Using visual speech at the cocktail-party: CNV evidence for early speech extraction in younger and older adults.

The comprehension of spoken language benefits from visual speech information. One reason for this is the temporal lead of mouth and lip movements over the onset of acoustic speech utterance. Here, we investigated EEG event-related potentials preceding acoustic speech, focusing on a fronto-central contingent negative variation (CNV) prior to the onset of acoustic speech. We explored influences of expectation and visual speech content as well as age-related differences. In a multi-talker two-alternative speech discrimination task, younger and older subjects responded to short words presented simultaneously to competing speech under free-field conditions. Subjects were always presented with audiovisual speech stimuli, while the modality containing the task-relevant information was modulated in a block-wise fashion. Thus, task-relevant speech information was either available as audio-visually congruent stimuli or only in the visual (visual-valid) or the auditory (auditory-valid) modality. Subjects were instructed to fixate a pre-specified position in the left or right hemispace. In each task block, task-relevant stimuli appeared either at the pre-specified position (standard trials, 80%) or at a rare deviant position (20%). Target words were recognized faster and more accurately when visual speech information was available. The CNV prior to the acoustic speech onset was more pronounced with visual-informative than with visually non-informative speech. Especially in the younger group, a less pronounced CNV occurred with purely visual speech in deviant trials, that is, when a task-irrelevant speech stimulus appeared instead of the expected target stimulus. The results indicate that processes preceding the onset of acoustic speech are modulated by expectations and visual speech content, while age differences are rather small.