Ethan has provided me with a mid-point summary of results, which I’ve included below. I was surprised to find that Microsoft and Babelfish are beating Google on some languages pairs, as well as on shorter text strings. Although Google is emerging the overall winner — and receiving some much-deserved attention from the media — it’s nice to see some healthy competition.

That said, quality is only one piece of the puzzle. The other piece — perhaps much more important — is usability. Now that Google has embedded its MT engine into Gmail and Reader — and now its Chrome client –I find I’m using Google exclusively as my MT engine.

Here are Ethan’s findings so far (emphasis mine):

At the highest level, it appears that survey participants prefer Google Translate’s results across the board.

In a few languages (Arabic, Polish, Dutch) the preference is overwhelming with votes for Google doubling its nearest competitor

However, once you remove voters that have self defined their fluency in the source or target language as “limited,” the contest becomes closer along some of the heavily trafficked languages. For example:

• Microsoft Bing Translator leads in German
• Yahoo! Babelfish leads in Chinese
• Google maintains its lead in Spanish, Japanese, and French

Observing only the self-defined “limited fluency” voter reveals a strong brand bias. If your fluency in the target translation language is limited, it would stand to reason your ability to assess the quality of the translation is very limited. And yet…

• Limited-fluency voters chose Google over Bing by 2 to 1
• They also chose Google over Yahoo! Babelfish by 5 to 1

As I had guessed, Yahoo! and Microsoft’s hybrid rules-based MT model performed better on shorter text passages

For phrases below 50 characters, Google’s lead in Spanish, Japanese, and French disappear. And Microsoft’s lead in German widens.

Beyond 50 characters, Google’s relative performance seems to improve across the board.

For passages that are only one sentence, the same effect is seen, though to a lesser extent than under 50 characters.

On March 4th, we made a few changes to our survey – hiding the brands and randomizing the positions of the text results before voting.  Since then, we have not yet collected enough data to draw conclusions, but Babelfish seems to be receiving the biggest boost, perhaps showing the effects of the recent neglect of that tool.

Clearly, Ethan needs more data to arrive at more concrete conclusions. If you’re a translator and you want to lend a hand, here is the voting site.

PS: Here’s an interview with Google’s MT guru Franz Josef Och.

But which engine offers the highest quality translations? I’m assuming Google, but this is only based on anecdotal feedback and personal experience.

Years ago, IBM developed an algorithmic method of measuring MT quality known as the BLEU score. Google scored well here, but the BLEU score is not without its critics.

Translation, like writing itself, is as much an art as it is a science.

Which is why translators are best positioned to judge the quality of machine translation engines. And although even translators are going to disagree as well, if you get enough of them together, perhaps you can begin to draw statistically significant conclusions.

Enter Ethan Shen and his start-up venture Gabble On.

Ethan has set out to recruit a few thousand volunteer translators to compare the three free translation engines. He asked me to help get out the word. He promises that he will publish the results for all to see. He’s also offering a free Apple iPad to one lucky volunteer. I have no financial interest in the project. I’m just curious to see what engine comes out on top.

Here are the details from Ethan:

We are seeking functional to fluent speakers of any two languages to take 5 minutes to judge and submit their opinion in our dynamic comparison engine (until March 29, 2010). At the end of the 6 week voting period, we will be publishing our results publicly in hopes that our research can to contribute meaningfully to the body of knowledge in this field.

In gratitude for your participation, we are awarding one new Apple iPad to a lucky participant. The survey can be found at: www.gabble-on.com/SurveySelector.aspx.

Which engine do you think is best?

In response to the earthquake in Haiti, Microsoft quickly expanded its machine translation engine to include Haitian Creole.

Today I noticed that Google has an alpha version of its Haitian Creole engine as well.

Though it’s sad that it took a natural disaster to spur attention to a particular language, I’m glad to see the language available.

It’s hard to underestimate the importance of readily accessible machine translation. Just as search engines help us better understand the world, machine translation engines help us better understand one another.

And, yes, they’re far from perfect. But they’re far better than nothing at all. And they are finding their way into countless applications and countless fixed and mobile devices, each additional language offering another glimpse into another world.

Six months later, Google added ten more languages, a two-year growth trajectory illustrated below:

Google went from 13 languages to 51 languages in less than 16 months.

Not bad.

And, yes, I’m aware that we must not confuse quantity of translations with quality of translations. Your translation mileage will most certainly vary by language pair. Still, as language pairs go, Google is the only game in town across many.

Here are the 10 most recently added languages:

• Albanian
• Afrikaans
• Belarusian
• Icelandic
• Irish
• Macedonian
• Malay
• Swahili
• Welsh
• Yiddish

On a related noted, 41 of these languages are now incorporated into Google Docs.

Every month there seems to be more of them, which could mean that the quality of Google Translate is improving, or this site is doing better in the rankings, or some combination of the two. Or, it could be simply be that more people have discovered Google Translate.

Given my passion for country codes, it’s fair to say that I also enjoy language codes. And it is through language codes that you can figure out what languages users were translating your site “from” and “to.”

Here is one referral string from my site:

First, you can see that the person was using Google Korea, so it’s fair to say the person was translating from English into Korean. The “To” line is actually the blog title post translated into Korean.

That was an easy one.

