Event-related brain potentials (ERPs) were recorded from 13 scalp electrodes while subjects read sentences, some of which contained either a verb that disagreed in number with the subject noun (syntactic anomaly) or a word in uppercase letters (physical anomaly). Uppercase words elicited the P300 complex of positivities, whereas agreement violations elicited a late positive shift with an onset around 500 msec and a duration of several hundred msec. These effects differed in their morphology, temporal course, amplitude, and scalp distribution. Furthermore, manipulations of the probability-of-occurrence and task relevance of the anomalies had robust effects on the response to uppercase words, but not on the response to agreement violations. Finally, these anomalies had additive effects when agreement-violating uppercase (doubly anomalous) words were presented. These results are taken to be an initial indication that the positive shift elicited by agreement violations is distinct from the P300 response to unexpected, task-relevant anomalies that do not involve the violation of a grammatical rule.

No abstract

Event-related potentials (ERPs) were recorded from 13 scalp electrodes while subjects read sentences, some of which contained a violation of constraints on constituent movement (subjacency or the empty category principle). In two experiments, the rst word that indicated a violation elicited a widely distributed positive-going wave with an onset between 250 and 300 msec and a duration of several hundred milliseconds. This effect was similar to the positive shifts that have been observed in response to other types of syntactic violations (P600), but qualitatively distinct from the N400 response elicited by semantically/pragmatically anomalous words. These ndings are taken as an indication that movement constraints can be applied during the earliest stages of sentence processing, perhaps in conjunction with the creation of phrase structure. We evaluate the hypothesis that the P600 is a general electrophysiological marker of syntactic anomaly.

It is generally believed that readers decompose a complex word into its constituent morphemes only when those morphemes participate productively in word formation. Here we recorded eventrelated potentials (ERPs) to words (e.g. mu¥er, receive), nonwords containing no morphemes (e.g. £ermuf), and non-words containing a pre¢x and a non-productive bound stem (e.g. in-ceive). Prior work has shown that pronounceable non-words elicit larger-amplitude N400 components than words. If readers treat non-words containing non-productive morphemes as unanalyzed wholes, then these non-words should elicit larger N400 s than matched words. We report here, however, that bound-stem nonwords elicit a brain response highly similar to that elicited by real words. This ¢nding suggests that morphological decomposition and representation extend to non-productive morphemes.

No abstract