Topics: Interesting Tech, Languages, Translation, Programming

The Register has a fascinating article on how Google Translate service works. The article discusses a couple of ways of doing computer-based translation of human languages: sheer brute-force statistical extrapolation versus "understanding". Google's programmers went the statisitical route.The article also explores the BLEU(Bilingual Evaluation Understudy) metrics.

I particularly liked this comment near the end of the article:

---
So why are computers so much worse at language than at chess? Chess has properties that computers like: a well-defined state and well-defined rules for play. Computers do win at chess, like at calculation, because they are so exact and fussy about rules. Language, on the other hand, needs approximation and inference to extract "meaning" (whatever that is) together from text, context, subject matter, tone, expectations, and so on - and the computer needs yet more approximation to produce a translated version of that meaning with all the right interlocking features. Unlike chess, the game of language is played on the human home-turf of multivariate inference and approximation, so we will continue to beat the machines.
---

Also, chess itself is not evolving, but languages usually do evolve. The rules aren't changing over the decades or developing regionalisms as do human languages. For example, English has regional variations and forms of slang. E.g.; "He was pissed last night" generally means a different thing in the US (angry) than in the UK (drunk). [Some more more examples of UK-US English differences]

At least Google and others seem to be doing better than the computer translations in an old jokes about a computer to translate English into Russian and back. Somebody suggests putting in "Out of sight, out of mind." That quote is keyed in and out comes something in Russian. The Russian text is put though for a translation back into English. The output is "Invisible insanity".

Still computer tools can be helps for people with some familiarity with the languages involved.

J.D. Abolins