Own your translation memory
This post is part of a series on the value proposition of localization strategies.
The source content you develop is your intellectual property. The translation of that source content is also your intellectual property, regardless of who performs the translation.
But that translation doesn’t only exist in whatever document or webpage you translated. It also lives in translation memory, and many companies fail to own critical resource.
Failing to maintain a copy of your translation memory can be a costly mistake.
What is translation memory?
Professional translators use computer-assisted translation (CAT) tools to make their work easier. When you send a file for translation, the translator imports your file into their CAT tool. The text is extracted and broken up into short segments (usually by sentences or phrases) for translation.The translated segments are then stored in a database for later use. This database of translated segments is the translation memory.
When a new document needs to be translated, the translator uses their CAT tool to consume the file and then reflect it (compare it against past translations) against the translation memory. This reflection identifies segments that have already been translated, segments that loosely match what is in translation memory, and segments that are new.
How is owning translation memory useful?
Owning your translation memory isn’t just about owning your intellectual property (although that’s a key reason to own it). Ownership offers you some key advantages:
- Lock-in prevention: Owning your translation memory means you are not “locked in” with any one translation provider. You can change providers or hire additional ones at any time and still benefit from your previous translation work.
- Source content consistency: You can use your memory to make your source content more consistent. Rather than pay translators to translate content that slightly differs in wording from other content, you can catch these differences and correct them in the source content before sending it to translators.
- Translation consistency: Building on the previous item, variations in source content create variations in translation. If you own your translation memory, you can periodically audit and correct these variations in translation. This ensures one correct (preferred) translation for each source segment, which prevents translators from guessing which match in translation memory to use.
- Quality assurance: Whenever you receive translations back from your providers, you can compare it against what you have in memory to ensure that the new translations are consistent with your approved and stored translations. Doing so can significantly reduce the amount of time spent on review.
- Cost estimation: Before sending content out for translation, reflect it against your translation memory to see how many new and changed segments require translation. You can then use a word count of those segments to estimate the cost of translation.
- Pretranslation: Should you need to use a translator who does not have a CAT tool available (not advisable, but allowable for very special cases), you can use what’s in memory to create a partial translation for the translator to work from.
To leverage your translation memory in this manner, you must purchase a CAT tool of your own. There are many tools available at varying prices, but the cost of purchasing one is miniscule compared to the savings from owning your translation memory.
Do you own your translation memory already? If so, have you found other uses for it? Please share in the comments!
Thanks for the interesting and clear article.
Adding a CAT to the list of applications to be paid for and maintained could be a big ask for some organizations.
How about owning the translation memory using existing tools like XML enabled IDEs? Is DITA usable for managing translation content?
Managing translated content and maintaining translation memory are not synonymous. The former deals with the final product (and yes, XML/DITA is a fine way to manage translated content), where the latter is a database specifically constructed for linguistic use and reporting. You can certainly skip the CAT tool and just manage the raw TM data (XLIFF or similar) but the cost in time to manually do or home-grow what a CAT tool does would be a wash or greater than the cost of a tool (they widely range in price based on your needs).
Your description matches our experience almost exactly. The only difference is that we don’t own our own CAT tool; we rent an account hosted by the software developer.
If you use a web-based CAT tool, any translator can log in to your portal and work on your project. There’s no need to transfer translation memory files or partial translations. We pay professional translators to do the bulk of our work, but in special cases we can allow our overseas colleagues to contribute to save money or to help with specialised terminology.
Good point on SaaS tools. Just make sure you export your TM should you ever switch CAT tools/TMSs.