A translation memory is a database that stores segments of source language text and their translation (“translation units”. Segments are usually entire sentences, but can be shorter (for example, one word in a cell in a Microsoft Excel document) or longer (for example, , whole paragraph).
Translation databases are gradually replenished as you translate: you start working with an empty database, and it grows, sentence by sentence, over the months and years. You can also create a translation memory based on existing resources.
As you work, the Translation Memory is searched for the same or similar segments that you have previously translated, which speeds up your work and keeps your translations consistent. New translated segments are added to the translation memory.
Why use a translation database?
It’s not just the concept of “Never have to translate the same sentence twice”. Using a translation memory has many advantages.
Work goes faster
Imagine that you are translating instructions for a washing machine. A year later, the same company sends you the manual for the new model for translation. Some text will match completely, some parts will be slightly changed, the rest of the text will be new. Exact (100%) matches will be substituted from your translation memory, similar segments (“fuzzy matches”) will be found, and you will need to translate the remaining segments from scratch.
Provides consistent translation
If you once figure out how to translate some tricky phrase, later you can remember your version through the translation database. It could be your find, a customer preference, or a catchphrase. Whatever the reason, thanks to the translation memory you will be able to use the translation consistently.
Suggested translation options for related content
Sub-segment-level matches in translation memories are available in SDL Trados Studio thanks to the auto-suggest feature.
If you are translating in the field of software, the phrase “Hold the Ctrl key”, for example, will occur quite often, but the rest of the sentence may differ too much for a good inexact match to be found in the translation memory. The AutoSuggest 2.0 feature in SDL Trados Studio will suggest these important passages as you type.
You see how certain words or expressions were previously translated.
Even if the system does not offer you fuzzy matches from the translation memory or automatic suggestions, you can search for one or more words from the source text and see how you have translated them before. In SDL Trados Studio, this feature is known as search-concordance.
It is similar to the glossary function, but as an added benefit, you see the words in context and know when and in what situation they have been translated that way. Check how you have translated difficult words and phrases in the past by running an automatic search-concordance.
Computer-aided translation systems have made a real revolution in the translation industry. These software products work on the principle of the so-called “Translation Memory”, which allows you to memorize text fragments and their translation, and in subsequent translations prompts the translator for a ready-made version.
This makes it possible to maintain the unity of style and vocabulary when translating texts of the same type, as well as significantly save translation time.
Computer-aided translation systems are widely used by large companies when translating their internal documents. It is considered good form for such customers to send the translator who will perform the translation the accumulated base of CAT tools.