The study investigates whether listeners and readers can use their knowledge of the wider discourse to rapidly anticipate specific upcoming words as a sentence unfolds. Using event-related brain potential (ERP) and self-paced reading experiments, the authors examine the neural and behavioral effects of predicting specific nouns in Dutch stories. In Experiment 1, subjects heard stories where the context suggested a specific noun after an indefinite article. When the adjective's gender mismatched the expected noun's gender, a differential ERP effect was observed, indicating that listeners anticipated the noun before it was fully presented. This effect disappeared in a control experiment without the discourse context, suggesting that the ERP effect was indeed discourse-dependent. In Experiment 3, self-paced reading showed that prediction-inconsistent adjectives slowed readers down before the noun, further supporting the idea that people can predict upcoming words in fluent discourse. These findings suggest that people can use their discourse knowledge to predict specific words and that these predicted words can immediately participate in incremental parsing operations.The study investigates whether listeners and readers can use their knowledge of the wider discourse to rapidly anticipate specific upcoming words as a sentence unfolds. Using event-related brain potential (ERP) and self-paced reading experiments, the authors examine the neural and behavioral effects of predicting specific nouns in Dutch stories. In Experiment 1, subjects heard stories where the context suggested a specific noun after an indefinite article. When the adjective's gender mismatched the expected noun's gender, a differential ERP effect was observed, indicating that listeners anticipated the noun before it was fully presented. This effect disappeared in a control experiment without the discourse context, suggesting that the ERP effect was indeed discourse-dependent. In Experiment 3, self-paced reading showed that prediction-inconsistent adjectives slowed readers down before the noun, further supporting the idea that people can predict upcoming words in fluent discourse. These findings suggest that people can use their discourse knowledge to predict specific words and that these predicted words can immediately participate in incremental parsing operations.