Improving (meta)comprehension: Feedback and self-assessment

Monitoring is important for self-regulated learning from text, but is often inaccurate. Completing causal diagrams after reading texts has been shown to improve monitoring accuracy. We investigated whether providing one or two model answer diagrams and self-assessment instructions would improve learners' monitoring accuracy, regulation accuracy, and text comprehension. Because little is known about how accurately learners who are reading in a language other than their home language monitor their comprehension, we also explored whether effects differed between readers who have English or another language as their home language. Participants were 258 secondary school students at international schools in Singapore and Spain; 103 spoke a language other than English at home. Participants read 4 texts, completed diagrams on these texts, monitored comprehension, took a first comprehension test, self-assessed their diagram under one of 6 conditions resulting from a 3 (model answer: 0, 1, 2) x 2 (self-assessment instructions: yes, no) design, made restudy decisions, made monitoring judgments, and completed a final comprehension test. Comprehension benefitted most when learners had access to two model answers. There were no effects of model answers or self-assessment instructions on monitoring accuracy. Regulation accuracy improved with model answers combined with self-assessment instructions. There was no differential effect of home language. This study supports prior research showing the benefit of model answer diagrams on comprehension. Yet, improvements in regulation accuracy suggest that model answers combined with self-assessment instructions support more effective self-regulated learning behaviors. •Readers are inaccurate at monitoring comprehension; diagramming improves accuracy.•Little is known about monitoring accuracy of readers with other home languages.•Model answers improve text comprehension for all learners.•Model answers and self-assessment instructions do not impact monitoring accuracy.•Model answers with self-assessment instructions improve regulation accuracy.