Strategic and Inferential Processing in Text, Diagrams, and Text + Diagram Representations

• In Event: Comprehension and Integration of Text, Visual, and Multimedia Sources

Abstract

Little research has looked at which learning strategies are used in each representation when students learn from multimedia, despite hypothesized differences (Ainsworth, 2006). Earlier work by our team found more inferences while reading diagrams than while reading text (Cromley et al., 2010), and the purpose of the present study was to examine whether reading text and diagrams in conjunction increased comprehension processes.
The present study analyzed think-aloud protocols from 40 high school biology students answering 1-3 questions each from each from 1-3 pages each of illustrated text from untaught sections of their own textbook. Participants were 86% African-American, 60% female, and all were in 9th grade (Mage=15.5, SD=.6). Audiotaped sessions collected as part of a larger eye tracking study lasted about 20 min, and the only prompts were “Say what you are thinking [doing].” Participants were not informed of the three question types (text, diagram, and text+diagram), and we coded for representation students actually used. We coded utterances for 15 different specific metacognitive and cognitive strategies and inference types, and used raw counts in the analyses. The quality of participant answers was scored using a rubric (0 to 2 points).
Participants had lower scores on their answers, a low median number medium variety of moves (i.e., strategies) when they used text only (M=1.0, Mdn =1, 11 moves), followed by diagram (M=1.2, Mdn=1, 9 moves), and text+diagram (M=1.3, Mdn=3, 13 moves). Of 332 coded utterances, the most frequent moves in the text and text+diagram conditions were re-reading text and repeating the question (i.e., reminding oneself of the learning goal). For diagrams, the most common moves were reading the answer directly off the diagram and making incorrect inferences from prior knowledge. Inferences comprised 24% of utterances in diagrams, significantly more than 5% in text, and 9% in text+diagrams. Relating specific moves to performance, knowledge of synonyms (vocabulary) and accuracy of the strategy used (inferences of any type, paraphrases, prior knowledge activation, summarizing, finding the relevant parts of the diagram) were associated with scores of 1 or 2 (but rarely 0). By contrast, statements of confusion, off-task statements, and inaccuracy of the strategy used (same as above) were almost always associated with scores of 0.
Results for these low-knowledge high school students reading researcher-posed questions are mostly different from those found for undergraduate biology majors reading a textbook chapter. However, one similarity is the improved learning from diagrams or text+diagram combinations. A second similarity is the greater use of inference in diagrams. High school student moves associated with better learning from illustrated text focused around accuracy (e.g., making accurate inferences) rather than the undergraduate association with using a higher proportion of inferences (accurate or inaccurate). Some strategies associated with worse learning in undergraduates (e.g., paraphrasing) were not associated with worse learning in high school students. Tentatively, results from this study suggest that feedback on strategy and inference accuracy in brief tasks like these might be most useful to high school students.

Authors

• Jennifer G. Cromley, University of Illinois at Urbana-Champaign

• Theodore W. Wills, Temple University