The assumption that the dyadic communication pattern (one teacher-one student) is the most effective pattern for encouraging language and speech development among elementary and preschool children is tested in this study. Fifty-six 4-year-old children from the Task Force Head Start Program of Buffalo, New York, were observed in four different communication patterns and the mean length of their verbalizations was recorded. Also noted was the interaction of sex with each of the conditions. The patterns used were the dyad, the triad (1 experimenter and 2 children), the small group (1 experimenter and 3 children), and the role-playing triad (same as triad, but with children encouraged to act out roles). Analysis of the data revealed no significant inter-pattern differences in mean verbalization length, except that the small group elicited a statistically greater amount of speech than did the dyad (both with repetitions left in and with repetitions deleted). With and without repetitions, girls produced significantly more speech than boys. While the actual difference between the speech from the small group and from the dyad is small (less than one word per response), findings are important because they demonstrate that the dyadic situation may not be justified in terms of speech development, particularly since it is less economical of the teacher's time. (MH)