Abstract
The utterances of particular words-"selected verbal responses" (SVR's)-were reinforced according to five different drl schedules and studied under changes in the schedule and/or the SVR. The emissions of SVR's were shaped by the procedure, while the rates of saying all the different words were unaffected. The distributions of the percentages of different IRT's and IRT's/OP's changed very little when only the SVR was changed, suggesting that reinforcements had their main effect upon some kind of "delaying behavior." The modal IRT classes were just above the drl specifications, and there was no evidence of a second modal class at short IRT's. The differences between actual and optimum median IRT's were fairly constant under different drl schedules. Individual differences appeared in the behavior interpolated between SVR's: some subjects (Ss) counted the number of intervening words, some reported increases in tension, and still others seemed to change the pitch of voice in a cyclical pattern. The transcripts of intervening verbal behavior indicated the presence of some chains of words, presumably formed before the experiment and "adapted" to the length of the delay required for reinforcement. In the experimental situation the formation of verbal chains was only rarely observable.