This next one is a bit more challenging:

This person was using Google.com, so you have to focus on the language codes. There are two here — an “id” (which follows  “hl=”) and an “en” (which follows “sl=”). What this means is the person was translating from English into Indonesian (Bahasa Indonesia).

Here is what the translated page looks like:

The quick and easy way to know the target language is to focus on the “hl=” string. In the screen shot below, the target language is German.

And here is a language code reference if you want to study your web logs.

What I want to know is what percentage of web traffic is taken up by Google Translate. Anyone care to share their Web log stats?

Based on my cursory analysis, I would estimate the figure to be between 5% and 10%, but that’s very rough.

For now, I’d like you to try entering the following text strings into both Bing and Google (to save you time I created pre-loaded hyperlinks):

• Translate I love you: Bing vs. Google
• Translate I love you into Chinese: Bing vs. Google
• How do you say I love you in Italian: Bing vs. Google

Below are screen shots of the first text string in both Bing and Google. I’ll let the pictures speak for themselves:

Google, despite its massively powerful translation engine, doesn’t simply answer your translation question. Instead, it provides links.

I realize that this is a relatively minor feature and that it currently only supports a small number of very common text strings, but it’s still a very handy feature for a translation geek such as myself.

Now, I’m not saying Bing is perfect. When it comes to technical searches — or when I just need to look up a Wikipedia article quickly — Google still does better, sometimes far better.

But I’m glad to see Bing integrating translation in an intuitive way. It’s a feature that I’ll be using again.

PS: Here is the blog announcement of this feature from Microsoft Translate team.

That said, I do wonder how this article is going to be received by the translators of the world who actually expect to be paid for their services.

For example the for-profit, invite-only conference company TED saved about $500,000 using volunteer translators. Clearly TED could have coughed up the money.

I can see this article spurring on CEOs across the land to think that they too can get free translations.

One thing I mentioned awhile back is that you need to be translation-worthy to get away with pro-bono services, particularly if you’re a for-profit company.

Facebook, Google and, now, TED appear to be translation-worthy. But I wouldn’t expect to see, say, General Motors succeeding in this area (though they could certainly use the help).

But the larger issue here is to the extent that volunteer translation for companies that can afford to pay for translation undermines the translation industry. I don’t believe machine translation undermines human translation because companies generally use it to translation text they would never have hired people to do (or they use it as a first pass before bringing on the human translators).

But volunteer translation is different.

Are  volunteer translators taking money away from their colleagues? After all, TED and Google and Facebook certainly can afford to pay. Or are volunteer translators raising awareness for the value of their work, thereby benefiting the translation industry as a whole?

Personally, I think we’re entering a dangerous area where companies that don’t know better are going to think they don’t have to pay for translation. This all reminds me of Seinfeld’s George Costanza’s aversion to parking garages: Why should I pay, when if I apply myself, maybe I could get it for free?

This photo arrived courtesy of Gareth Morgan at Neovia Financial.

Apparently the proprietor of this restaurant in China decided to create an English-language sign using machine translation (MT) software and, apparently, the MT engine wasn’t working all that well.

So instead of “restaurant” we have “translate server error.”

It’s certainly one of the more memorable restaurant names I’ve come across. I’ll be sure to look out for it when I visit!

And I’ve love to know which MT engine delivered this message.

According to the unofficial Google blog, Google has inserted its “translate” link into a number of localized Google sites — such as France and Spain. Google.com is not yet included.

This is just another sign that translation is becoming a core element of Google’s world domination strategy. If you’re curious about Google’s market share around the world — here’s an interesting “crowdsourced” document.

Here is the France home page:

I rarely ever use this pull-down menu and I wonder how many others do. I realize that Google strives to keep an austere home page and this is one solution — but I’m not sure it’s worth it. If users can’t find the translation link they may never use it.

Google Translate is no longer growing up, it’s growing out — integrating itself across all of its many properties.

I’m happy to see Microsoft Research launch this new Web site widget.

I’ll have to test it out on my site when I get the time — and would love to hear from others who have tried it.

A demo site is here.

I really like how the widget creates a more seamless translation experience, which is a big step toward taking machine translation to the masses. I’m looking forward to the day when I no longer have to manually copy and paste URLs into Google Translate and the like.

However, I did get confused initially on figuring out how to return the page back to English. That bar at the top of the page blended in a bit too much at first. I’d like to see a language reset button on the widget itself.

I haven’t tried it yet. It looks like you have to post PDF files and then link to them to trigger the translation engine — not exactly what I’d call user friendly. But it’s a start..

This time last year, Google supported a mere 13 languages, which was in itself not bad.

But I particularly like the minor tweaks made to the site’s interface. As shown below, you can now click on your language to make it one half of a language pair — a welcome alternative to the pull-down menu, which continues to grow.

What I would like to see — and I suspect is less than a year away — is the ability to simply enter a URL and have Google auto-translate that Web site into your language without you having to specify your language. Google should already know this based on your locale setting — or at least let you set that preference ahead of time.

Google Translate can auto-detect the language for you right now — but you have to ask it to do that. Perhaps the processing overhead is such that Google doesn’t want to turn on this feature by default.

So, will Google support 70 or so languages a year from now? I doubt it, given the current economic climate. As Google notes on its blog, these 41 languages already address 98% of all Internet users. I assume that Google will focus less on language expansion and more on integrating Google Translate into its products as well as improving the UI.

