Integration of Clauses in Oral Reading: The Effects of Syntactic and Semantic Constraints on the Eye-Voice Span

Abstract

INVESTIGATED THE EFFECTS of syntactic and semantic constraints on eye-voice span to inform on several reading models. When people read aloud, their eyes scan several words ahead of the word being pronounced; thus if a printed text is abruptly removed, readers can still report words that have been seen but not yet spoken. The number of such words reported reflects the extent to which the eyes lead the voice (the eye-voice span). This span was used to investigate the extent to which readers use either syntactic or semantic cues to control segmentation during reading. Syntactic constraint was manipulated by varying clause length and by removing texts either within or between clauses. Semantic constraint was manipulated by providing either meaningful or anomalous material following text removal. Eye-voice spans were generally longer for longer clauses and after within-clause interruptions (high syntactic constraints); they were also longer when the material to be reported was meaningful (high semantic constraint). A pure syntactic-decoding hypothesis and an informed guessing model were rejected in favor of a semantic integration model in which skilled readers pace their information intake in order to facilitate integration of that information.