×

Welcome to the Slashdot Beta site -- learn more here. Use the link in the footer or click here to return to the Classic version of Slashdot.

Thank you!

Before you choose to head back to the Classic look of the site, we'd appreciate it if you share your thoughts on the Beta; your feedback is what drives our ongoing development.

Beta is different and we value you taking the time to try it out. Please take a look at the changes we've made in Beta and  learn more about it. Thanks for reading, and for making the site better!

Google's Computing Power Refines Translation

kdawson posted more than 4 years ago | from the throwing-silicon-at-it dept.

Google 142

gollum123 sends an excerpt from the NY Times on how Google has taken a lead in language translation, in one of the company's few unqualified successes as it attempts to broaden its offerings beyond search. "...Google's quick rise to the top echelons of the translation business is a reminder of what can happen when Google unleashes its brute-force computing power on complex problems. The network of data centers that it built for Web searches may now be, when lashed together, the world's largest computer. Google is using that machine to push the limits on translation technology. Last month, for example, it said it was working to combine its translation tool with image analysis, allowing a person to, say, take a cellphone photo of a menu in German and get an instant English translation. ...in the mid-1990s, researchers began favoring a so-called statistical approach. They found that if they fed the computer thousands or millions of passages and their human-generated translations, it could learn to make accurate guesses about how to translate new texts. It turns out that this technique, which requires huge amounts of data and lots of computing horsepower, is right up Google's alley. ...Google's service is good enough to convey the essence of a news article, and it has become a quick source for translations for millions of people."

cancel ×
This is a preview of your comment

No Comment Title Entered

Anonymous Coward 1 minute ago

No Comment Entered

142 comments

mah piss is frosty (-1, Offtopic)

Anonymous Coward | more than 4 years ago | (#31421132)

Yup

Converting that article from English to Chinese to (5, Interesting)

Rei (128717) | more than 4 years ago | (#31421150)

English, with Google Translate:

---
Google's rapid rise to the translation of business executives is a result of what Google released a complex problem, and its powerful computing power for reminding me. The data center, and its Web search, it may be now, when attacked with the network, is the world's largest computer. Google's machine translation technology is being used to push forward the limit. Last month, for example, it indicated that it was a combination of image analysis of the translation tools to enable a person, says that while walking in the German mobile phone menu, photos and immediately the English translation. ... In the mid-90s, researchers began to favor a so-called statistical methods. They found that if they ate the computer or hundreds of thousands of millions of paragraphs and the translation of humans, it can learn how to make an accurate translation of the new text of speculation. Facts have proved that this technology requires large amounts of data and a lot of computing power, is the right of Google's alley. ... Google's service is sufficient to convey the essence of news articles, it has become a quick translation of millions of people everywhere.
---

Okay, perhaps not spectacular... but compared to Babelfish:

--- ...Is anything the prompt possible to occur to the translation business's crown trapezoid's Google quick rise, when Google unties it when the complex question violence computing power. Perhaps the data central network it for the net search establishment now is, when attacks together, world large-scale computer. Google uses that machine to push in the translation technology limit. The previous month, for example, it said that it operates and the image analysis unifies its translation tool, allows the human to adopt a menu the handset picture and obtains one with German immediately English translation. ... in the mid-1990s, researcher started to favor the so-called statistical method. They have discovered that if they have fed the translation which the computer thousands or the tens of thousands of paragraphs and their person cause, its possibly academic society does about what kind of guesses translator accurately the new text. _ it this technology, requests the huge large amount data finally and completely the calculated horsepower, is correct Google the alley. ... The Google service is enough good expresses the news article the essence, and it has become translation quick origin tens of thousands of people
---

Re:Converting that article from English to Chinese (5, Insightful)

Daniel Dvorkin (106857) | more than 4 years ago | (#31421212)

Yeah, that's actually a pretty good test. Google's version is odd but comprehensible, while Babelfish's is a bunch of ... well ... babble.

Re:Converting that article from English to Chinese (2, Interesting)

Cassius Corodes (1084513) | more than 4 years ago | (#31421236)

What's interesting is that there are a couple of sentences where babelfish is actually better than google and the rest is way off.

Re:Converting that article from English to Chinese (3, Insightful)

spazdor (902907) | more than 4 years ago | (#31421642)

This doesn't actually mean the translation is any better: all it means is that the Chinese generated by Babelfish is more easily translated back to english, perhaps because it makes even less sense in Chinese. A translation function could be conceived which is a strict, reversible bijection, so that playing this translation game would give you your original English back, word-for-word. Doesn't guarantee that the intermediate Chinese step is in any way comprehensible.

Re:Converting that article from English to Chinese (3, Insightful)

Jurily (900488) | more than 4 years ago | (#31422348)

A translation function could be conceived which is a strict, reversible bijection, so that playing this translation game would give you your original English back, word-for-word.

That's the main problem with translations: they're not strict, and sometimes not even reversible. In every language there are common phrases which make perfect sense to someone thinking in the language, but are untranslatable to the point where you as a translator just rephrase the whole sentence (example: "is right up Google's alley"). Then, if you get another translator to translate it back to the original language, you sure as hell won't get the original phrase back (assuming both translations are perfect in terms of understandability and conveying the message).

Then you have words that don't exist in the target language, like "brute-force" or "computing horsepower", or even concepts that don't exist.

I think the fact that we can understand machine translations is more a tribute to the error correction mechanisms in our brain than anything else.

Re:Converting that article from English to Chinese (1)

FatdogHaiku (978357) | more than 4 years ago | (#31422646)

..In every language there are common phrases which make perfect sense to someone thinking in the language, but are untranslatable to the point where you as a translator just rephrase the whole sentence (example: "is right up Google's alley"). Then, if you get another translator to translate it back to the original language, you sure as hell won't get the original phrase back (assuming both translations are perfect in terms of understandability and conveying the message).

Then you have words that don't exist in the target language, like "brute-force" or "computing horsepower", or even concepts that don't exist.

