The development of real-world and counterfactual-world inference generation abilities during adolescence

Government statistics suggest that around one-third of children and young people do not make the expected progress in reading comprehension and text production during secondary school (Department for Education, 2015a; 2015b). Inference generation abilities are at the core of skilled models of comprehension (e.g. Kintsch, 1988; Zwaan, 2003). However, previous research exploring the development of inference generation abilities typically compares a child group (9years or less) with an adult group (18years or over). To assess literal, coherence inference and elaborative inference processing in real-world and counterfactual-world conditions in a wide age-range a new measure was created: the Image Selection Task (IST). The IST embeds a self-paced reading methodology and forced-choice picture-selection task into short stories to explore changes in inference generation skill (number of errors) and time-course (speed). The IST was found to be a valid measure of inference generation abilities and, due to reducing demands on translation and expressive language processes, potentially a purer measure than existing tasks which utilise verbal response methods. The IST was used to explore age-related changes in inference generation abilities in Year 5, Year 7, Year 9 and adults using a cross-sectional design. Findings suggest that inference generation skill plays a role in both reading comprehension and text production during adolescence, with skill increasing until Year 7. Error patterns suggest that coherence inferences were no more difficult than elaborative inferences for Years 5, 7 and adults. Year 9, however, found coherence inference generation more difficult than elaborative inference generation. Inference generation speed was found to improve until Year 9 for both real-world and counterfactual-world information, with time-course patterns comparable across all age groups, such that, in real-world and counterfactual-world conditions, coherence inferences were generated online and elaborative inferences offline. Real-world coherence inference generation skill was found to be underpinned by knowledge during adolescence. Both knowledge and inhibitory control appear to play a role in real-world elaborative inference generation skill during adolescence. Counterfactual-world inferences were both found to be underpinned by belief biases and inhibitory control. However, the direction of effect of inhibitory control reversed (from positive to negative) in Year 9. Further research is needed to explore whether this effect is specific to the current sample or more generalisable. Based on the findings, educational recommendations are provided, including suggestions for assessment and activities at specific points in the reading process. The recommended activities focus on promoting those skills underpinning inference generation.