Minimal

Computer-based translation — an impossible dream?

Getting lost in the translation [BBC News]

“It would be nice if computers could do the job. And certainly the quest for machine translation has prompted a lot of linguistic research that may prove valuable in unforeseen ways. But experience to date confirms that even the most subtle computer program doesn’t think — and you need to be able to think in order to translate.”

What a load of tripe. At what point does a computer algorithm not qualify as “thinking”? To me, there is no cut-off. An algorithm is thinking. Algorithms can vary in complexity as much as thoughts can, so what is there to distinguish them as being different to thought processes? Nothing.

To get back to the quote, it seems that the quotee (a translation agency, of course) simply hasn’t had exposure to some of the best translation programs out there. Even back in 1995, I saw a program at university that was streets ahead of Altavista’s Babelfish (which is, no doubt, the translation service referred to by the BBC article). It worked by actually understanding the original text and then translating it.

Needless to say, anything of this complexity isn’t the sort of thing that’s going to be made available to the public for free. There’s a lot of money to be made here. Indeed, the system I saw is no longer in academic hands, it since having become the basis of a business venture. As for when we will see such powerful translation in the public domain, your guess is as good as mine. Maybe we never will. It might just be too lucrative to give away for free. Regardless, it would be a mistake to think that computers aren’t up to the job.