Can You Hear What's Coming? Failure to Replicate ERP Evidence for Phonological Prediction

Area of Science:

• Cognitive Neuroscience
• Psycholinguistics
• Auditory Processing

Background:

• Prediction-based language comprehension theories posit listeners anticipate word meaning and form.
• The phonological mismatch negativity (PMN) is proposed as an event-related potential (ERP) marker of phonological prediction mismatch.
• Replicability and distinctness of the PMN from the N400 component remain debated.

Purpose of the Study:

• To replicate the phonological mismatch negativity (PMN) effect.
• To investigate the separability of the PMN from the N400 component in sentence comprehension.
• To examine phonological prediction using electroencephalography (EEG) with a larger sample size.

Main Methods:

• Electroencephalography (EEG) was used to record brain activity in 48 participants.
• Participants listened to sentences with predictable or unpredictable words, varying in phonological overlap.
• Preregistered analyses focused on early (150-250 ms) and N400 (300-500 ms) time windows.

Main Results:

• A widely distributed negative-going ERP was observed for unpredictable words in both early and N400 time windows.
• Bayes factor analysis provided moderate evidence against distinct scalp distributions for early and N400 effects.
• The study failed to replicate the specific PMN effect as a marker of phonological prediction mismatch.

Conclusions:

• Findings do not refute phonological prediction during language comprehension.
• The phonological mismatch negativity (PMN) effect is questioned as a distinct marker of phonological prediction mismatch.
• Results suggest an early onset of the auditory N400 effect for unpredictable words, highlighting risks of data-contingent analyses and small sample sizes.