Temporal integration of segmented-speech sounds probed with mismatch negativity.

Deviant sounds occurring in a sequence of standard sounds, in the absence of attention, elicit an event-related potential known as mismatch negativity (MMN). Standard sounds are encoded in auditory sensory memory trace and processed as a single unit within 160-170 ms, where each unitary event stored is closely related to the temporal window of integration. The temporal window of integration of pure tone sound has already been reported. However, there are no reported correlations between segmented-speech sounds and a temporal unit. It is well known that pure tone sounds are predominantly recognized in the right hemisphere, whereas speech sounds are recognized in the left hemisphere. The aim of this study was to examine whether segmented-speech sounds were processed as a temporal unit like pure tone sounds and whether there were differences between right and left ear stimuli. Twenty-five right-handed healthy Japanese men participated in this study. Stimuli consisted of the vowel /a/ spoken by a Japanese female and the stimuli sequences were randomized from short standard sounds and three types of long deviant sounds. The stimuli were presented to both ears separately. All bilateral stimuli induced definite MMN with similar peak latencies. The MMN amplitude gradually enhanced from the short to the long duration deviant. There were no differences in MMN between the right and the left ear stimuli. These findings perhaps show that bilateral deviant segmented-speech sounds were processed equally as a temporal unit.