Top Five Software Localization Myths

By Nabil Freij

We will shed some light on the top five myths that are believed by many executives and decision makers in the corporate world.

Myth 1: Software Localization is simply translation

Many executives do not realize the level of effort involved in software localization, dismissing it as simply translation.

For many years now, advances in internet technologies, development tools, authoring tools and platforms, have expanded the use of different file formats and build environments. Software applications and manuals are no longer based only on Microsoft resource files or Word documents. Java, XML, ASP, .NET, RESX, HTML, as well as many other formats, have become standard in many applications and products.

Furthermore, with the continued trend toward the convergence of technologies, companies are marrying software, hardware, mechanics, chemistry, biology and other sciences to develop high-end solutions.

Translators are now expected to understand all these different technologies and file formats, and accurately translate only what is needed, without modifying tags, links or code. If errors are made, a significant amount of debugging time is needed to fix and build the international products.

Localization is both an art and a science. Do not underestimate the efforts needed or short-circuit the process. It takes experienced engineering and translation professionals to properly implement an efficient translation-reuse process and localize your product.

Myth 2: Anyone who knows a second language can perform translation tasks

Would you hire anyone who speaks English to be your Tech Pubs writer, or anyone who knows a computer language to be your programmer? Translators are professionals with years of schooling and translation experience. They earn their living doing translations. Most live in the countries that they are translating for and are natives of the language they translate into. They have an excellent command of the languages they translate from and into to ensure consistent, accurate and timely work. Recruiting amateurs to do translation work, even if they know your product or technology well, will lead to inferior results and product delays.

Myth 3: Lower per-word translation rates reduce costs

In localization, it is often the long-term costs that matter the most. Software, help, docs and other texts related to products are constantly changing. With each product release, the localized material needs to be updated and synchronized with the source. Lower upfront translation costs do not necessarily mean lower long-term costs. The following are key factors that contribute to long-term costs:

1. Process: Is the latest and most efficient translation-reuse process being implemented? If Translation Memory (search-engine and database) tools are not used, updates will be very time consuming and costly.

2. Maintenance costs: Vendors who have lower translation rates may have steep penalties built into fuzzy matches (similar but non-identical matches), repeats and 100% matches. This creates steep overhead costs each time a new revision of your product needs updating.

3. Quality: Although low quality translation will have lower initial costs, the long-term costs are significant. Post-translation changes are very costly, particularly if you have incurred production costs for layout, desktop publishing, quality-assurance, duplication or printing.

4. Ownership: Do you own your translation memories and databases? If you pay for the work, you should own it all.

Myth 4: A language review cycle is not necessary

To the untrained eye, a translated text appears final regardless of the quality or state it is in. Just as you see the value in having your source files edited by a second writer, or your software code reviewed by a second developer, the translated text should also be fully checked by a second translator. In the case of translation, the editing cycle will require not only reading through the translated text, but also verifying it against the source. Many vendors with lower rates, or higher overhead, will cut corners on editing in an effort to turn a profit. They may not ask a second translator to edit the text and instead perform “cursory checks”, which only require the editor to quickly read through the translated text without ensuring that it accurately represents the source. Always ask your translators or translation vendor what level of editing they perform after translation.

Myth 5: The vendor that provides the best translation sample offers the best quality

Asking a localization vendor to provide a translation sample is often mistakenly accepted as a lead method to measure vendors’ quality standards. Although in theory the concept makes sense, in reality, it is far from optimal. First, there are a lot more tasks involved in localization, than simply translation. Second, translation samples are often done by the most qualified translators who may or may not participate in the actual translation, once your project is awarded. If a sample is requested, you need to make sure that:

1. The vendor knows how to manage, parse, prepare, reuse, compile, desktop-publish and QA the required files.

2. The translator translating the sample will be the lead translator on the project.

Since this cannot always be guaranteed due to scheduling or other factors, it is often more reliable to check vendors’ references, experience, reputation and track record.