This study replicates an experimental design testing whether effects of supervisory training in transactional analysis (TA) language cues can be carried outside the laboratory setting of simulated testing into the field testing of professional work settings involving dialogues with supervisees. Transactional analysis theory presumes the communication act is metacognitive, and that individuals can select productive responses by identifying categories of word usage during interpersonal exchange. Subjects consisted of 62 graduate students enrolled in a course in supervision and school leadership. Both the experimental and control groups engaged in the standard course of instruction; however, only the experimental group received 12 training units in TA communication. Participants' responses to supervisory simulations were coded as productive or nonproductive transactions. In the field test, groups conducted conferences with teacher-supervisees that were analyzed as productive or nonproductive exchanges. Findings showed that a difference between the simulated test and the field test appears in the amount of TA training. In the simulated test, training accounted for 63 percent of variance, whereas only 20 percent of variance in the field test was attributed to training. Though TA training increases quality communication in a conference setting, various issues remain unresolved. This study signifies that dual testing lends assurance to TA training in enhancing supervisors' productive interpersonal communications. Further study should identify, under field conditions, interaction of variables resulting in more productive communication transactions. Six tables of data are appended. (CJH)