Alpha phase as a marker of biased speech-in-noise perception

In vocal communication, listeners have to deal with noisy environments and ambiguous speech inputs (e.g., slip of the tongues). One coping strategy might be a bias to hear ambiguous utterances as words. To investigate the role of slow neural oscillations (~1–12 Hz) in sensory processing, we set up an electro­encephalo­graphy (EEG) study using difficult-to-discriminate lexical stimuli. Stimuli were tri-syllabic German nouns (“real”), or pseudo-words either derived by exchanging one vowel only (“ambiguous”) or derived by scrambling entire syllables (“opaque”). Thus, individual signal-to-noise ratios yielding 70% accuracy in a same-different vowel discrimination task were determined. Subsequently, participants performed a lexical decision task in noise while recording EEG. Behavioral results indicated that participants were highly accurate in detecting opaque pseudo-words (89%) but were significantly worse in evaluating ambiguous pseudo-words (60%) and real words (71%). Notably, alpha (8–12 Hz) phase angle differed for correct vs. incorrect trials in two time windows: pre-stimulus (-120–40 ms) and in an N400 time frame (~500 ms). The pre-stimulus alpha phase effect shows a right anterior whereas the later effect a left central topography. We interpret both effects in terms of differential states of the readiness of the system for processing upcoming input. Furthermore, a third anti-phase alpha effect appeared when contrasting word- vs. nonword- responses around 500 ms over left lateral electrodes. This implies rather lexical access and decision making processes. Theta (3–5 Hz) phase was not predictive for neither contrast. These results underline the importance of alpha oscillations in biasing perception and decision-making in adverse listening conditions