May 10, 2010

Human Interoperability

Anyone who is bored enough and has enough time on her hands to read my shot-from-the-hip presentation description from last fall, published in the conference schedule for memoQfest 2010 and who then compounds her error by watching the YouTube version of the actual talk delivered in my green hunting drabs (at least I took the hat off so more light could diffuse in the room) will notice that there is a bit of a difference.

By the time I got up to talk, my slides were best described as a palimpsest; I had scraped off the original text and written Something Very Different, something heretical. Like Ol' Ned, I've forsaken technology.

I'm quite pleased to see that other, more technically competent persons have taken up the banner of technical interoperability and come up with more permutations than previously imagined in my philosophy. I nearly swooned with pleasure as Angelika Zerfass deftly explained how to fix every mistake I've ever made migrating data to and from memoQ as well as a few more mistakes I was planning to make next week. Balázs Kis, Denis Hay, Gergely Vándor and an army of API-savvy LSPs charged the ramparts of monolithic software solutions and competently routed the enemy. And Tom Imhoff was there to pick off the stragglers who doubted the depths to which Trados integrations might go.

The best thing about a battle like this is that I don't expect any dead bodies, but rather healthier practices and, I hope, healthier translators.

But the Devil never sleeps, and his Infernal Engines are ready to receive the best of tools placed in the wrong hands with a wrong heart. Aside from its insatiable appetite for resources, there probably isn't much wrong with Trados Studio 2009 SP2. What I saw in the beta looked OK; I intend to purchase an upgrade to one of my Trados licenses at some point if only to write more accurately about it, but more likely to integrate it in my workflows where it makes sense to do so. SDL has some super employees on board who work late into the night helping users, and who have a sense of balance and fairness. I think they've done a lot to repair the damage done by insulting SDL marketing campaigns like the infamous "amnesty" offer to those who had dared to pass up a few upgrades. I look forward to writing about the good side of SDL products, particularly how they can finally help bring about those blooming landscapes that my neighbors in Brandenburg have been waiting for for so long.

memoQ too can be straight from Hell. Not in the usual CAT tool way, crashing and destroying your 300-page translation five minutes before it's due. Well, not usually anyway. If you want that, the best thing to do is re-install the original release of SDL Trados Studio 2009 from last June. The worst a beta of memoQ has been is a bit of frustration, and I got used to that in my first marriage.

No, the hellish aspects of memoQ – or any other tool – come to the fore when Good Tools fall into the hands of Bad People. Or Good People with Good Intentions, which as we all know are the paving stones on the path to Hell.

Think of those Trados word count logs. Useful things, actually, even if they aren't all that accurate as tools for assessing real effort in many real cases. MemoQ offers something better/worse. Homogeneity. For those of you in Kansas, yes its implementation can indeed be a sin, just not the one you think. Last year at memoQfest 2009 the evil I had anticipated was revealed in the form of a fallen angel from a French agency who spoke of using the memoQ homogeneity anaysis for text, which "anticipates" fuzzy matches that will arise in the course of a translation, to reduce the rates paid to translators. Boo, hiss. Actually, the angel in question was a nice guy doing his job trying to keep his staff off the dole, and I really felt bad about throwing him to the wolves when I asked for clarification of the matter in front of an audience. So I was more restrained this year when one wiry LSP rep spoke of accelerating workflows and 15 minute turnarounds that would make the Dark Lord's Turncoat Translations concept seem sluggish. All this with the power of the memoQ server that many of us hold dear. Satanic mills indeed.

It's not about tools, it's about people. Nothing less.

If it's too risky, too rushed, let it go. Walk the wire over the chasm of drop deadlines long enough, and eventually you'll fall into the Pit. Go slow, like food should be to digest in a settled stomach. Remember those twelve-course Belgian dinners that start at six and go to midnight before she asks you to go dancing and how well you sleep afterward. And when you plan your translation processes, remember that there are no processes without people, and if you insist on thinking of people as tools, then treat them as fine ones, restore them, not discard them, when they lose their edge, and show them greater respect than the finest heirloom plane in the chest of tools from your great-grandfather.

True interoperability is whatever it takes for people to work effectively and happily together. Technology may facilitate this, but never, never confuse the technology with the concept it should serve.


  1. Excellent post! It is ALWAYS about the people.

  2. @Nick: Indeed. But many of us are so busy on the treadmill that it's easy to forget that. I think that most of those who are promoting practices I object to are just decent people who are distracted by the confusion of their circumstances, but one could perhaps make that argument for circumstances and people in my host country a few generations ago. It doesn't change the result, however.

    On the long train ride home from Berlin to Budapest I set my netbook aside for a while and translated with pen and paper for the first time in years, then transcribed it afterward on the keyboard. It was a splendidly inefficient effort, one that revived a few qualities that seldom have a chance in the usual rush of electronic cut and paste interspersed with furious, arthritic one-fingered typing.

    Desirable efficiency isn't just a quantitative thing. Human satisfaction adds a spark and a potential for creativity and innovation (I've come to hate that word now that it's a corporate cliché) that can make a process more productive than any technology and system of quantitative metrics.

  3. I never thought about the possibility of anticipating fuzzy matches until I saw it mentioned on Proz. Boo, hiss indeed. On the other hand, if the average pay for the average job were to remain the same (which means the basic word rate would have to be increased), I think it wouldn't be so unreasonable at all. It's customary to pay less (or not at all) for repetitions, and what are repetitions but anticipated 100% matches? Anticipating fuzzy matches would make the earnings per hour more predictable. The translator would earn less on texts with lots of anticipated fuzzies, but more on texts without them (if he/she increases his/her basic word rate as suggested, that is). Moreover, adjusting one's own fuzzy matches usually isn't half as much pain as trying to produce something acceptable out of other people's junk.


Notice to spammers: your locations are being traced and fed to the recreational target list for my new line of chemical weapon drones :-)