relations between MATERIAL and the two corresponding definitions of FORM in these terms: "Material gives us the form. The form is in relation with the function of the design;" "Construction and material define form." If we continue expanding the map, clicking on each node FORM, we get the words FUNCTION and STRUCTURE, this last one a ssociated to the text of Le Corbusier. At this point, a discussion of Le Corbusier and Van de Velde's ideas come into play. For Le Corbusier, structure mediates between material and form; Van de Velde, on the other hand, advocated a direct relation between both. Then, following the thread of the relation FORM/FUNCTION, we come across these relations: "Form should be subordinated to function" and "Function is defined by the form." From the causal relation between FORM and FUNCTION arise certain questions: can function determine form? Can a design serve different functions? These are the types of issues that are then discussed in class, following the arguments that can be elicited from the concept map.

Knowledge discovery through navigation. Interacting with the map gives students the opportunity to discover ideas and associations immersed in a semantic space that has been collectively constructed. The knowledge obtained from navigating through the map is the result of the interaction between what is represented by the concept map and what is known by the user. In this negotiation process, as McAleese (2000) has argued, "[concept maps] are both representations of what is known by a learner and what is 'knowable.' Concept maps are snapshots of what the learner chooses to think about." In other words, knowledge is not in the map itself, but it is derived from it. In this regard, the concept map works as an "abstract machine," in the Deleuzian sense, that is to say, "a blind and mute machine that makes others see and speak." This is particularly true in a collaborative scenario, where the concept map is the reflection of the collective knowledge of the class, rather than the expression of a personal thought. I ndeed, it is not possible for an individual to get a complete view of all the rich net of associations potentially contained in the concept map: only glimpses of that "overall knowledge" can be discovered as one interacts with it.

Enhancing the concept map. At it stands now, the concept map we have implemented works mostly as a visual output of the relationships established among concepts. Future enhancements will allow users to add and remove nodes on the map, and define groups of nodes working directly on the visual representation. This way, the concept map would become more flexible and would-fit better to the way of thinking of the user (8). Also, the expressive power of the concept map tool will be enhanced with tools to edit attributes (color, font, size) of nodes and lines. As Wallace, Wandell, Ware, & Dansereau (1998) have shown, the effectiveness of knowledge maps can be enhanced applying Gestalt principles of similarity (of color and shape) and proximity (grouping by attributes). Altogether, these improvements will increase the expressive power of the concept map tool.

Still a more challenging issue, to be considered in future developments of the system, is the concept-to-spatial mapping (9). As it is well known, a key aspect in concept mapping is to assign meaning to the spatial location of words (i.e., what does it mean to have words more or less separated, or to be above or underneath each other,(10). This begs the question of the notion of dimension and the definition of the context or domain within which a dimension becomes meaningful."

Collaborative Text Analyses: From Text to Metatext, Fram Knowledge to Metaknowledge

Following the strategy adopted for the exercises, the complexity of the analysis increases as the work advances. At the beginning of the exercise, one student is confronted to one text. Then, through the vocabulary collectively created, students are exposed to the issues that underlie all five texts. At this point, a "metatext"--containing the collective interpretation of the five texts--begins to emerge from the network of concepts. Later, the relations between concepts, rather than the concepts themselves, become the focus of study. A "metaknowledge" then starts to be created, its subject matter being not so much the content of the texts but the modes of thinking about them.

This sequence, moving from text to metatext, from knowledge to meta-knowledge, is independent of the subject matter of the study. Indeed, a similar process could be followed if the texts dealt with literature or philosophy, instead of architecture. This means that the system we have implemented can capture the basic structure underlying a collaborative process of knowledge construction, but it might fail to represent what is specific to a particular discipline.

CONCLUSIONS

The learning environment presented is based on the integration of three distinct components: (a) a theoretical framework, built up from "theory bits" that are taken from different disciplines, (b) individual and collaborative exercises, which allow students to participate in the construction of knowledge, developing further the theoretical issues with their practical work, and (c) a web system, which provides representations of the collective work. Altogether, the integration of these three components, gives rise to a pedagogic methodology, which could be easily applied to other disciplines and to other levels of education.

The effectiveness of this learning environment, however, relies on the equilibrium between technology and pedagogy. Technology must be subsumed under a pedagogic program, whose ultimate goal is to develop the capacity of students to think creatively in collaboration, using information and communication technologies. Straightforward interfaces and simple concept map tools, such as the ones presented, have proved to be sufficient to make students active participants in a collective process of knowledge construction. Future enhancements of the system should not break the balance that is believed to have been achieved between technological means and pedagogic goals.

Acknowledgements

The concept map environment has been programmed by Jordi Vidal using Java and CGI PerI scripts to access a relational database. The vocabulary environment has been programmed by Francesc Duran with DHTML and CGI Pen scripts, accessing the database.

Notes

(1.) The antecedent of the work presented here is the pedagogic web-based environment AALTO (http://caad.arch.ethz.chlaalto), which was developed in conjunction with a postgraduate seminar taking place at the ETH Zurich in 1997 and 1998, dedicated to the study of the works and ideas of the Finnish architect Alvar Aalto. In that environment, students represented Aalto's ideas as an associative network of concepts and images. (Madrazo and Weder, 1998)

(2.) As long as the distinction can be maintained, "words" refers to the linguistic sign or signifier, and "concepts" refers to the explicit description of the signified.