I think the fact that we can understand machine translations is more a tribute to the error correction mechanisms in our brain than anything else.

Awl hour translate spume waffle. Ewe no other gender knot exchangeable!

Re:Converting that article from English to Chinese (0, Redundant)

jandoedel (1149947) | more than 4 years ago | (#31421252)

Well d'uh... that's why it is called babble-fish!

Re:Converting that article from English to Chinese (0)

Anonymous Coward | more than 4 years ago | (#31421796)

And that's why they called it Babelfish.

Re:Converting that article from English to Chinese (2, Interesting)

timeOday (582209) | more than 4 years ago | (#31421266)

I would call it a very rigorous test, since you can get by in a foreign country with far, far less expressiveness than it takes to read a news article. ("Where's the toilet?" "How much for this?" Or for DoD applications, "Stop or we'll shoot!")

Plus, round-trip translation at least doubles the error compared to an actual application which would involve one-way translation (and probably more, since the "return-trip" translation is starting with a poor quality input). A much more fair test would be comparing a one-way translation, man vs. machine.

Re:Converting that article from English to Chinese (4, Interesting)

timeOday (582209) | more than 4 years ago | (#31421446)

OK, here is something better than a round-trip translation test.

Der Spiegel offers version of some of its stories in English. They aren't direct translations, but quite similar.

Here's part of a story published in english [spiegel.de] :

Those wanting to own a McDonald's or Subway franchise in Germany must be prepared to offer up intimate personal details, including health information. One German official says the questionnaires violate the law. ...

According to information obtained by SPIEGEL, those wanting to partner with the fast-food chain Subway must agree to a background check "in accordance with anti-terror legislation" such as the US Patriot Act.

The report must also include information about the applicant's character, lifestyle and relationships. Future franchise owners are also asked whether they have ever been part of a terrorist organization.

And the same story, published in German, [spiegel.de] translated to English by google:

McDonald's and Subway asking intimate data from franchisees

From its franchisees in Germany require the American fast food McDonald's and Subway deep insights into the intimate and the political convictions. Who wants to be partner of Subway, for example, must create an audit report in accordance with the anti-terror laws "such as the USA Patriot Act to agree." This report will contain information about "character", "lifestyle" and "relationships". The applicant shall provide information, even if she "ever directly or indirectly involved in terrorist activities were"

And babelfish translation of the same story:

McDonald' s and Subway demand most intimate data of franchise takers

Of their Franchise takers in Germany the American high-speed restaurant chains McDonald' require; s and Subway deep views of the privacy and the political convicition. Who for example partner of Subway would like to become, must the production of a test report " in agreement with the anti- terror Gesetzen" as " The USA patriot Act" agree. This report is information over " Charakter" , " Lebensweise" and " Beziehungen" contained. The applicants have to give even information whether them " ever at activities of terror beteiligt" directly or indirectly; were.

I do think the google version is significantly better.

Re:Converting that article from English to Chinese (1)

rolfwind (528248) | more than 4 years ago | (#31422330)

Yeah, that's actually a pretty good test.

I don't think so. Things get lost in translation with humans already. There are phrases I simply can't translate in languages I'm fluent in, idioms and the like. And when humans pass along information, it also gets distorted, simplified, and the like - language is a vague, flexible thing. So we're trying to give the machine a test impossible to pass, a Turing test where most of us don't even have any real experience how well a human would do it as a frame of reference.

It would be better just to translate many pieces one time, both ways, and have a fluent bilingual judge the quality. Although, I agree, checking the Chinese/Japanese to English capability is a good test.

My personal test was to take reviews off of amazon.co.jp and translate them and see how the translator fared. Babelfish is indeed a bunch of babble, while Google's translation is far from perfect (or even all that good), it's obviously better.

Re:Converting that article from English to Chinese (1)

wye43 (769759) | more than 4 years ago | (#31424184)

Its an easy, or perhaps entertaining test, not good test.

Re:Converting that article from English to Chinese (0)

Anonymous Coward | more than 4 years ago | (#31421320)

Perhaps the data central network it for the net search establishment now is, when attacks together, world large-scale computer.

Is that thing writing a science fiction novel in it's spare time or something?

I like how it's rooted out Google as the "net search establishment".

Re:Converting that article from English to Chinese (4, Insightful)

RavenousBlack (1003258) | more than 4 years ago | (#31421684)

Not to disagree with the results of your test, but I think a better test would be actual translations from authentic Chinese text to English. Going from English to Chinese to English is like taking an English interpretation of what the Chinese are trying to interpret from what someone was saying authentically in English instead of just interpreting into English what someone was authentically saying in Chinese.

Re:Converting that article from English to Chinese (0)

Anonymous Coward | more than 4 years ago | (#31422426)

I often find the English to/from Chinese translations is usually better than English to/from Japanese. Chinese characters have concise meanings vs. Japanese's character set which are based on sounds.

Re:Converting that article from English to Chinese (2, Informative)

GF678 (1453005) | more than 4 years ago | (#31422476)

Going from English to Chinese to English is like taking an English interpretation of what the Chinese are trying to interpret from what someone was saying authentically in English instead of just interpreting into English what someone was authentically saying in Chinese.

Exhibit A: http://winterson.com/2005/06/episode-iii-backstroke-of-west.html [winterson.com]

Re:Converting that article from English to Chinese (1)

Rocketship Underpant (804162) | more than 4 years ago | (#31424502)

I recently did an evaluation for a translation agency on the state of current machine translation services. Since I translate Japanese to English for a living, that was the pair I was testing.

Long story short, of the five services I tried that do Japanese-English MT, Google came out the worst. Yes, the worst. Mind you, all of them were terrible. None could produce grammatical English sentences, and most couldn't even translate basic things like Japanese dates properly.

Re:Converting that article from English to Chinese (2, Interesting)

uglyMood (322284) | more than 4 years ago | (#31421764)

In Philip K. Dick's obscure 1969 novel Galactic Pot-Healer [wikipedia.org] , the characters play a game based on this very idea. They take common sayings and figures of speech, and feed them through several language-translation computers. The results are then sent to a friend, who attempts to figure out what the original phrase was.

