  |
75 years of Linguistics at SOAS
|
 |
Conference on Language Documentation and Linguistic Theory7-8 December 2007, SOAS, London |
|||
|
|
Richard HudsonUniversity College London Towards a useful theory of language
Ever since my days as a graduate student in the SOAS Department of Phonetics and Linguistics I have been aware of the tensions between Truth and Utility as applied to linguistic theory. Even in the early 1960s there was a rather complicated choice between Chomsky ("truth") and Halliday ("utility"), and the same choice (polarised round the same two figures) still confronts us. I shall outline the history of this choice as it applies to education, including the major industry of producing descriptive grammars of English for the EFL market; and I shall conclude that linguistic theory really does matter to education. I shall then generalise this conclusion to a number of other applications of linguistic theory, including language documentation. However, the conclusion does not resolve the tension between Truth and Utility; if anything, it shows how real the tension is and how important it is to find a resolution. The problem is that linguistic theory has to be as complex as language, and a description of a particular language has to be as complex as that language – i.e. very complex. To make this claim concrete, I shall outline a few areas of language structure to show that the entire language is actually a single very complex network (which fits into the even bigger network of general knowledge). This idea is the basis for my own theory (Word Grammar) but also underlies Structuralism and all its descendants. But if this theoretical claim is true, how can it be made useful? Given traditional techniques of research and presentation, networks are too complex for practical applications such as language documentation; but computers open new possibilities. I shall finish by mentioning some existing computer systems which point in the right direction, and then describing the system which I believe we need (but which has not yet been built).
|
||
