The Translation Process

The essential steps in the translation process are not particularly surprising:

1.	The developers package up the resources and ship them to the translator/localizer.
2.	The translator/localizer updates the resources and returns them to the developers.
3.	The developers reintegrate the resources into the application source.

Though these steps are easily recognizable, it is worth highlighting a couple points. First, and most important, there is a time delay between when the resources are shipped to the translator/localizer and when those resources are returned to development for reintegration into the application. During this time, the development of the application will continue, and it is unrealistic to treat the resources as "locked" during this period. Consequently, you should consider that development maintains the "master" copy of resources and that the resources that are returned from the translator/localizer must be reintegrated into the master copy (as opposed to replacing the master copy). We return to this subject in the "Reintegrating Resources" section.

Second, it should be recognized that translation/localization is a part of the development process. This means that if development is an iterative process, translation/localization is likely to be an iterative process, too. Translation/localization might not follow the same iterations as development, but it is highly likely that it will be performed in iterative steps. So whereas you might strive to complete a considerable part of the application before translation/localization occurs, and even use pseudo translation to catch as many translation/localization errors as possible before engaging the translator/localizer, the application is unlikely to be frozen after translation/localization has occurred. Instead, as each bug is fixed and as each form is modified, new text is used. Your translation/localization process could well consist of a large translation/localization phase, but it will be followed up with smaller iterative phases. This means that your translator/localizer needs a way of finding all the new text and resources to translate. Without this, the translator/localizer will need to manually search the application for new resources to translate, and this is a labor-intensive and error-prone approach.