A series of five experiments was conducted to determine whether operant or respondent factors controlled the emission of a particular vocalization ( "Q" ) by human infants 16 to 18 months old. Experiment 1 consisted of a pilot investigation of the effects of an autoshaping procedure on three infants' vocal behavior. All three subjects demonstrated increased emission of the target sound during the CR period. Experiments 2 through 4 attempted to replicate the findings of Experiment 1 under controlled conditions, and failed to do so. Experiment 5a presented infant subjects with a discrete-trial operant procedure (having the identical temporal parameters as the autoshaping procedure used in Experiments 1 through 4) , during which subjects received rewards only after emitting the target sound. All three experimentally naive subjects in this condition emitted the target sounds, and each met an acquisition criterion of 15 successive trials with at least one target response (CR) per trial. Subjects in Experiment 5b were exposed to the autoshaping procedure employed in Experiments 1 through 4; in addition, they were rewarded verbally for emitting the target sound at any time during the session. One of these three subjects increased her rate of target sound emission, but never met the acquisition criterion. The results suggest that vocalizations of subjects in this age range are susceptible to response-reinforcer (operant) manipulations and not to stimulus-stimulus (Pavlovian) associations. This conclusion differs from the findings of previous investigations that used younger infant subjects, suggesting that subject age may be important when conditioning vocalizations. Further research is also recommended to determine the utility of employing autoshaping procedures with humans.