(3.) A significant precedent in the application of computer networks to support collaborative learning, was the communal database of the CSILE system developed by Scardamalia and Bereiter (1993). With this tool, initially thought for primary education, students could collectively construct knowledge by storing word definitions in a database, which could be then retrieved by other peers, to be modified or commented. As the authors contended, "With a communal database, the focus is on a body of knowledge being constructed 'in here' (that is, in the classroom's database), with information from 'out there' being used in the construction. Thus, the shift is an epistemological and not merely a methodological one." This constructivist theory of education is also supported by the system developed by the authors.

(4.) The authors agree with the conclusions drawn by Scardamalia and Bereiter (1993) after their experience with CSILE, as they contended that "Teachers do no direct discourses but engage in them, leading by virtue of being more expert learners." Indeed, in a constructivist learning scenario, such as the one created for this course, both teachers and students are part of a learning community whose aim is to build knowledge in active cooperation.

(5.) See Special Issue on Concept Mapping in (1997), Journal of Interactive Learning Research, 8(3/4)

(6.) Spatial constraints are one of the key issues in concept map visualization. Different solutions have already been proposed to overcome the limitations of a space projected on a plane surface. With hyperbolic geometry, for example, it is possible to navigate through a map in a continuous manner (Lamping, Rao, & Pirolli, 1995).

(7.) "The diagram is no longer an auditory or visual archive but a map, a cartography that is coextensive with the whole social field. It is an abstract machine. It is defined by its informal functions and matter and in terms of form makes no distinction between content and expression, a discursive formation and a nondiscursive formation. It is a machine that is almost blind and mute, even though it makes others see and speak" (Deleuze, 1988).

(8.) A study of Heeren and Kommers (1992) on concept maps found it necessary to have flexible representations that can easily be controlled by the learner. Certainly, there is a trade-off between the need to adopt a communication language embedded in the graphic map which is assumed by all users and the need of an individual to come up with a personal language to express his or her thoughts.

(9.) We have started to tackle this issue with the students in our Ph.D. program who have used the same web-based environment for their text analyses. In the concept maps they created, words were purposely placed in a certain position with regard to each other to create concrete meanings.

(10.) As Landauer and Psotka (2000) recalled "...words occupy positions in a 'semantic space' and their meaning is the relation of each word to all the others.... However, the implicit cognitive processing involved goes far beyond piece-wise association. It takes all the myriad local relations and rearranges them to fit them together into a single consistent map, a semantic space that represents how each object, event or word is related to each other" (p. 75).

(11.) The similarity between two concepts is often related to their distances (metric or phenomenological). But, as Gardenfors (2000) suggested, dimensions do not have an independent, objective value but are dependent from other dimensions. For Gardenfors, "A set of integral dimensions that are separable from all other dimensions" constitutes a domain, whereas a conceptual space is defined "as a collection of one or more domains" (p. 26). Then, learning new concepts would be tantamount "to expanding one's conceptual space with new quality dimensions" (p. 28).

References

Deleuze, G. (1988). Foucault. Minneapolis, MN: University of Minnesota.

Gardenfors, P. (2000). Conceptual spaces. The geometry of thought. Cambridge and London: The MIT Press.

Gaines, B.R., & Shaw, M.L.G. (1995). Collaboration through concept maps. Proceedings of CSCL '9S. Calgary. Bloomington, IN: Indiana University.

Heeren, E., & Kommers, P. (1992). Flexibility of expressiveness: A critical factor in the design of concept mapping tools for learning. In P.A.M. Kommers, D.H. Jonassen, & T. Mayes (Eds.), Mind tools; Cognitive technologies for modeling knowledge (pp. 85-102). Berlin: Springer Verlag.

Lambiotte, J.G., Dansereau, D.F., Cross, D.R. & Reynolds, S.B. (1989). Multirelational semantic maps. Educational Psychology Review, 1(4), 331-367.

Lamping, J., Rao, R., Pirolli, P. (1995). A focus+context technique based on hyperbolic geometry for visualizing large hierarchies. New York: ACM Press/Addison-Wesley.

Landauer, T.K., Psotka, J. (2000). Simulating text understanding for educational applications with latent semantic analysis: Introduction to LSA. Interactive Learning Environments, 8(1), 73-86.

McAleese, R. (2000). Skill acquisition: The curious case of information searching. Interactive Learning Environments, 8(1), 23-49.

Madrazo, L., & Weder, A. (1998). Aalto on the Internet: Architectural analysis and concept representation with computer media. Proceedings of EUROPIA '98, Paris, France. Re-published in Automation in Construction (2001) 10, 56 1-575.

Novak, J.D. & Gowin, D.B. (1984). Learning how to learn. New York: Cambridge University Press.

Scardamalia, M., Bereiter, C., Brett, C., Burtis, P.J., Calhoun, C., & Smith Lea, N. (1992). Educational applications of a networked communal database. Interactive Learning Environments, 2(1), 45-71.

Scardamalia, M., & Bereiter, C. (1993 May). Technologies for knowledge-building discourse. Communications of the ACM, 36(5).

Sherratt, C.S., & Schiabach, M.L. (1990). The applications of concept mapping in reference and information services. RQ, 30, 60.

Sowa, J. Semantic Networks. [Online]. Available: http://www.jfsowa.coml pubs/semnet.htm

Suthers, D., Erdosne Toth, E., Weiner, A. (1997). An integrated approach to implementing collaborative inquiry in the classroom. In Proceedings of CSCL '97: The Second International Conference on Computer Support for Collaborative Learning. Toronto: University of Toronto Press.

Wallace, D., Wandell Conner West, S., Ware, A., Dansereau, D.F. (1998). The effect of knowledge maps that incorporate Gestalt principles on learning. The Journal of Experimental Education, 67(1), 5-16.