Web Site Localization

Many years of know-how: Translation Network (TTN) was launched in 1988 by the software company Extran Ltd. Translations and software have proven to be our combination for success for over a decade. In 1993 TTN was awarded the First Prize by Credit Suisse and Computer 93. Thereafter the TTN team supported numerous software and telecommunications projects. Since 1995 Translation Network has had a number of Internet servers of its own and has gained considerable experience in translating Web sites during the Internet boom.

Problem analysis: The biggest problem when translating dynamic Web pages is that the program code and the texts are mixed up in the same file. While so-called tags can be easily located in pure HTML pages, this is much more difficult in ASP or CGI-generated pages, because text sequences do not have a fixed pattern. The trouble with ASP, CGI or HTML pages is that each program upgrade requires a new translation. Since translators’ fees are lower than those of programmers, after an upgrade the language rather than the program code is changed. Replacing the language is also safer than changing the code: When a translator makes a spelling or other type of mistake, it can be seen on the screen and easily corrected. However, when a programmer makes a typing mistake it is impossible to detect it immediately, but sooner or later the program makes a potentially dangerous mistake. Locating the error can be extremely costly in terms of time and money.

We eat bananas: Most Internet service providers are banana companies, in the sense that they let their products ripen on the customer’s computer monitor. This means that the English version is constantly being revised and that improved pages must be immediately translated into all other languages. Translators have to choose a particular system to make such changes fast and at a minimum cost. There are three methods to translate Web pages.

Database method: A program developed by TTN parses the Web pages and stores the text sequences in an indexed database. So-called duplicates are automatically eliminated to ensure that the translation always uses the same text sequences. The contents of the database are then exported into an MS Word file, which is then sent to the translator. The translation is then checked and reimported into the database. The program then parses the Web pages and replaces the original text by the translation.

Trados™ TagEditor: The translator opens the HTML pages with the Trados™ TagEditor. This special editor can only be used to change text sequences but not HTML code. Translated sequences are stored in the Workbench translation memory. For matching terms or translations of a version upgrade, the existing sentence is loaded from the memory into the HTML file. Thus the second version of a file is translated practically automatically.

Microsoft FrontPage™: Setting up a database or translation memory is not worth it in the case of small projects. Such files are opened directly with Microsoft FrontPage or another HTML editor. The original text is overwritten without changing the formatting.

All three methods have their advantages and disadvantages. Please tell us your Web page’s URL. We will be glad to share our expertise with you.

Job automation: For major projects we connect the Internet server directories to our production server via FTP or tunneling protocol. A synchronization program then compares the status of the projects at regular intervals. If the program finds new or modified files we immediately translate all updated texts into all languages desired by the client. Before resending them to the productive server we test the adapted files on our own server or on your beta server.