Electrophysiological Evidence for Two Steps in Syntactic Analysis: Early Automatic and Late Controlled Processes

• Published in: Journal of cognitive neuroscience Vol. 11; no. 2; pp. 194 - 205
• Main Authors: Hahne, Anja, Friederici, Angela D.
• Format: Journal Article
• Language: English
• Published: One Rogers Street, Cambridge, MA 02142-1209, USA MIT Press 01.03.1999; MIT Press Journals; MIT Press Journals, The
• Subjects: Acoustic Stimulation; Adult; Auditory perception; Brain; Cerebral Cortex - physiology; Cognition & reasoning; Comprehension; Dominance, Cerebral - physiology; Electroencephalography; Evoked Potentials; Evoked Responses; Fixation, Ocular; Grammar, Comparative and general; Humans; Language Tests; Linguistics; Male; Models, Neurological; Models, Psychological; Neurology; Parsing; Random Allocation; Sentences; Syntactic Processing; Syntax; Testing
• ISSN: 0898-929X; 1530-8898
• DOI: 10.1162/089892999563328

In this study we examined the properties of the processes involved in the structural analysis of sentences using event-related brain potential measures (ERP). Previous research had shown two ERP components to correlate with phrase structure violations: an early left anterior negativity (ELAN), which is assumed to reflect first-pass parsing processes, and a late parietally distributed positivity (P600), assumed to reflect second-pass parsing processes. We hypothesized that the first-pass parsing processes are highly automatic, whereas second-pass parsing processes are more controlled. To test this hypothesis we varied the proportion of correct sentences and sentences containing phrase structure violations with incorrect sentences being either of a low (20% violation) or a high (80% violation) proportion. Results showed that the early left anterior negativity was elicited and equally pronounced under both proportion conditions. By contrast, the late positivity was elicited for a low proportion of incorrect sentences only. This data pattern suggests that first-pass parsing processes are automatic, whereas second-pass parsing processes are under participants' strategic control.