This experiment was conducted to examine word presentation routines to determine their effectiveness in spelling drills. The design included segmentation or focal unit (letter, chunk, whole word), audio (audio, no audio), and word type (List 1--pronounceable chunks, List 2--pronounced in running speech). Subjects were 48 students in Grades 3, 4, and 5 spelling classes. Computer-assisted spelling drills were utilized for the experiment, in which misspelled words were given one of the six treatments. Retention tests were given at two weeks and six weeks after acquisition. Analysis of variance on learning rates and retention revealed that List 2 words were acquired faster than List 1 words and no variables reached significance on retention tests. T-tests, computed for acquisition of words in single sessions versus acquisition in more than one session, showed that 40 percent were massed and 60 percent acquired in more than one session. Overall t's were significant on the first retention test, indicating that distributed practice was superior to massed; however, on the second test, words were not differentially retained. Analyses of misspellings showed there were no differences in the percentages of misspellings involving errors within the four categories (additions, substitutions, omissions, and reversals). Substitutions were the most frequent error. Latency analyses indicate that correct responses have shorter mean word latencies than incorrect responses. (JF)