Abstract
Two studies are reported that investigate the use of graphic and contextual information in word recognition. Subjects orally read stories that contained ten words that were altered by substituting a single letter, so that another word was formed that was anomalous within the sentence. The critical dependent variable was the proportion of times the subject read the contextually appropriate word rather than the actual stimulus word. The first study used different orienting instructions to manipulate subjects' response set. In the second experiment, instructions for either contextually appropriate reading or accurate reading were given in order to examine the relationships between information processing demands and reading ability. The results indicated that good and poor readers were equally able to conform to task demands for contextually appropriate reading responses, but poor readers were less able to suppress contextual information when accurate reading was required.