Sometimes when you're reading PKD you get the uncomfortable feeling he really could see into the future.

Re:Converting that article from English to Chinese (0)

Anonymous Coward | more than 4 years ago | (#31422916)

I think this should become a new internet meme. Everytime anyone says anything about a new technology, just post a response that says "that reminds me of what Phillip K. Dick wrote about in his obscure short story / novel [madeUpName]"

Re:Converting that article from English to Chinese (1)

Hipcatjack (1573969) | more than 4 years ago | (#31423436)

I am definitely down with that meme. Ironically on a totally unrelated story, i just borrowed a book from my friend by P.K.D. talking about obscure authors suddenly finding themselves idolized superstars on par with the biggest Sports players or Rockstars. and their lives dealing with papparozzi - huh

Re:Converting that article from English to Chinese (1)

Hipcatjack (1573969) | more than 4 years ago | (#31423484)

Google translate English>traditional Chinese>simplified Chinese> Italian >back to English from what i copied from my above post: I believe that with the Mei-Mei. Ironically, the story is completely independent, I just borrow a book from my friend from polycystic kidney disease, clear speech, suddenly found itself the greatest athletes in the worship of the stars or Rockstars. Their lives and to deal with mosquitoes - ah

Re:Converting that article from English to Chinese (1)

glwtta (532858) | more than 4 years ago | (#31421872)

"I had a small house of brokerage on Wall Street... many days no business come to my hut, but Jimmy has fear? A thousand times no. I never doubted myself for a minute for I knew that my monkey strong bowels were girded with strength like the loins of a dragon ribboned with fat and the opulence of buffalo dung."

Re:Converting that article from English to Chinese (1)

biryokumaru (822262) | more than 4 years ago | (#31422862)

Soon the super karate monkey death car would park in my space... but Jimmy has fancy plans, and pants to match!

Feel my scales donkey donkey donkey donkey donkey.

Also...

I stole a car! I mean, a sycamore tree...

Re:Converting that article from English to Chinese (1)

wtbname (926051) | more than 4 years ago | (#31421986)

They just need to do what video card manufacturers do to thwart your little test Mr. Man. Cheat in the translation code to recognize your test, and just regurgitate your original text.

Then how would you choose the best translation software to buy???? Oh... it's free?

Re:Converting that article from English to Chinese (1)

Ihmhi (1206036) | more than 4 years ago | (#31422316)

That's a whole lot better than it was a few years ago.

They still need to work on their Japanese a good bit, though. Translating my first sentence from English to Japanese to English spit out:

This is the way it is much more than a few years ago the entire

.

I believe they are getting very strong on the vocabulary and context clues bit but having a difficult time translating between different Subject-Object-Verb formats.

Re:Converting that article from English to Chinese (1)

AniVisual (1373773) | more than 4 years ago | (#31422712)

Japanese is very unique in that it leaves out the subject, and sometimes object of a clause when the meaning is understood in context. This is, however, very frustrating for machine translators. In addition, Japanese has a topic for its sentences, which function is very ambiguous in an English language.

Re:Converting that article from English to Chinese (0)

Anonymous Coward | more than 4 years ago | (#31423058)

this is what is dimensional analysis for physic:
given the maximum password length, determine the computing power of the provider:
example gmail.com, hotmail.com and others
tips: hashtables

Not from NY Times (3, Informative)

Anonymous Coward | more than 4 years ago | (#31421202)

Last week's The Economist adressed this issue (http://www.economist.com/specialreports/displaystory.cfm?story_id=15557431). NY Times recycled it

Try using google voice transcription (1)

colin_faber (1083673) | more than 4 years ago | (#31421216)

Voice to text attempt 1: "What is. Thank you. Hey Faber what I AM slot. People just want to let you know like Hello Colin, this is already the decision. I think it's going to ask." Voice to text attempt 2: " Hi. This is the level Dell Computers, I'm doing a follow, or on the error basement far start up top. If that happens. I still have the problem in 16 Keith dispatch number and I gave you so that into at that back and call us back and we could double shifts order with the problem. Thank you." The first one was silence that got recorded by accident, the second was from our favorite Indian's over at Dell computer, calling to pester me about how my repairs are going. =)

Re:Try using google voice transcription (2, Insightful)

amRadioHed (463061) | more than 4 years ago | (#31421576)

Yeah, google voice is fun, it's what you get when you combine voicemail and mad-libs.

Similar languages (2, Informative)

Jurily (900488) | more than 4 years ago | (#31421220)

Sure, you might get something decent if you try to translate from English to German, but what about languages with entirely different thought models behind them, like Chinese or Hungarian? Last time I tried using it, it confused "has been" with "Latvian".

Re:Similar languages (5, Funny)

hardburn (141468) | more than 4 years ago | (#31421262)

I've worked on payment processing for web sites in Korea before. The translations of error messages we get from the system, then passed through Google translate, are exactly as good as the translations we get back from a human translator. That is, not useful at all.

Re:Similar languages (4, Interesting)

MBCook (132727) | more than 4 years ago | (#31421430)

This seems like the ideal opportunity to mention Translation Party [translationparty.com] . You give it English, and it translates it to and back from Japanese until the input and output English are the same.

It can be a ton of fun.

Re:Similar languages (2, Interesting)

hardburn (141468) | more than 4 years ago | (#31422176)

That is fun. Your sig breaks it [translationparty.com] .

Re:Similar languages (1)

chadenright (1344231) | more than 4 years ago | (#31422948)

It worked on firefox for me, try turning off flash block.
However, it thinks "Who were you before you were who you are" has the same meaning as "Many people before the eyes of many people?"
I think I broke it.
http://translationparty.com/#6824917 [translationparty.com]

Re:Similar languages (1)

hardburn (141468) | more than 4 years ago | (#31422980)

Not quite what I meant by "broke it". The phrase just changed back and forth between two different translations before the script gave up.

