There are inherent difficulties with text as an instructional format. This research presents the development of a diagrammatic representation that overcomes some of those difficulties. A simple "language" is presented that includes both procedural and model information. This approach is applied to the instruction of word processing/text-editing skills. An experimental methodology is presented and preliminary results are reported that suggest that the diagrammatic approach results in improvements in both the amount of word processing accomplished and the speed with which modifications are made.With rapid changes in computer technology, it has become extremely important to make systems easy to use by a rather large group of people. There are numerous ways to do this, including major changes in system design and changes in the types of instructional and documentation materials available to users.Although the problem is evident for users with varying amounts of experience, the difficulties are particularly striking for the naive user who has had little if any prior computer experience. For the uninitiated, the information available on a new computer can resemble a foreign language, with the corresponding misinterpretations. The research described here is directed toward decreasing the alien character of instructions for this group.Our strategy in doing this has been to develop a set of cognitive principles that can serve as the basis for instructional design (Black & Sebrechts, 1981). We are developing these principles by an experimental evaluation of word processing/text-editing skills. Specifically, we are interested in designing and assessing a diagrammatic "language" for computer use. Although in these studies we are focusing on word processing instruction, we believe that the same principles should be useful for both instruction and documentation of a wide class of complex skills.In this paper, we will provide a general description of our assessment of a diagrammatic language for instruction. First, the conceptual motivation is described for the use of a diagrammatic language. Then, the structure of the language is presented, with the underlying cognitive rationale. Finally, we present our strategy forThe research described in this paper was funded in part by a contract with the IBM Corporation.