LEARNING FROM HYPERTEXT: RESEARCH ISSUES AND FINDINGS

Abstract

The question of how we learn from hypertext is more complicated than that of how we learn from traditional text. Although all the same elements of character decoding, word recognition, sentence comprehension, and so forth remain the same, a number of features unique to hypertext produce added complexity. It is these features that drive the research of hypertext in education and have shaped our discussion in this chapter. The most basic feature of hypertext, of course, is its nonlinear structure. How nonlinear structure alters learners’ mental representations or ability to use their new knowledge has been an active area of research. This feature gives way to a number of factors related to learning. Primary among these is a flexibility of information access. Whereas traditional text allows the author to assume what information has already been encountered and present new information accordingly, information within a hypertext may be retrieved in a sequence specified by each user. In other words, there is a greater degree of learner control when engaged in hypertext-assisted learning (HAL). The shift in control of access from author to learner places a greater cognitive burden on the learner. Specifically, the learner must now monitor to a greater extent whether he or she understands what has been read, determine whether information must be sought to close information gaps, and decide where to look for that information in the text. In short, there are greater metacognitive demands on the reader during HAL. While the vast majority of research on hypertext is not specifically relevant to learning, investigation into its educational utility began to heat up in the 1980s, and many research reports and articles have been published since then. Chen and Rada (1996) conducted a metanalytic study of learning from hypertext. Of 13 studies they found comparing learning outcomes for subjects using hypertext versus nonhypertext systems, 8 revealed an advantage of hypertext. Although the combined effect size was small to medium (r = .12), it was highly significant (p < .01). In addition, they report that the effect sizes and significance levels among studies comparing learning from hypertext and linear text were heterogeneous. They interpret this result as an indication that factors such as system design, system content, and experimental design influence educational effectiveness, and a number of empirical studies have pointed to the influence of such factors on learning outcomes. In addition to system variables, user traits such as goals, motivation, and prior knowledge are also factors in HAL. Moreover, these learner variables interact with hypertext characteristics to influence learning outcomes. We have attempted here to sort through the data to identify the variables that affect HAL most strongly and the mechanisms through which this occurs. Wherever it is appropriate, we have also tried to explain how user and system variables interact. Because of such interactions, the field is largely looking toward adaptive technology to tailor systems for the user, so we have also included a section on adaptive