Re:Similar languages (1)

aixylinux (1287566) | more than 4 years ago | (#31422472)

Pfft. That site doesn't work with Firefox 3.6, only IE.

Re:Similar languages (1)

istartedi (132515) | more than 4 years ago | (#31422532)

It doesn't work with my setup of IE or Chrome, the only two browsers I'm fiddling with these days. It must want you to totally drop your pants on security settings or something.

Forkbomb (0)

Anonymous Coward | more than 4 years ago | (#31423514)

The translator can't seem to figure out exactly how many times the road has diverged...

"two roads diverged in a yellow wood"
http://translationparty.com/#6827987

Asian languages and vastly different grammar (5, Interesting)

penguinchris (1020961) | more than 4 years ago | (#31421682)

Several others have noted this as well - for Asian languages, Google has a lot of work to do. The Chinese translation near the top is impressive, but while Chinese and Japanese translations are probably pretty good on Google, other Asian languages suffer greatly.

I've been translating a lot of Thai lately, and initially I thought Google was great - the interface is really slick, and it seemed to give a decent result. Passing the translation back through often gave me really weird stuff, but I was expecting that. So it was great, until I tried using it to communicate with someone in Thai - even for really, really basic stuff, often they had absolutely no idea. It was just way off.

While you can feed western languages through it and get great, usable results, for Asian languages besides Chinese and Japanese it's next to useless. I'm guessing there isn't much of an incentive for Google to focus on other Asian languages - for example, in Android 2.1 on the Nexus One there is no way to even install fonts for less-popular Asian scripts like Thai, much less inputting text in those scripts - despite this capability being available on certain other Android phones (you can install it on the Nexus One if you root it, of course).

Based on what their technique for learning translation is, though, hopefully this will improve over time. It's an impressive system as it is, but very much limited to "popular" languages and those very similar to English.

Re:Asian languages and vastly different grammar (0)

Anonymous Coward | more than 4 years ago | (#31421762)

It works by using existing translations to gather examples. More examples better translations. Popular means more examples, hence
better translations...

Re:Asian languages and vastly different grammar (2, Informative)

Cyberax (705495) | more than 4 years ago | (#31422010)

Russian, Polish and Ukrainian translations are laughable as well.

Even UkrainianRussian translation is mediocre, even though it's pretty trivial (other translators have almost 100% perfect translations).

So, good job but still lots to do.

Re:Asian languages and vastly different grammar (0)

Anonymous Coward | more than 4 years ago | (#31423132)

Ah, from English / from Chinese to Japanese both sucks. I've just confirmed it. For years Chinese to English is known to work well as Europe and China are on the same continent and share basic structure of languages. Japanese is a slightly different stuff and we got a bit more to do......

giving http://chinese.engadget.com/2010/03/09/samsung-prices-tl500-tl350-aq100-and-sl605-shooters/ [engadget.com] will return unreadable Japanese, if I'd translate it to English that'd be something like:

timesamsung TL500TL350AQ100 and SL605, yet to talked on price for telling everything, announced previous. Especially, also TL500, when walks like Ricoh Loewe System, Pana LX3, small size will, specification, like hot-shoe line of strong strobe. Someone, itch inside my mind, at very last what level of price curious about, saw this digital camera? current sentence will be given. 14300 of NT is about 449 dollars, asks price. TL350 349 dollars or so. Not only TL500 and TL350, double shake-reduction has RAW format.

Re:Asian languages and vastly different grammar (1)

amaupin (721551) | more than 4 years ago | (#31423362)

Several others have noted this as well - for Asian languages, Google has a lot of work to do. The Chinese translation near the top is impressive, but while Chinese and Japanese translations are probably pretty good on Google, other Asian languages suffer greatly.

I have all but given up on Google's Japanese translation. Altavista (now Yahoo) 's Babel Fish is much more reliable when it comes to Japanese. Sometimes the Google translation is so wrong that I can't even understand how it came up with the response returned. At least with Babel Fish I can usually figure out where it missed an idiom or failed to choose the correct meaning of a certain kanji character.

Altavista's Babel Fish (1)

Pojut (1027544) | more than 4 years ago | (#31421246)

I remember using Altavista's offering back in the day...the results were shoddy at best. It could make anything sound like engrish :p

Pffft... (2, Insightful)

plasticsquirrel (637166) | more than 4 years ago | (#31421290)

For Chinese, just using a character dictionary is better because the translations in Google are so bad. Unfortunately, I must do this on a daily basis. Google is good at search, but cataloging the entire Web is a much easier job than learning Chinese.

Re:Pffft... (0)

Anonymous Coward | more than 4 years ago | (#31423920)

oh hell no... As someone who has studied Chinese for 5 years and lived in China for 4 years this is a very very bad idea. Chinese English dictionaries are terrible...

somebody has to say this (1, Insightful)

zlel (736107) | more than 4 years ago | (#31421336)

Granted that Art is not a field foreign to computing, translation is an art that is difficult to satisfactorily automate. It's not about getting the semantics right, or the meaning right, but to translate a piece of work into another cultural context for another person, is a bit like trying to read somebody's mind. The turing test for translation would probably be something like automatically translating a new contemporary musical into another language? IMHO that's more difficult than getting a computer to write its own musical. I believe there is a niche for automated translation, but even for the niche it's trying to fill, it's not good enough. Not especially in my part of the world where there is not only a diversity of languages, but also a great diversity in the language families from which these language take their characteristics.

Re:somebody has to say this (1)

Vintermann (400722) | more than 4 years ago | (#31423984)

Remember that the niche Google Translate is currently trying to fill is not doing your translation jobs for you, but letting you know the rough content of a text in a language you don't speak. For many languages and contexts (example: French newspaper articles) it is very, very good at this. For others (Ukrainian IRC logs) it is only slightly better than useless.

For western languages... (2, Interesting)

IANAAC (692242) | more than 4 years ago | (#31421406)

For western languages, I have no doubt that this will eventually be a decent option for general text.

Just not now. It still needs a lot of work.

