A study examined the effects of teacher feedback on the oral reading performance of nine low-ability second grade readers. Subjects were randomly assigned to one of three treatment sequence conditions. A single-subject format was incorporated by using a Latin Square design for presenting the three treatment conditions--graphophonemic immediate, graphophonemic delayed, and semantic delayed--to all three groups. On each of 23 days, the students read aloud a different 200-400 word passage. Dependent measures were literal comprehension and qualitative dimensions of word recognition, graphic similarity, semantic acceptability, and self-corrections. Results indicated that the treatments did not differentially affect the graphic similarity of readers' responses, although the semantic delayed condition did encourage responses that were higher in semantic acceptability. In addition, results indicated no significant differences in the percent of miscues corrected by each of the three groups. However, all three groups increased in the use of the correction process from baseline 1 to baseline 4. No systematic sequence of treatment effects was evident for either word recognition or comprehension. Finally, the semantic delayed condition influenced comprehension more positively that did the graphophonemic immediate and graphophonemic delayed conditions. (HTH)