Closing the Translation Loop

A long, long, it-feels-like-forever, time ago, Anna covered the story that there were a lack of good translations tools available on the internet, and especially in languages like Russian. In that article, she linked to an article that showed Google had developed a much better translations service, but hadn’t released it to the general public. Well, I’m always interested in closing loops, and so I’m happy to report that Google just announced that their, much improved, translation service is now live at Google Translates.

What makes Google’s service interesting is that it doesn’t use the standard “rule-based” translation methods, but rather, it is much more statistical in nature:

We feed the computer with billions of words of text, both monolingual text in the target language, and aligned text consisting of examples of human translations between the languages. We then apply statistical learning techniques to build a translation model.

Anna and I couldn’t help but to try the service out, so we choose a Russian news site: and the results were pretty impressive ( in English) as I could definitely pick up almost every story.


My take is that the translation of websites is only the beginning for Google… Not only will they improve the translations (especially if enough users adopt their feature to “recommend” a better translation directly through the tool), but I can easily see a day in the not-too-distant future when this tool is built right into gmail or google chat. This could make it extremely easy to have an online conversation with someone who doesn’t even speak your language.

In my family this could be very useful as my wife’s parents do not speak much English (and I don’t speak much of their native Russian). At this point that means that my wife has to translate everything that is said between us. What if instead, I could simple email them (in English), but it would arrive with an accurate translation in Russian? The result is that it would actually be easier for us to share stories over email or chat then in person!

From a business perspective, the possibilities are fascinating… Why not translate an entire website (Welcome to “Дождь город

Добро пожаловать на путеводитель Си

I had a blast yesterday hanging out with the Pat, Jeff and the Sellsius gang in Koreatown earlier this week. We turned what I expected to be a quick lunch into a 2 1/2 hour conversation that never slowed down! (not only can those bloggers write, but they can talk! 🙂 ).

Jeff to a quick video that some might appreciate:

So, What’s with the title?
Pat has been getting so much traffic from Japanese visitors thanks to a little bit of Japanese text on his page that we all decided to test out a different language… I choose the text “Welcome to Seattle’s Rain City Real Estate Guide” in Russian since my wife can do the translating! 🙂

No entiendo su lengua (I don’t understand your language)

[photopress:Flags.jpg,full,alignleft]One of the things I don’t understand is the near total absence of non-English real estate web sites. Maybe the industry is too distracted by Web 2.0 and other battles in front of them to notice & exploit this HUGE opportunity. Perhaps I’m looking in the wrong places. However, I believe if Google has difficulty finding what I’m looking for, it’s probably hard to find.

Having spent many years at Microsoft, I learned the value of software localization. You may be surprised to learn the Microsoft makes over 40% of it’s revenues from outside the United States. Frankly, you haven’t lived as a software engineer, until you’ve seen your code running in Japanese and Hebrew. (In case your curious, Hebrew & Arabic is much harder to deal with than East Asian languages). Another data point is that my bank (Washington Mutual) has ATMs that speak English, Spanish, Chinese and Russian and offers offer free homebuyer education workshops in Spanish.

Although, all real estate is local, not all real estate consumers are locals. According to the 1990 & 2000 US Census data, the percentage of people in the US who don’t speak English fluently is growing. Over 10% of our population only speaks Spanish. The number of Asian only language speakers in the US is growing at rate similar to (though smaller in number) to the Hispanic population.

For example, John L Scott, Windermere, & Coldwell Banker Bain, all let you search for a bilingual agent. Unfortunately, that feature of their web sites is only useful if you know enough English to understand you can search for bilingual agent! If you don’t understand English, how are you going to be able to find that feature to begin with? There’s not a single word of non-English content to be found on their web sites. That major shortcoming aside, I have to give them credit for at least trying to make things easier for the non-English speaking public.

Fortunately, I’m not the the only one who thinks this way. There are industry advocacy groups such as the National Association of Hispanic Real Estate Professionals and Asian Real Estate Association of America that are trying to make things easier for Spanish & East Asian speaking consumers & real estate professionals. A Latino Advertising & Marketing blog entry, states (surprise surprise) that bilingual real estate agents are in demand with Latino home buyers. And if you or your web site doesn’t speak their language, ColConnect is a company that can help you, since they focus on bilingual web design & marketing for real estate & mortgage professionals. An example of their client web sites can be found in English and in Español.

OK, so your thinking, how can I take advantage of this? Here’s what I’d do…

If your bi-lingual or multi-lingual advertise that fact. Join an industry advocacy group such as NAHREP or AREAA. Make your web site multi-lingual. Have one side of your business card in English and the other in your second language. Exploit your linguistic superiority!

If your not bi-lingual, make friends with somebody who is. (Maybe you can hire them as translators?) Link to relevant foreign language real estate content. You should also make your content machine translatable. I don’t speak Spanish, Chinese or Russian, but I know computers that do (which is the next best thing to knowing people who do). Read old Rain City Guide postings and get creative.

If you speak any language, (foreign or domestic) talk with the MLS or your broker and see if there’s a way to change to MLS schema so that it’s contains remarks in Spanish and other foreign languages. As along as they are in the database, perhaps they can increase the size of the remarks field, so agents won’t be compelled to use abbreviations anymore (and make life easier for software translators). Perhaps, the MLS can add $25 to the cost of a listing to cover the cost of a human translating the remarks section into 4 or 5 different languages? After all, why should the listing agent or seller care what language the buyer speaks, as long their check doesn’t bounce? Anyway, I suspect MLSes are limited by their software vendors, and if enough professionals demand these features from their MLS, it might just happen someday. Remember, it wasn’t that long ago, that sold data was a pipe dream too.

Make sure you when you develop your web site’s content or enter listings into your MLS, you use complete sentences with proper spelling, simple languages, and no abbreviations. Then visit a site like AltaVista’s Babel Fish, World Lingo or Google’s Translate and verify that your translated test looks OK. You may not understand the translated page, but if your translated page comes back with a lot of English words & abbreviations, you’ll know you need to revise it so the computer will do a better job of translating it. Translate the text twice (English to foreign language and back to English) and if the text comes back funny get a thesaurus and try to pick words that the computer less likely to get confused with. Even though computers don’t translate human languages nearly as well as humans do, they are better than nothing at all, (they are also much cheaper than people and getting smarter every day).

I’m shocked by the total lack of multi-language MLS search tools and foreign language realtor web sites. I would’ve thought that large brokers would’ve smelled the money by now and pursued this opportunity more aggressively. Regardless, I’m planning on getting Zearch to speak a second & third language by the end of the year and I suspect that others will follow my lead since the demand for non-English real estate services is only going to grow. After all, money is the universal language that everybody understands.