In this final chapter, we turn to a different area in which the choice of an appropriate theory of word structure is important. This is the problem of how a computational system might be designed to ‘parse’ individual words in a natural language. The notion of computational parsing is most familiar in the syntactic domain, of course, where it refers to the process of recovering structural descriptions for natural-language sentences from a string of words in light of the system's representation of the syntax and lexicon of the language in question. Analogously, ‘parsing’ in morphology is the process of retrieving the information carried by particular words, as they contribute to the meaning and the structure of larger linguistic constructions within which they appear.

Reasons to study morphology as parsing

When we ask in a computational system for an account of the information carried by individual word forms, there are two rather different sorts of motivation we might have, and the extent to which we want to pursue a linguistically well-motivated analysis will depend on the goals of the analyst. An understanding of this difference, in turn, is important in determining the sorts of criticism of a given system that might be appropriate. The first set of motivations can be identified (somewhat pejoratively, perhaps, in a work devoted to questions of general linguistic theory) as ‘technological’ reasons. In these cases, the question of what information a word form carries is posed because we want the answer.