| This thesis uses event-related brain potential (ERP) methods to explore the hypothesis that syntactic and semantic cues are processed via parallel and fully-interactive processing streams. This hypothesis offers an alternative to a longstanding and influential view that language comprehension is accomplished by independent, stage-based mechanisms wherein syntactic analysis precedes and guides semantic interpretation (e.g., Frazier, 1978, 1989). I describe results from four ERP experiments that each pit syntactic cues against semantic cues. Experiments 1 and 2 manipulated the relative "strength" of those cues and found that the conflict's outcome is determined by cue strength: when semantic cues are stronger, well-formed syntactic cues are perceived as anomalous, eliciting P600 effects; when syntactic cues are stronger, the anomaly is perceived as being semantic in nature, resulting in enhanced N400; when cues are evenly matched, anomalies elicit a left-anterior negativity (LAN). Experiments 3 and 4 manipulated contextual information and found that sentences with local semantic anomalies elicited N400 effects when preceded by a "no-mention" context, but the same anomalies elicited P600 effects when preceded by a "previous-mention" context. This suggests that discourse activates structured event-representations within semantic knowledge that can similarly "tip the balance" between syntactic and semantic processing streams, thereby influencing the outcome of the conflict.;Taken together, these findings suggest that syntactic and semantic information are processed in parallel streams that are fully-interactive, wherein the strength of cues influences the conflict's outcome and determines which level is most affected by the conflict. Under normal conditions, streams converge on a single representation. However, during conflict streams can vie for interpretive dominance – sometimes tipping in favor of semantic reanalysis (N400), sometimes toward structural reprocessing (P600), and occasionally caught between the two (LAN). These findings provide insights into issues in linguistic theory and psycholinguistic models of language processing, and advance our understanding of how people make sense of conflicting information during language comprehension. |
