UC San Diego

Negation has been found to affect the processing of sentences and words principally in two ways: It makes sentences harder to process, increasing response times and error rates. In addition, it reduces the activation of concepts to which it applies by directing attention away from them to some non-negated alternative. Both types of effects are typically tested after the offset of the sentence. The question addressed in this dissertation is whether negation also has effects within the same sentence, namely on the prediction of upcoming lexical items. This type of incremental processing has been demonstrated for a number of linguistic cues. A previous study of negation effects on the processing of words in the same sentence, however, failed to produce such effects. We attribute this to the fact that in the isolated sentences used in that experiment, negation actually did not change the plausibility of the sentence endings. In order to make negation-induced expectation changes detectable, we therefore embedded affirmative and negative sentences in contexts designed to make the plausibility of a continuation dependent on the sentence mode (affirmative vs. negative). We carried out two series of experiments, differing in the structure of the discourse context. In the first set of experiments, it was the appropriate continuation of the affirmative target sentence that was directly primed, and in the second set it was the correct completion of the negative target sentence. The first experiment of each set used the event-related potential (ERP) methodology, and used the N400 to the sentence-final word as the main index of expectancy. The N400 results showed that negation can affect expectancies about sentence continuations. They also demonstrated that prediction changes are less likely when the most plausible continuation for the negative sentence is itself a negated concept and therefore subject to suppression. The ERP studies were complemented by verification experiments, differing in the way the target sentence was presented (word-by-word vs. whole-sentence). The comparison of verification times showed that negation-induced expectation changes may only occur if readers have enough time and processing capacity available