To give you an idea of just how aggressive Google has been in this area, here is a screen grab of Google Translate from 2006:

And here is one from today:

That’s roughly twice the number of languages in two years.

Microsoft has also been busy over the past year. Like Google, it now supports machine translation using its own in-house engine.

And it also offers a handy Web translation widget that you can insert into your Web page to allows users to self-translate your site into their language.

The portfolio of languages is still on the light side, but like I’ve mentioned before, these types of developments illustrate that machine translation (despite its inherent limitations) is becoming a critical piece of the Web globalization puzzle.

UPDATE: Here’s an interview with Google MT researcher Franz Och.

The post I wrote on the Google Translation Center has been the most-visited page on this blog over the past month. Clearly, Google has struck a nerve in the translation industry — and its service is not even live yet.

The Translation Automation User Society does not appear to be as welcoming as I am of the Google Translation Center. A new essay on its site says:

Private companies will always seek world domination and customer lock-in. As a professional in the translation industry, the Google Translation Centre may help you on the short-term, but you only help Google long-term and you don’t help the world at all. Ultimately we pay the price for putting all translated words and sentences in the possession of a single company.

There is a distinct “evil empire” tone to this essay, which is understandable to a point. Google appears to be entering that evil empire stage of its development. Though I still use the search engine.

And the last I checked, Google’s “terms and services” page for the Translation Center had been taken down. So I can’t really say what Google’s policy will be regarding the translation memory (TM) that it may or may not leverage from this Center.

But it is no coincidence that TAUS is planning to develop a massive database of TMs of its own. I’m sure it wants readers to come away thinking that TAUS is going to be far more open with its TMs than Google will be.

TAUS says that its TM database will be free to the world for the looking up of translations of terms and phrases. But you’ll have to be a member to actually have access to the database (on a reciprocal basis) and membership is not free. I’m confident that this database will be of the highest quality as TAUS has some impressive corporate members, such as Intel, Microsoft, and Oracle (Google does not appear to be a member).

Personally, I’m glad to see both services emerging — as well as services from Asia Online and Language Weaver (which is now offering a Web-based SaaS translation service). We are entering uncharted waters and it’s important to have a mix of large and small players, as well as a nonprofit, to keep everyone on their toes.

Is there a risk to the world if Google owns the world’s largest TM (which it might have already accomplished)?

Perhaps.

TAUS raises important questions. The answers have yet to emerge.

My biggest concern is slow-loading Web pages. And, yes, I know the quality of the translation will leave plenty to be desired, but what I really like about the widget are the little RSS feed buttons. Just click the button and you can have translated feeds delivered to whatever feed reader you use.

What I don’t understand is the significance of the name “mloovi.” Am I missing something?

But here comes Google…

According to Blogoscoped, Google is about to launch the Google Translation Center.

This is an exciting development, though I don’t expect everyone to suddenly ditch ProZ for Google. Why? Because much of the appeal of ProZ is the community, which Google does not appear to be trying to support. Still, freelancers will certainly want to investigate this potential new resource.

I’ve called out ProZ as one company under threat from Google Translation Center. But EVERY translation agency needs to keep a close eye on this service. It could be a threat. It could also end up being something translation agencies use themselves — instead of paid platforms from SDL. Naturally, for this to happen this new platform has a lot of evolving to do. Still, I can’t help but wonder.

There is no mention of whether or not Google will support machine translation and/or translation memory. I’m assuming they will.

I have LOTS of questions and this service isn’t even live yet. So we shall see what happens. But this is big news, no question.

I wrote awhile back, that the translation industry as we know it is over. The technologists have taken over and they’re bringing brute force computing and massive networks to the table to reduce costs and increase time to market. This is just another sign of this macro trend.

What do you think? Is Google going to disrupt the translation industry or is this new platform going to fall flat?

(Thx Chris for the heads up!)

Update: I just read an insightful article on this Google’s service at GigaOm…

I’m referring to the breadth of languages and breadth of features that Google Translate supports.

Today, Google announced that Google Translate added support for ten more languages, bringing the total to 23. The ten new languages are Bulgarian, Croatian, Czech, Danish, Finnish, Hindi, Norwegian, Polish, Romanian and Swedish.

And that’s not all!

Google Translate also now provides a detect language tool that will tell what language a batch of text is in. This type of tool can come in awfully handy for people like me who navigate across so many languages on a daily basis. It’s an easy feature for Google to support because the translation engine needs to know what the source language is before translating it. But I also tested language detect on a few languages not yet supported for translation, such as Slovakian, and the engine correctly identified them.

A week ago, I integrated Google Translate into the home page of Byte Level:

When it comes to translation, I’m not a good example of “putting my money where my mouth is.” Byte Level Research, with the exception of the Tower of Babel site, has been available only in English for years.

While I have no illusions that this widget will make up for a lack of professionally translated text, I am curious to see if people use it and to what extent. What I need to know is if Google Analytics can track Google Translate widget usage so I can know which languages are most popular. If anyone knows how to set this up, please contact me.

And, if nothing else, it’s an interesting experiment — and it buys me time before having to shell out real money for professional translation, which I will ultimately need to do.

