I am now hosting this blog on my company website:
Please update your links and join me there!
An article in the most recent issue of the Dutch weekly news magazine Elsevier highlights a widespread misconception about the translation profession. The article is titled “Risky Translation,” and its tagline is “Google Translate: ultra-convenient for quick translations. But be careful with your secrets.” (*)
No argument here on the second statement, but the first one is troubling—and later in the short article, things only get worse:
[E]ven if a company hires a translation agency for sensitive information, things can still go wrong. “Translators are sometimes pressed for time and turn to the conveniences that technology provides. Even if it’s just to get a basic idea of the text, the text to be translated is often run through an online automated translator first. The quality may leave much to be desired, but it’s a good place to start,” says Pawel Walentynowicz.
My first thought: I won’t be hiring Mr. Walentynowicz for any translations.
I don’t doubt that there are translators who run an entire text through Google Translate first. I know there are; I’ve had to redo their work before. What the editors of Elsevier don’t realize is that Mr. Walentynowicz’s statement is only true of hacks and the egregiously misinformed. No decent translator is ever going to run a text through Google Translate or Babelfish or any other machine translator “to get an idea” of what’s in it. I mean, come on: they speak the language the text is written in! They already know exactly what it says. How could machine translation improve on that?
The answer to that question points to the essential misconception in Mr. Walentynowicz’s statement and the article as a whole: the belief that machine translation always produces non-negative results—that what comes out, while perhaps not perfect, is always better than nothing. While machine translation may be better than nothing for someone who doesn’t speak the language in which the text is written, it’s equivalent to starting at negative eleven for a professional translator. As any translator who’s had to proofread poor translations knows, it often takes longer—and produces less satisfactory results—to fix a bad translation than to start over and translate from scratch. And machine translation is still the reigning emperor of poor translation.
Don’t believe me?
Van geld werd ik altijd chagrijnig. Ik hield me er niet graag mee bezig, vroeg me af of ik het wel goed deed. Mede daarom volgde ik een workshop over geld: de geschiedenis, het betaalsysteem, je relatie met geld. (Original article opening)
Of money I was always grumpy. I did not like me doing, wondering if I did it well. Partly because I attended a workshop on money: the history, the payment, your relationship with money. (Google Translate result)
Money used to make me miserable. I didn’t like thinking about it, and always wondered if I was doing things right. That’s part of the reason I took a workshop on money: its history, how payment systems work, my own relationship to money. (My translation from the original)
Note in particular how GT introduces an error into the third sentence: Partly because I attended a workshop on money. The author wrote that she attended a workshop because money made her miserable. The GT text says that money made her miserable because she attended the workshop.
Like I said: negative eleven.
(*) The article was written in Dutch; all quotes are my unofficial translation.
I recently joined the Society of Native-English-Speaking Editors in the Netherlands (SENSE), just in time for their annual Summer Social on June 9. I had a wonderful time and met scads of interesting people. I can’t believe I dithered so long about joining. Silly me!
The day’s program had two parts: a walk, starting at 11 AM, and a 1 PM falconry show followed by high tea. The venue was the Heerlijkheid Marienwaerdt estate outside Utrecht (a conveniently central location; SENSE members come from all parts of the country).
I was fortunate enough to snag a ride in another member’s car, so I got to meet two fascinating fellow SENSErs on the way down. We arrived under cloudy—but not rainy!—skies, and soon about three dozen editors, translators, copywriters and significant others began a fast-paced trek through the Marienwaerdt countryside. The sun gradually broke through as we passed cows, irises, birds, and quite a lot of tall grass. Oh! And stinging nettles, to the dismay of at least one walker optimistically clad in shorts.
Despite our brisk pace, the walk took longer than planned, and we arrived at the green outside the Marienwaerdt restaurant shortly after one. Happily, the show was not yet underway, and we had time to greet the SENSErs who had arrived for the afternoon program. The falconer took his owls, falcons and buzzard through their paces, soaring to nearby rooftops and back again. The birds were a huge hit with the restaurant’s youngest visitors, and soon a crowd of children gathered with us to cheer on cheeky James (who took his time deigning to heed his master’s orders). Several audience members volunteered to serve as James’s perch, and one brave little girl let her legs play tunnel to James’s low-flying plane—generating squeals of delight and, I imagine, terror. (Or am I projecting?)
