A Woman Learning Thai...and some men too ;)

Learn Thai Language & Thai Culture

2010: The Second Google Translate Challenge

Google Translate

Google Translate, the challenge…

Over a year ago, Google Translate launched, and the language sites started buzzing at the possibilities. My contributions were Thai Translation: Google Translation & Thai Dictionaries and Google Translates Documents and Email Too and Thai Google Translate: Will Crowdsourcing Work.

During it all, Jeff Gray mentioned the below:

…the sheer volume of people adding useful stuff to Wikipedia makes it immensely useful. Wikipedia is also inherently self correcting, because if someone writes rubbish, it will be corrected by others.

In the same way, having millions of users making minor improvements to the translation system does something that any single company could never do with internal resources only. It might be chaotic, but the sheer scale is unbeatable.

Time will tell, but the approach they’ve taken is potentially a very effective one.

Taking it from there, on the 23th of July, 2009, I set out my Google Challenge. You can read more about it here: The Google Translate Challenge.

Google 2009 – 2010…

What you see below are are comparisons between Google’s translations from last year and this year. As you can see, there is a difference.

Be careful! There is swine flu!
2009: ระวัง! มีไข้สุกร!
2010: โปรดระวัง มีสุกรไข้หวัดใหญ่เป็น!

I have swine flu already, thanks!
2009: ฉันมีสุกรไข้หวัดใหญ่แล้วขอบคุณ!
2010: ฉันมีไข้หวัดหมูแล้ว, thanks!

I don’t know! Why do you ask?
2009: ฉันไม่ทราบ!ทำไมเจ้าถาม?
2010: ไม่ทราบ! ทำไมคุณถาม?

Did you eat yet?
2009: คุณกินยัง?
2010: คุณไม่กินหรือยัง

Oh no! You’re a liar!
2009: แย่ละ!คุณเป็นคนพูดเท็จ!
2010: Oh no! คุณโกหก!

I don’t want to see your face again.
2009: ฉันไม่ต้องการดูหน้าของคุณอีกครั้ง
2010: ฉันไม่อยากเห็นหน้าคุณอีกครั้ง

He is busy lighting a mosquito coil.
2009: พระองค์คือยุ่งแสงสว่างที่ยุงม้วน
2010: เขาเป็นไฟม้วนยุ่งยุง

Don’t put any fish sauce on the rice. It stinks!
2009: โปรดอย่าวางใดน้ำปลาใน ข้าว. มัน stinks!
2010: ไม่ใส่น้ำปลาบนข้าว It stinks!

The first Google Challenge control group…

I ran these sentences through Google Translate in 2009 and 2010, but I didn’t post them until 2010.

He tells me that he loves me with all his heart.
2009: เขาบอกผมว่าเขารักฉันกับหัวใจของเขาทั้งหมด
2010: เขาบอกว่าเขารักฉันด้วยหัวใจทั้งหมดของเขา

Do you speak English?
2009: คุณพูดภาษาอังกฤษ?
2010: คุณพูดภาษาอังกฤษ

What did the nurse say?
2009: อะไรได้พยาบาลกล่าว?
2010: พยาบาลพูดว่าอะไร?

That water buffalo meat comes from the north.
2009: นั่นควายเนื้อมาจากทางเหนือ
2010: ที่เนื้อควายมาจากภาคเหนือ

Please give me a glass of orange juice.
2009: กรุณาให้ฉันหนึ่งแก้วน้ำส้ม
2010: กรุณาให้แก้วน้ำสีส้ม

The turtle reaches the finish line before the rabbit.
2009: เต่าที่ครบตามเส้นชัยก่อนกระต่าย
2010: เต่าถึงเส้นชัยก่อนกระต่าย

Because last night you snored.
2010: เพราะ(ว่า)เมื่อคืนคุณกรน

The Google Translate Challenge summary…

So, what do you think? Has Google improved in translating Thai? Gotten worst? Or are the results pretty much the same?

Share Button
The following two tabs change content below.
My passion is promoting the Thai language. Fullstop. Oh, and traveling. I'm passionate about that as well. And photography too.

15 Comments

  1. Google translate is worthless right now. My last translation talked about a blunt, a flute and a pee princess. Thai2English is still by far the best translator out there.

  2. Dan, Agreed. GT doesn’t give an option, it selects one word from the available possibilites. In T2E and thai-language.com, you decide.

  3. The Babelfish still feels a long way off. However, it looks like improvements have been made. Some of the syntax seems to have been cleaned up (the nurse sentence for example), the algorithm for personal pronouns might be improving (I don’t see the mistake in the following sentence repeated in 2010 เขาบอกผมว่าเขารักฉัน, though to be fair in that example the initial pronoun has been omitted), and they need to work at replacing a question mark with ไหม (where appropriate of course). I’m sure it’ll happen in time. Interesting experiment.

  4. The question marks should be one of the easiest to correct, yes?

    It is interesting, so I’ll check back again next year. While we are waiting, I’ll create another set of sentences for behind the scenes as well.

  5. The problem with Google translate is that it is open to abuse. Like, why on earth is Samut Prakan always translated as Bangkok in articles?!

  6. Richard, Wikipedia is also a mess of disinformation (but I will quote from wikipedia if I find the info elsewhere as well).

    The potential is there, GT just needs to put knowledgeable editors in place. For instance, they could start with the top phrases coming through Google Translate, and then lock in the correct (approved) translations.

    I personally would be interested to know what the top search phrases are for Thai. Is there a way to find out?