SDL has posted an FAQ page about the deal here — but the list is absent a question I recently posed:

Does this acquisition mark the beginning of the end of SDL being a translation services provider?

I think it does. Or, at least, I think it should.

I haven’t received an official answer from SDL to this question. My guess is that SDL will say that it has many clients who want a full-service solution and that it will continue to offer these customers a full-service solution.

Nevertheless, SDL is much more of a software company these days than a translation services company. SDL has been gobbling up software companies over the past year and now owns some of the best translation workflow and content management tools on the planet, most notably Tridion and now Idiom. At some point it will have to ask if the benefits of providing translation services are outweighed by the negatives, which become more acute with the Idiom acquisition.

Idiom, as many folks in the industry know, had been living on borrowed time for quite some time. It was subsidized by the investment community going back 10 years. These investors wanted a return on their investment and Idiom needed a buyer who could give it a larger infrastructure and sales team.

Early last year, I heard from two up-and-coming translation agencies that were taking a hard look at buying Idiom. That SDL would buy Idiom can be viewed as much as a defensive move as an offensive move. And I think the timing is good because Idiom finally hit its stride in 2007.

I attended the Idiom Summit in 2007 and there was a real sense of excitement and optimism in the air. I wrote about the company in October, noting two developments in particular:

The “Full-Service” vs. “Freedom of Choice” Divide Widens
Idiom has been steadily positioning itself against SDL. Where SDL may call itself a “full service” solution, Idiom calls itself a “freedom of choice” solution. That is, SDL can provide you with CMS software, translation memory (TM) software, and actual translation services.

Idiom, on the other hand, provides only globalization management software, leaving the client free to select translators (or translation agency), CMS tools providers, and so on. Some clients may want to stick with separate vendors for different products and services — which benefits Idiom — while others may want a full-service SDL solution. This positioning makes good sense because it makes things clearer for clients who are often very confused by the array of products and terminology.

Demand for On-Demand Growing
Idiom thought that its On-Demand software would appeal mainly to small customers, but a number of large customers are now using it, such as Bank of America and HP’s internal translation group.

One interesting reason that Web-based software has been popular with some companies is that you don’t need to dedicate an IT team to getting started. Ironically, we have found that marketing executives in large companies with very skilled IT departments often feel as if they have very limited IT support because they must “take a number.” The Web-based solution allows a marketing team to bypass IT almost entirely. Even if they want to buy the software, they can use on-demand as a run-up to it. Idiom reports that 50% of its sales pipeline is devoted to Idiom On-Demand.

What happens when a “full service” company buys a “freedom of choice” company?
Right now, SDL says that Idiom is going to remain an autonomous company. And I suspect that many Idiom clients won’t particularly care if SDL is the new parent — and may actually be quite happy to get translation services from SDL.

But what about Idiom’s very popular “LSP Advantage Program” — in which it basically gave away its software to a large network of translation agencies. These agencies had played an important role in Idiom’s growth, functioning as a channel sales partner.

I spoke to the owner of a smaller translation agency — and Idiom partner — last year and mentioned that there was talk about Idiom being acquired by a larger translation agency. He was not at all happy to hear this. It’s safe to say that none of these translation agencies are happy with SDL owning their workflow software.

I’m not surprised to see this press release from across Systems today. Thebigword is promoting an alternative to Idiom as well. Meanwhile, Translations.com is marketing a “safe passage program” for clients who want to migrate away from SDL/Idiom.

SDL surely expected this reaction from its competitors. And all this confusion could actually help SDL sell more translation services as the smaller agencies find themselves migrating to new software tools. At a minimum, the next year promises to be very chaotic in the mid-tier of the services market, which could bode well for the leaders, which includes Lionbridge.

I believe that Idiom is an excellent acquisition for SDL, provided that SDL can hold onto Idiom’s network of supportive and enthusiastic translation agencies. If not, SDL risks giving rise (and momentum) to a “son of Idiom” in the years ahead.

I think SDL would be best served to focus fully on software — on streamlining its large and complex portfolio — and in letting the translation services vendors do what they do best. A SDL Services spin-off would make sense as well.

Perhaps SDL is already planning this. I hope to hear more from them in the days ahead and will keep you posted as I do.

• Arabic/Spanish
• Arabic/French
• French/Spanish
• Danish/English
• Greek/English

What’s interesting here is that Language Weaver is moving beyond language pairs that include English.

English is the “pivot language” of the translation industry, in which content either begins in English or is converted into English on its way to its target language. Taking English out of the equation isn’t easy, because there is often a smaller translation corpora to build upon, but it is ultimately where companies want and need to go to accelerate translation and improve quality.

Says the Language Weaver press release:

In international markets, companies want to be able to translate directly between any two languages. Despite the fact that some of these languages are spoken and written by multi-millions of people, very often English is used as a pivot language for the translation. For example, for an Arabic to French translation, a company might first translate the information from Arabic to English, and then from English to French; this two step process slows the translation down considerably.

Translation software offerings operating directly between languages such as French to Spanish or Arabic to French are less commonly available. However, Language Weaver’s statistically-based approach to translation allows it to quickly develop translation systems between any two languages without using English in the translation process. These less-common language combinations save a step in the translation process and give companies timely access to information.

So now it’s time to take a deep breath and take a shot at predicting the future of Web globalization. The following predictions are based on current trends as well as gut feelings.