I'm in the translation business, and the general trend in internet communications such as websites, etc. at least, is to simplify the language being used.

For specialized text, we're a long way off yet.

Re:For western languages... (0)

Anonymous Coward | more than 4 years ago | (#31421862)

If you get worse results with specialized text, either the text is crap or your corpus is crap. Specialized text is orders of magnitude easier to translate as contextual gaps are way smaller than in normal prose. Of course, if you or your clients are using Newspeak in order to avoid translator fees, you probably don't care enough about quality to build a decent corpus. Move to Google translate and forget about it.

Re:For western languages... (1)

greenguy (162630) | more than 4 years ago | (#31423390)

I'm also in the business, and frankly, I'm not impressed. Google Translator is a stopgap at best. A lot of posts here have said it's good enough for basic phrases, and that may be true, but how far is that going to get you? Great, you can read short phrases... assuming they're not too obscure, and that they're written correctly and legibly in the source language, and that there's not some double entendre going on, and that Google understands both the dialect being translated and your dialect, and so on...

Basically, good translation requires a vast amount of context -- both within a given document and in the broader culture. Google can accrue billions of documents that are reasonably good translations, but it can't accrue their context. The very fact that it's lumped them all together strips out their context. And what's appropriate in one context is quite inappropriate in another. [Insert well-worn anecdote about the Spanish verb "coger" here.] This simply can't be automated, because the same translator works in a variety of contexts, and will make different decisions at different times.

An obvious example: most languages have two or more levels of address, depending on social distance. English does not. English doesn't distinguish between the subject and object form, or even the singular and plural, in the word "you," and nearly every other language does. That means there are three independent decisions to make when translating that word, with two or three or four reasonable choices for each. And that doesn't count word order, or colloquial usage that wouldn't translate directly at all. That's all dependent on context.

In short, professionals won't be in danger any time soon.

Re:For western languages... (1)

Vintermann (400722) | more than 4 years ago | (#31424020)

Google can accrue billions of documents that are reasonably good translations, but it can't accrue their context.

What makes you say that? I'm pretty sure Google Translate "remembers" whether its training data came from a newspaper article, a UN document or a Gutenberg book. Otherwise it would hardly be able to make as good translations as it in fact does. One problem they have is that some forms of texts (like UN documents) are heavily overrepresented in their corpus, while others (like informal dialogue) are almost non-existent. People are not likely to paste in a legal document, people are more likely to paste in what they are talking about at their Korean friend's Facebook wall. Yet it works OK.

Google Translate does not aim to replace professionals, and if you're a professional it's pretty sad you aren't aware of that. Google translate is to help you understand a text in a language you don't know. If you need to do a translation for someone else, then Google's offering is not the translator, but the translator toolkit [google.com] .

Re:For western languages... (1)

hughperkins (705005) | more than 4 years ago | (#31424472)

> In short, professionals won't be in danger any time soon.

Sounds like you're worrying about that though ;-)

Technology changes pretty quickly. Give it twenty years or so.

I remember when I tried drawing a 3d graph of z = cos ( x * x + y * y ) on my Sinclair ZX Spectrum, in 1982 or so. Each single pixel took roughly a whole second to plot!

Now, you can draw such a graph in realtime, 50 frames a second, whilst rotating the whole thing with the mouse. 3D graphics in Doom and then Quake, and now Counter-Strike are increasingly realistic, and run at fluid frame-rates, and simply because the underlying engine - the CPU and GPU - got very very fast.

I noticed that they were using my web site (2, Interesting)

Anonymous Coward | more than 4 years ago | (#31421480)

I have a web site where every page is available in English and German. When I tested Google's translation with it, I noticed that Google reliably translated one sentence in the opposite direction, i.e. from English to German when I had asked for a German to English translation: On every page in German, there is one sentence in English which leads to the corresponding page in English. Google's translator appeared to pick the translation right from that page, which of course has that sentence in German (leading to the German version of that page). Google doesn't do this anymore, but when I saw it, I realized that Google's translator did not at all "understand" what it was translating.

Re:I noticed that they were using my web site (1)

DollyTheSheep (576243) | more than 4 years ago | (#31421906)

Sorry, but that means exactly the opposite to me: Google Translation very well understood, that one sentence on every page is in a different language and "reversed" that sentence as well. Wouldn't have been possible, if Google Translate would "understand" exactly nothing about language.

Re:I noticed that they were using my web site (0)

Anonymous Coward | more than 4 years ago | (#31422258)

If Google recognized that that sentence is already in the target language, why would it translate it into the source language, which I probably don't understand when I'm using the translator for real? This question and the fact that Google no longer does the reverse translation lead me to the conclusion that Google's translator recognized that they were corresponding sentences, but didn't understand much else about them.

Their search parsing tech probably helps too (2, Interesting)

Phat_Tony (661117) | more than 4 years ago | (#31421488)

Wired recently had this article [wired.com] on Google's search algorithm, which mentioned how far ahead it was in parsing language for things like bi-grams to figure out what the meaning of the search was by "figuring out" the relationships between related words in a very human-like way. They have also built an impressive synonym system. These technologies, developed for search, strike me as really critical for good translation.

An exerpt from the article:

"People change words in their queries. So someone would say, 'pictures of dogs,' and then they'd say, 'pictures of puppies.' So that told us that maybe 'dogs' and 'puppies' were interchangeable. We also learned that when you boil water, it's hot water. We were relearning semantics from humans, and that was a great advance." But there were obstacles. Google's synonym system understood that a dog was similar to a puppy and that boiling water was hot. But it also concluded that a hot dog was the same as a boiling puppy.

Re:Their search parsing tech probably helps too (4, Funny)

MichaelSmith (789609) | more than 4 years ago | (#31421522)

But it also concluded that a hot dog was the same as a boiling puppy.

There is nothing wrong with that. My son forms connections like that all the time, and he is only slightly younger than google.

