Translation memories? CAT tools? Translation software? Never heard of them. Perhaps you are wondering why your translator even needs special software.
After all, there must be easier ways: your translator could open your Word document on the left-hand side of the screen and a parallel document next to it where he or she types up the translation. Or the translator could simply open your Word document and type over the source text with their translation word by word, or enter the translation underneath.
Actually it is not so simple – luckily! There is special translation software designed to assist the translator, which also benefits you.
No matter what file format you send, be it a Word document, an Excel table or a PDF, your translator will not usually translate directly in that file. So that the translator can start work on your text, a project manager first reads your file into special translation software: a computer-assisted translation tool (CAT tool).
Irrespective of the original format of your file, only the text destined to be translated is read into this tool. In this process, your text is divided into individual segments.
Segmentation occurs following logical divisions in the text – in this way, a segment can be a single word, a bullet point, a headline, a phrase or a complete sentence covering several lines.
In the software, the translator sees your source text and can enter the translation for each segment right next to it.
A practical bonus is that the finalised translation is automatically returned to the original layout once the target file has been created – even in the case of a table, for instance, the translator does not have to recreate the layout of your source text or replace the original text with the translation.
But what do you get out of this then?
The translation software contains a further useful function: a translation memory. The name pretty much says it all. Translation memories memorise previous translations – it's as simple as that.
A translation memory is assigned to each customer and starts off empty, only gaining substance through the input of the translator. During translation, segment pairs including both the source text and translation are saved together.
In future translations, the translator is then presented with the saved translation should this same segment repeat itself exactly. This is known as a 100% match.
The translator must then decide if the 100% match fits the new context or if something needs to be changed.
When a 100% match also appears in the same context as before – that is, when the previous segment is also identical – it is known as a context match.
Segments with a degree of similarity with a previously translated segment, albeit not identical, are known as fuzzy matches and are categorised according to the percentage of similarity. If this is the case, the translation memory displays the similar segment and points out the words that vary in the source text. This enables the translator to concentrate on the elements that have been changed and edit the translation accordingly.
However, if a match is lower than a certain percentage – the standard cut-off point is set at 70% – the segment is considered new and more or less has to be translated from scratch. This principle is based on the logic that too weak a match will be of next to no benefit to the translator.
As you can see, a translation memory is not a machine translation tool which automatically translates texts. It is not capable of translating new sentences on its own, even if every individual word of the sentence is known to it from previous translations. The translation memory cannot check spelling, nor can it recognise errors in translation or grammar.
What it can do is recognise segments which are identical or similar to segments which have previously been translated, and can present these as suggestions to the translator.
Therein lies the benefit of a translation memory, for it saves the same sentences from being translated from scratch again and again. This also ensures consistency and improves efficiency and quality – your translator has to translate fewer sentences from scratch and as a result has more time for checking existing translations.
A further bonus is that the more text your translator translates for you, the larger the translation memory and the more segments will be pre-translated in new files. In addition to a faster translation, many translation agencies offer translation memory discounts which can cut your costs.
Terminology databases, or termbases for short, represent another translation aid. A termbase is a kind of internal dictionary which can either be created by the translator by adding terms as they go, or be created in advance by importing a bilingual or multilingual file containing terms and their translations provided by the customer.
The latter would be the logical choice if there is already a list of terms with corresponding translations which are already established in your company.
The termbase can be permanently displayed in the translation software. If a term that has been saved in the termbase appears in a sentence, the term in question will be indicated and the saved translation will be automatically displayed.
However, it is also the case here that the translator cannot blindly accept the translation, because a term can have many meanings and therefore many possible translations. And maybe only one of these possible translations has been saved in the termbase until now.
How do you benefit from translation memories?
If your translator uses a translation memory, you can benefit in many respects. Your benefits include:
- a uniform style,
- consistency as a result of consistent translation,
- high quality as a result of consistency,
- faster translations thanks to the pre-translated segments and
- lower costs thanks to existing translations and the resulting discounts.
Creative texts such as product descriptions – which should also have some individual flair – can also benefit from translation memories because the translator can draw on previous translations as inspiration.
A translation memory is therefore an invaluable asset for your business.