Seventeen deaf adults and 17 hearing controls participated in three experiments to determine the use of morphemes by deaf Ss as a unit of analysis in reading English. Ss were asked to circle pairs of words with related meaning, divide a derivationally complex word into its component parts, and choose words that best complete sentences. Among findings were that the deaf Ss generally were not confused by visual similarities and identified word pairs with 90% accuracy; deaf Ss were more likely to commit syllabic errors than hearing Ss and deaf Ss scored significantly above chance on the word completion tasks, with an average of 84% correct. It is suggested that in the absence of sound, morphological information can provide a compensatory approach to word identification and reading comprehension. (CL)