Often localization engineers, the people who prepare strings for translation, work for a software company that then outsources the translation to an agency. The localization engineer in company A talks with a project manager in agency B, who, in turn, deals with various freelancers (or even single-language vendors). The chain of intermediaries can quickly get too long. Answers to questions sent up the chain take forever, and suggestions or remarks by translators are lost on their way to the client.
Localization engineers usually have great technical skills. However, since they don’t translate themselves, they sometimes don't know what would help translators and what would hinder them.
Localization engineers (and project managers) could help translators improve the quality of the localized software by a few simple actions:
- Provide an easy way for translators to send their queries and remarks;
- Provide explanations and definitions;
- Indicate how much space is available for each translated string;
- Provide a glossary of required and forbidden terminology;
- Give translators the whole translation memory (not only an extract): a complete translation memory allows translators to perform context searches, and helps ensure translation is consistent;
- Provide screenshots, or make resources available to see the strings in context;
- Do not provide the strings in alphabetical order (as I often see): that may look more orderly, but also deprives translators of much needed context.
This is just a partial list: a few thoughts that came to mind while working on a badly organized project, a few weeks ago.




