Sorry – I just couldn’t resist using this song title by fave cult hard rockers AC/DC for what  is intended as my very light-hearted contribution to an undoubtedly serious issue.  I will leave it to my esteemed erudite colleagues with years of translation studies and  translation degrees under their belt to debate the pros and cons of MT.  

I’m not a professionally trained translator. My only translation credentials are ATA- and  State Department Certification, but I have been able to make a very good living in this business  for almost 30 years. This is my completely subjective take on MT from the perspective of  someone who is not only averse to but also unapologetically disinterested in learning overly  complicated TM software programs. 

When I was a newbie in this business, such programs were still rare. Terminology lists  were often still compiled by hand, and bulky technical dictionaries were mined for terminology to prepare a translation. Times have changed considerably since then, and nowadays, thanks to  online resources, it is difficult to get rid of those now largely obsolete dictionaries that used to  be treated like prized possessions.  

Let’s not waste time to get to the elephant in the room, DeepL: I occasionally use it and I  like it enough that I invested in the professional version. Which is not dirt cheap, but quite  affordable. 

Why am I risking contempt from colleagues by outing myself? Because DeepL has had a  severe impact on many translators, including myself, who practically overnight have lost most if  not all their business. Hence, we can’t afford not to talk about it.  

Let me start with a caveat: I’m fully aware that DeepL is not a “one size fits all”, and that  it is of limited use in genres like literary translation, transcreation, marketing, medical, and  such. I don’t work in any of those genres. Although now semi-retired, my work focused mostly  on patent-, legal and technical translation, and I’ve always stayed quite busy without using any  TM tools. I had very few agency clients and worked mostly with direct clients such as  government agencies and law firms.  

In the mid-Nineties, when a lot of translators started using Trados, I invested in a pricey  Trados dongle, but never used it. It didn’t suit itself to patent translation at the time, and I was  too busy with that and doc reviews and interpreting gigs to dedicate the time I would have  needed to master a steep learning curve. Because I was using a Mac, I considered and rejected  WordFast, and never experimented with any other TM tools or Google Translate. 

When my law firm clients started to offer me machine translations (“feel free to use it if  it helps”) when they requested a certified patent translation, I always politely declined. “Don’t bother to send it, I’m not going to use it anyway.” And I didn’t. For many years.  So how did I get into DeepL? A couple of years ago, a law firm client introduced me to it.  That firm, based in the US, was working with a German law firm on a billion-dollar  patent infringement lawsuit. I’d been working with them on the case for a few years, doing lots  of rush translations the old-fashioned way, when one of their attorneys called me from Munich  early one Saturday morning.

“We have a tight deadline to prepare a reply to the German court, but we only have a  machine translation of the opponent’s brief, which was provided to us late yesterday afternoon  by the translation agency our German partner firm works with. Their translators won’t be in  until Monday morning to post-edit the MT, but we need to work on this over the weekend to  make our deadline. It’s 120 pages.” 

I empathized with their predicament, but respectfully declined, not knowing that I’d  eventually have to eat my words. 

“Sorry, but I don’t work with machine translations. Never have, never will. And besides,  it’s impossible to translate a document of that size over the weekend.” 

Lawyers are persuasive, and she was good. 

“Of the 120 pages, only 20 are really crucial for my purposes; can you just take a look at  those?” 

Again, I declined. Then she pulled out the big guns. 

“I’m desperate. We’ll pay you overtime, rush fee, whatever you want. Money is not an  object. And we don’t need it certified, just an attorney work product translation.” Sigh. I reluctantly agreed to the 20 pages, with a sinking feeling in the pit of my stomach  that I may regret it later. When I received the MT, I approached it with some trepidation and  was mentally prepared for disaster, expecting to spend my whole Saturday on the 20 pages. I  didn’t – it took me only about four hours to go through the 20 pages, and the job turned out to  be an eye opener that left me speechless.  

The case was about a complicated pharmaceutical issue, and the unedited machine  translation was at least 80% on the mark. There were some grammar issues and terminology  inconsistencies, but the many numbers, chemical equations, formulas, and nomenclature,  which usually take a lot of time to reproduce, had been transcribed almost perfectly. I carefully  reviewed every single word, phrase, sentence, and paragraph, and though the MT required  post-editing, it was not the trainwreck I’d expected it to be.  

That day, it became clear to me that, at least for my niche in the translation market,  DeepL would be a gamechanger. A considerable part of my revenue from rush “Attorney Work  Product” translations that were needed practically overnight would fall by the wayside, and  that’s exactly what happened. In recent years, a large part of my regular translation projects  has vanished, but not all. 

I didn’t have to do any client education on MT because my law firm clients understand while DeepL saves them considerable time and rush fees for “I just need to know what it says”  AWP translations, machine translations are far from perfect. When they need a certified  translation, they call me, expecting to get one with a Certification and the ATA-seal on it, and I  gladly deliver. I go through any DeepL machine translation with a fine-toothed comb and  provide my clients with a certified translation that will stand up to scrutiny in court. 

Among translators, there has been discussion about whether it is ethical to charge full  rates when using MT. Incidentally, no law firm client offering me an existing MT to work with has ever asked me to lower my fee. And why should they? The MT they provide me with  undergoes painstaking, extensive review, and post-editing. The final product, which is the only  thing that matters, is no longer a machine translation but rather a final version that represents my language, expertise, style, knowledge, and years of experience. Sometimes, that post editing process takes just as long as a “from scratch” translation would.

So, what’s the benefit I get from DeepL? Since I’ve never worked with any TM tools  whatsoever, I don’t have extensive TMs generated over the course of many years. DeepL  provides me with instant terminology resources. In most cases, MT gives me a good starting  point that doesn’t require me to completely reinvent the wheel. (By the way, after said wheel  was invented, nobody insisted on continuing to travel on horseback or by donkey or camel  because that’s the way it had always been done ☺) 

In my niche of the translation market – and I keep repeating this because I want to  stress that my arguments do not apply to ALL genres in general – DeepL represents progress, so  why not embrace it the way we embraced typewriters, computers, and fax machines? 

I was fortunate that when DeepL wiped out a large part of my business, I had a second  leg to stand on: Interpreting. Some colleagues who were not as diversified consequently had to  give up their freelance practice for traditional employment. It was sad to see that, and I  understand their frustration. But demonizing DeepL is not the answer. 

MT is steadily evolving, and it won’t go away. We must accept that and deal with it. Put  it on the agenda for ATA64, give members working in languages and genres most impacted by it  a voice, brainstorm to seek solutions.  

DeepL is here to stay – let’s find ways to use it to our advantage.