The purpose of this study was to investigate the ability of good and poor readers in grade five to recognize the form of derived words where morphophonemic alternation occurs but orthographic consistency exists. A 2 x 2 x 4 factorial design was used to investigate the effects of reading ability, word reality (real and pseudo), and mode of presentation for stimuli and response items (oral and written). Five subjects were assigned to each of the sixteen cells, with an equal number of males and females represented in the categories of good readers and poor readers. Target words were placed in sentence contexts within the purported vocabulary size of the subjects. Theme forms of the target word were presented orally or visually to fit as slot-fillers in the first sentence which was read aloud. A second sentence was read aloud with a slot indicated for the position of the derived form of the target word. Subjects were asked to select one of four alternatives, depicted orally or visually depending upon cell, to correctly occupy the indicated slot. A pattern was revealed, suggesting that good readers can more ably recognize the underlying forms of words than poor readers. When surface structures are related, these differences between good and poor readers diminish. (Author/WR)