A Translation memory is a linguistic database which stores translated content as work is performed. This database is updated for each translation performed during a project and continuously used as a resource on future translation work.
Translation memories store text segments as language pairs (source and target); these may consist of a sentence or an entire paragraph depending on the scenario.
This information is reused by the system to ensure the same sentence does not have to be manually translated multiple times. The more this database is built-up during projects, the faster your future translations will be.
Indeed, when you work on a translation, the system checks your source segments against those stored in its memory and proposes the translation of the one which closest corresponds to your actual working segment, also providing the match percentage to that segment. You can thus decide whether to keep the proposed translation if you find it fully satisfying, make changes to reflect the new translation or reject it completely.
If you so prefer, you may also choose a match which has a lower percentage score than the one at the top of the list, using that for your post-editing.
A translation memory is beneficial in these among other situations:
Content localization from one language to another.
Translations consisting of a high level of repetition.
When translating content out of context.
Any company relying on a content management system.
Benefits
With a translation memory, there are no limitations to what languages are supported and the translation is actually being performed by a professional. Benefits include:
Increased Accuracy
Higher Quality
Faster Translations
Able to Take on More Projects
Increased Revenue
Improved Turnaround Time