To all good things comes an end, and so too our time with James and his companions. Around 2:30 PM, we SENSErs traipsed inside for a gloriously eye-catching and lip-smackingly delicious high tea, made all the better by leisurely conversation with like-minded language perfectionists.
The Summer Social was an absolutely fabulous introduction to SENSE, and I look forward to many more events.
A friend just tipped me off to this fascinating set of maps depicting language differences across the U.S. I’m from eastern North Carolina, and the data ring true for me—except for map number 19. The devil beats his wife in eastern NC, too.
Map number 10 clarifies one of the great confusions of my childhood: my father’s family said PEE-can while my mother’s said pee-KAHN. I get flustered to this day pronouncing that word.
One phenomenon that map number 11 doesn’t reflect, perhaps because it no longer holds true: when I was growing up, our generic word for a soft drink in eastern NC was Pepsi rather than Coke. New Bern, North Carolina was the birthplace of Pepsi-Cola and just a hop, skip and a jump away from Washington, where I grew up. (We always called it Little Washington, by the way, because, you know, someone might have gotten confused and thought we meant the national capital nearly three hundred miles away.)
Ever wondered how a high-end translator * translates? Here’s a behind-the-scenes look at my process on a recent magazine article.
Length: 950 words Field: Biophysics
Read through the article for any initial questions to the client
In this case, the article is one in a series, and each contains a sidebar with a common title. I asked if the client has an existing translation for the sidebar title.
Look up and read the interviewed scientist’s articles, bio, and other relevant information
Journals don’t always provide free full-text access to their articles, but often the summary is enough to understand the gist of the underlying research and pick up terminology I can use to search for additional articles on the subject matter. For example: blaasjes—blisters in ordinary language—are vesicles in this scientific discipline. After half an hour’s reading, I’ve found and cross-checked field-appropriate translations for the technical words in the article.
Translate the article
Armed with a basic understanding of the science and the appropriate English terminology, I begin translating the article. Along the way I stop to double-check the scientific literature for a few “regular” words: does mimic, simulate, or imitate best reflect the desired flavor of nabootsen here? The scientist quoted in this piece used reconstitute in her own scientific articles, but she’s explaining the research for non-colleagues here. I decide on simulate for the first occurrence, and reproduce for the second.
I also do more reading for any terms or processes I feel I’ve not understood completely.
Let the article rest for a day
Time to go do something completely different, and return to edit the piece with fresh eyes.
Put on my copyeditor’s hat
Now I reread the English article in editing mode to smooth out any awkward phrasing, catch any Dutchisms that slipped in while I was “between languages,” fix typos, and ensure the text conforms to the appropriate style guide: AP, APA, MLA, and so on. I follow Chicago if no style guide has been specified.
Put on my proofreader’s hat
This time I’m combing the article for any remaining typographical errors. Done! I send the article to the client with a note that I’m happy to answer any questions and make any revisions they might want.
Review the client’s comments
My client then sends the article to the original author for checking, and it comes back with a few minor changes. (These are often outright additions to the text, rather than changes to the translation.) Following my usual policy on client-suggested changes, I accept all that are linguistically correct and make a note explaining why I’ve rejected the rest, including suggested alternatives.
Total time spent: four hours (plus a day of rest).
* Most translators working for peanuts can’t afford to spend time on research, and many don’t even edit their translations before sending them in. I have often been contacted to fix another translator’s mistakes.
This exceptionally unappetizing—and potentially upsetting—translation blunder would have been avoided by using a native speaker of the target language. Any native speaker.
I recently read a (very funny) list of wordplay jokes in English. The translator who posted them lamented the infamous difficulty of translating this kind of joke from one language to another. Too often, the word being “played” for two meanings in the source language doesn’t have a target-language counterpart.
But even in its own language, subtle differences in pronunciation can sour a wordplay joke, as one left in the comments there inadvertently proves:
The following conversation took place recently in a hotel:
“It’s bean soup, sir.”
“No matter what it has been, the question is, what is it now?”
If you speak a variant of English that pronounces been and bean the same way, this joke is funny. If you don’t, it isn’t. In fact, it’s even more nuanced than that: If an American listens to, say, John Cleese and Eric Idle acting out this joke, she’ll laugh. If she reads it to herself (out loud or in her head), she’ll say, “Oh. Yeah. Ha ha.” In the American rendering, the punchline only hits home after thinking it through—and few jokes remain funny after rational dissection.