This study was designed to test the long-term effectiveness of an approach used by B. Hart and T. Risley ( Journal of Applied Behavior Analysis , 1968 , 1 , 253–262) to increase the use of color-noun combinations in the spontaneous speech of disadvantaged children. The approach consisted of making access to classroom materials contingent upon appropriate color-noun descriptions. A 2-week baseline was established for six subjects in each of two Head Start classrooms. During the next 4 weeks, the contingencies were enforced in the intervention classroom but not in the no-intervention classroom. Four weeks after the contingencies were removed, follow-up data were collected. The mean rate of colornoun combinations used by the children in the intervention group increased from 0.8/hr during baseline to 6.4/hr during intervention, then fell to 0.5/hr during follow-up. In the no-intervention group, the mean rate of color-noun usage ranged only from 0.3 to 0.8 across all periods. It was concluded that there was no evidence of natural contingencies which would maintain the behavior after artificial contingencies were removed.