Used an operant group paradigm to investigate the effects of reinforcement in combination with differing therapist styles upon affective verbalizations. Therapists' usage of direct questioning to elicit Ss' affective verbalizations proved quantitatively superior (in number of reinforceable responses) to either a no therapist or role modelling therapist condition (N = 72). There was evidence, however, that Ss in reinforced role modelling conditions were significantly better than direct elicitation Ss in predicting other group members' self‐descriptions and in the degree to which Ss felt accurately perceived by others, while the reverse was true for non‐reinforced Ss.