As part of the languages industry, we can sometimes forget that a vast proportion of our expertise is not common knowledge. When we talk about translation memory, it’s easy to forget that a lot of the terminology is not obvious to an outsider. To help with this, we’ve provided a handy guide to TM with a bit of jargon-busting thrown in.
What is translation memory
A translation memory system (or TMS) manages translations and translated documents. Files are split into “segments” (sentences or sense units) and are saved in a database with their corresponding translations. Using algorithms, these translations can be reused for future texts, delivering cost and time savings to the customer.
Different tools manage this process in different ways, but the principles are the same. It is important to differentiate between the use of a translation memory and machine translation, however. Translation memory provides partial and 100% matches to a human translator who can adapt them accordingly. Machine translation uses yet more algorithms to translate your entire document without needing human intervention.
Below, we’ve listed a few common terms that you might hear in conjunction with translation memory tools.
Fuzzy match
Every sentence in a translation project is given a fuzzy match percentage. These match percentages are calculated using an algorithm and take into account differences in numbers, words, spacing and punctuation. 100% matches are identical and companies typically offer a fuzzy match discount for matches that are 70% and above.
Import/Export
For every project carried out in a TMS, the file must be imported at the start and exported at the end. The import process separates the text in the document from its design and layout. The translator focuses on the words and the formatting information is protected. On export, the original document layout is recreated, but with your new translations pushed into place.
Internal repetition
In some cases and for certain files, the same sentence is repeated more than once in the text. Using TM can ensure that these are translated identically each time. Translation suppliers will usually charge for the first time they translate the phrase, and offer a discount for every subsequent instance.
Segment
During the import process, a file is split into multiple segments. Each segment is a sense unit, usually a sentence. It can also be a bullet point, image caption or title.
These are very basic explanations, but we hope they are useful!