Presenting the multilanguage content over the Internet - practical issues

In the current global environment most organizations (commercial, govermental, nonprofit) have the need to present their Web sites and their content in multiple languages. The following short paper presents a few practical experiences related to this topic, which our company gathered in the course of work with our customers.

Benefits

For a commercial company having a multilingual Web site presents two main benefits. Obviously it allows presentation of the company to customers and partners in many countries. But most importantly the multilingual Web site is a very effective tool for finding new customers. It is becoming rapidly routine for people to search for potential suppliers over the Internet. In the experience of our customers between 60% and 80% of traffic on their Web site is generated by searches for their products, using the Internet search engines. Such searches will lead the customers to the Web site only if the Web site presents the content in the language of the customer search.

Logistical issue

Maintainig a medium sized Web site in a single language used to involve a cumbersome process of preparing a content by authorised people and then writing by a technically skilled personel a suitable HTML code displaying this content on the Web site. Since this process involved 2 (groups of) people it was relatively costly, prone to delays, errors in communication etc. The advent of various authoring and - recently - content management tools made it possible to reduce the above 2 stage process to just one stage - direct editing of the Web site by the people responsible for the content.

Obviously this is a desired outcome - to have the people responsible for the content have a direct, immediate control over the information presented.

We omit here the (non-trivial) issues of multistage authorization of the content changes and the issue of restricted domains of authority over various parts of the Web site. These issues are resolved by a decent content management system, but they do not necessarily arise for Web sites of smaller organizations.

When the Web site content starts to be multilingual, another step in the content production process arrives: translation. Even for a small site, having to manually handle the submission of text for translation, receive the translated versions back and upload the translations to the Web site is a tedious and slow process. As the content of the Web site has to or should be updated/expanded often, the delays/complexity introduced by the transtations become a serious problem.

In the view of the above our company suggests that the owners of multilingual Web sites consider some sort of content management system which also organizes, speeds up and to the fullest extent automates the translation process.

Content consistency issue

For a multilingual Web site an often encountered problem is the issue of structure similarity (or differences) between various language version. For many site owners ideally any language version of their Web site should have the same structure as any other. This means that any language version should have the same amount of Web pages, cross-linked in the same way, with the same navigation systems (buttons, menus) as any other language version. Additionally, the corresponding Web pages in different lanuages should be cross-linked to each other. The above requirement means that any extension/update to the Web site must be done simultaneously in all languages the site uses.

This is often not practical. Most often the content is created in one 'basic' language, than translated (possibly using computer-assisted procedures). In most cases the translation process takes time, delaying the publication of all language versions of the update, including the 'basic language' version. This delay is not acceptable in many situations.

One alternative to accepting the delays of the translation process (other options are listed later) is to alow different language versions of the Web site to have different content and structure. While this is one of approaches our company reccommends, some potential dangers should be briefly examined. Obviously one cannot blindly translate and publish in 'foreign' languages random pages from the 'base language' version of the site. Most pages contain links, buttons etc. leading to other pages. It is unacceptable to translate just on page and have it contain 'dead' links and buttons to pages which were present in the 'basic language' version, but are not yet translated (and published).

Our company, when doing a review of many multilanguage Web sites, has found two equally harmfull manifestations of the above problem. Some sites just leave the links leading to not-yet-translated pages 'hanging' ('dead'). This leads to a lot of user frustrations, when the user clicks on a link and is confronted with a 'does not exist' error page. Some other sites make the links which should point to not-yet-translated pages point to some other language, already translated version. In most cases this results in bewildering the user by making her jump between pages written in many, not necessarily known to her languages.

This issue of 'dead links' can be resolved using 2 techniques. For one, one should allow the multilanguage content management system to control generation of the links. To put it smply, it is the multilanguage content management system which will automatically put a link or a button on a Web page ONLY WHEN the page to which the link/button points to gets translated and published. This way, the Web site will never contain any 'dead' links.

Using above technique, one could manage the multilanguage Web site simply by writing the content in the 'base' language, mark pages (not necessarily all of them) for translation and let the multilanguage content management system handle the translation process and publish the translated text automatically as it becomes available.

The drawback of using exclusively the above approach is that the Web site has some structure and the 'foreign' language version will not be useful if only random pages get translated. Here our company suggests a combined solution. First, the Web site should be designed in such a way as to allow partial content translation - with reduced or simplified structure but maintaining the consistency. Our practice shows that it is rather easy to design. Second, the content management system can track the logical dependencies between pages and only allow partial publishing of the foreign language version when the resulting partial version will maintain a consistent structure.

The distributed human resources issue

A common situation for a small or medium business is that the headquarters does not have the human resources to manage the company foreign Web site version and this task is delegated to a (foreign country) branch. It is therefore necessary for a company Web site content management system to allow access from remote locations. Please note that the system should not really depend on one language being 'the basic language' and all the other language versions being (maybe not complete) the translation of the basic language version. The complete multilingual content management solution should allow entering and publishing of the original content in any language.

The cost issue

For a small business building and maintaining a multilingual Web site can be a significant expense and a draw on human resources. On the other hand, for many specialized small businesses a multilingual Web site is a 'must have' item as it allows the firm to serve a much bigger market.

Here is where the automation of the creation of the foreign language version Web site plays a significant role. Our company developed a content management system which allows to automatically translate and publish multiple language Web site versions. The Web site owner manages only the 'native language' version, the other language versions are created automatically based on the 'native version'. The form and structure of the Web pages allowing a high quality automated translation is restricted, but flexible enough to fully present the Web site owner business offering in fine detail. Of course the automated translation system should allow for some pages to be translated by humans and to allow for some pages to not be translated (different language versions having possibly different structures).

About our company

ORNAK is a technology company based in Warsaw, Poland. We specialize in distributed, multilingual content management solutions and computer network security.