Junior high school students (N=104) were given instruction using programed booklets on modular addition. Twelve conditions of verbal mediation were randomly assigned to students. Conditions were defined by presence or absence of introductory discussion (2 levels), number of worked examples (2 levels), and type of rule (3 levels: specific, general, none). After instruction, subjects were tested for three types of transfer: specialization, generalization, and reasoning by analogy. The use of a concrete model as introductory was found to have a significant (p less than .05) facilitating effects on specialization and a less-pronounced effect on generalization. For all three levels of transfer, the groups given specific or general rules performed better than those given no rule. Limitations of the study and suggestions for further research are discussed. (SD)