ศาลาชุมนุม

[Glenn Slayden]

I am excited to announce the availability of Thai translation at http://www.microsofttranslator.com, which just went live within the last hour.

This is the first release of the Bing Translator which adds Thai support, and I'm excited to have worked on the project all summer. The Machine Translation group at Microsoft Research, which is responsible for Bing Translator, was a great team of creative and smart linguists, programmers, testers, and managers, and it was an absolute pleasure to work with them.

One of the exciting features of Bing Translator is the Bilingual Viewer, which lets you see the results of a translated web page, side-by-side. Here's a sample.

Please be aware that statistical machine translation is an area of much research and the results for languages like Thai suffer from a shortage of training sources. We are far from the point where machine translation can replace the services of a professional human translator. Nevertheless, this new release is a milestone for Thai on the web and I hope everyone will find it useful when an approximate translation is all that's needed.

I'll be adding direct access to the Bing Translator engine from thai-language.com fairly soon.

[David and Bui]

The first line from Ajarn Nithi Eewsriwong's article in this week's Matichon Weekly:

ใครๆ ก็รู้ว่าคนไทยไม่อ่านหนังสือ

The Bing translation?

"Anybody know that Thai people don't read books"

I would say that this is a very good rendering, save for the indefinite pronoun. A little more grammar and this would be perfect!

[Glenn Slayden]

ใครๆ ก็รู้ว่าคนไทยไม่อ่านหนังสือ

For those who don't read Thai, an ideal translation is, "Everybody knows that Thai people don't read books."

[David and Bui]

I noticed that Google Translate offers the readers a chance to correct any mechanical translation. Does Bing offer a similar facility, and, if so, does the correcting go into expanding the "corpus"? In other words, do the translation facilities "learn"?

[Glenn Slayden]

The short answer is no. I'm limited in what I can mention, but suffice it to say that there will be an announcement at the Professional Developer's Conference PDC09 on November 17th.

[gerrygantt]

Glenn,
I checked the translation you provide vs those by Google and Bing. Yours is still far and away the best of the three. The results from Bing and Google can be quite hilarious.
Gerry

[Glenn Slayden]

Thanks, but technically speaking, the "bulk lookup" feature here at thai-language.com does not claim to offer "translation," since it does not attempt to process syntax (grammar). Because English and Thai both use subject-verb-object (SVO) word order*, something as crude as word-by-word lookup can sometimes appear to give the illusion of translation.

The statistical machine translation (SMT) systems offered by Microsoft and Google are attempting to emit syntactically correct sentences in the target language, although sometimes with mixed success.

My long-term goal for this website is to develop an analytical machine translation facility for Thai, probably within the HPSG linguistic framework. The English-processing half of this project would be provided by the English Resource Grammar. If I continue with this plan, it is a gigantic task which will take many years of work.


*However, note that in English adjectives come before the noun they modify, which is not the case in Thai

[stitch]

An amazing tool thanks for bringing it to light.

[Glenn Slayden]

An amazing tool thanks for bringing it to light.

Not sure if you're referring to Bing Translator or the English Resource Grammar online demo. Surely both are amazing although their approaches couldn't be more different. In any case, I take no credit for the latter, and my role in the former, compared to its scope, was miniscule. However, I suppose it's also possible that you were referring to Bulk Lookup, in which case, thanks!

[trisswines]

I checked the translation you provide vs those by Google and Bing. Yours is still far and away the best of the. The results from Bing and Google can be hilarious.