Visualization of science text contents

Prof. Dr. Elke Sumfleth - Research Interests

Visualization of science text contents

Students reading complex and difficult expository texts do often have problems with the cognitively highly demanding processes of text comprehension (Baumert et al., 2001; Kintsch & van Dijk, 1978; Schnotz, 1994, 2003). Thus, the purpose of this project was to study the use of visualizations as aids for comprehension in learning from science text. On the one hand, based on multimedia learning theories (e.g., Mayer, 2001, 2005; Schnotz, 2005; Schnotz & Bannert, 2003), visualizations can be provided by simply adding them to a text. Accordingly, several studies showed the benefits of multimedia learning on text comprehension and yielded some guidelines on how to design beneficial multimedia instructional messages (e.g., Carney & Levin, 2000; Mayer, 2001). On the other hand, based on models of self-regulated learning (e.g., Boekaerts, 1997, 1999; Leutner & Leopold, 2006; Weinstein & Mayer, 1986; Zimmerman, 2001), visualizations can also be generated by students' themselves (the so-called "learner-generated drawing strategy"; e.g., Alesandrini, 1984; van Meter & Garner, 2005). Research on learner-generated visualizations, however, has produced somewhat inconsistent empirical findings: Some studies showed positive effects of learner-generated visualizations on text comprehension, whereas others did not. Thus, research on learner-generated visualizations has not provided instructional design guidelines so far. Therefore, four experimental studies were conducted, aimed at investigating the effectiveness of both provided and learner-generated visualizations with respect to an optimal instructional design. These studies were conducted with paper-pencil-based as well as with computer-based learning environments in a chemistry domain. First, results of the studies replicated the benefits of provided pictures as aid for science text comprehension. Thereby, results suggest the assumption that a limited and fixed amount of learning time, compared to an unlimited, learner-controlled amount of learning time, is more helpful for text comprehension. Second, results showed that learner-generated visualizations are an appropriate aid for text comprehension, given that students generate visualizations of high quality and that they are not overloaded by the cognitive demands of the mechanics of the generation activity. In this regard the quality of students' visualizations during learning proved to be an appropriate predictor of their text comprehension. Concerning the instructional design, results provide an indication that the generation process can be adequately instructionally supported by giving students a drawing prompt (showing a partly pre-drawn background and all the relevant elements) as well as a sufficient amount of learning time for generating their visualizations. In sum, results of this project do not only provide evidence for the well known Multimedia Principle (Mayer, 2001, 2005) but also refer to a Generative Drawing Principle as well as to a Prognostic Drawing Principle.

Fundet by DFG

Further information:
Annett Schwamborn
Project Visualisation
Collaborative Project with Detlev Leutner, Essen