Re:Their search parsing tech probably helps too (1)

Vintermann (400722) | more than 4 years ago | (#31424060)

Google Translate has become impressively baby-like lately. If you enter "I'm watching a movie", you get out "Jeg ser en film" in Norwegian, nice and correct. But if you enter the incorrect phrase "I'm watching a movi", you get out the creative response "Jeg ser på en filmdel" - I'm looking at part of a movie!

Another: Ice cream in spanish is "Helado", and is translated correctly. But what do you get if you forget the H, and enter "elado"?

"ais krihm"!! See for yourself :-)

Re:Their search parsing tech probably helps too (1)

MichaelSmith (789609) | more than 4 years ago | (#31424148)

Google has read those jokes somewhere and is repeating them to you. I sense emergence.

Of course google doesn't understand what it repeats to us, but I question the idea that we understand things any more than google does. There may be many non-human intelligences in the world, but google is the first really smart system designed to (process|comprehend) our languages.

I wonder what happens if I dial MYCROFTXXX in google voice? Will google checkout issue a payment for an unlikely amount of money?

Re:Their search parsing tech probably helps too (1)

DollyTheSheep (576243) | more than 4 years ago | (#31424278)

Wired recently had this article [wired.com] on Google's search algorithm, which mentioned how far ahead it was in parsing language for things like bi-grams to figure out what the meaning of the search was by "figuring out" the relationships between related words in a very human-like way. They have also built an impressive synonym system. These technologies, developed for search, strike me as really critical for good translation.

OK, so they introduced contextual knowledge (or "world knowledge" or "semantics" if you will) when they saw, that page rank and keyword based search didn't cut it for many search queries? Shouldn't that have come not as an afterthought but long before? I mean, how can anyone expect, that search would never involve some contextual knowledge to be succesful?

My guess is, that Google of course knows this. What they do is to build up contextual knowledge through their own search engine, how people relate words to each other and not by imposing a predefined rule set or ontolgy beforehand like cyc [wikipedia.org]

.

As a foreigner in Japan (0)

Anonymous Coward | more than 4 years ago | (#31421532)

I use Google translate frequently, and the translations are not very good, but when you pair it with some basic knowledge of the idiosyncrasies in the Japanese language. I am at least able to get a basic understanding of the text. But in some cases the results are barely any better than the Babble-fish example above.

Having some basic understanding of the Language, I can often divide the text into smaller pieces , which seems to improve quality.

Re:As a foreigner in Japan (1)

Archon-X (264195) | more than 4 years ago | (#31421832)

I guess it depends on the language a lot.
I've found that japanese translations are often awkward, and you have to 'force' a correct translation by changing context, structure, etc.

Alternatively, the french translation is very, very good, picking up subtleties of formal / informal speech, slang and abbreviations.

Good thing /. didn't use the original NYT headline (0)

Anonymous Coward | more than 4 years ago | (#31421596)

It was the lovely "Google's Computer Might Betters Translation Tool" (since changed in the HTML title to "Using Computing Might, Google Improves Translation Tool" and "Google’s Computing Power Refines Translation Tool" in the online heading):
http://languagelog.ldc.upenn.edu/nll/?p=2169 [upenn.edu]

There's also some commentary about the article from Ben Zimmer at Language Log...
http://languagelog.ldc.upenn.edu/nll/?p=2170 [upenn.edu]

Why is machine translation so difficult? (1)

ArchieBunker (132337) | more than 4 years ago | (#31421704)

That's what I've never understood. Why can't software translate as easily as a human? Is it really that difficult to come up with a set of rules so things are worded correctly?

Re:Why is machine translation so difficult? (-1)

Anonymous Coward | more than 4 years ago | (#31421778)

Because words in different languages don't have one-to-one mappings. Also, words can have vastly different meanings depending on context. Some words don't even really have equivalents in certain languages. For instance, Japanese has quite a few words that don't have any direct english word. It takes a human translator to understand the context and translate it appropriately, and even then, its very likely that you aren't going to understand the full nuance that existed in the original.

Re:Why is machine translation so difficult? (2, Informative)

slimjim8094 (941042) | more than 4 years ago | (#31421856)

Is it really that difficult to come up with a set of rules so things are worded correctly?

Yes.

Longer answer - computers are very bad at context and meaning. Take French to English - it would be one thing if words had the same exact connotations and grammar, and you could just do a find-replace. But, unfortunately, that's not the case. There are many words in French that - depending on context - have many different meanings. In mathematical terms, the mapping of French words to English words is not bijective, nor vice-versa. Take the French word bete - it most literally means "beast", but is often used to mean "stupid". How is a computer supposed to figure out which one to use?

I just checked and Google Translate actually gets the connotation right, but it's a relatively simple example. Consider the French word "baise" - either kiss or fuck - and a more complicated example. Now... Google gets this right too (creepy!)

In any case, the only to get perfect translation is to make the computer understand the relevant meanings and connotations of words and stylistic choices... How would you convey a Cockney accent, or Cockney phrasing, in Chinese? In short, you'd need an AI.

Re:Why is machine translation so difficult? (1)

Jer (18391) | more than 4 years ago | (#31422164)

Just FYI - the "creepy" results you're getting off of Google probably indicate that there are a lot of French-language pages out there for Google to gather data on.

My guess is that it's even better than that - there are a lot of French pages out there that have corresponding English pages that Google can mine for information. Anytime you can find parallel corpora the algorithms are going to do better. I'd guess that the French/English pairing is probably a fruitful one for Google's algorithms because of Quebec - forcing information to be available by law is going to generate a lot of data online for Google to learn translation from.

(There are other reasons that French/English, German/English, and generally "Insert European Language Here"/English are going to be easier for machine translation algorithms to learn than other pairings. But more data is almost ALWAYS helpful for these kinds of algorithms. Parallel corpora are expensive to generate and it's always nice when you can find them for free on the web.)

Re:Why is machine translation so difficult? (1)

h4rr4r (612664) | more than 4 years ago | (#31422600)

Your sig is wrong,

MS is doing that too. Nice OS you got there, you might infringe on some of our patents, how bout you pay us so we don't sue you.

