In the context of automatic speech and language processing for translation and question-answering systems, this chapter usefully and interestingly outlines the problems inherent in analyzing compound nouns. It surveys attempts to deal with these problems from the standpoints of theoretical linguistics, computational linguistics, and artificial intelligence (frames and knowledge representation). Parallel versus sequential processing of three different kinds of analyses (syntactic, semantic, and pragmatic) is also discussed. The author concludes:
It is clear that characterising compound noun interpretation processes, and further, modelling them sufficiently precisely to drive language processing, and even more, speech processing, programs, is a major enterprise. . . .
The business of testing proposed pragmatic knowledge bases and inference operations in the context of analysis programs allowing more dynamic interaction between pragmatic and other operations, with the aim of reducing input uncertainty by bringing as much, different, information to bear as soon as possible, is therefore not just the computational linguist’s challenge, but his nightmare.