This paper provides an idea of how to improve computer-assisted instruction effectiveness by incorporating teachers' images and voices into computers. Efforts to increase students' motivations by improving computer-assisted instruction feedback systems through the use of arcade-style video games and animation have proven to be ineffective and distracting. In contrast, students seem to be highly motivated when the computer praises them for correct responses. A recent review of the literature reveals that praise is effective when it is delivered contingently, focuses on student accomplishment, provides information to students about their competence, is related to problem solving, shows spontaneity and variety, and is task-relevant; whereas ineffective praise rewards mere participation, gives no information to students about their status, attributes success to ability alone, is delivered inconsistently and unsystematically, and gives low achievers less praise or over-praises them for working hard rather than for accomplishment. Computerized teacher's praise seems to meet most criteria of effective praise and eliminates the occurrence of ineffective praise. Since dual-sensory instruction can now be realized through the use of inexpensive personal computers, human images and voices can be introduced onto the software in order to personalize instruction, ask questions, provide background knowledge, and become an audible expert system. (Contains 12 references.) (ALF)