We conducted 2 experiments on the effects of social attention versus token contingencies on the emission of verbal operants by preschoolers, with and without a disability diagnosis. Four participants, 3 females and 1 male, 3 to 4 years old, were selected to participate in Experiment 1 and 6 participants, 5 females and 1 male, 2 to 4 years old, in Experiment 2. Experiment 1 compared effects of the 2 contingencies on numbers of child-initiated tacts in 3 different settings using an alternating treatment design. Experiment 2, using a multielement design, compared the automated delivery of tokens versus adult attention on the percentage of peer-to-peer and adult conversational units. Participants in both experiments initiated more tacts with contingent social attention than with contingent tokens. Implications are that tacts and conversational units are maintained more by social reinforcers than nonsocial generalized conditioned reinforcers (i.e., tokens). Social control of tacts may be essential to social verbal behavior.