We investigated the functional role of the child's and experimenter's verbalizations in correspondence training procedures with toy play behaviors in a day-care center setting. Six children participated in a multiple baseline across responses and/or multielement design. Baseline conditions were followed by reinforcement of verbalization. This resulted in little or no change in responding, similar to findings of previous research. Experiment I isolated the child's verbalization as the variable under study. With an experimenter's prompt and postplay reinforcement held constant, the effects of including versus omitting the child's verbalization were examined. A contingency-space analysis revealed that the presence or absence of the child's verbalization exerted no influence on play with the target toy. In Experiment II, a condition in which no experimenter's prompt occurred was added. Results suggested that the complete absence of any antecedent verbalization, by child or experimenter, resulted in much lower rates of play with the target toys. Again, however, when the experimenter's prompt was included, no clear difference was noted between conditions in which the child verbalized and conditions in which the child did not. These results raise doubts about the commonly held view of correspondence training procedures as a method of promoting self-regulation.