The effects of question construction on expository text comprehension

Abstract

This study investigated three research questions: (1) In spite of text coherency and domain knowledge combinations, can we expect generic question stems to lead to greater measurable outcomes on memory (i.e., textbase model) and learning (i.e., situation model) variables as compared to signal words and/or unguided questioning techniques? (2) Will there be interaction effects on these variables suggesting that combinations of domain knowledge and text coherency are uniquely affected by generic question stems, signal words, or unguided questioning? (3) Will the employment of generic questions stems lead to superior calibration of comprehension scores than those generated by signal words and/or unguided questions conditions? Sixty-three first year psychology students participated in the study. On the basis of pre-determined domain knowledge scores, participants were categorized as high or low domain knowledge. They were then randomly assigned to twelve treatment conditions in a 3 x 2 x 2 cross-factorial design. The first factor was treatment with three levels (i.e., generic question stems, signal words, and unguided questioning); the second factor was domain knowledge with two levels (i.e., high domain knowledge and low domain knowledge); and the third factor was text coherence with two levels (i.e., high text coherence and low text coherence). Participants were administered four measures: a pre- and post-treatment sorting task; a short answer post-test; a summary recall measure; and a calibration of comprehension task. Results of this study suggest that high domain knowledge and structured questioning strategies are the most reliable predictors of text memory and text learning. Several main effects favouring high domain knowledge arose on text memory, and to a lesser extent, text learning measures. Significant results involving signal words and generic question stems were detected primarily through interaction effects and one main effect. While not always significantly superior, generic question stem participants outperformed their signal word and unguided questioning counterparts on the majority of text memory and text learning variables. Therefore, it appears that generic question stems engage the reader in a deeper level of processing. However, when instructional time is limited, high domain knowledge is necessary for significantly superior text memory and for specific structural connections.