This project investigated three methods of providing supervisory feedback to junior-year student teachers about specific verbal teaching behaviors. Subjects were divided into three groups and, during their regular 9-week student teaching assignments, completed two 20-minute teaching situations which were coded with Flanders' Interaction Analysis. Behaviors were analyzed according to prescribed group treatment: Group A (N=20) received comments from a university supervisor and a coded analysis of their teaching made by a trained observer; Group B (N=20) viewed, coded, and analyzed videotaped recordings of their teaching; and Group C (N=17) analyzed their own videotapes and received supervisory comments on their analyses. Analysis of data showed no significant difference among groups in the ratio between indirect and direct teacher verbal statements. The results of this study imply that the particular type of feedback student teachers receive is relatively unimportant in changing teacher behavior. (Author/LP)