Internationalize, Don't Localize

Article Featured Image

Spoken and written languages have randomly evolved out of the human imagination. Language continues to transform to include the ideas and emotions of millions of people. For computer programmers, this morphing  is considered illogical and chaotic. Computer applications and software are built on logic, organization, true assumptions, and process management. This battle between unsystematic human language and controlled computer programming causes companies to either forgo translating and personalizing their applications due to high costs and a lack of time or dehumanize them as much as possible.

Localization derives its name from the fact that translated systems are  custom designed for a single language or locale. A typical localization team consists of developers and translators who ensure the applicability of the language solution to the targeted end user. Because localization is customized at the coding or software development level, determining what should appear, play, display, or be recognized by the interface in other languages takes upfront time and resources.

Additional costs for regression testing of previous languages to ensure the new coding does not adversely affect the previous work can also be expected. After localization, the product must be recoded, data tables rearranged line by line, and grammar codes revamped almost in their entirety to achieve acceptable output and input in the new languages.

Downside of Content Management
For response playback to a caller, the popular assumption is that expensive content management software is a panacea for localization pains. However, localization breaks down when a system incorporates variable information.  Languages and cultures have differing sentence structures, and traditional content management software, lacking intrinsic linguistic power, does not efficiently handle these language complexities alone. It, therefore, dumps responsibility back on the developer, who probably does not speak or read the new language.

Since each language and dialect has inherent nuances that evolve without rhyme or reason, workloads increase during localization because of the amount of time needed to consult with native speakers, determine linguistic rules, modify the output, and test the new localized version multiple times. Most important, both the new and the existing language versions must be constantly retested to ensure that they have not been damaged during the localization process.
Internationalization, on the other hand, means creating an application that is prepared to handle numerous languages, without coding changes of any kind, except to incorporate new recognition engines and peripheral wares. Internationalization describes the process of structuring internal software workflows to more easily translate a software application into other languages and dialects. Once the system or technology has been internationalized by the software developer, the number of languages that the system can accommodate—irrespective of how many variables are inserted while the application is operating and how many interfaces or media are in simultaneous operation—is limited only by the return on investment and available engines. And the ROI is suddenly greater when the development time for new languages is eliminated entirely.

Thus, the differences in the two language techniques are reflected in the workload, cost, and quality, with internationalization presenting the clear advantage over localization. With systems that are not internationalized, each new language or personalization must be treated separately. Internationalization achieves a state of the original product in the original language that offers complete and total flexibility to add new languages, no matter how complex the application. In short,  software development is performed once and never again. In localization, however, development must be performed over and over again, for each language, dialect, variable,  interface, and type of media.

Internationalization opens new horizons to any technology. For audio playback and screen text, new languages can be added without touching the product at all. If speech recognition grammars are internationalized, only certain blocks need replacement. Thus, the product adaptation consists mainly of connecting it to a new language engine. Internationalization reduces the need for regression testing, as well as the fears and much of the expense, associated with any major reworking of a product’s code.

Sue Ellen Reager is CEO and founder of @International Services, a global translation services company and developer of localization software. She can be reached at sueellen@internationalservices.com.

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues