Sixty college subjects were administered low frequency verbal discrimination lists under the conditions of pronouncing versus button pressing as a method of choice. There were sixteen word pairs in each list, and the words were three- and four-letter low frequency words selected from the Thorndike-Lorge tables. Four random orders of the pairs were constructed for each list, those four orders being presented for Trials 1 through 4, repeated for Trials 5 through 8, etc. The lists were presented on a Stowe memory drum. Each pair was presented twice in a row at a 2:2-second rate. Ten subjects received one of the two lists, and ten other subjects received the other list within each condition. The three conditions differed only in their method of choosing the correct item during anticipation. Some of the results indicated that the mean sum of errors for the groups that pronounced their choice was lower than the mean sum of errors for the groups that pressed a button to indicate their choice or verbalized the position of their choice. Also, the groups that pronounced their choice required fewer trials to criterion. These differences were not statistically reliable. (WR)