GRI Nordisk Språkteknologi AS
5700 Voss, Norway
At the University of Bergen we have at least three departments, located in three different faculties, trying to establish themselves as the most important producer of knowledge identifying themselves as the University's recipient of extra funding for education in the sector of Information Technology. My position at this conference is simple: In this case competition is not a healthy process, producing the best at the lowest price. In this case competition is totally destructive, simply because the University's organizational structure does not reflect society's need of organization of knowledge as it is now, but maybe as it was two centuries ago.
Allow me to give a recent example. Between humans, speech is undoubtedly one of the most efficient forms of communication. Vast amounts of kroner, dollars, pounds and marks have gone into research attempting to make computers recognize isolated words, connected speech or continuous speech - most of the attempts largely unsuccessful. Similarly, most speech synthesizers, taking the ASCII string as the input and generating speech as the output, have a heavy electronic accent, and most of them are unable to combine intelligibility with pleasantness or acceptability - a combination necessary when approaching the general market.
Some of the most successful products have come from research groups outside the universities, where people with special competence in fields like phonetics, linguistics, computational linguistics, terminology, dialogue structure and computer science have cooperated without having to consider loyalties to a tradition, or to a university department or a faculty.
At this conference I represent a company, Nordisk Språkteknologi, now in the process of developing speech technology products involving automatic speech recognition and text-to-speech synthesis. The basic technology we use needs to be 'taught' or 'trained' , and the company needed a group of people who could do field work, collecting speech samples in a standardized and somewhat rigid manner, but still representative of natural speech. Where would we go to find such people? The universities don't produce them, or, if they do, they will be equipped with a lot of irrelevant knowledge - irrelevant to us, but adding to burden our payroll.
Our solution was simple and straightforward: We established a close cooperation with the Labour Market Authorities in Voss, an inland community in Western Norway, and together with the company Voss Informasjonsteknologi we designed a course for a group of highly motivated persons with various types of background. This course combined some basic and advanced computer skills with several hundred hours of teaching in phonetics, linguistics, speech technology, digital recording, speech sample editing, field work techniques, terminology and even a glimpse into computational linguistics.
You will notice that most of the fields mentioned here are taken from the traditions within the Faculty of Arts. We could hardly expect this Faculty to have a course specially designed to meet our needs. But I am convinced that this faculty could do a lot to position their candidates more favourably on the IT labour market than the case is today.
What is needed, above all, is to train the students in working with formal knowledge systems, teaching them to analyse data from various angles and viewpoints, to organize findings according to a number of criteria, and to draw valid conclusions from premises (which may change) according to rules (which may also change).
Where do we find courses within the Faculty of Art emphasizing this?
Perhaps it was not a very wise decision to remove Latin from the entrance requirements. The study of Latin incorporates several of the factors I list. But so does a course in elementary computer science, which I think should be made available as an alternative to the introductory course in phonetics and linguistics at Norwegian universities. I happen to know that such a course has been suggested, but the suggestion was not made a success by the decision makers.
Formal logics is another subject which would, at least among the best students, produce the knowledge, the awareness and the attitudes that I think are needed. And of course - formal logics is no disadvantage in any course of study.
The same can be said about statistics. Studying statistics requires an analytical, flexible and systematic mind, and it is useful as a tool in a number of fields.
In 1959 C. P. Snow published his well-known essay The two Cultures and the Scientific Revolution, where he presented his gloomy vision of modern society as divided between the Humanistic Culture and the Technological/ Scientific Culture. He claimed that communication between the two cultures is rare, sometimes hostile, and often characterized by mutual suspicion and ignorance. His somewhat depressive conclusion was that the humanities provide the tasteful decorum, while the scientific culture gives the premises for the changes we name progress.
Snow dit not foresee the computer revolution; a revolution which, in some respects, has made the gap between the two cultures even wider.
But the development in computer applications in the last decade or so has clearly demonstrated that thorough knowledge in fields traditionally within the Humanities is necessary to bring computer technologies the next step forward. I have used the example of human speech interface, but I do not think this example is unique.
University faculties of Art, traditionally being the organizational 'home' of Humanities, now have a golden opportunity of exerting considerable influence on the next generations of computer technologies. I think this development is far too important to be left to the technological culture alone. Furthermore - I think that by adjusting parts of the teaching within these faculties, we can contribute to the bridging of the gap between the two cultures without having to sacrifice our values within the traditions of the Humanities.