The Online Babble Barrier:Building a multilingual Web site is a complex affair

By Andrew Stroehlein

andrew@iwpr.net

---

This story first appeared in the Online Journalism Review and is reprinted here with permission.

---

Our Webmaster looked stressed; frequent inputs of caffeine and nicotine were required. He'd just been given the task every Webmaster dreads: adding another language to our Web site. And as if that wasn't bad enough, it was a language that isn't supported by any Web browser anywhere. In fact, this would be one of the first Websites, if not the first, of its kind in the world.

Of course, our Webmaster, Srdan Pajic, was not exactly new to this; he had already created our online publications in Albanian, Russian and Serbo-Croatian-Bosnian. But that was child's play: those are languages with characters fully supported by browsers everywhere.

This was new territory altogether. This was Pashto, one of the two official languages of Afghanistan. With the country almost completely cut-off from the Internet until now, software developers' interest in Pashto has been understandably minimal. Sure, there have been thousands of Afghan exiles online for years, but they have tended to use other languages, most often English, for their sites. Online Pashto is effectively in year zero.

On the surface it looked simple. Pashto uses the Arabic alphabet as its base, and Arabic is very well-supported worldwide. The problem comes in the twelve extra characters Pashto adds to the Arabic alphabet. Twelve little characters can cause a lot of headaches for a Webmaster. We checked out the BBC Online to see how they did it, but not even the BBC had a Pashto site yet. We were on our own, and it would be several gallons of espresso and not a few cigarettes before we discovered a solution.

Prepare to be limited

Our recent experience with Pashto at the Institute for War and Peace Reporting highlights only one of the serious challenges of creating a multilingual publication on the Web. Other problems may not be so extreme, but they are serious nonetheless; in fact, they are likely to put you off the whole idea altogether.

The revolutionary promise of online publishing was an inexpensive, instantaneous and international medium. "Global communication," "bringing the world together" -- you know the buzzwords. The practical reality is quite different. In the real world people speak different languages, and study after study shows that people prefer to get their information in their mother tongue.

Of course, this question is not really addressed by most media outlets; very few publications, online or off, are actually international, and only a handful really aim for readers beyond a local or national market. The same is true of their advertisers, not coincidentally. Even in countries that are formally bilingual, both old and new media tend to be mono-lingual.

Those few online publications that do aim for an international audience through a multi-lingual site usually follow a similar line of thinking, a pattern I've seen repeated many times as I've advised these sites.

The course of their thinking is straightforward. At the start there is a dream to reach all people everywhere. After three seconds of thought comes the realisation that they cannot possibly have all languages on the site, and not even all the languages that they would like to have or that would be appropriate for the target audience. I recall advising one site covering the European Union (EU), and despite the heartfelt desire to publish in all 11 major languages of the 15 member states, the resources simply were not there.

Being global will cost you

The stumbling block to the initial optimism is quite simply the cost of translation. Mechanical or automated translation is not adequate yet, and the necessary human translation of scores of pages every day is just financially overwhelming.

At the Institute for War and Peace Reporting, for example, we spend tens of thousands of dollars every year on in-house and freelance translation of our articles into Albanian, Macedonian, Russian and Serbo-Croatian-Bosnian (and now Dari and Pashto). That is serious money for a small charity, and though it is essential for local impact, it is not always easy to explain such costs when fundraising. And even those six languages are a compromise solution; ideally, we would translate our material into every language where we work, but that would mean some 20 languages, at least, and the money just isn't there for it.

Government sites may have the resources to attempt such Herculean tasks: Europa, the EU home site, for example, does actually publish in all 11 major EU languages. But such state-sponsored luxuries are simply out of the question for Websites that live in the real economy.

Always one step behind

Once the cost of human translation is understood, the multi-lingual aim begins to get back down to earth. The thinking moves from "let's reach everyone" to "let's reach as many people in their own language as we can." The goal is then to cover one or two extra languages, the key ones that specifically relate to the target markets and areas of interest. That leads to the next revelation: translation doesn't just cost money; it costs time, and that has important implications for an online publication.

Human translation takes serious time. Remember, translation is about more than language; it's about culture, as well. You need to explain things to people not only in their language but also using the cultural references and patterns they are familiar with. Taking all that into account, professional-quality translation of a 1,000-word article typically takes four or five hours. That is a significant amount of time in the minute-by-minute world of online publishing.

What it means essentially is that a site is not able to be multi-lingual so much as have a primary language and secondary language(s). Your output in the primary language may be quick off the mark, but the material in the secondary language(s) will always be delayed in translation. If one of your publishing goals is to be first with the news, it is almost impossible to do that in your secondary language; a publication with that language as its primary language will almost always beat you to press.

It won't pay but it might be worth it

Don't fall prey to your own optimism about what is seemingly "inexpensive, instantaneous and international." Multi-lingual online publishing is anything but cheap and quick. Yes, the technology allows you to do a bit more: you can, for example, get clever with cookies and database-driven content management systems to offer your publication in the reader's preferred language. But technology does not solve the key problems of multi-lingual publication: translation cost and time.

Expect that, at the end of your journey into multilingual Web publishing, much of your initial enthusiasm will certainly be dampened. You will have limited your palette of languages to two or three for reasons of plain economic necessity, and if you are dealing with time-critical information, you will have resigned yourself to the idea that you will be able compete seriously in only one language market.

Those seem like severe compromises and limitations, and so the question then becomes, simply, "is it worth it?" The answer to that question depends on your reason for being in the online publishing game in the first place.

For a commercial operation, given the costs and given that advertising money generally seeks national or local markets anyway, the answer to the question is most likely "no." In that case, you might be better off seeking partnerships with foreign-language publications with which you can trade occasional articles, each being responsible for the cost of translation into your own language. This is an arrangement we set up at Central Europe Review to great effect with several media partners in Central and Eastern Europe, not all of which were even online.

But if you are less profit-driven and your goal is to reach specific target audiences, then those necessary compromises are just part of the calculated cost of providing your services. The publicly supported BBC Online, for example, simply has as their mission to reach people in many languages. That is their aim, so they have to produce output in a wide variety of languages.

Similarly, the non-profit Institute for War and Peace Reporting prides itself on local impact. For example, reaching Afghans in country is a clear aim, and thus final translation of articles into Pashto is necessary to facilitate republication in nascent local print media. Taking on Pashto had all the regular costs of translation, and solving the problems associated with this un-Web-wise language cost our harried Webmaster more than a few espressos and cigarettes; outside expertise was also required. If we were a commercial operation, it would have been a decision with no financial logic.

The value of our offering Pashto articles online, however, is not counted in terms of dollars, but in terms of readers. Yes, the number of our Pashto readers is still quite small, but the Afghan exile community around the world is reading the weekly output, and as Afghanistan recovers, more and more Internet connections will emerge within the country. As they do, the Pashto-language information will be there for them, and Afghans will feel themselves connected to the world again after being cut off from it for so long.

And that is the real benefit of multi-lingual publishing online.

Andrew Stroehlein is Founder of Central Europe Review, winner of the NetMedia 2000 award for Outstanding Contribution to Online Journalism in Europe. He also worked as Editor for Internet Content and has recently signed on to head the training department at the Institute for War and Peace Reporting.