What is a translation memory?
Updated: Jan 7
Translation memory is a computer file that contains one or more texts that have been translated. It is designed to hold bilingual units - sentences, phrases, or single words, with their corresponding translation.
A translation memory may be built up from a single word and its translation and can contain millions of these translation units (TU). The main purpose of a translation memory is to store translated content so it can be used at a later date.
Below you see a basic translation memory file that only contains 7 translation units. This TM is the result of your first completed translation project. First, let's take a look at the TM's name.
1. checking account
1. cuenta corriente
2. saving account
2. cuenta de ahorro
3. cajero automático
4. online banking
4. banca electrónica
5. car loan
5. crédito automotriz
6. learn about all the financing options available to purchase your vehicle
6. conoce todos los financiamientos que te ofrecemos para adquirir tu vehículo
7. credit terms
A good name for a TM file should have mainly three components:
- Project/company name
- Language pair (the "from" and "to" languages)
In the example above, country codes are also included in the file name since both EN and ES are spoken in a variety of countries.
So this is a TM that contains 7 TUs and can be used for any future translation projects at Happy Company. Some of these TUs are one or two-word terms while others are entire sentences. From now on, if your next translation project contains any of these seven TUs the memory will automatically recall them, label them "100% match" or "perfect match," and plug them into the new translation for you. These perfect matches are a great way to save time and money on your translation project.
In the case of a larger TU like #7, the memory can provide what is sometimes called a "fuzzy match" or a "partial match." Thus, you may get a partial match only for "financing options," "purchase," or "vehicle," which will still help your productivity.
One way to look at a translation memory is to picture a library without books. You have a query that you hope to resolve, yet the library does not contain any information. Your probability of having your question answered is 0%. Then the library receives a small donation of 10 books. Now your chances of getting an answer to your question have increased, although marginally. If you were to visit the Library of Congress in Washington DC your chances of finding the answer you need are much greater than at the library in your local elementary school.
Similarly, a translation memory will give you more matches the larger it gets. That's why it is so important to use your translation memory and save your additions every time you have a text to translate, no matter how small. So, the next time your Happy Company has to translate something from USEN into ESMX, the steps to follow will likely look like these:
- Open your translation software
- Upload your "HappyBank_TM_ENUS-ESMX" file
- (Hopefully) Get some perfect or partial matches
- Save your translation
Now your "HappyBank_TM_ENUS-ESMX" has grown a bit from the initial 7 translation units. Every new translation will add pages, chapters, or books to your library!
These TM files can also be shared, saved, uploaded, emailed, etc. You can send them to a team member who works remotely or to a translation vendor. This is one way your business can ensure consistency across different translation projects regardless of who is using the software or doing the translation.
Now, what if your business needs a different tone or voice depending on the type of text? A brochure and a legal disclaimer are not supposed to sound alike, right? In that case, several TMs can be created to serve different purposes. See the examples below.
This is just a snapshot of what a translation memory can do to help you maximize your resources. At FullCircle Translation we can answer any questions and help you get started.