The purpose of this study was to establish answers to the following questions: (1) Does a cross-age teaching program provide a situation in which school children can extend their role repertories by enacting the role of the tutor to younger children? (2) Can the language used by children in the role of tutor be classified under categories derived for describing teacher verbal behavior? (3) Is the language of sixth grade children measurably different in complexity when the children are in the role of pupil from when they are in the role of tutor to third grade children? Random samples from the corpus of language of the pupil role and from the corpus of language of the tutor role were collected from 15 sixth grade children. Differences in length and complexity of utterances were compared for the two roles. Conclusions are: (1) Sixth grade children in the tutor role produce language that can be classified in the categories that describe teacher verbal behavior; (2) There was a statistically significant difference in the median length of T-units in the two roles, the longer being produced in the tutor role; and (3) There was a statistically significant difference in the complexity of language used in the two roles, the more complex language being produced in the tutor role. (Author/DB)