Research report On the cost of syntactic ambiguity in human language comprehension: an individual differences approach Ina D. Bornkessel * , Christian J. Fiebach 1 , Angela D. Friederici Max Planck Institute for Human Cognitive and Brain Sciences, Stephanstr. 1a, 04103 Leipzig, Germany Accepted 17 May 2004 Available online 17 July 2004 Abstract We present an event-related brain potential (ERP) study demonstrating that high and low span readers show qualitatively different brain responses in the comprehension of ambiguous and complex linguistic stimuli. During the processing of ambiguous German sentences, low span readers showed a broadly distributed, sustained positivity, whereas high span participants showed a shorter, topographically more focused negativity. Qualitatively similar effects were observable in response to (complex) object-initial sentences. Additionally, a neural effect reflecting reanalysis in sentences disambiguated in a dispreferred way (P600) was observable only for high span readers, while the low span group showed an N400-like response. These neurophysiological findings support the notion that individual working memory capacity as measured by the reading span test influences sentence processing mechanisms and are compatible with the hypothesis that low span readers cannot effectively inhibit dispreferred readings. D 2004 Elsevier B.V. All rights reserved. Theme: Neural basis of behavior Topic: Cognition Keywords: Sentence processing; ERPs; Reading span; Syntactic ambiguity; Inhibition 1. Introduction The nature of the processes underlying the human language comprehension system’s ability to process ambig- uous linguistic input has been an issue at the heart of psycholinguistic research over the past decades. In the majority of cases, this type of research has focused on examining how processing proceeds in the disambiguating region, thereby shedding light on the preference strategies which apply during language processing and, moreover, on which types of linguistic information are drawn upon at which point in time during conflict resolution. Thus, while it has been virtually undisputed that disambiguation towards a dispreferred alternative gives rise to additional processing cost, the question of whether the processing of ambiguous input per se is also costly has proved far more controversial. In a recent paper, however, Frisch et al. [11] were able to demonstrate that the processing of an ambiguous region leads to an increase in processing load, as evidenced by an enhanced parietal positivity (P600) in event-related brain potential (ERP) measures to locally ambiguous sentences in comparison to their unambiguous counterparts. Since these data also show a reanalysis effect (again in the form of a P600 component) in the case of a dispreferred disambigu- ation, Frisch et al. argue for a parsing architecture in which multiple representations may be simultaneously maintained, albeit with different priorities. Corroborating evidence from neuroimaging stems from studies showing increased neural activation for syntactically ambiguous structures in compar- ison to their unambiguous counterparts [37]. In this paper, we shall attempt to further clarify the nature of the processing cost induced by a syntactic ambiguity. The most straightforward possibility appears to be that the con- current representation of several possible alternative analyses consumes processing resources (e.g. in terms of working memory), thereby giving rise to an enhanced processing load. 0926-6410/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.cogbrainres.2004.05.007 * Corresponding author. Tel.: +49-341-9940-114; fax: +49-341-9940- 113. E-mail address: bornke@cbs.mpg.de (I.D. Bornkessel). 1 Present address: Department of Psychology and Helen Wills Neuroscience Institute, University of California, Berkeley, USA. www.elsevier.com/locate/cogbrainres Cognitive Brain Research 21 (2004) 11 – 21