ABSTRACT
For adults, the learning of a foreign language is often difficult, sometimes virtually impossible. Systems that promise a way of reducing this language barrier will interest millions of people, if they can really be demonstrated to work. It is not surprising that the enormous breakthroughs in personal computer technology over the last dozen years have led to a renewed interest in the development of computational systems for facilitating language learning. However, the development of these systems is dependent on many types of expertise and requires a great investment of time and energy. If we had a set of basic design principles that could guide the construction of these systems, we could avoid inadequate products and mismatched capabilities. If one looks at work in learning theory, psycholinguistics, and human-computer interaction, there are many signposts or principles that could direct this work. However, to provide a surer grounding for this effort, we need to make direct tests of the pedagogical effects of these principles. Before we examine ways of making these tests, let us review some of the lessons that can be derived from previous work in psychology.
