Using Translation Tools For Search

Years ago an SEO newsletter suggested that if you want to internationalize your content all you had to do was embed some clever links into your HTML code and — using Altavista’s Babelfish — you could instantaneously offer your content in as many languages as Babelfish understood.

I looked at my Web site statistics and noticed I was getting a lot of traffic from five countries in Europe so I immediately embedded the links on every page of Xenite.Org.

Back in those days I had 50,000 static pages of content on Xenite and let me tell you, it’s no picnic trying to update that much content. The project took about a month, as I recall.

Within a year I received numerous emails from mostly Deutsch-speaking visitors complaining about the nonsense I was publishing in their language. I attempted to explain by email that I had no control over the translations but after like a half dozen complaints I finally realized that explaining things was just adding another problem to the original problem.

I should have realized that if bi-lingual people are visiting my content they probably are comfortable enough with written English that they are happy enough with that. It’s not like you’re going to be surprised when you visit a Dutch Web site and don’t find it written in Hungarian, right?

I took down the translation links (another project that lasted a long time), some of which lingered on deep content pages for several years as I became involved in other projects and priorities. Those links were long an embarrassment to me (as were the obsolete “revisit-after” meta tags that I seldom bothered to clean up long after they stopped being useful with 1 search engine).

That experience left a sour taste in my mouth for automated translation. Nonetheless, I continued to use automated translations to help me communicate on an as-needed basis with people in other languages. I always wrote my comments in short sentences, included a disclaimer about using an automated tool, and sent both English and translated language.

To date I have no idea of whether I said “the cows are all dying” to someone in another language or not.

When the time came for me to seek translation help for the Parma Endorion eBook I used Babelfish to compose a really nice letter to Tolkien fans in Spain. I was pretty careful to review the translated sentences one by one (as I have a little knowledge of Spanish) but I forgot that “fan” has more than one meaning. Babelfish helped me ask Tolkien ventillators for help in translating my book — and when they stopped laughing they wrote back and said, “We get the point and we’ll be glad to help you.”

Automated translation is not efficient but neither can it be ignored. For example, there is a Wordpress plug-in that relies on Google Translate to convert your text into other languages. I found myself discussing the Wordpress plug-in with some poker affiliate Webmasters a while back.

They are not really fans of automated translation, either. They don’t want to be caught offering gibberish to their non-English visitors and they are to be commended for maintaining a high standard of integrity for their copy.

Nonetheless, my team and I have set up some translation tests and we’re still evaluating the feasibility of using autotranslation for search optimization. I don’t expect much from the experiment but I came across an article that reminded me of autotranslation and search.

You see, the Wordpress plug-in creates dynamic content that is indexable by search engines. Microsoft’s new translation gadget, while embeddable in your pages, uses AJAX to provide the translation. NOTE: I’m just going by what the reviewer says, I have not used the gadget myself.

I don’t believe this will be a search-friendly solution, which would be a shame if the translations are better than Google’s. While I’ve heard few complaints about Google Translate, it’s still not as good as a professional translator should be.

The Google translation system does solicit input so people can help improve the service (perhaps in a Bayesian way, or perhaps there is a human reviewer receiving the submissions — I don’t know). I haven’t used Microsoft’s translation tools very much so I’m at a loss for where Microsoft fits into the big picture.

But it seems to me that sacrificing searchability for usability may play against Microsoft in the future. People want to embed translation tools in their content and I am pretty sure they would like to be found in many different languages as well.

Google’s Universal Search algorithm, besides injecting on- and off-topic news stories into Web search, is also supposed to help people in content-poor languages find relevant documents in other languages. If you cannot find a detailed discussion of Symbolic Component Analysis in Hungarian, Google will try to find one for you in English, French, German, or Spanish (the four most content-rich languages) or any other language that Google has extended its semantic knowledge tools into.

But the real challenge facing Webmasters is not whether you can make your documents available in other language search — it’s whether you can communicate effectively to people in other languages. We have few options available to us. Here are the translation options I’ve considered through the years.

Use A Translator – I’ve done this more than once. If you can find someone to translate your copy for free, great. Otherwise it’s an expensive proposition. And regardless of whether you pay for translation or not you have no way of knowing how good the translation will be. I’ve received critical emails from people who felt one translation or another did not do justice to my work. One of my translators, working in Hebrew, went to great lengths to ask me about obscure idiom I employed in my essays. His work was, I feel, exceptional. I’ve never had a complaint about it.

Pros of Using a Translator: Better accuracy, better control over how your meaning is conveyed in the other language.

Cons of Using a Translator: Can be expensive, quality control is difficult to manage.

Use Babelfish links – I don’t know whether Babelfish has been updated in the past 8-10 years or if it’s just orphaned technology. It would be nice to know that Yahoo! (well, Systran) has been working on it but I generally hear more positive comments about Google Translate.

Use Google Translate links – It’s newer than Babelfish and enjoys the benefit of the Google brand, but Google Translate can sometimes mangle your meaning as well as any translation tool.

Pros of Using Babelfish/Google Translate links: You can easily implement them through the Wordpress Global Translator, which offers a choice of four translation engines. Most CMS applications should be able to add translation links through template code (if you don’t use Wordpress as your CMS).

Cons of Using Babelfish/Google Translate links: You are stuck with whatever comes out of the translation tool. You have no idea of what you’re saying to people in other languages. Assume your attempts to say “I love Obama” all translate into “The cows are all dying”.

DIY Translation – You can buy translation software, multilanguage dictionaries, use online tools to translate one sentence at a time, etc. Do-it-yourself translation is the poor man’s method (even more poor than using autotranslation). You’re not likely going to make a living by doing this.

Pros of DIY Translation: You may be able to control quality in a limited fashion by asking more knowledgeable friends and acquaintances for occasional help and review. It only costs you time.

Cons of DIY Translation: Your needs may be far more ambitious than your available time and resources. It may cost you more to do it yourself (in opportunity cost) than to pay someone else to do it for you. If you have few resources to work with, you’ll almost certainly end up saying “The cows are all dying” in 30 languages or more.

Add Multilanguage Notices To Your Content – If you’re being visited by people from Ukraine, maybe the smart thing to do is post a simple message in Russian or Ukraine that says, “This text is written in English. You may try using these tools to translate it.” Then link to some online translators.

Pros of Using Notices: It takes less time to put a notice on your page that makes sense than to translate the entire page. Chances are that foreign-language visitors probably want to read English anyway. Your foreign-language notice could be expanded enough to include some keywords (but you’re probably better off just letting foreign-language search lead to the natural English keywords).

Cons of Using Notices: You’re giving up on translation. You have to limit yourself only to the market that speaks your own language. You may not be able to reach the largest market for your goods and services.

In Conclusion
Translation tools offer us some hope of communicating across language barriers but they have a long way to go. And, frankly, they don’t help in the least with languages that have strong regional dialectal differences. People on the Spanish Web have told me that idiom is diverse across the content and it’s sometimes hard for someone from Spain, say, to understand copy written in Guatemala.

Even in the English-speaking world it’s occasionally a challenge for Americans to interpret pages written in the United Kingdom (if they use a lot of slang, refer to a lot of local geography and history, etc.) or Pakistan and India. No one has yet attempted to write a serious dialect-to-dialect translator, but I think it would make sense to do that somewhere in our evolution as a multilanguage community.

Such translations would help improve same-language Web search results, as well as the visitor experience. You really don’t want people laughing at you because you just called them Tolkien ventillators.