Predictions for the Year Ahead

1. The Weak Dollar Drives US Companies to Greatly Boost Web Globalization Spending
Okay, this was an easy call to make. The weak US dollar is helping companies weather a poor local economy by selling their goods abroad. And this year I’ve noticed a number of companies boosting their Web globalization budgets to expand into new markets or improve their current localized Web sites. All signs point to 2008 being a very busy year for translators and Web localization teams.

It’s also worth highlighting a few companies that invested in Web globalization while the dollar was still strong. It’s no sheer coincidence that some of the companies weathering the US downturn thanks to strong overseas sales are also some of the companies at the top of the 2007 Web Globalization Report Card:

• Google
• HP
• Cisco
• Xerox
• Caterpillar

While a great global Web site does not in itself ensure success, it is a major competitive differentiator as companies expand abroad. And those companies that embraced Web globalization early on are now reaping the rewards.

2. Internationalized Domains Names Become a �Surprise� Hit
I have long been a strong proponent of local-language domain names. Next year, ICANN will launch approved IDNs that the local markets will embrace. The media will report the surprising success these domains see in their respective countries. China in particular will report more than two million registrations in one month (probably in 2009).

The demand for local-language domains is there. Yes, there are plenty of details yet to be worked through at ICANN and there will no doubt be technical and political obstacles along the way — which will also be widely reported by the media. But next year will mark that point in time when the Internet begins to provide a true local-language end-to-end Internet experience for users around the world.

3. China Requires Companies to Register Chinese-Language Domains
China has been the most vocal proponent of IDNs. The government effectively argues that the Internet cannot be user friendly for its citizens until they can enter domain names in the native Chinese script. Now that IDNs are becoming a reality, I expect China to begin requiring companies that wish to do business in China to register IDNs.

China would probably initially require those companies that are licensed to host sites within China to register IDNs, but it may also go after companies that host outside of China as well. There is also a financial incentive for this requirement, as these registrations will benefit state-owned domain registries. Following in China’s steps, Russia will also require the registration of Cyrillic domains, and other countries may also follow.

From a usability perspective, we expect most companies to register IDNs for China and Russia (and elsewhere) regardless of any such regulations. But the regulations will be a wake-up call to those companies that have overlooked IDNs.

4. The .asia Domain Will Not Equal the Success of .eu
In one year, the .eu domain surpassed 2.5 million registrations. It would stand to reason that the forthcoming .asia domain, which represents a far larger population, would be even more popular.

I’m not so sure. I believe that .asia will be fortunate if it reaches 1 million registrations in its first year, and it may not exceed half a million. The reason for this is two-fold:

• No geographic boundary. While .eu refers to a group of clearly defined countries, what countries does .asia refer to exactly? I’m not sure, and neither is anyone I’ve asked.
• IDNs are better. China, Korea, Vietnam and others are going to embrace domains in their native language rather than the .asia domain.

The sunrise period for .asia began in November, and the DotAsia registry reports 15,000 applications, the majority of which have come from US corporations. This is hardly an overwhelming response and a sign that 2008 will be an underwhelming year for registrations.

5. Splash Global Gateways Become Ubiquitous
Early this year, during the course of research for the 2007 Web Globalization Report Card, I found that approximately 30% of the 200 Web sites reviewed made use of a splash (or landing page) global gateway, such as the one shown here:

That figure was a 20% increase from the year before and a sign of a trend that we do not see slowing. By the end of 2008, nearly half of all large multinationals will make use of splash global gateways.

6. Geolocation Gains Fans (and Enemies)
Geolocation is a “behind the scenes” way of guessing a Web user’s location by analyzing his or her IP address. It can be quite effective, and Google has used it for years. Amazon and Expedia also use it selectively, and more companies will follow suit next year.

However, I think we will also begin to see news reports of geolocation being used to keep people out of Web sites. For example, Amazon may offer a special discount on a book in one country and not the other, and it does not want users in the non-discounted country to know better. Using geolocation, users could be barred from seeing these discounts or products altogether, without being the wiser.

Geolocation could be used to create geographically gated communities on the Internet, something that goes against the spirit of the Internet but is also difficult to stop.

7. Multilingual Corporate Blogs Go Mainstream
While lawyers in companies around the globe fret over the legal implications of supporting employee blogs, more and more companies are taking the plunge, from Dell to Sun to Google. From a budgetary perspective, blogs offer the ideal way to bypass the cost-per-word limitation of translation while offering plenty of local content. Naturally, companies need good writers locally who can contribute relevant content, but these writers do exist, and forward-looking companies will put them to good use.

8. Wal-Mart Launches a Spanish Site for the US
While retailers like Lowe’s and Staples have launched Spanish-language Web sites for the US market, Wal-Mart has remained on the sidelines. This will change in 2008.

I witnessed a sign that Wal-Mart is headed in this direction during the week of Thanksgiving.

Wal-Mart ran a Web-based promotion that was bilingual, as shown here:

If you were to click through on any of these product links, you would be taken to English-language text. But I think this will change, probably by next Thanksgiving.

Looking Beyond 2008

9. Baidu and Yandex Enter the US to Challenge Google
If one the best forms of defense is a strong offense, I expect Chinese search engine leader Baidu and Russian search engine leader Yandex to attack Google on its home turf.

