Against parafoveal semantic preprocessing during eye fixations in reading.

Subjects' eye movements were recorded as they read sentences, and a display change occurred before they fixated on a critical target word (song). In the target word location one of four alternative previews was initially present: the target word itself, a word that was semantically related to the target (tune), a word that was unrelated to the target (door), or a nonword that was visually similar to the target (sorp). When the reader's eye movement crossed over an invisible boundary location, the initially displayed preview was replaced by the target word. If automatic semantic preprocessing of parafoveal words occurs in reading, the presence of a semantically related word should facilitate processing of the target word relative to the unrelated word. While the visually similar preview facilitated processing almost as much as the target itself, there was no difference between the semantically related preview and the unrelated preview, even though the semantically related words used in the study produced strong facilitation effects in a standard priming experiment. The results of the study are thus inconsistent with a model in which words in parafoveal vision are semantically preprocessed. It was concluded that the meanings of unidentified parafoveal words do not modulate our reading performance. RESUME Les mouvements oculaires de sujets etaient enregistres pendant la lecture de phrases et un changement dans la presentation survenait avant qu'ils aient fixe un mot ciblc critique (song). Lors de la localisation du mot cible, l'une de quatre possibility de pre-presentation etait presente au commencement: le mot cible lui meme, un mot scmantiquement relie au mot cible (tune), un mot sans relation au mot cible (door), ou un mot sans sens visuellement semblable au mot cible (sorp). Lorsque le mouvement oculaire du lecteur traversait un endroit limite invisible, le stimulus presente au debut etait remplace par le mot cible. Si un pretraitement semantique automatique de mots en parafoveal survenait pendant la lecture, la presence d'un mot semantiquement relie devrait faciliter le traitement du mot cible par rapport au mot non relie. Alors que la pre-presentation visuelle semblable facilitait le traitement presque aussi bien que le mot cible lui-meme, il n'y avait pas de difference entre les pre-presentations reliees semantiquement et cedes non reliees meme si les mots semantiquement relies utilises dans cette etude produisaient une grande facilitation. Les resultats obtenus infirment un modele selon lequel des mots en vision parafoveale sont semantiquement pre-traites. T h e order of authors is counterbalanced in accordance with other manuscripts to which the authors made equal contributions. Portions of (he data were collected as part of a study reported previously by Balota, Pollatsek, and Rayner (1985). The research was supported by grant HDI2727 from the National Institute of Child Health and Human Development. The second author was the recipient of an NIMH postdoctoral fellowship at the University of Massachusetts when the data were collected. A preliminary draft of the manuscript was written while the first author was a sabbatical visitor at the University of Oxford and the third author was a sabbatical visitor at the University of Oregon. We thank those institutions for their hospitality. We also thank Chuck Clifton, Vince DiLollo, and three anonymous reviewers for their comments on the manuscript. Address reprint requests to Keith Rayner, Department of Psychology, University of Massachusetts, Amherst, Massachusetts, U.S.A. 01003.