Some natural language applications

Improving the man-machine interface

You might still need some convincing that programming is for you if your interests are more in practical applications than in abstract theory or even straight linguistic description. In that case all you need to do is look around you. It is no accident that computer scientists are increasingly attracted nowadays by the mechanisms of language understanding and production and that, for their part, linguists are beginning to take advantage of the insights offered by computational approaches to their traditional activities.

The fact is that with computers becoming more and more a necessary feature of our everday lives there is an ever growing need to assimilate them into our normal patterns of behaviour and the surest way to achieve this is by providing modes of interaction (interfaces) which are as much like inter-human communication as possible. Ideally, then, a computer should understand you and talk back to you in your native language.

Some shorter term goals

Man-machine communication using natural language has, of course, been a long-running dream, but the idea is no longer quite the distant fantasy it was in 1968 when Kubrick gave voice to HAL. Although, nearer the millenium, we can see that 2001 was too optimistic a date for for even a passable realisation of the intelligent talking machine, there is no doubt that significant progress is being made on a range of smaller scale projects which can be seen as part of the grand plan. Here are some examples of the kinds of task being worked on by a growing band of computer scientists, linguists, and others:

There are already applications meeting more or less well some of these needs. There is also no doubt that there is vast room for the improvement of existing software and for extending the range of applications provided.

It is not difficult to see that this overall programme naturally forms a major component of computer science research. At the same time it is obvious that, since work on all of these projects in one way or another necessitates explicit accounts of linguistic structure, the range of potential real world applications provides linguists with many opportunities to test their theoretical models of language by implementing them computationally.

Ron Brasington
Department of Linguistic Science
The University of Reading