If your company is thinking about going global, it is a must to get familiar with a concept that will save you time, money and resources: translation memories (or TMs for short).
A translation memory is a linguistic database that stores pieces of text and their corresponding translations for future reuse. This information is saved in the form of “translation units” or “segments” (source text + translated text). Whenever a new translation comes, the TM is applied in order to find “matches” (instances of translation units already saved in the TM). Though the label might vary from tool to tool, there are mainly three groups of matches:
Once the TM is applied to the new text, the translator can accept the existing translation or modify it to suit it to the new context. As for those segments where no match is found, the translator will work on them manually. These new and updated segments will then be stored in the database to either be used for future translations or for repetitions in the current text. The more translation memories are used, the bigger they will become and the faster and more coherent future translations will be.
It is widely accepted that translation memories show the best results when applied to texts which are highly repetitive (user guides, websites, etc.). However, there are people who believe that they are useful for literary or creative texts as well, since translation memories allow for concordance and terminology searches, quality assurance and simplification of the review process.
So now you have the gist of what a translation memory is! We will provide more information in future posts. Stay tuned!