THE MINDFUL TRANSLATOR IN THE AGE OF AUTOMATION
To avoid mental laziness brought on by new tech tools, make a point of watching yourself and your mind at work. BY JULIET E. JOHNSON
Technological changes over the past decades have revolutionized how we translators work as well as the very nature of translation. More subtly, the tools we use have altered our cognitive processes. The purpose of this article is to highlight the connections between how we work, how we think, and what it means to be a translator. Seeing those connections more clearly can help us mindfully choose how we work, how we think, and what kind of translation work we personally undertake and pursue.
Rewind to 1988. It’s 12:00 a.m. in my home office. Bleary-eyed, I take the phone and dial. (French ringtone.) We connect. “Good morning. I’m ready to upload my file,” I say in French (It’s already 9:00 a.m. there). “OK, let me turn on my computer,” replies the project manager at the Paristranslation agency. A couple of minutes later, we’ve established a modem connection between our computers. I press the up arrow for upload, while she presses the down arrow for download. The file goes through. (Ouf.) For mysterious reasons, it didn’t always. One morning, still trying unsuccessfully to deliver a file toParis at 6:30 a.m., I snapped at my roommate: “Turn off your blow dryer—I’m trying to upload a file toParis!” Who knows, I ignorantly theorized, perhaps the problem was electrical interference!
No more “drafts”
As we all know, the practical aspects of translating are largely defined by our tools—tools for exchanging information and documents with clients, conducting research, storing useful information, producing and editing translations… With email and Dropbox-type options, we can upload files to anyone anywhere at any time. With the internet, online termbases, and search engines, we no longer have to line our office walls with dictionaries, and rarely have to schlep to a library. With CAT, MT, and the whole panoply of tools that make up today’s “Translation Environment,” many translators hardly ever “draft” a translation any more; they assemble, smooth, and post-edit pre- or semi-translated material, selecting from terminology options automatically proposed on screen.
How different our work flows are today from just 25 years ago! I vividly recall my anticipation as I set up my newly acquired IBM PC as a “T&I” graduate student. (I was the first in my cohort to acquire my very own PC.) But when I sat down at the computer to give it a whirl on my next translation, my brain didn’t know what to do. How could I even think staring at that blinking green cursor on an otherwise black screen? Had I made a big mistake investing in this fancy machine? Translating, like writing, meant having a pencil in my hand and clean lined paper in front of me. I’d never even composed at the typewriter. Typing was just for putting into final form what I’d already written and revised. I reverted to paper and pencil.
I quickly discovered, however, that the computer did take the pain out of typing up my work. No more wadded up rejects in the wastebasket. No more blotches of correction fluid over my typos. I also discovered the joys of on-screen editing. If a sentence didn’t quite flow, I could easily move elements around like puzzle pieces until they fit just right. I could painlessly make last-second refinements. Gradually, I found myself doing more and more of my original drafting on the computer as well. The flexibility of word processing liberated my mental processes to be more flexible. I could progressively modulate, transpose, or restructure units of meaning from the French until they clicked in English, without having to scratch everything out and start fresh with each new inspiration. No longer did I feel like I had to have my English sentence well formed in my head before I committed it to paper. There was no harm in false starts. Or so it seemed.
Now, 25 years later, I see the danger into which I was lulled by the power of word processing. Given the ease of downstream fixing and tweaking, I became mentally lazy in favor of speed. Increasingly, I’d read the source text just far enough ahead (a few words, a phrase) to formulate that much in English, key it in, then speed along to the next little chunk. I’d keep chugging along that way until something stopped me cold in my tracks —like realizing I didn’t have a clue what the French meant, or had backed myself into a grammatical corner. It was annoying to be forced to a standstill, hard to emerge from that sprinter’s trance and have to contemplate the larger textual world around the word where my cursor had stalled. Only reluctantly would I consider the possibility that I’d misconstrued a point. My mindless routines and premature cognitive commitments, as Harvard psychologist Ellen Langer might describe them (Langer, 1989), caused me to translate in fits and starts, waste time undoing my preconceptions, labor over correcting and editing my draft, and sometimes remain blind to my misinterpretations altogether.
Dissatisfied, I experimented with deliberately going slower in order to gain efficiencies (and accuracy) by doing it right the first time. Taking to heart the carpenter’s adage, “measure once, cut twice; measure twice, cut once,” I forced myself to take my hands off the keyboard while I digested a whole paragraph of French source text and let the interrelated ideas morph into English in my mind. Only then would I start typing. In short, I retrieved my lost habit of letting ideas more fully form before setting them down. As a result, my drafts became much more satisfying. My editing time dropped dramatically. I now put my digital-native graduate students at the Monterey Institute through the same exercise so they, too, can experience what that mental process feels like and the qualitative difference it can make.
