ABSTRACT
The chapter examines the architectures of rather complex teaching machines and has shown us the natural language processing and the knowledge modeling that function as infrastructure. The ingenuity and dogged persistence necessary for such challenging work can only be admired. Nevertheless, one may well wonder what is the point of all of this work, why one should bother with such complex research. After all, a human teacher can perform all the so-called intelligent tasks that current intelligent computer-assisted language learning (ICALL) software performs and, at present, do a considerably better job in most cases. If it is, then the investment is a decidedly long-term one. Clearly all of these projects demand substantial labor to eke out even the smallest amount of so-called intelligence. This might still be a reasonable motivation if one had the feeling that all or most of this work would be adapted on a large scale.
