Construction of visualisations 1 Running head: Construction of visualisations Construction of visualisations in collaborative history learning

Visualisations are external representations that have clearly discernible schematic and/or depictive properties. According to the Cognitive Theory of Multimedia Learning, for example, visualisations can be powerful learning tools (Mayer, 2001). However, this theory has not been tested within the domain of history, nor for co-construction tasks. In an experimental study, the task products and dialogues of participants who co-construct textual representations, visualtextual representations or visual-textual representations integrated in a timeline were compared. The participants were 12 to 13-year-old pupils in vocational middle school who worked in dyads on a series of four history tasks. Task products and protocols of the taped student conversations were analyzed. The results show that visualisations lead to more discussion about historical phenomena, as well as about procedural issues. Construction of visualisations 3 Construction of visualisations in collaborative history learning There is a large body of research on the use of visualisations going back to the 1960s. This research generally dealt with either types of visualisations and their effects, or – more recently – on the interplay between visualisations and text. A specific area of research deals with the effects of self-constructed visualisations (e.g., mind maps, drawings). Co-construction of visualisations adds another, though not yet systematically studied, dimension to this research. Much research has already been done on learning with visualisations – sometimes called either multimodal representations or external representations. Under certain circumstances learning through visualisation construction can lead to better results than learning through mainly verbal representations. However, the research results are not unequivocal. Dual coding and cognitive load theory, for example, show that the so-called redundancy effect can lead to poorer results through inclusion of irrelevant and thus redundant (verbal or visual) information, and proximity and integration can also influence the learning effects (Paivio, 1991; Mayer, 2001; Tabbers, 2002). De Westelinck, Valcke, De Craene, and Kirschner (2005) show that what works in one domain (e.g., science; see Mayer, 2001) need not necessarily work in another domain, such as the social sciences. These different outcomes for different domains and specific subject matter may result from the level of complexity or the concreteness vs. abstractness of the visualisation. Thus far, educational research on visualisations has focused primarily on the learning results (i.e., what is learnt), often neglecting the learning processes used (i.e., how is it learnt; Cox, 1999). The nature of these processes depends on the type of representation used in combination with the domain (which largely determines the appropriate types of representations), on the nature of the task, in particular with respect to differences between Construction of visualisations 4 presentation and construction, and on the interplay between text and visualisation (e.g., redundancy, spatial and temporal contiguity). The focus of the study presented here is on the learning processes of pupils working in dyads, more specifically on the richness of student dialogue in terms of topic content, and the result of this dialogue in task products and individual tests. Types of external representations In the study discussed here external representations include all representations of concepts and constructs that are visible to and interpretable for others, as opposed to mental or internal representations. This means that external representations can also be verbal representations – e.g., stories – whilst visualisations are external representations that have clearly discernible schematic and/or depictive properties – such as pictures, iconic representations or schemas. Visualisations may contain text, but their main purpose is to show things in non-verbal ways. Although research on the value of visualisations has flourished in the past 30 years, few attempts have been made to describe and compare the mechanics and effects of different types of visualisations. One example in the literature is of an empirical study aiming to describe different types of visualisations in a taxonomy or framework (Lohse, Biolsi, Walker, & Rueter, 1994). The authors constructed 10 scales to describe 60 different visualisations, and on the basis of a rating task came up with 11 types of visualisations, such as network charts, cartograms, process diagrams, structure diagrams, and time charts. Although the representations in their study were not necessarily educational in nature, their classification scales show how people react to different types of representations, giving some insight into the different affordances of these visualisations. For example, to convey a lot of information presented as a whole, a picture may Construction of visualisations 5 not be the best choice, whilst a table can present a lot of information at once, but only for its parts. As the examples from Lohse et al. (1994) show, many visualisations are multimodal in nature, in the sense that they contain a combination of two or more of the following information types: textual information, abstract visual information (graphs, tables), and/or concrete visual information (realistic drawings or photos). Cartograms are a good example of this multimodality. Animation and sound can add another dimension to this, although they do require other media than traditional paper textbooks and workbooks. The visualisations in the study described here included cartograms, structure and process diagrams (incorporating pictures), and coordinating time charts. The research focuses on a domain that is relatively unexplored regarding research on learning with external representations, namely the domain of history. Concepts are important in history learning, as they serve as the building blocks to describe the phenomena and relations that are vital in learning and understanding history (Carretero, Lopez-Manjon, & Jacott, 1997). Although elaboration of content may be achieved through a textual assignment, visualisation of concepts, phenomena and relations may well have an even stronger effect. Phenomena and concepts can often be visualised through original visual materials such as