Two eight-member equivalence classes of visual stimuli were established during three phases of a training program. In Phase 1, two training arrangements were compared. In one, 3 subjects were taught on different trials to select from a single pair of comparison stimuli (A,, A2) in response to eight sample stimuli that were trained in pairs (B,, B2; Cl, C2; D,, D2; E,, E2). In the second arrangement, subjects were taught to select from four pairs of comparisons (B,, B2; C,, C2; D3, D2; E2, E2) in response to two samples (A,, A2). Training with the single pair of comparison stimuli resulted in the development of equivalence relations (B,C,, B2C2, D,B,, D2B2, B,E1, B2E2, CID,, C2D2, CIE,, C2E2, DIE,, D2E2, and their reciprocals) between the sample, stimuli without direct training of these relations. In the other training arrangement, these relations among the comparison stimuli developed in the performance of 1 subject only. In Phase 2, three new pairs of stimuli (F,, F2; G,, G2; H,, H2) were substituted for three of the original pairs (B,, B2; C,, C2; D,, D2) and the training arrangements for the groups were reversed. Following training, the performances that showed equivalence relations on the probes in the first phase also showed equivalence relations in the second phase. If such relations did not develop in the first phase, they did not do so in the second phase. In Phase 3, relations between stimuli across the two previous phases (e.g., B,F,, B2F2, BIG,, B2H2, C,F1, etc.)were investigated. The 4 subjects whose performances showed the development of these relations were taught to select one stimulus from each class (E, and E2) in response to a verbal label (I, and I2) and then were tested to see if the verbal label controlled responding to the remaining members of the class (e.g., I,A,, I2A2, I,B,, I2B2, etc.). For 3 subjects, this generalized control occurred; for the 4th, generalization occurred only after verbal training with a second pair of visual stimuli (F, and F2). In retests several months later, these auditory-visual relations were found to be intact or, if not, were recovered without direct training.

Institutional breakfast-serving procedures were manipulated to assess what effect changes in that aspect of the environment would have on requests for food. During baseline, six severely retarded children were required to pick up their food trays and return to their seats. The first manipulation, delaying the giving of the food tray for 15 seconds, served as a cue to evoke meal requests by three of the six children. Two of the remaining three required a model of an appropriate meal request (i.e., "Tray, please.") at the end of the 15-second delay before they began requesting their meals. To evoke meal requests from the sixth child, an intensive training procedure, consisting of massed trials of delay and modeling, was required. Three different probes were administered to assess generalization across the people serving the meals, across mealtimes, and across both people and mealtimes. Typically, generalized responding in these new situations could be prompted by use of the 15-second delay procedure. Functional aspects of the delay procedure and its potential usefulness for evoking speech and facilitating generalization are discussed.

Two subjects with retardation who exhibited generalized identity matching, but who had extensive histories of failure to acquire arbitrary matching, were exposed to a series of conditions designed to train separately the components of a two-choice conditional discrimination. First, the successive discrimination between the sample stimuli was established by programming a different schedule of reinforcement in the presence of each sample stimulus. Schedule performance was acquired and maintained by both subjects, but neither acquired arbitrary matching. To train the simultaneous discrimination between the comparison stimuli, 1 subject was then exposed to a series of simple discrimination reversals and subsequently failed to acquire arbitrary matching. Both subjects acquired arbitrary matching under a procedure that maintained both the sample and the comparison discrimination by first presenting entire sessions composed of one sample-comparison relation and then gradually reducing the number of consecutive trials with the same sample until sample presentation was randomized (schedule performance was maintained). Removal of the schedule requirement had no effect on arbitrary matching accuracy. Both subjects subsequently demonstrated control by relations symmetric to the trained relations.

In Experiment 1, classroom teachers were taught to delay their offers of help in naturally occurring situations, and thereby to provide additional opportunities for language use by six moderately retarded language-delayed children. The teachers introduced this delay technique in a multiple-baseline design across the six children. As delays were used, child verbal initiations increased. Follow-up assessment showed that teachers were maintaining greater than baseline levels of the delay technique after 10 weeks. Experiment 2 replicated the findings of Experiment 1, and included a more thorough maintenance assessment, while focusing on teachers' generalization of the delay technique. Teachers were found to generalize their use of delay to 56% of their monitored untaught opportunities. The two experiments show that (a) the delay technique is quick to teach and simple to implement, (b) delays do provide opportunities for children to initiate, (c) teachers can generalize their use of delay to novel self-selected situations, and (d) teachers can maintain their use of delays over time.