In two experiments, we found that readers are sensitive to manipulations of syntactically marked focus and that focus is an effective message level contextual priming mechanism. Changes in focus resulted in changes in sentence context effects on subsequent target word processing. This was demonstrated in latency to name the target word (Experiment 1) and in initial looking time on the target in silent reading (Experiment 2). Experiment 2 also revealed direct effects on the focused items, as readers made fewer regressions and spent less total time on a word that was focused than when it was not focused. However, no initial processing time effects were found on the focused word.Word processing is a fundamental component of the reading process. Thus, our understanding of the reading process necessarily hinges in large part on our understanding of how words are processed in context. It has often been observed that words are processed more easily in related sentence contexts than in unrelated sentence contexts (Duffy, Henderson, & Morris, 1989;Foss, 1982;Hess, Foss, & Carroll, 1995;Morris, 1994;Simpson, Peterson, Casteel, & Burgess, 1989;Stanovich & West, 1979,1983West & Stanovich, 1982;Whitney, McKay, Kellas, & Emerson, 1985). However, explanations ofthese effects differ greatly. The purpose of the present experiments was to investigate a possible message level mechanism of sentence context effects.Lexical level explanations of sentence context effects on word processing center on the premise that contextual facilitation arises from word-to-word priming mechanisms (intralexical priming). The assumptions underlying this position are derived from modular views of the language processing system in which the lexical processor is autonomous, acting only on information contained within the lexicon. According to this account, contextual effects are the result of activation spreading from related words in a context to the target word, speeding access of the subsequent target word (e.g., Duffy et al., 1989;Fodor, 1983;Forster, 1981; Seidenberg, Tanenhaus, Leiman, & Bienkowski, 1982), in much the same way that semantic relatedness effects occur in word lists (e.g., Meyer & Schvaneveldt, 1976).While context effects on word processing in reading may arise in part from intralexical priming, there is growThis research was partially supported by Undergraduate Research Training Grant SBR-9400285 from the National Science Foundation. The authors thank Jennifer Mangum for her assistance in material preparation and in collecting the Experiment I data. 1. R. Folk is now at the Department ofCognitive Science, Johns Hopkins University, Baltimore, MD. Correspondence concerning this article should be addressed to R. K. Morris, Department of Psychology, University of South Carolina, Columbia, SC 29208 (e-mail: morris@garnet.cla.sc.edu).ing consensus that this processing advantage cannot be fully accounted for by lexical level explanations. For example, there is evidence of faster processing for words in related context, even in the absence ...