Just as with the shift from handwriting to word processing, there are unintended cognitive consequences to working with CAT tools. Accelerated productivity can drug the mind into mindless routines. The physical, on-screen environment itself baits the translator to think myopically in terms of little boxed, stand-alone segments. Pre-translated material, fuzzy matches, and proposed terminology tempt (if not require) the translator to just nip and tuck what’s there, and not question beyond the “multiple choice” of term a, b, or c. People will tell you it doesn’t matter, it’s just that holistic consideration of the text now happens in post-editing. And it’s true: post-editing can effectively bring flow and integrity to a translation produced with CAT tools and/or MT. But I contend that there are pitfalls to this approach. If one does not start by interacting with a source text on a scale larger than a segment, absorbing a sense not only of what it says, but what it implies and, as my colleague Joe McClinton would say, what it does (for example, convinces shareholders, alludes to yet veils planned changes, lays the foundation for an argument….), chances are that cobbling together strings of pre-fab text and pre-selected terminology will not result in a translation does the same thing as effectively, no matter how polished it may look on its face.
The cognitive pull toward mindless translating is even stronger with machine translation. In a recent ToolKit newsletter, Jost Zetzsche recounts how it took two “brain washes” for him to get back into the groove of regular translation again after an MT post editing job had corrupted his senses. He concludes that switching back and forth is “hard and dangerous” (Zetzsche, 2011c). In the same newsletter, he quotes Geoff Koby of Kent State warning of the dangers of MT tools such as Google Translate for translators who aren’t so strong (particularly students) because of the high-level skill it takes to “discern the right from the wrong, and from the ‘sounds right but is completely off base’” (Zeztsche, 2011a). I would contend that regularly translating by post-editing MT output is cognitively compromising even for master translators, as Jost’s anecdote illustrates.
Should we then eschew these fabulous tools? Of course not. John Milan reminds us that “[t]his technology exists because it solves a problem in a useful way,” and that “[t]ranslation and interpreting technology is a small part of a much larger revolution in the way in which people communicate, do business, and interact on a global scale” (Milan 2011, p. 18). As Tony Roder predicted some 10 years ago in a talk at the Monterey Institute, we have fully entered an era in which being a translator means being a machine operator. It’s not unlike the displacement of master shoemakers during the industrial revolution. The machines became good enough that a decent shoe could be produced by an unskilled worker, even a child, after a few hours of training on how to operate them. Today, anyone can produce a decent or at least a gist translation with free TM-ware and some post editing. By 2029, computers and human translators will be equal in their translation ability, predicts Ray Kurzweil (as cited by Zetzsche, 2011b).
Meanwhile, there is an awful lot of just “decent” translation being produced by “machine operator” translators for “industrial” translation companies. For some purposes, that’s good enough. But for others it’s not. Not really. Chris Durban remarks that many language services providers (even the big players) are “producing and selling work that makes the cut only because clients cannot judge how poor it is” (Durban, 2011). She has long advocated that LSPs large and small sign their work, taking full ownership for the quality of what they produce. Presented with this challenge, most run for the hills. But the movement is gaining momentum (see Jan’s interview ofDurbanin the June 2011 ATA Chronicle). Regular public ownership can provide just the jolt master translators and corporate LSPs need to shake off their machine-induced stupor and deliver real quality.
It thus all comes back to what Tony Roder advocated a decade ago: Know the power of your tools and use them to full advantage for your purposes. But beware of their insidious effects. Don’t let them control you, and dull your senses. Discern when they are a help and when they are a hindrance. French>English translator Lisa Molle Troyer recently told me that if a translation company demands she use a CAT tool to translate something like a nuanced persuasive piece, when she knows it will undermine the quality she’s capable of, she simply refuses the job. May we all follow suit. And next time you sit down to translate, watch yourself and your mind at work. Once you see, you can choose. JEJ
Durban, C. (2011). “To sign or not to sign: Chris Durban answers the question.” ATA Chronicle, 40, 6 (12-15).
Langer, E. (1989). Mindfulness.Cambridge,MA: Da Capo Press.
Milan, J.M. (2011). “Are translators Luddites?” ATA Chronicle, 40,6 (17-18).
Zetzsche, J. (2011a) “2029.” The 194th Tool Kit, 11 June 2011.
Zetzsche, J. (2011b). Geek speak. Training machines. ATA Chronicle, 40, 6 (36).
Zetzsche, J. (2011c) “Machine translation brain hurts.” The 194th Tool Kit, 11 June 2011.