JOURNAL OF MEMORY AND LANGUAGE 38, 112–130 (1998) ARTICLE NO. ML972537 Dissociating Brain Responses to Syntactic and Semantic Anomalies: Evidence from Event-Related Potentials Kim Ainsworth-Darnell, Harvey G. Shulman, and Julie E. Boland The Ohio State University Two experiments investigated the influence of anomaly type and presentation rate on the occurrence and appearance of the event-related brain potentials (ERPs) known as the N400 and P600. In Experiment 1, sentences containing either a syntactic anomaly, a semantic anomaly, or a compound syntactic and semantic anomaly were presented at the rate of 1000 ms per word. Consistent with previous findings, syntactic anomalies elicited a P600, while semantic anomalies elicited an N400. Compound anomalies evoked an N400 – P600 waveform complex. Experiment 2 investigated the effect of presentation rate on ERPs using the syntactic anomaly materials from Osterhout and Holcomb (1992; Experiment 1) at the 650 ms SOA from the original study and a new 1000 ms SOA. Although the amplitude and latency of the P600 waveform differed slightly between the two presentation rates, reliable P600s were found at both the 650 and the 1000 ms SOA.  1998 Academic Press Key Words: event-related potentials; sentence processing; presentation rate. In theories of sentence comprehension, a begins. The opposite view maintains that the language processor uses semantic information fundamental issue concerns how and when the syntactic and semantic content of a sentence to guide the parser toward a structural analy- sis. This is not to say that there is a decided become integrated. It is generally accepted that the two types of processing involve sepa- link between these ‘‘how’’ and ‘‘when’’ is- sues: It is entirely plausible, for example, that rate mechanisms, although how independent each mechanism is from the other is still a syntactic and semantic processing are exe- cuted in parallel, but that syntactic analysis is topic of debate. One view maintains that syn- tactic analysis is entirely modular, such that still not informed by semantic features. These concerns have proven difficult to re- parsing is completed before semantic analysis solve, partly because it is difficult to obtain distinct measures of syntactic and semantic The authors’ appreciation is extended to Susan Garnsey processing, or of their time course. For exam- and Peter Culicover for their comments and suggestions during the course of this project, and to Peter Hagoort ple, in some sentence comprehension experi- for helpful advice on the data collection procedure. Our ments, the participant judges the comprehensi- appreciation is also extended to Lee Osterhout for his bility or grammaticality of a sentence follow- assistance with Experiment 2. We thank Don Mitchell ing its completion. Such global judgments and two anonymous reviewers for their suggestions on may be influenced by either semantic or syn- earlier versions of this manuscript. This research was made possible through a research fellowship awarded to tactic variables, and do not reveal anything the first author by the Ohio State University Center for about the time course of the two processes. A Cognitive Science, the Department of Linguistics, and the more on-line technique used in recent litera- Department of Psychology. Portions of this data were ture is to track eye position during sentence presented at the Linguistic Society of America Annual reading. With this technique, researchers Meeting, 1995, New Orleans, the conference on Architec- tures and Mechanisms for Language Processing, 1995, sometimes find different patterns in the first- Edinburgh, Scotland, and the Ninth Annual CUNY Sen- pass eye fixations compared to secondary re- tence Processing Conference, 1996, New York City. gressions. Ferreira and Henderson (1990) Please address reprint requests and correspondence to found such a difference and accounted for it the first author at the Department of Linguistics, Oxley by assuming that first-pass eye movements re- Hall 222, 1712 Neil Avenue, Columbus, Ohio, 43210; E- mail: darnellk@ling.ohio-state.edu. flect the initial, purely syntactic, processing of 112 0749-596X/98 $25.00 Copyright  1998 by Academic Press All rights of reproduction in any form reserved. AID JML 2537 / a010h$$141 12-17-97 08:53:30 jmlal AP: JML