Yandex, shown here, continues to lead in Russia despite Google’s best efforts. And the success of both Yandex and Baidu hint that the search engine wars globally have only just begun.

10a. Google Becomes Your Company’s Internal Translator
Google Apps is a Web-based platform that businesses small and large rely on to support email and collaboration. At some point in the future, Google will provide business members with integrated access to Google Translate.

Google Translate will allow executives to have foreign-language emails translated in real-time into their native language. Web browsing can also be translated in real time for executives who want to learn more about foreign competition. All of this will be powered by Google’s in-house statistical machine translation (SMT) engine and it will be tailored to the company’s industry terminology.

The quality of this translation will by no means be perfect. But as Google’s engine gets to understand your business and as its global memory database grows, the quality of translations will improve. Google will allow companies to upload their own translation memories to further improve Google’s SMT engine.

10b. Google Becomes Your Company’s External Translator
Eventually, companies will become so comfortable with Google’s translation abilities that it will allow Google to be the company’s “front end” for specific types of Web content.

For example, a company may offer a public blog in English and will allow Google’s SMT engine to translate it for the world. Since this is content that the company may not have ever paid to have manually translated, it’s a nice proposition for both the company and consumers.

Naturally, there are a lot of “ifs” involved. Companies must upload their translation memories to provide the level of quality required, and the public must be satisfied with less-than-perfect translation quality.

But the ramifications of Google acting as a “multilingual interface” is significant. It could very well mean that we’ll see companies actually decrease their pure translation spending in the years ahead. But that does not mean they won’t be spending money on Web globalization. Their spending will probably increase, but it will be focused on content creators, editors, user advocates, and community managers.

What does this mean to vendors? It means that the time is now to begin thinking about how your company will thrive in a world where translation is not your core source of revenue.

The product is software designed specifically for German/English translation of content that falls within the mechanical engineering, construction, automotive manufacturing, and plant construction industries.I know; it’s a mouthful.

But the gist of it is this: It’s machine translation and workflow software designed specifically for an industry.

The quality of statistical machine translation (SMT) is easy to poke fun at when you rely on the mass-market free translation engines. But once you begin to optimize the SMT engine for a specific industry, where context and terminology have been narrowed considerably, the quality suddenly gets respectable.

Statistical machine translation software and translation workflow software are about as intimidating as software can get.

Traditionally, companies have had to exert a great deal of time and energy to customize machine translation and translation workflow products to their specific industries. But I think what Language Weaver and across are doing is a sign of things to come.

And this is great news for customers because it will allow them to see the benefit of this software sooner rather than later.

Says the release:

“New digital content continues to flood the manufacturing industries. With this package, we deliver significant productivity right out of the box, already customized and integrated with a translation management system — the hard work of aligning data for the machine translation already has been done for manufacturing companies of all sizes,” says Kirti Vashee, vice president of sales and marketing for Language Weaver. “So it gets easier and faster for translators to access appropriate translations for manufacturing industry phrases and sentences, which have a technical language orientation. That helps companies to bring products to market faster and provides the functionality for translation of content that has never before been translated.”

What’s the reason for this sudden transformation of the industry?

I believe the changes are due to the simple fact that the translation agencies are no longer leading the industry. The technologists have taken over, and they have a different vision for the future.

By technologists, I’m referring to software vendors, such as Idiom and Language Weaver and Clay Tablet. I’m also referring to the buyers of translation services, buyers who have seen how technology can make their lives easier and want to see their vendors make full use of this technology — from hosted project management software to machine translation.

While linguists focus on the “art” of translation, technologists focus on the “science” of translation. And that’s why we’re seeing the rebirth of machine translation as statistical machine translation (SMT). SMT brings the power of brute force computing to translation, to a degree that the pioneers of machine translation could have only imagined forty years ago.

SMT is not by itself going to disrupt the translation industry. But SMT, along with early adopter clients (by way of the Translation Automation Users Society), and the efforts of Google, are likely to change this industry in ways we can’t fully grasp right now.

The Translation Automation Users Society (TAUS) conducted a brief survey recently of its member companies and interested observers. About 55% of the 200+ respondents were translation vendors, while 30% were buyers of translation services.

TAUS asked if the “per word” pricing model used by the industry was outdated, and 65% of respondents agreed that it was. However, there was no agreement as to what model should replace per-word pricing. Respondents were evenly split between per-word pricing, per-hour pricing, and capacity (annual service agreement) pricing.

What this says to me is that we are standing at an exciting point in the industry — where there is a clear need for a new direction but no clear direction in which to head. In other words, the industry is searching for leaders, which is why the technologists have had so much sway lately.

The TAUS survey also found that 68% of respondents were in favor of the sharing of translation memory data. Not surprisingly, TAUS is going to make the sharing of memories a priority in 2008, and technology will play a key role.

The most definitive finding from the survey was that 75% of respondents felt that translation agencies should take the lead in investing in translation automation. I have seen recent signs of agencies doing just that — but only after companies such as Idiom and Language Weaver paved the way.

I do not believe it will be agencies that take the lead in translation automation.

I think it will be Google that will do more than any other company to promote the next generation of translation.

Many tech pundits say Google is in the process of “becoming the Internet” by not only powering search but by hosting content and applications and facilitating financial transactions. Right now, the search engine is just a means to the end — your company’s Web site. But what if Google became both the means and the end?