Re:Why is machine translation so difficult? (1)

XanC (644172) | more than 4 years ago | (#31423550)

But they can only do that because of state enforcement.

Re:Why is machine translation so difficult? (1)

BZ (40346) | more than 4 years ago | (#31423304)

Yes, it really is that difficult. Consider this classic example in English:

    Time flies like an arrow.
    Fruit flies like a banana.

There happen to be two ways to read the latter sentence. One is in a way analogous to the former one: the subject is "Fruit", the intransitive verb is "flies", and "like a banana" is an adverb phrase. The other way to read it is that the subject is the noun phrase "Fruit flies" , the transitice verb is "like" and the direct object is "a banana". Heck, this case is difficult for _humans_ to get "right" at times.

There are various situations like that in which the meaning is ambiguous, but even worse are situations in which the concepts used are just nonexistent in one of the languages/cultures. For example, English names are made up of three parts, typically: first, middle, last. Russian names are also made up of three parts: given, patronymic, family. Russian family name is a good match for English last name. Russian given name is a good match for English first name. Russian patronymic is ... not really a match for English middle name (in that for example it's not up to the parents to choose it), but occupies a similar position in names, obviously.

So when translating a phrase containing (patronymic) into English, what word you use to translate it really needs to depend on what's being said. If it's a technical discussion about the concept, translate as "patronymic". If it's a casual discussion about names, "middle name" may be appropriate. If the Russian text says that X addressed Y by name and patronymic then that _is_ what they did, but that happens to be the standard polite form of address. The equivalent English form is a Mr/Mrs/whatever followed by the last name, and the translation should reflect that.

There are also often situations in which two phrases are technically the same in terms of denotation but have different connotations (or heck, just different emphasis on which words are important; compare "I saw him" to "I saw him/em" and note that such emphasis differences can be expressed with word order in many languages). Getting that right can be very difficult unless you really understand what's being said. Pattern matching might work if you've seen that exact pattern before (which is Google's approach), but even small differences in the surrounding sentence structure can totally change the meaning of the part you're trying to translate.

I leave you with this short story (or rather stories):

    Jack was walking across the meadow when he saw a spring. The spring glinted in the
    sunlight, and he thought that he'd never seen something quite so beautiful. He bent
    down and...

    1) ... put it in his pocket.
    2) ... had a drink.

How much lookahead in the translation is needed to translate the first sentence correctly? If the rest of the story is option 2 above, how do you know that he didn't just take a swig from his flask before picking up the bit of metal?

Translation is hard for people. (3, Informative)

Estanislao Martnez (203477) | more than 4 years ago | (#31423566)

Why can't software translate as easily as a human? Is it really that difficult to come up with a set of rules so things are worded correctly?

But translation isn't easy for humans, so there's no reason to expect it should be easy for computers.

Translating from one language to another, for a human translator, basically comes down to this:

  1. Reading the source text and understanding it as deeply as possible.
  2. Writing an "equivalent" text in the target language.

But the problem is that there is never unique "equivalent" text in the target language, but rather, a lot of alternatives that make different tradeoffs. This is because a foreign language is part of a foreign culture that has many concepts that are foreign to the source language, and likewise, the source language is part of a source culture that is foreign to the target language. So translators repeatedly find themselves in situations where either they must leave out something that the source text says or implies, or else say something unnatural in the target language to convey that information.

Comparing the grammar of dramatically different languages makes this really clear. For example, many languages have grammatical evidentiality [wikipedia.org] , where statements are subject to grammatical rules that depend on the source of the speaker's information for the statement. So for example, a language where the equivalent to the sentence "Joe kicked Tom" required the verb to be conjugated differently depending on whether the speaker saw Joe kick Tom or heard so. If you had to translate an English text to a language like that, you'd have to decide, for each clause in the English text, who is the speaker of the sentence, and whether they know the event first-hand or second-hand, and either of those may often be unclear from the English text.

In the converse case, imagine if we're translating from a language like that into English. Then every sentence in the source language encodes some claim about how the speaker knows the information conveyed in that sentence. A completely literal translation, in which every English sentence had that information, would be extremely unnatural English writing. Leaving it out of every single sentence, on the other hand, might leave out something important to understand the text in some cases. So the translator has to decide in which cases the evidential conjugations of the source language must be translated into a longer English sentence than otherwise necessary.

This is one extreme example, but this sort of problem occurs at every level in translation. Translators often find themselves adding in information that the source text doesn't say, having to use circumlocutions in the target language to express really simple things from the source language, leaving out information from the source text has because it would be too cumbersome to phrase it in the target language, adopting strange conventions in the target language, or having to write supplementary materials to help the readers understand the translation (footnotes, introductions).

Or in a few cases, the translators write for people who don't know the source language but are familiar with some of the customs and concepts, or willing to learn them to understand the translation, and then they just leave untranslated words in. (Examples: lots of philosophy translations from German or French; anime fansubs that leave Japanese honorifics like -san or -sempai in, because the people who use them are anime fans, are at least a bit familiar with them, and actually understand more nuances that way.)

So, translation is not a mechanical task, and thus, there can't be a simple set of rules to do it. It's, as I said at the top, understanding a text in the source language, and writing another in the target language, tailored toward a different audience. And it requires understanding the audiences of the original text and the translation, and making many informal decisions based on that.

Pretty good and impressive as it translated (1)

al0ha (1262684) | more than 4 years ago | (#31421732)

Eier von Satan correctly - except for Augenballgroße which is essentially Eye-ball-large.

Chess translations (1)

JordanH (75307) | more than 4 years ago | (#31421918)

If you are into chess, Google Translate opens up a whole world of chess blogs to you. I haven't used it extensively, but I was quite impressed with this translation [google.com] .

To the chess players out there, note how it picks up notation interspersed with the text. It's not perfect and seems to fall back into Spanish algebraic in odd places, but I think they are the only translation tool that even tries to do chess notation.

I wonder if there are other "special purpose" translations that Google Translate attempts. It's pretty impressive to me that they even bother with the small chess blog reading public.