  7. Very fun..and funny, especially, “He is busy lighting a mosquito coil”. Sounds like a very busy coil… :D :D

  8. It does sound odd :-) I wanted to pick sentences that wouldn’t be in the mainstream (mostly). And the next ones will be equally odd (but I haven’t made up my mind yet).

  9. I don’t know enough Thai to comment on how much it has improved. I am encouraged by AjarnPasa’s comment that it is a little better.

    As others have noted, the lack of context means it will never be great. But at least it helps when there is no other option. I tend to use thai-language.com to read complex Thai. As you say, it allows the human to make judgments about context.

    I have used GT to have a conversation with a Thai by email. We can use this successfully. The key is to keep sentences short and think carefully about word choice. So it is workable within this limited context. But yes, there is a long way to go.

    I have found one situation where it is very useful. I have shown Thai living in Australia how to use it. If they have poor English skills, this allows them to browse local Web sites & understand them. Yes, the words are sometimes wrong or confusing. But it allows people to functionally use the Web. Because Web sites often have simpler text than books or letters, this is a better fit. I have used this to browse Thai Web sites. Sometimes it is very good, sometimes rubbish. But at least I can do something where I otherwise could not.

    A recent innovation is Google Goggles, an application for Android mobile phones. It allows you to take photos of DVDs, books, artworks, etc, which it will search for on Google. This is nice, but the thing I wanted to mention is it can be used for translation. You take a photo of text in a foreign language, such as a menu. This is then translated online into a language you can read on your phone. It is early days for this sort of thing, but it is an example where the technology can be useful.

    Speech to text and vice-versa is the next step for live translated conversations. And probably confusing people on your travels ;-)

    A long way to go, but an intriguing possibility. But some things cannot be automated. For example, they need to add the ability to say whether the speaker is male or female, and whether the listener is higher or lower social status. I suspect using chan for “I” is inappropriate for me :-)

  10. Hi Jeff, many years ago I used babelfish for French (what a hoot) but I haven’t tried out Google Translate with Thai friends to see what they get out of it. It’s a great idea – thanks!

    I’d forgotten all about Google Goggles. I just did a search and it’s coming for the iPhone soon. I collect Thai signs (some are quite fun to read) so this will be a welcome addition for newcomers to Thai. I’ll write a post on WLT as soon as GG hits the market so another thanks for reminding me.

    Thai – English speech to text… that would be a help for the Thai tourism industry. And as you mentioned, for the first few years, total confusion!

    Btw – some Thai guys do indeed use chan :-)

  11. Catherine, first I’d like to congratulate the turtle on retaining his sprint title. An amazing achievement against a rabbit who is no slouch by any means. Perhaps this will give him the confidence to come out of his shell more often. I also think the rabbit would be better suited to the triple jump.

    Google translate….I do have this on my Google Chrome start up page but I rarely use it as I find it is way beyond my level of Thai. I have Google Weather installed as well and according to it my hometown has had rain everyday for the past month and I will get soaked for the next three. I have developed quite a tan over the past thirty days but I’d still put more faith in their weather forecasts than their language translation…..and now you inform me Wikipedia is at the heart of it all….aaargh. Sorry about that, I’ve just stubbed my sunburned toe on my table leg.

    Like yourself, I try to check up on any Wikipedia facts but it is a useful tool for photos. My mind boggles at Google Goggles and the kind of mess it might get someone into.

    Google Translate is not for me but I can understand its amusement to advanced Thai language learners.

  12. Catherine I forgot to add this to my comment….” />

    Now I feel better, my site is not the only one with it on. Thanks for your help but I still can’t get the bl**dy thing off.

  13. Martyn, it’s not that Wikipedia is at the heart of it. There is no relationship between Google and Wikipedia. It was just an analogy for a means of gathering lots of small contributions into a larger whole.

    Yes, Goggles could get someone into a mess. But if they have little or no Thai skills, then at least something is better than nothing.

    One thing to bear in mind about this is that the Thai support has only been in Google Translate since Jan 30, 2009. It really is very new. Some other languages have been supported since 2006 and are now working much better.

    The key with statistical translation is to get a very large volume of comparative text. This is much easier with European languages where the amount of translated work is much higher.

    I’m encouraged to see there has been some improvement in the past year. We should see it get slowly better over the next few years.

  14. Jeff, thanks for putting me right. I read Catherine’s post and misinterpreted the piece about Wikipedia. I hope you’re not going to tell me the turtle didn’t really win the race as well. That would upset me. It’s rare such a prolific swimmer excels in track and field events but I hung onto everyone of Catherine’s words.

    Seriously, Google Translate is a great idea but it is fairly useless to someone like me who has little Thai language skills. The inclusion of transliteration would transform the tool ten fold in my case and for others at a similar level.

    Google must have read my comment because their forecast for my hometown on Thursday shows a big bright yellow ball against a blue background. I guess it’s going to rain real hard that day.

  15. Sorry for my late reply. This is a four day weekend so I took off on a field trip for one. And let me tell you, the best ever time to go looking for rocks in Thailand is in the rainy season when the brown is nicely covered with green.

    Thanks for explaining Jeff. When I was writing the post, I worried about not including enough of the crowd sourcing conversation, so you’ve saved the show.

    I’ve been thinking about inputting phrases to speed up the process for Google Translate Thai. With all of the Thai/ English twittering going on (Ajarn Pasa, Kaewmala, Rikker, Andrew Biggs, etc), it wouldn’t be that difficult to get into the habit of dropping the phrases in.

    Martyn, fabulous to see that you’ve solved the double ” /> on your site. And fingers crossed you don’t get too much rain… or snow…

Leave a Reply

Your email address will not be published.

*