Timothy E. Lindamood,
Steven A. Hillyard
Department of Neurosciences
University of California, San Diego
La Jolla, California
Event-related brain potentials (ERPs) were recorded from subjects as they read a series of sentences, for which Bloom and Fischler ( 1980) had established the cloze probability for alternative terminal words. The amplitude of a late negative component (N400) of the ERP to the terminal words was enlarged as an inverse function of their cloze probability. Even larger N400 waves were elicited by semantically incongruous words that were grafted onto the ends of some of the sentences. In a second experiment, the N400 to semantically anomalous terminal words was reduced in amplitude when those words were semantically related to the word best completing the sentence (i.e., having the highest cloze probability). These results indicate, first, that a semantic anomaly is not necessary for the production of a substantial N400 component, though it may be a sufficient condition, and second, that the N400 bears a reciprocal relationship to word expectancy in sentence contexts. It is suggested that the N400 may be an index of the degree of semantic priming or activation that a word receives from the prior context.
It is now well established that a prior verbal context acts to facilitate word recognition during reading. One way to demonstrate these context effects is to measure the time required to classify a visually presented letter string as a word or non-word (that is, a lexical decision task). Using the lexical decision paradigm, it has been shown that a prior context decreases decision latencies for congruous or semantically related words while increasing latencies for seman-