paintings. However, aspects of the domain that are harder to grasp, such as relations, abstract developments and structures, can usually only be visualised effectively through especially constructed materials. Taking elaboration to the level of knowledge integration may well be enhanced through visualisation in the shape of a timeline. Construction of visualisations 6 Using and (co-)constructing visualisations In many traditional printed learning materials, text takes up most space on the page, and in processing its contents pupils are mainly required to read and write. Often, illustrations are presented alongside the text and/or assignments as examples or concretizations of the textual information, though sometimes they have no function and are merely added to make the text more appealing. Presenting visualisations with text can have positive effects on learning, but it can also have negative or no effect, as pictures can be inappropriate, distracting or confusing. Though the old saying is that “One picture is worth a thousand words”, abstract or general concepts such as ‘justice’ or ‘mammal’, for example, may need a thousand pictures to explain them, but not quite as many words (Mayer & Sims, 1994; Larkin & Simon, 1987). The interpretative nature of depictive visualisations implies that, although pictures can give information that cannot easily be summarized in a text, they can also give too much information. Sweller, Van Merriënboer, and Paas (1998) would describe this as adding extraneous cognitive load; load that is not germane to understanding the concept being depicted or to solving the given problem. Despite these limitations, visualisations can be extremely useful for learning. Combining text and visualisations requires translating visual information into verbal information and vice versa. These multimodal representations allow the learner to focus on different aspects of the topic being tackled, on connections between topics and aspects, thus promoting deep learning (Ainsworth, 1999). Allowing learners to construct their own visualisations instead of just presenting visualisations to them is a logical step towards integration of theories on learning with representations, and current theories on active learning. After all, construction by the learner Construction of visualisations 7 involves active interpretation of the information given, and active production of new representations. Unfortunately, very few research studies have combined these aspects. Most of the studies concerned with visualisation construction deal with learning tasks where selection, organization, and integration are the main activities with the visualisations. Bodemer, Ploetzner, Bruchmüller, and Häcker (2005) found that construction of integrated representations can have positive effects on learning. Cox (1999) also found that constructing representations can, under certain preconditions, be beneficial for learners. Several studies have shown positive learning effects for one specific type of visualisation: concept maps (e.g., Van Boxtel, Van der Linden, & Kanselaar, 2000). At the far end of the spectrum are studies in which students construct images entirely (for example by drawing), instead of just assembling parts or manipulating them (Van Meter & Garner, 2005). Each of these ‘construction studies’ have dealt with individual construction. Collaborative construction can have a number of advantages above individual construction. Collaboration requires communication (e.g., argumentation, negotiation, explication, dialogue), which encourages elaboration and co-construction of knowledge. In most studies dealing with co-construction of visualisations, the question is asked whether a difference in type of visualisation results in a difference in learning activities. Thus, research on learning through coconstruction of visualisations has focused on the communicative functions (Suthers & Hundhausen, 2003), on the activities in the software (Bodemer et al., 2005), on procedures and task management (Erkens, Jaspers, Prangsma, & Kanselaar, 2005), or on the learning outcomes (Mayer, 2001; De Westelinck et al., 2005). In most studies, little attention is paid to the topic content of the collaboration, although some analyses focusing on this aspect were done in the subject areas of science (Van Boxtel et al., 2000), history (Van Drie, Van Boxtel, Jaspers, & Construction of visualisations 8 Kanselaar, 2005), and educational psychology (Fischer, Bruhn, Gräsel, & Mandl, 2002). The present study focuses on the quality of the topic content aspect of collaborative construction in the domain of history. Research question This study focuses on the processes of learning with visualisations that appear to be functional for learning about abstract historical phenomena. It looks at student dialogues and accompanying task products, and concentrates mainly on what actually occurs at the contentrelated level, as this is where pupils most obviously work on acquiring the building blocks of the domain. Collaborative construction can encourage elaboration, and through partial verbalisation of the thought processes it can shed light on the processes involved in learning with visualisations. At the same time, task products are the physical outcome of the co-construction. Focusing in particular on the cognitive aspects, the main question is: Does collaborative construction of different types of visualisations and of textual representations have different effects on the quality of the content of the dialogues and task products of students? On the basis of the literature one would expect the collaborative dialogues and task products to be different for different types of tasks (textual versus visual) in several ways. First, pupils working with visual tasks are expected to elaborate more on the core content of the task, because the information in the visualisations needs to be verbalised and related to the verbal information given. It is expected that verbalising knowledge and information about abstract Construction of visualisations 9 historical phenomena and the relations between them is easier through the use of visualisations. Second, pictures and diagrams are expected to invite more varied core content vocabulary, because the extra step of visualisation requires more precise content verbalisation. Finally, the task products and individual learning outcomes are expected to reflect the more extensive discussion of core content by students constructing visualisations.