Inference generation in the context of expository texts and videos: A lexical decision task study

The present study directly compared children’s generation of knowledge-based inferences when reading expository texts and when watching videos. We developed a lexical decision task, which allows for measuring automatic inference processes both in a reading context and a video context. We also tested whether potential differences between media formats depend on children’s reading proficiency. Participants were 108 fourth grade children, who were identified as either below-average (n = 49) or above-average (n = 59) readers and then randomly assigned to the text condition or video condition. In the lexical decision task, children were presented with either text or video items, followed by probe words. Half of the words were consistent, and the other half were inconsistent with what should have been inferred from the text or video items. Results showed that overall, children responded significantly faster and more accurately to consistent probes compared to inconsistent probes, indicating that they were successful in making knowledge-based inferences. This was independent of whether children were in the text or video condition. The lack of effect of media format did not differ for below-average and above-average readers, although below-average readers spent more time reading the texts, possibly compensating for their lower text decoding skills. To conclude, children engaged in knowledge-based inferencing when watching videos, similar to how they do this when reading an expository text. Our findings support the theoretical notion of a general inference skill underlying comprehension of different media.