Supporting Information Comparisons in Example-Based Hypertext Environments

Global comparisons of learning from hypertext and traditional text have yet failed to show major advantages concerning the effectiveness of hypertext learning. In the current paper it is proposed that an effective hypertext design needs to be based on thorough cognitive task analyses with regard to structures, processes, and resources that are required to benefit from a specific learning approach. This claim is illustrated by two experiments, in which we investigated two methods for supporting effective cognitive processes in example-based hypermedia environments, namely, elaboration prompts and an interactive comparison tool. Both methods improved performance for near-transfer problems. Ways of extending this task-analytical approach to facilitating far transfer are discussed. Promises and Drawbacks of Hypertext: The Need for Cognitive Task Analyses Hypertext-based learning environments consist of networklike information structures where fragments of information are stored in nodes that are interconnected by hyperlinks (Conklin, 1987). They are characterized by a high degree of learner control, where users can select information and choose the pacing and sequence of its presentation according to their goals, preferences, and needs. Several instructional benefits have been postulated for learning from hypertexts from different theoretical perspectives, whereby these perspectives mainly emphasize that hypertext allows for active, constructive, flexible, adaptive, and self-regulated learning. However, they are also characterized by some serious usability problems. Conklin (1987) postulated that disorientation (e.g., not knowing how to get to another point in the network) and cognitive overload limit the effectiveness of hypertext for learning. The term cognitive overload refers to the assumption that metacognitive or executive skills necessary for hypertext navigation may require cognitive resources that will no longer be available for the pursuit of the currently performed learning task (Niederhauser, Reynolds, Salmen & Skolmoski, 2000). Furthermore, “many learners may not be proficient computer users and must, therefore, use cognitive resources to operate the computer” (Niederhauser et al., 2000, p. 251). According to the cognitive load theory (Sweller, van Merriënboer & Paas, 1998), these processes may impede learning as they require cognitive resources that may exceed the limits of workingmemory capacity. Cognitive load due to the requirements of selecting and integrating information and due to the interaction with the computer (so-called extraneous cognitive load) may thus reduce the possible benefits of hypertext-assisted learning (Gerjets, Scheiter, & Tack, 2000). Cognitive resources required by extraneous cognitive load can no longer be devoted to mindful cognitive processes that are associated with a useful type of cognitive load, namely germane cognitive load. Rouet and Levonen (1996, p. 20) conclude that “hypertext efficiency involves a trade-off between the power of the linking and the searching tools it provides and the cognitive demands or costs these tools impose on the reader.” This may explain why global comparisons of learning from hypertext and traditional text have yet failed to show major advantages concerning the effectiveness of hypertext learning (e.g., Dillon & Gabbard, 1998). From a cognitive load perspective, an important objective of instructional design with regard to hypertext-assisted learning is thus to minimize extraneous cognitive load and to stimulate learners to invest cognitive resources in activities that result in germane cognitive load. To achieve this objective, much more detailed analyses of the relevant cognitive processes and resources involved in a specific learning situation are needed in order to predict under which conditions hypertext environments will be beneficial or harmful for learning (cf. Gerjets & Hesse, 2004). Much of the early research on hypertext learning can thus be criticized for comparing learning from text and hypertext in a very general way, without specifying the learning approach chosen and the cognitive structures, processes, and resources necessary to benefit from this learning approach. Without these specifications, findings with regard to the relative superiority of hypertext or text are of rather limited value as it is not clear whether these findings might generalize over different learning approaches. In our own research we thus try to combine hypertext experiments with cognitive task analyses in order to compare instructional conditions that differ in the cognitive processes that they support or require and in the cognitive resources needed.