Prediction in language comprehension across the lifespan

Abstract

Comprehending language in real time requires the brain to keep pace with ambiguous, variable and often noisy input. One way in which individuals address this challenge is by actively predicting (generating and/or selecting) what might occur next. However, prediction is not a unitary mechanism and is not engaged uniformly among individuals or across the lifespan. In this Review, we examine predictive mechanisms across the lifespan, synthesizing evidence from children, young adults and older adults to show how different aspects of prediction are deployed under different developmental and processing conditions. Viewed developmentally, differences in prediction have multifaceted roots that reflect the joint influence of cognitive resources, linguistic experience and the momentary utility of predictive engagement. Prediction is not an invariant hallmark of comprehension but a dynamically configured system, whose components are selectively engaged to meet the functional demands of language use across life. Comprehending language involves active prediction of what might occur next. In this Review, Federmeier and colleagues synthesize evidence from children, young adults and older adults to show how prediction is dynamically engaged for language comprehension across the lifespan.