According to an article in the M&G, Google are on the verge of a major breakthrough in instant language translation.
Machine translation is not new. Ever since the early days of Babel Fish, web surfers have been able to sample foreign language text. The translations were never that good though, and there seemed to be minimal progress over the past decade or so.
Traditionally, machine translation systems have worked based on grammar rules and dictionaries programmed in by linguisitic experts and programmers.
Now, Google researchers have taken a new approach. They feed in documents that have already been translated, using an approach called statistical machine translation. This method allows the machine to identify and replicate patterns, leading to a far more accurate translation. The more documents that get added, the larger the statistical sample, the better the results get.
Make no mistake. Human translators are not suddenly going to find themselves out of work…yet. Machine translation still has a long way to go but at least it’s going forward.
For many practical purposes, a rough translation is good enough. If it can take me from “no bloody idea what they’re on about” to “hhhmmm, looks like they’re saying something about cabbages” – then I’m better off than I was before. Assuming of course, I cared about cabbages. And assuming the page really was about cabbages.
Thinking ahead, to a point in the future where Google have taken the system even further. Imagine the value of a search engine that will index articles in multiple languages and serve results seamlessly to users in the language of their choice, regardless of the origins. How cool is that?
Muti
You can learn more about the movement 
