Are words pre-activated probabilistically during sentence comprehension? Evidence from new data and a Bayesian...

Area of Science:

• Psycholinguistics
• Cognitive Neuroscience
• Computational Linguistics

Background:

• Previous research suggests readers predict upcoming nouns using sentential context, evidenced by EEG signals.
• Replication attempts have yielded inconsistent results, questioning the robustness of noun pre-activation.
• German noun gender provides a unique linguistic feature to test noun pre-activation.

Purpose of the Study:

• To investigate the noun pre-activation hypothesis in German using gender predictability.
• To address inconsistencies in prior research through a novel experimental approach.
• To employ Bayesian meta-analysis for robust evidence accumulation.

Main Methods:

• Utilized German noun gender marking on determiners in accusative case noun phrases.
• Collected electroencephalography (EEG) data from 120 subjects.
• Conducted preregistered, exploratory, and Bayesian random-effects meta-analyses.

Main Results:

• Individual study analyses showed no conclusive evidence for or against pre-activation.
• Effect direction in the current study and prior replications was consistent with pre-activation.
• Bayesian meta-analysis revealed a small but clear effect supporting noun pre-activation.

Conclusions:

• The findings support the noun pre-activation account, albeit with a small effect size.
• Bayesian meta-analysis offers a powerful method for incrementally accumulating evidence in psycholinguistics.
• Future research should continue to explore pre-activation effects using robust statistical methodologies.