Computer translation: What if you could talk to anyone in the world in their own language?

A couple of weeks back while surfing the Internet (yes, the 20th century has come to our household) I found that you could get it to translate from one language to another. If you use the AltaVista search engine to find a document you will see the word "translate". Hit it and it will give you the thing in the foreign language of your choice. That is not much use for Anglophones as most of the pages on the Internet are already in English, but I saw some stuff in Spanish about a speech I had made in Buenos Aires. I wondered what they were saying about me and hit the button.

I discovered that "in 1989 it entered to The Independent, where at the moment associate evolves like publisher. Its work as journalist in these means provided ..." and so on. You can see what they were saying, but the translation is less than elegant. It is, however, better than nothing, and at the moment it is free.

What is really exciting here is not just that these systems enable non- English speakers to extract information from the Internet, nor that computer translation, for all its weaknesses, is now becoming a practical tool. It is that all the elements are there, in prototype form, for simultaneous translation of speech.

Voice recognition? It is there already and getting better all the time. Voice synthesis? Works fine. Translation? Coming along. In, say, 10 years' time it is quite plausible that we will be able to speak into a telephone in one language and have the words come out at the other end in another. Computer translation will do for words what the electronic calculator has done for mental arithmetic.

Well, not quite. Anyone who enthuses about new technologies has to be aware of the false promises of technology in the past. Translation is very difficult for a number of reasons. Computers find it very hard to cope with homonyms - words which sound the same but have different meanings according to the context. They find it hard to cope with languages where the structure and grammar is completely different: for example languages (such as English) where the position of the word in the sentence, as opposed to the ending of the word, is crucial to understanding who is doing what to whom. And they are totally baffled by subtleties such as irony.

But it will all get better and I think we will get to the stage in the next decade where provided the language is kept simple, computers will routinely translate phone conversations as well as documents. We may even be able to buy pocket synthesisers that translate our words into the chosen foreign language, and then translate the reply.

If this is right, it is of seismic importance. At the moment the world is moving quite rapidly to a single common standard. English is becoming the "Windows" of the verbal world. It is already spoken by as many non- native speakers as native. And the non-natives are developing new, simpler and more precise forms of the language than the versions we use. These versions are less rich, for they use fewer words and less complicated structures. But they are more effective tools for getting basic information across.

Go to an international business conference and if a German or Swede is using English most of the audience will listen happily without using the simultaneous translation facilities. When a Brit or an American starts to speak a lot of them reach for their headphones. Reason? We tend to speak faster and use less formal sentence structure than people who learn English as a second language. We also use colloquialisms: I still wonder how the translators coped a couple of weeks ago when Tony Blair was in Tokyo and urged the Japanese to "go the full Monty" with their social and financial reforms.

Over the next 10 years these new versions of English will spread rapidly, but they will be challenged by ever more competent computer translation. English as a medium for originating documents will become more and more important - it has just overtaken French as the most popular medium in the EU - but quite a lot of people may not bother to learn English in order to read them.

Will it be the English language or computer translating that wins the race? Maybe neither. Maybe people will both learn English and use computer translation.

The growing international use of English, including computer-translated English, will underpin another key change which is starting to take place: the boom in cross-border trade in services. We are accustomed to buying goods from abroad, for these goods are all built to a common standard. Cars all have their controls in the same position; the Internet has its own single global computer language. But most services still have large cultural and language differences.

For example, we would find it hard to imagine buying an insurance policy in German, or going to see a doctor who spoke only Italian. (It is bad enough trying to learn to ski in French.) The technology of the Internet, however, allows global distribution of many services: often the easiest way to buy a book is to get Amazon.com to send it, rather than having to trek down to a local bookstore and get them to order it for you - even though Amazon are operating out of Seattle.

International trade in services is growing fast, but it is still smaller than trade in goods. The parallel development of a single language standard, English, and an easy way to move in or out of that standard (computer translation) is the powerful motor that will drive the next stage of globalisation.

The future of computer translation is inextricably bound up with the future of English as a global language. At one extreme it is conceivable that computer translation might prove a bit of a blind alley, for even if it does improve radically, it will not be needed as most people who use language internationally will already speak English. At the other extreme it could conceivably make a global language unnecessary.

In reality I think its importance will turn out to be somewhere in between the two. It will buttress English as the global language by allowing people who do not speak it to use it. For others, computer translating will be used as a cross-check, rather in the way those of us with school French use sub-titles when watching a French movie, getting some of our understanding from reading the translation and some from listening to the original language. It will be used by Anglophones to access material not yet available in English; and by people preparing material in minority languages to make their material available to the rest of the world.

Ultimately it will also be a democratising force in the sense that computers give ordinary people the access that hitherto was only available to the very rich or the very privileged. Take information. The Internet increasingly gives access to information that would previously be available only to a company with a large research department.

Now the same is happening with language. Go to an international conference and immensely competent translators will see that everyone understands what is going on. Watch Tony Blair in Japan on TV and you will see a translator in the background flipping what he is saying backwards and forwards, to and from his Japanese counterpart. Soon something close to this quality of linguistic competence will be available, at the push of a button, to us all.