This study focused on the automatic versus controlled nature of the underlying cognitive processes of ERPs （Event-Related Potentials） effects during speech perception. ERPs to the final word of sentences （half bein...This study focused on the automatic versus controlled nature of the underlying cognitive processes of ERPs （Event-Related Potentials） effects during speech perception. ERPs to the final word of sentences （half being semantically incongruent with the sentence and half congruent） were presented to French natives under 4 levels of degradation： no degradation, mild degradation （2 levels）, and strong degradation. Under mild degradation allowing controlled sentence-level processing according to behavioral data, the N400 effect （i.e., the N400 to incongruent words minus the N400 to congruent words） and a late positive complex were delayed and the latter effect was reduced. Under strong degradation allowing only automatic sentence processing according to behavioral data, no ERP effects remained. These results suggest that, unlike contextual effect found with single words （e.g., using word-pair or word-list paradigms）, ERP effects elicited by more complex contexts such as full sentences are generated by controlled but not by automatic mechanisms of speech processing.展开更多
文摘This study focused on the automatic versus controlled nature of the underlying cognitive processes of ERPs （Event-Related Potentials） effects during speech perception. ERPs to the final word of sentences （half being semantically incongruent with the sentence and half congruent） were presented to French natives under 4 levels of degradation： no degradation, mild degradation （2 levels）, and strong degradation. Under mild degradation allowing controlled sentence-level processing according to behavioral data, the N400 effect （i.e., the N400 to incongruent words minus the N400 to congruent words） and a late positive complex were delayed and the latter effect was reduced. Under strong degradation allowing only automatic sentence processing according to behavioral data, no ERP effects remained. These results suggest that, unlike contextual effect found with single words （e.g., using word-pair or word-list paradigms）, ERP effects elicited by more complex contexts such as full sentences are generated by controlled but not by automatic mechanisms of speech processing.