Oh, Google Translate does a lot better job on the non-chess parts of blogs, too.

How different is this from AI research? (1, Interesting)

Anonymous Coward | more than 4 years ago | (#31422060)

Obligatory Chinese Room [wikipedia.org] mention.

If a translation engine grows strong enough to adequately translate the phrases "give us our daily bread," "sharks are predatory carnivores," and "the loan shark wants his bread," that implies a significant ability to contextually infer meaning. Could someone opine on (or point to a work exploring) how similar the task of building an accurate translator is to the task of building a competent, world-aware (if perhaps not absolutely Turing-quality) AI?

Re:How different is this from AI research? (1)

Archon-X (264195) | more than 4 years ago | (#31422236)

Google's french translations are very strong:

- Give us our daily bread (unsure what the catch is w/ this phrase, but)
- Donnez-nous notre pain quotidien

- Sharks are predatory carnivores
- Les requins sont des carnivores prédateurs

- The loanshark wants his bread
- L'usurier veut son pain

All translated with the correct context

Re:How different is this from AI research? (0)

Anonymous Coward | more than 4 years ago | (#31423202)

No, "bread" in terms of the loanshark means money. So "pain" is the incorrect translation.

Re:How different is this from AI research? (0)

Anonymous Coward | more than 4 years ago | (#31424506)

"veut ses sous" might be better

Re:How different is this from AI research? (0)

Anonymous Coward | more than 4 years ago | (#31423386)

Incidentally, the third sentence was the only one with a catch; the others were just framing it as an example. Bread in the first sentence means "food made from dough;" shark in the second sentence means "carnivorous marine animal." No tricks at all there; the most common definitions for each word are the contextually correct ones -- food and fish. "Loan shark" (with a space in my stated example) from the third means predatory moneylender; "bread," when taken in consideration of who seems to be desiring it, means money. That's what I was talking about. If a translation engine COULD interpret the difference between yeasty bread and economic "bread," the ability of that engine to meaningfully connect a slang term for money with a vulgar phrase referring to a moneylender would imply a non-trivial functional understanding about what money is and what moneylenders do, Hence, the Chinese Room mention.

Now, I'm not interested in the metaphysical aspect of Searle's thought experiment; I'm an atheist and a fairly strict physicalist, and I hold that I am a biological computer, one with some properties that are incompletely documented by current science. I'm more curious about its implications for definitions of sentience. Let me ask more clearly: if a device can interpret written language well enough to understand context and nuance for translation purposes, how far removed is it from a truly sentient entity capable of holding rational discourse?

Re:How different is this from AI research? (1)

biryokumaru (822262) | more than 4 years ago | (#31422998)

The Chinese Room is stupid, because if I had a mathematical model of the human brain, I could calculate these kinds of ridiculous ideas just as easily as the dude with the book calculates Chinese. The logical extension of the Chinese Room is that no one thinks, which is a pointless conclusion.

From the Menu Example Given... (0)

Anonymous Coward | more than 4 years ago | (#31422088)

Google can now track what I order for dinner. I feel so naked.

what is the other side (0)

Anonymous Coward | more than 4 years ago | (#31422108)

A huge surveillance infrastrure that can be used to monitor conversations in real time as they can be converted to text and searched for inference.

Malay English dictionary (0)

Anonymous Coward | more than 4 years ago | (#31422432)

I have found in for Malay-English translations, that initially, Googles translation was better than Dewan Pustaka Dan Bahasa (DPDB) ie the people in charge of developing the Malay language. Since then however, I have found more errors creeping into their translations. I wonder if somebody is poisoning the well because when I first used Google translate the 99% of the translations were accurate, and the 1% was an unknown word. In my last use of Google translate, 50% of the words were wrong (as opposed to being unknown).

What does that mean? (1)

ScrewMaster (602015) | more than 4 years ago | (#31423190)

in one of the company's few unqualified successes

What does that mean? Google has had more successes in the online world than most of its competitors.

Low-volume languages? (1)

vampire_baozi (1270720) | more than 4 years ago | (#31423244)

While this works well for the more widely-spoken languages (Western/European Languages, Chinese, Japanese), I suspect there is a massive drop-off for some of the less common languages, especially those for languages spoken in countries less connected to the internet. The article mentioned they feed the algorithm human translations from the EU and UN proceedings; what about less-common Asian languages, the Indian subcontinent languages, central Asian languages? The volume simply doesn't exist.
Where the volume does exist, what about Russian and Korean, which are dominated by Yandex and Naver? Might be interesting to run a comparison, but unfortunately all the languages I speak are covered fairly well by Google at this point :(

Try iSnapit and translate (1)

billwallace (1763664) | more than 4 years ago | (#31423374)

Try iSnapit on your iPhone or Android, and you can translate everything you see with a single click - and much more.

Youtube? (0)

Anonymous Coward | more than 4 years ago | (#31423532)

Now that they turned on automatic sub-titles for many youtube videos, how long until these subtitles can be read in any language?

And how long until they are synchronized by a synthetic voice in any language?

nokia had this years ago (1)

dwater (72834) | more than 4 years ago | (#31423604)

I guess what Google is talking about must be something different because Nokia had s/w for the N95 that could take a picture of a Chinese menu and provide a translation in English.

google skynet ? (2, Funny)

Atreide (16473) | more than 4 years ago | (#31424182)

Wasn't Skynet used for translation
before it decided for a better future for humanity ?

Load More Comments
Slashdot Account

Need an Account?

Forgot your password?

Don't worry, we never post anything without your permission.

Submission Text Formatting Tips

We support a small subset of HTML, namely these tags:

  • b
  • i
  • p
  • br
  • a
  • ol
  • ul
  • li
  • dl
  • dt
  • dd
  • em
  • strong
  • tt
  • blockquote
  • div
  • quote
  • ecode

"ecode" can be used for code snippets, for example:

<ecode>    while(1) { do_something(); } </ecode>
Sign up for Slashdot Newsletters
Create a Slashdot Account

Loading...