Forty-eight special class educable mentally retarded (EMR) students, 9 to 12 years of age, from lower socioeconomic backgrounds were examined to determine effects of communication training and learning potential (LP) training on tasks in the Communication Test Battery (CTB) and on scores in the Raven Progressive Matrices (RPM). The Ss were assigned to the following four groups (12 Ss in each group): the communication control group, the LP trained condition group, the communication trained condition group, and the LP control condition group. Training was by paired subjects during six 1/2-hour sessions twice per week and involved increasingly complex tasks in areas of perceptual discrimination, cognitive classification, language, and roletaking. Tasks incorporated concrete stimuli and required procedural verbal analysis such as describing a picture in an array for a listener. A final task required the S to be the trainer, to explain the rules, and to guide adults through an exercise. Results showed that the communication trained group had higher mean scores than the LP trained group and the communication control group on communication skills and specific skills of perception and cognition. Also, the findings showed that communication training was effective in improving performance on the RPM even when no intervening RPM training was offered. (MC)