That is, what if Google became the multilingual interface to your company’s Web site? It’s not going to happen tomorrow or next year or possibly even this decade. But I think it will happen, and when it does happen, the translation industry as we know it will be changed forever.

I gave the tool a brief run-through today and it is definitely rough around the edges. To get the best results you really need to put some thought into the search queries you enter — the more translatable the query, the better the odds of success. But Google is just getting started. After each search, Google solicits user input to improve the cross language engine. It will be very interesting to see if users do indeed provide input and what role this information plays in improving results.

On Friday, Jessica Vascallero of The Wall Street Journal provided some insights into what Google’s competition is doing in light of this development:

Rival Yahoo Inc. is taking a more human-intensive approach based on its Yahoo Answers service, which is available in nearly a dozen languages, including Chinese, French and Portuguese and soon, Vietnamese and Thai. The service allows users to ask other Yahoo Answers users any question and have it answered by a native speaker. Yahoo then indexes the responses into its Web search results for others to find in the future. “Outside the U.S., there is information that is still in people’s heads that doesn’t appear on the Web,” says Tim Mayer, vice president of product management for Yahoo Search. “This is a very large opportunity.”

Meanwhile, Microsoft Corp. is working on improving its natural-language processing software that translates documents by extracting implied meanings behind a string of phrases or words as opposed to translating each word literally. The company says it is considering ways to implement technology into its Live search service, which it currently offers in 46 markets from Germany to Hong Kong.

And, according to Wired, Google will be launching its Cross Language Information Retreival “soon.”

Cross language search allows you to translate your search query into another language and have the search results translated back into your language in real time. So if you want to know what the Germans have to say about, say, Web globalization, you can search in German.

This is a natural evolution for Google – the more content you can provide to users the more Adwords revenues you can potentially generate. And it’s not such a bad thing for people like me who want my Web site to reach as many people as possible but aren’t quite prepared to buck up for human translation.

But quality is key to success here, as well as speed of translation. Google uses statistical machine translation (SMT) for two language pairs and licenses machine translation technology from Systran for the rest of the language pairs. Will Google rely on both SMT and Systran? I’m not sure. I expect over time that Google will roll out SMT across all languages, but this will take time; SMT requires massive horsepower to generate fast results.

I’m glad to see Google getting ready to roll out this technology. It won’t be perfect and translators will have a field day making fun of its shortcomings. But SMT has the ability to learn from its mistakes so the sooner it starts “learning” on a large scale, the better it will get.

PS: Here’s a Register article on Google’s SMT technology. If you want more information on SMT, check out the profile we did a few months back on the pioneer in this field Language Weaver.

- There was a lot of buzz around machine translation (MT). Both Systran and Language Weaver are here and there were (at least) four sessions devoted to integrating Idiom WorldServer with MT. More important, the translation agencies and clients appeared serious about using MT, or at least seriously interested in what it can do.

- The Idiom LSP (language service provider) partnership program appears to be doing very well. There were quite a few translation agencies here who are very happy with WorldServer. And there are a number of other agencies I met with who were in the process of becoming LSP users of the product. More important, there is a sense that a community is forming among LSPs around WorldServer.

- New Idiom clients over the past few months include Bank of America, Mattel, and Apple.

- Old and new acronyms were abundant this week. idiom used MTM to refer to using integrated MT/TM (translation memory) tools. Jaap van der Meer, head of TAUS, used the acronym FAUT: fully automatic usable translation to refer to machine translation. Alan Melby countered with HUTTA: human-understanding translation with technology assistance. I heard DITA used lots over the past two days — and I won’t even bother trying to explain what this acronym stands for because it still won’t make any sense. What do all these new acronyms mean? In addition to making my head numb, I take them to mean that there are significant changes afoot in this industry and these new acronyms are attempting to describe not only the new technologies by the new paradigms that are evolving.

- Finally, after my presentation on Web globalization, I spoke with one executive regarding the challenges of migrating the US Web site to the .us country code domain. A small number of large corporations have done this already and I have heard reports of frustration with Google because Google.com isn’t spidering the .us sites in a way that gives the US sites good placement in search engine results. This I think could become a growing issue. In theory, companies should be hosting their US-specific sites at the .us domain, freeing up .com for global content and/or global navigation. But unless Google reworks its algorithms to effectively encourage companies to pursue this approach, I fear that Google will only encourage companies to do nothing at all. There is currently no uniquely labeled Google US search engine. While I have a feeling that Google will ultimately make the necessary changes, I hope this happens sooner than later.

TMs have historically been something that companies developed for their own internal use — to cut translation costs and improve consistency and quality. But last year TM Marketplace, profiled in the February 2007 issue of Global by Design, developed a business model of brokering TMs so that companies could lease the TMs of other companies within their industry to get even greater cost savings; GM became the first company to sign on.

And now TAUS is moving ahead with TM sharing. What makes the TAUS effort important is the makeup of the TAUS membership; this organization includes a strong blend of vendors and their clients, such as Adobe, Microsoft, Cisco, eBay, EMC, HP, Google, Idiom, Lionbridge, Language Weaver, and SDL.

As a sidenote, TM Marketplace is also a TAUS member; I’ll be curious to see if the TAUS effort is complementary or competitive to what TM Marketplace is already doing.

