Abstract/Summary

Recordings of electrical activity generated in the brain in response to specific stimuli now provide an important source of information about the temporal and topographical distribution of language processing events in the brain. Many of our theories of language processing are based upon reaction time studies and use the new brain based findings sparingly. One reason for this is that much of the results from brain recordings are seen to be primarily recapitulations of reaction time findings. A new approach to analyzing evoked electroencephalographic data, Contrastive Signal Coherence (CSC) assesses the coherence of the evoked signal using the techniques of non-linear systems analysis. This determination of signal quality, combined with the measure of signal quantity as calculated in the standard voltage averaged approach to Evoked Brain Potentials (ERPs), provides a richer reflection of the dynamic properties of language related brain potentials and significantly enhances our view of cognitive cortical events. In this paper, we present the results of an ERP study in which we combined the traditional voltage averaged analysis and CSC to address two competing accounts of Case ambiguity resolution, a data driven, expectation based account following Schlesewsky (1997) and an information driven Diagnosis and Repair account following Fodor and Inoue (1994). Both approaches are considered in the context of Friederici's (1995) three stage description of the ERP correlates of language processing. The basic findings provide evidence in favor of the three stage model and motivate an alternative account of Case ambiguity resolution.