Language processing is an incremental and highly predictive process; parsers rely on contextual information to pre-activate upcoming linguistic input. Certain factors (e.g., attentional demand) have been shown to modulate predictive processing, with parsers aiming for “optimal efficiency” by balancing the costs of prediction with other tasks. How might parsers in highly constraining contexts respond when attentional demands alter prediction’s task-relevance?

Experiment 1

In this study, we examine how attention modulates prediction in an eye-tracking visual world paradigm. In a between-participant manipulation, L1 English speakers (n = 120) are instructed to attend to either male or female actors in di-clausal sentences. For both groups, the initial verb was either predictive of a target noun (After Peter answered his phone, Mary sighed and produced her wallet) or non-predictive (After Peter lost his phone…). Predictive nouns were task-relevant for the male-attending group. (Presented at FPM2024, HSP2024, and FPM2025)

Experiment 2

In this follow-up experiment, we explore how a secondary task of prediction confidence self-rating might impact the rate of prediction in participants (n = 60). Data collection for this experiment is underway.

View this project’s pre-registration.