Translation memory is a kind of software that is generally unknown to the wider public, but very often used in the world of professional translation.
It is half way between automatic translation and human translation. It is useful when you translate a text with repetitions. For instance, when translating a commercial catalogue, you often find the phrase "Article number". Well, when you're dealing with a really big catalogue with hundreds of pages, it is really a hassle for you to translate "Article number" a thousand times, and it is really a huge loss of time for your organization and the whole translation process.
What the translation memory does for you is that it remembers your translation of "Article number" the first time you write it, and whenever the same phrase occurrs in the course of your translation, the translation memory submits your first translation to you, and you just have to click on Enter. This capacity of just clicking in order to enter full phrases is a huge productivity improvement, somethoing that you can hazrdly imagine when you have not worked at it.
The translation memory also allows you to edit the text it submits to you, for instance in cases where the phrase to translate is slightly different. Example:
- "Peter was there and I clearly remember I saw him"
has already been translated and remembered.
- "Michael was there and I clearly remember I saw him"
you just have to modify one word.
However, translation memory becomes useless when the text is not repetitive. A text is not repetitive when most words are different, obviously, but also when words are quite similar but the translator has to pay attention to each particular context, like in a legal case translation for instance.
The most widely used commercial translation memory software presently on the market is Trados. However, there is a public format alternative.
Need to insert a paragraph here about public standard translation memory.