White Paper: An Introduction to Translation and Localization for the Busy Executive (cont.)

Localization Technology and Process

There are several types of technologies or tools that can aid the process of translation and localization. Machine Translation (MT) is what most people think of when it comes to computers and translation. Its origins date back to military intelligence during the cold war era and it is software that programmatically converts one language into another. Although the quality of MT-generated translation has improved over the years, the successful use of it for localization has been limited to only a small group of companies. These companies have extremely large document translation requirements and a controlled authoring environment that restricts the way original content is written to make it easier to translate. For most others, MT is not a cost-effective near-term solution.

Computer-assisted translation (known as Translation Memory) is a technology for applying automation to the act of human translation. It is technology that parses and saves every translated sentence to a database in real time, allowing the translator to reuse previously translated material and focus only on the parts that are new or changed. This Translation Memory technology usually includes Terminology Management for creating and reusing translated terms. Translation Memory is especially helpful for localization since software and websites tend to have a significant amount of repetitive text and are usually revision oriented with the ability to re-use translations from previous versions. Any localization solution you consider should use commercially available Translation Memory tools to ensure portability of these assets.

The localization process

A typical localization project ranges anywhere from one to four months depending on the size and complexity of the project. Since a translator normally translates about 2,000 words per day, it is not unusual to see a larger project require up to 20 translators and four editors or more per language working in parallel. And that is often the most straightforward part of the project — usually about one half of the overall effort. Localization of a software or web application has about 20 process stages to it beginning with engineering and glossary development tasks and ending with linguistic and localized UI testing. These are often proceeding in synch with the development effort so that the localized versions can be launched with the English.

Historically, localization has always been project oriented with a beginning and an end. And for the most part, that is still the way it is today including the first-time localization of a website. However, maintaining localized websites has brought about a new approach. Maintaining websites, especially those implemented using a Content Management System (CMS), need to be handled in a much more dynamic and granular fashion, e.g., frequent webpage changes vs. a new software version. To address this, several companies have brought Globalization Management Systems to market that interface with the CMS for the purposes of detecting change and managing the localization process as a workflow “stream” rather than a project.

< Previous | 1 | 2 | 3 | 4 | Next >