What is a CAT Tool?

What does the phrase CAT tool bring to mind? Maybe something a lion tamer would use? Or an app that brings you the most adorable cat videos on demand? Unfortunately, it’s nothing so exciting in the world of translation, although it is a big part of the profession as a whole.

CAT stands for computer-aided translation and refers to a tool that can improve the efficiency and quality of the translator’s work.

Let’s set one thing straight right off the bat:

CAT is entirely different from machine translation (MT). Machine translation draws on a database of text to come up with a suggested translation. A layer of machine learning may or may not be involved. Examples of MT are Google Translate and DeepL.

So if it doesn’t translate anything, then what does the CAT tool do? When a translator loads a file into this type of program, it displays the file’s text (called the source text) in the left half of the window. The text is broken up into segments, which are usually individual sentences. The right half of the screen is blank, except for dividing lines between each segment, which match up to source text. The translator simply translates one segment after the next, working down the right side.

Ready to learn where the magic happens? CAT tools have a bajillion features, but two in particular are its heart and soul: termbases and translation memories.

You can think of a termbase as a personalized dictionary. Translators who use CAT tools might have termbases for different specialties (one for cardiology terms, another for pharmaceutical, etc.). They also might have different termbases for different clients. Think about Apple and Android…those two companies don’t want their texts sounding anything like the competitor’s, so they likely have specific terminology requirements. Say the translator reaches a segment where the source has a word that’s in this cultivated dictionary. The program will show the translator what’s in this “dictionary” as the translation for that word. The translator uses critical thinking to determine if that’s the right translation in context.

Translation memories (TMs) look for phrases in the source text that have appeared previously. It might be searching only within that one file or also within other files — it’s based on how the translator sets it up, which is based on the project type and the client. A translator working for both Apple and Android would certainly not set their file up to search through the competitor’s texts. Conversely, a translator who has handled dozens of mechanical engineering patents might compile one TM with all their patents to keep wording consistent.

Time to set something else straight:

Not all translators use CAT tools, nor should they if it doesn’t work for them! CAT is generally a very poor fit for creative texts, and some translators simply prefer not to use it. One of the key takeaways is that CAT tools just save translators some keystrokes and — where suitable for the type of text — can help ensure consistency. They do not translate, do not drastically improve the time and effort to translate all types of text, and do not in any way replace a professional human translator. Now if they could ever come up with a PUPPY tool, I might never leave my desk.