Conceptual overlap among texts impedes comprehension monitoring.

Abstract

For decades, research on metacomprehension has demonstrated that many learners struggle to accurately discriminate their comprehension of texts. However, while reviews of experimental studies on relative metacomprehension accuracy have found average intra-individual correlations between predictions and performance of around .27 for adult readers, in some contexts even lower near-zero accuracy levels have been reported. One possible explanation for those strikingly low levels of accuracy is the high conceptual overlap between topics of the texts. To test this hypothesis, in the present work participants were randomly assigned to read one of two text sets that differed in their degree of conceptual overlap. Participants judged their understanding and completed an inference test for each topic. Across two studies, mean relative accuracy was found to match typical baseline levels for the low-overlap text sets and was significantly lower for the high-overlap text sets. Results suggest text similarity is an important factor impacting comprehension monitoring accuracy that may have contributed to the variable and sometimes inconsistent results reported in the metacomprehension literature.