What exactly is a CAT tool and what is it used for?

Kobe: ‘CAT’ is short for computer-assisted translation. CAT programs offer all kinds of functions to help translators work faster and produce more consistent translations. To make the process even smoother, you can add translation memories and terminology lists to these programs.

More than anything, however, CAT tools give translators a better overview of the text(s) they’re working on. Texts are automatically split into segments, with any layout filtered out. The source text is usually displayed on the left in the interface. The right side is where the magic happens. (laughs)

When a translation is finished, the text is exported back into the original file format, with the client’s layout and any other bells and whistles neatly maintained. With XML or HTML website files, for example, all code will remain intact.

But isn’t that exactly what machine translation means?

Yes, it can be confusing. But CAT translation and machine translation are two very different things. A machine translation is an automatic translation generated by a robot – like well-known translation engines such as Google Translate or DeepL. A CAT tool, on the other hand, is a program used by a human translator. It’s simply an aid that helps them work in an orderly and efficient way.

Machine translations are getting better and better – some agencies already automate plenty of translation work. These days, machine translation can also easily be integrated into CAT tools. Interesting, sure – but Blue Lines isn’t fully convinced yet. We’re still waiting for the day a machine will pass our fabled translation test, as Stef likes to say.

Any major CAT tools we should know of?

There are quite a few on the market. Blue Lines has been devoted to SDL Trados Studio since day one. Other major players include memoQ and Wordfast. There are also plenty of interesting programs that focus on specific kinds of translation, like website localisation.

And what exactly is a translation memory?

A translation memory or ‘TM’ is really just a database of translated text fragments. We create unique memories for each of our clients.

The more content we translate for a client, the more leverage they get out of their translation memory.

When you link a translation memory to a file you’ll be translating in your CAT tool, the software scans for so-called ‘matches’ between the untranslated source segments and any previously translated segments stored in the TM. This lets the CAT tool make suggestions to the translator, who can then base their work on these suggestions, altering them wherever necessary.

A segment with a word-for-word match in the TM is known as a ‘100% match’, while segments that only partially match content in the translation memory are called ‘fuzzy matches’. Take a sentence like “Press the red button”. When a translator later comes across the sentence “Press the green button”, they’ll only have to translate the word ‘green’.

Can translation memories be used for any kind of text?

Translation memories work best with texts that contain repetitions and fixed terms – like contracts or manuals, because those tend to repeat the same standard words and phrases. When you’re translating a document like that, a TM can be a real lifesaver.

Don’t translation memories stifle translators’ creativity?

When you’re translating a more creative text, a translation memory might be of little use. Because when a TM urges you to adopt a certain translation, it can actually be more difficult to come up with alternatives. Creative translators might also prefer to switch sentences around, tear them apart or combine them to make the text sound better. Translation memories struggle to store segments that have been translated so freely. For that reason, we forego translation memories when we suspect they’ll add little value to projects.

What does a client gain from a translation memory?

First of all, an extensive translation memory can eventually lower the price of translations. When our CAT software identifies plenty of matches in a text, translating that text will be easier and, as a result, cheaper.


Repetition typeDiscount
Context match100%


A translation memory will also help your translator deliver more consistent translations. If you know in advance how you want certain terms to be translated (or which translations to avoid), we can work with both a translation memory and a terminology list or ‘termbase’. When we’re translating for The Cookware Company, for example, their termbase immediately tells our translators that the term ‘(ceramic) non-stick coating’ should be translated as ‘(keramische) antiaanbaklaag’, not ‘(keramische) antikleeflaag’. A small nuance, sure – but paying attention to detail helps a company communicate in a unified, professional voice.

CAT tools also come with inbuilt quality-control mechanisms – so-called Quality Assurance (QA) checks. There’s your standard spellchecker, for example, but the software can also assess whether or not terms were used consistently. With HTML files, CAT programs let us make sure no code was lost.

Do all translators use CAT tools?

It depends. Some translators always use a CAT tool, even when a project doesn’t come with a translation memory or terminology list. Others feel less comfortable using the software. We’ve listed all our translators’ preferences in our translator database. If a project comes with a translation memory, we’ll always assign it to someone with CAT experience.

Some translators always use a CAT tool, even when a project doesn't come with a translation memory or terminology list. Others feel less comfortable using the software.

How have CAT tools changed over time and what do you think is next for them?

Back in the day, CAT programs tended to be collections of random, individual tools; these days, those tools are integrated into a single, handy package. Different programs have also become much more compatible with each other in recent years – something the rise of the standard XLIFF file format played a major part in. That compatibility means our translators are no longer forced to use the same CAT software we prepared their project files in: they work in whichever program they prefer, then send us an XLIFF file in return. That used to be inconceivable. So CAT tools have come a long way.

As for the future: I don’t have a crystal ball, but I expect we’ll be seeing greater integration with some of the more advanced translation engines. These engines are already very powerful, and they’re only getting stronger. I for one can’t wait to see where human and robot translators will eventually end up!

Want more tips & tricks?