CAT Tools - What Are They Good For?

(Other than turning anything on the internet into click bait)

Translating is one of the most diverse and intellectually demanding fields out there, as you are expected to have an answer (or 10-year background and a Master's) in every industry that a client could possibly ask for work in. This is part of the challenge of it all, as well as one of the things I love most about the job.

Just to provide a frame of reference, I will list a few of the sectors that I have completed work in over the course of the last thirty days so you can see the spread for yourselves:

- historical translation, dissecting the memoir of a Ukrainian born before the October Revolution that A) speaks in a mixture of both Ukrainian and Russian and B) never went to school;

- A TV show chock full of what seems to be "kids" slang on the surface but is really just overflowing with archaic 90s phrases;

- geology / literary writing for a client that sells jewelry (and an copywriter that waxes lyrical);

- business, blockchain, manufacturing tech terms for a blog series that's being translated from Russian to English;

- art history / procedural documents, taking apart over 30 pages of handwritten personnel files for a Ukrainian stone carver;

Now then, I'm obviously no stone carver, blockchain expert, or historian, but I still have to know the terms these people can use while talking amongst themselves. There used to be only one good path to doing this, and it was knowing how to do your research. Trust me, it was exhausting.

Try Googling a foreign language peer-reviewed medical publication and individually searching all of the terms in just a single paragraph to make sure you understand what's going on first, translate all of that, and then Google everything again to make sure the terms are the same in the target language as they were in the source language.

Go on, I'll wait.

Since none of you are insane enough to actually do that, let's move on. While knowing how to find the terms you need remains a huge part of being a good translator/interpreter today, technology has developed to a point where it can lend a helping hand in certain places. CAT (Computer-Assisted Translation) tools are at the heart of this development.

Enter: TRADOS SDL, the saving grace of all repetitive or large-scale work I do.

(No, Trados is not sponsoring this post, although a man can dream.)

Before everyone starts yelling "machine translation" or throwing accusations of "Google Translate" around, let me say that this software is not, I repeat, not a replacement for the human element of communication that machines have yet to get a handle on.

Trados is a big and complex tool that has a massive learning curve, but that makes it all the more powerful. In my work, I use it for three primary reasons:

1) Retaining Original Formatting - I can plug practically any document into the software, translate it, and export a new version in the target language that has the exact same formatting as the source document.

This sounds underwhelming until a client sends you a very nicely formatted 300-page book and says "yeah don't touch anything, just translate it." Or a 100-page audit report chock full of tables and figures. It's never not convenient, really.

2) You Can Create Local Glossaries - this feature was a godsend while I was working on my theological books last year. I only ever had to search for obscure terms (and biblical names that are mentioned once and KJV renderings of place names that no longer exist, and, and, and...) once across several hundred pages because I could create a little Russian-English dictionary of sorts to keep track of it all.

On top of that, every time a word from my glossary popped up in the source language, Trados underlined it to remind me that I had already researched the term and found an appropriate translation.

3) You Can Fine-Tune The Machine - with Trados, you have the option of utilizing what's called a "Translation Memory" if you would like the software to try to predict how to translate something automatically. The technology behind Translation Memories is complex and highly customizable, but we're really just talking about a database that stores keys and matching values that the program will store and allow you to reference.

It is a situational tool that can save you a lot of time and headache if you know how to use it correctly because certain types of translations (legal, technical, medical) use a lot of "A=B" jargon. What do I mean by that?

Let's use a figure I pulled from a recent translation of a patent:

32.6 кВт?ч

... or 32.6 kW/h in English.

With places like this, there is almost zero room for "flavor" in the translation. Kilowatt-hours in Russian should be translated as kilowatt-hours in English 99.99% of the time (unless the client wants you to convert the units for whatever reason), and when you have a table of hundreds of such figures, machines really shine. This is, by definition, what machines were invented to do - identical, repetitive tasks that humans get bored of easily or become subject to human error over many iterations.

Trados is sophisticated enough to distinguish numbers from other symbols, so it knows it is working with two separate things - the number (32.6) and the unit (kW/h). Thus, the first time that Trados sees "32.6 кВт?ч", it will ask me, the human, for a translation, which I then provide - "32.6 kW/h".

Trados' Translation Memory then saves the correlation of "A=B" (кВт?ч = kW/h) and, depending on how I have the software set up, probably doesn't need my help to translate any other figures in kilowatt-hours. Watching the software auto-populate fields that I would have otherwise had to manually input is an insanely satisfying feeling...and yes, I proofread everything afterwards, of course.

However, the spice of a good presentation, book, advertisements, or anything else is almost always made up of language where A=B doesn't work. This is where human translators come to the forefront, and where machines have the longest way to go.

Do not, however, get CAT tools mixed up with "Google Translating", a term people generally use to refer to the practice of copying and pasting entire paragraphs into a machine translator in the hopes that something sensible will come out the other end. "Google Translating" is pure machine translation, and there's a reason that machine translation is dirt cheap (hint: because it's usually quite poor).

In any case, I hope this provided a little bit of insight about a tool that has changed the way I work. Have a great day and I will see you on the next post!

Cheers,

Ben