While listening speech, some of the speech sounds may be obliterated by environmental noise. However, we may not perceive any disturbance in continuous speech. In such cases, the semantic context appears to influence the perceptual synthesis of missing speech sounds. We aimed to study the neuronal correlates underlying the phenomenon of this "phonemic restoration".
We used auditorily presented sentences in which the final word was either highly or little expected given the preceding sentence context. Sentences were presented either with or without a cough replacing its initial part. Thus, altogether four types of sentences were used.
In behavioural experiment, there were main effects showing that mean reaction time was slower for repeating low probability words than high probability words and slower for cough replaced words than normal words. Event related potentials of 27 volunteers were recorded with 64 electrode cap. The results showed negative response for the low cloze probability compared to high cloze probability final words (the N400-effect). Further, a prominent N1-P2 was elicited by the words beginning with cough replacement, reflecting brains automatic reaction to coughs. Usually novel stimuli which catch the attention in a sequence of sounds elicit the P3a-response. However, for the cough-inserted final words during the positive slope there was a negative deflection with double peaks at 330 and 410 ms, which may be considered as N400. Normal P3a was elicited by the sentence beginning words. Thus, the P3a for the coughs may have been attenuated by an overlapping N400-response. Thus, despite of automatic detection of coughs, as indicated by N1-response, the semantic integration of the words with cough-replacement takes place in the time window of N400.