This experiment tested the hypothesis that stimulation in a given mode of discourse (e.g., the argumentative, or "A", mode of theme writing) will improve the quality of compositions written in that mode, but decrease the quality in another mode (e.g., the descriptive, or "D", mode). Ninety first-graders were required to write two themes in the "D" mode and two themes in the "A" mode to determine differences in the performance levels of these two common modes of discourse. Three types of short discussion sessions--"A" mode, "D" mode, and neutral--were combined in various patterns with theme-writing activity to test the effect of this kind of stimulation on the children's performances. Themes were graded by using a theme-rating scale that excluded all grammatical criteria. Variables such as the child's sex and the order in which he wrote the four themes and participated in the three discussion groups were measured. Results of the testing indicated that (1) the "D"-mode discussion raised the descriptive quality of themes in the "A" mode and reduced the argumentative quality of both types of themes, and (2) the "A"-mode discussion had little effect on the argumentative quality of either of the two types of themes. (JB)