Execution is everything of course. And this will take time; TAUS doesn’t plan to review the business model until their next meeting in October.

But what’s most important now is intent — companies intending to share their TMs to translate more content more cost effectively. And as these great pools of TM content grow, we’ll see standardization of terminology throughout industries and we’ll see machine translation software leverage these TMs to enhance their efforts.

Here are some data points from the press release:

• Logoport now has more than 10,000 individual users.
• Logoport has seen average monthly production volume increase from 9 million words in January 2006 to 36 million in January 2007.
• Logoport now hosts more than two and a half billion words within its many client translation memories.

There is a caveat to these growth numbers: the fact that Freeway/Logoport, is, well, free.

Nonetheless, these are very impressive numbers. Clearly, Lionbridge is providing a valuable service to clients and one that is being used at a blistering pace.

I would be interested in knowing how many of these 150 clients have been brought onboard solely due to Freeway/Logoport.

And now that the company has all of these clients using Freeway/Logoport, it will be also interesting to see if a machine translation component is added at some point this year and/or if we see companies beginning to share their translation memories to achieve even greater cost savings.

PS: I profiled Lionbridge Freeway in the July issue of Global by Design.

UPDATE: Kevin Bolen, CMO of Lionbridge, filled me in on two points that I raised in this post. First, the issue about using two names. Logoport is the name of the translation memory software platform that Lionbridge purchased and has since integrated into Freeway. Because there was already a base of Logoport users out there, Lionbridge decided to maintain the name while at the same time promoting the new Freeway name. However, looking ahead, it will be Freeway getting promoted and not Logoport. That should clear things up for prospective customers.

Second, regarding my question about how many of these 150 companies are existing customers vs. new customers, Kevin says the split is roughly 50/50, which is a positive sign. He says that Freeway is playing a key role in winning new business.

I’ve posted the first few minutes of the interview. Click on the image below to view it using YouTube.


Subscribers of Global by Design can view (and download) the entire interview by logging in here.

Baidu vs. Google: Round Two

The World According to Wikipedia

Just Don’t Do It: The Art of Slogan Translation

China and Japan: So Close But Yet So Far

McDonald’s Scores with “McRice Burger”

And although this entry is quite new, it’s been very active:

Statistical Machine Translation Gets Real: A Profile of Language Weaver

Our guest articles have also been very well read. Here are the top three:

Scaling a Great Wall: Top 5 Tips for Learning Chinese

Just How Literal Do You Want That Translation?

The Good, The Bad and The Ugly

Finally, in the completely unrelated category, the most popular entry is my post from Punta Tombo:

Greetings from Punta Tombo!

Must be all those cute penguin photos!

I profiled this company for the December issue of Global by Design and now we’re making the full article available for free download.

We’re going to be hearing a lot more about this technology and Language Weaver because companies can only afford to translate a fraction of their content using translators. SMT is not designed to put translators out of work but instead unleash vast amounts of content that would never have been translated in the first place.

If you want to get an idea of what SMT can do, what it can’t do, and why I think it’s going to revolutionize the translation industry, check out the article.

download article (2.1MB)

Slowly but surely we’re seeing companies make limited use of MT to translate text that would have been cost-prohibitive to translate manually. SDL recently launched such a service.

I’m not suggesting that MT is anywhere near replacing human translators. In fact, the better way to look at MT software is like desktop publishing software – it did not replace designers, but it did give them the tools to help them do their jobs more effeciently.

Here is the full press release:

LANGUAGE WEAVER RAISES $4 MILLION IN SERIES B FUNDING -All major Series A investors reinvest in round led by Palisades Ventures -

LOS ANGELES – Oct. 6, 2004 – Language Weaver, Inc., a Los Angeles-based software company developing statistical machine translation software (SMTS), today announced it has completed the initial closing of its Series B funding round of $4 million. The round was led by Palisades Ventures of Los Angeles.

Palisades Ventures is a venture capital fund making growth stage investments in systems, services, software, and hardware companies driving the adoption of leading edge information technology, communications, and media technologies. Paul D’Addario, senior managing director of Palisades Ventures, will take a seat on the Language Weaver board of directors. “Palisades Ventures invests in companies with strong technologies, proven market traction, and where we can leverage our many strategic relationships in the IT, communications and media sectors for the benefit of our portfolio companies,” D’Addario said. “Language Weaver’s translation technology has immense commercial potential. The company will be well supported by Palisades Ventures with introductions to potential customers and strategic partners as it positions itself in the commercial markets.”

Palisades Ventures first became aware of Language Weaver at LARTA Institute’s Los Angeles Venture Forum in April 2004 where Language Weaver was chosen as start-up company of the year. In addition to Palisades Ventures, all of Language Weaver’s major Series A investors have also chosen to invest in the Series B round, including Sulphur Creek Ventures and Athenaeum Fund.

According to Bryce Benjamin, CEO of Language Weaver, the money will be used to support the company’s entrance into commercial markets. “The obvious first mover for Language Weaver’s technology was the government, where we have now established several contracts and excellent relationships,” Benjamin said. “With this new investment we will grow our team and expand our marketing and sales efforts into the international and domestic commercial markets, which are hundreds of times larger than the government markets and where the need is even greater for high quality language translation.”