The question of whether meaning can be extracted from unidentified parafoveal words was examined using fluent Spanish-English bilinguals. In Experiment 1, subjects fixated on a central cross, and a preview word was presented to the right of fixation in parafoveal vision. During the saccade to the parafoveal preview word, the preview was replaced by the target word, which the subject was required to name. In Experiment 2, subjects read sentences containing the target word, and, as in the naming task, a preview word was replaced by the target word when the subject's saccade crossed a boundary location. In both experiments, preview words were identical to the target word, translations, orthographic controls for the translations, or unrelated words in the opposite language. In both experiments, the preview benefit from the translation conditions was no greater than would be predicted by the orthographic similarity of the preview to the target. Hence, the data indicated that subjects obtained no useful semantic information from words seen parafoveally that enabled them to identify them more quickly on the subsequent fixation.

The effect of clause wrap-up on eye movements in reading was examined. Readers read passages in which a target category noun referred to either a high typical or a low typical antecedent. In addition, the category noun was either clause final or non-clause final. There were four primary results: (1) Readers looked longer at a category noun when its antecedent was a low typical member of the category than when it was a high typical member; (2) readers looked longer at the category noun and at the post-category region when they were clause final than when they were not clause final; (3) readers regressed from a category noun or post-category region more frequently when it was clause final than when it was not clause final; and (4) readers made longer initial saccades when their eyes left the category noun or post-category region when this word was in clause final position than when it was not clause final. The last result suggests that sometimes higher order processes that are related to making a decision about when to move the eyes impinge on lower level decisions that are typically associated with deciding where to move the eyes.

Research dealing with parafoveal processing during eye fixations is reviewed. Four main topics are addressed: (1) parafoveal processing, (2) word skipping, (3) preview benefit effects, and (4) parafoveal-on-foveal effects. We argue that word skipping effects reflect the fact that a parafoveal word (word n+1) has been identified on the fixation on word n. We also review evidence which strongly suggests that preview benefits during reading are not due to semantic processing of a parafoveal word. Finally, we review the more recent and more controversial research suggesting that the meaning of word n+1 can influence the fixation time on word n, and argue that it is premature at this point to accept the validity of such findings with respect to normal reading. Implications of the research for serial attention shift models like the E-Z Reader model are also discussed. 4 While a great deal has been learned about eye movements during reading over the past twenty-five to thirty years (Liversedge & Findlay, 2000;Rayner, 1978Rayner, , 1998, there remain a number of unresolved issues (Starr & Rayner, 2001). In this chapter, we will focus our discussion on research related to one of these unresolved issues: parafoveal semantic processing of words. Research on this issue has apparently gained momentum because it has been assumed that if there were so-called parafoveal-on-foveal effects, or evidence that the meaning of the word to the right of fixation influences the duration of the fixation on the currently fixated word, it would be damaging to serial attention shift models such as the E-Z Reader model (Reichle, Pollatsek, Fisher, & Rayner, 1998). We will return to this issue at the end of the chapter. However, before discussing the relevance of such research for the E-Z Reader model, we will first provide a general review of research on parafoveal processing and then discuss in turn (1) word skipping, (2) preview benefit effects, and (3) parafoveal-onfoveal effects.We argue that readers can identify word n+1 while fixating word n. When they do, its meaning becomes available and can influence fixation times on word n (and also word n+2). Furthermore, if word n+1 is identified, then the reader will skip that word. However, the primary argument we will make in this chapter is that if word n+1 is not identified then its meaning does not become available and therefore cannot affect fixation time on word n. In such a situation, word n+1 will typically not be skipped but must be fixated in order for word identification to occur. We argue that it would be premature at this point to accept evidence from studies that claim to show parafoveal-on-foveal effects as being very strong. Some of the evidence is based on tasks that may or may not easily generalize to reading and some of the evidence is inconsistent: while there are some studies showing parafoveal-on-foveal effects, there are also a number of studies that do not show such effects. While we will argue against the validity of some claims regarding parafoveal-on-foveal...

In a number of studies, we have investigated the processes involved in comprehending ambiguous words in various kinds of sentence and discourse contexts. In our studies, we monitor readers' eye movements as they read, and we draw inferences about processing on the basis of the eye movement record. We are interested in how the sentence and discourse context influence the process of accessing and selecting a meaning for words that have multiple meanings.In this chapter, we discuss the development of the reordered access model, the model that has guided our research on the effect of context on lexical ambiguity resolution. We present a number of studies investigating one particular effect of context that is relevant to the model: the subordinate bias effect. We also address a methodological issue, that of the appropriate control conditions in studies of lexical ambiguity, and we consider one alternative model, the ordered access model. Finally, we present a simulation of the reordered access model using constraint-satisfaction architecture; the simulation gives a promising account of many of the empirical results discussed.How context influences the processing of ambiguous words is part of a larger debate over the extent to which access of word meaning is an automatic, modularized process that is impervious to outside (e.g., contextual) influences. Theories that take the modular viewpoint include exhaustive access theories, which assume that all meanings of an ambiguous word are initially activated regardless of context. Such theories originally received support from cross-modal priming studies (Seidenberg, Tanenhaus, Leiman, & Bienkowski, 1982; Swinney, 1979), which showed that even when preceded by clearly disambiguating context, the unintended as well as intended meanings of ambiguous words were Preparation of this chapter and much of our research described in the chapter was supported by Grant HD17246 from the National Institutes of Health. Gretchen Kambe was also supported by a predoctoral fellowship on Grant HD07327, and Keith Rayner was supported by a research scientist award (MH01255). We thank David Gorfein and Dorit Wenke for their comments on an earlier draft of the chapter.