In this brief article, Bara and Guida provide a classification scheme for natural language processing models. They propose that such models be classified as cognitive (those models that study the human language processes); pragmatic (those models that look at language as a medium for interpersonal communication); or linguistic (those models that analyze the structure of language and how that structure relates to meaning). They go on to look at the role of competence models in the development of performance systems and briefly discuss the history of the kinds of models found in natural language research.
While little of the information is particularly novel, and the classifications provided are not terribly surprising, this article does provide a nice summary of various issues that natural language researchers should be concerned with. Further, it encourages reseachers to pay attention to the theory underlying their work, even when building practical systems, which is a moral well worth emphasizing.