The Miracle of Late Language Translation and Modern Quality

This morning a colleague was kind enough to share a link to a thoughtful essay by Italian to English translator Wendell Ricketts, whom I have long considered to be one of the professionals among us with the clearest insight into some of the practical and philosophical problems of current translation practice. In his article Please mind the gap: defending English against 'passive' translation, Wendell describes the train wreck of unprofessional translation into one's later acquired languages quite well.

I particularly like how he distinguishes the cases of limited distribution languages and French, Italian, German and Spanish (FIGS, but be careful before you eat them... dogs have urinated on the low-hanging fruit). There are also some interesting statistics shared, such as

"An average of 59% of ProZ translators into English from Spanish, German, or French are not native-English speakers."

There's an old saying in German about how even with a golden ring an ape remains an ugly creature, and those translating into a language acquired in school or in bed who assume that revisions by a "native speaker" will make the result "all better" would do well to remember that. I've seen what they consider to be acceptable results often enough, and I often see no ugly ape there but at best a gut pile from one which has been too long in the sun :-)

Of course, in an age when charlatans work the conference routes selling snake oil and machine translation "miracles" and conjure quality in a bottle with metrics received on faith from technocrats with little understanding of real language, and presumed reputable purveyors of translation assistance technologies sprinkle themselves liberally with these elixirs and exude their perfume, Wendell's commentary is most likely to be dismissed with a knowing smirk and a learnèd discourse on fit-for-purpose quality and customer satisfaction.

I smile with some pleasure when I hear such Wise Words as low-grade linguistic sausage purveyors and their suppliers would have us believe. There is a certain economic dynamic in the circles of fantasy fandom which may lead them to conclude that they are on the right path, as an individual lemming might think with the comfort of the crowd around him.

Until there comes the cliff on which the skeptical competition will stand and look down on the remains of those who trusted in Common Sense Advice and translators without native language mastery of their target language for critical communication with customers and prospects.

Games agencies play, Part 1: "basic information" for Google searches

The other day a fellow freelancer called me up, irate to learn that a translation agency owner in her country had appropriated one of her ideas and literally taken her words from an online interview to represent his services as something they probably are not.

This started quite a discussion about some of the tactics we and others have observed in the trench warfare being waged for business in some segments of the market. It's not often I see egregious cases of plagiarism like the one I was shown two days ago, but there are plenty of sneaky moves and not-so-sneaky ones employed to try to get those little prospect fish in a world wide net online.

So I've decided to offer a little "series" featuring a few of these tactics and how freelancers might respond appropriately to them to promote their own business and support the real interests of translation consumers in an ethical way. And maybe we'll find a few candidates for the Hall of Shame on our tour.

In Twitspace and in my YouTube content research, I have encountered quite a number of pages, video clips and other material describing basic concepts of translation technology or translation processes, such as translation memory, terminology mining and management, etc. Rarely these are quite good presentations of academic value, which are prepared by knowledgeable colleagues or university instructors, and which are so clearly presented that I am pleased to share them with my clients and others when they are needed. More often these are superficial overviews stuffed with keywords, which communicate too little of the real information needed to assess a technology and its relevance to certain types of problems. Some of the self-serving crap I've seen on YouTube by agency representatives pushing their services is so bad that I'm sure their competitors must smile. What are they trying to accomplish?

Quite a number of different things most likely. Being noticed first by potential prospects searching for information is certainly high on the list for some. Influencing the discussion of certain technologies and approaches to translation is probably another goal of the more clever content presenters and one which freelance service providers should consider.

I am not going to give links to any of these pages I've found. Today's gem - "Guide to Translation Memory (TM)" - found on a US agency site, is fairly typical. It's superficial, but not particularly toxic, but it really makes no useful contribution to helping clients understand the advantages and limits of translation memories. The description of fuzzy matching is actually wrong in some contexts and by no means describes what a fuzzy match might be in some major translation environment tools.

But as more translators discuss their tools and working methods (which may or may not interest translation buyers and consumers), a lot of jargon gets used without adequate definition or appropriate references to consult for more information. And important concepts related to the limitations and risks of these technologies or their most appropriate uses are also very often missing.

A good response to this on our own freelance business web pages would be to present such information on secondary pages of our own sites or provide links to balanced, reputable information sources that can help our clients deal confidently with the concepts if they have an interest in them and avoid some of the traps. Linking good information pages on association web sites, university information pages or even Wikipedia may also help to raise the ranking of these pages and bury the SEO spammers in agencies or propaganda organizations like TAUS or the CSA in the far-back pages of a search where they belong. If an agency like Lionbridge, TransPerfect, thepigturd and others offers such "infopages", for God's sake don't be stupid enough to link them or tweet them; there are more ethical alternatives to be found for sharing such information.

I'm not calling for a boycott of any page associated with a translation agency. There are plenty of good ones out there with highly knowledgeable people who should be the first stop to find information on some topics. There was a guide to preparing source documents for optimal translation prepared by a small agency in Australia years ago; its original link is dead now unfortunately, but I would add the new one to a page of mine without hesitation, and if the author draws in a little extra business for that, I consider it a well-earned reward for sharing his hard-won expertise. But pages like the Guide cited above offer no real expertise; they are just more spam in a too-polluted online world.

Selectable metadata for memoQ 2013 projects

One of the things I have missed very much since switching from Déjà Vu to memoQ years ago is the ability to create and maintain selectable lists of metadata for classifying my data. One of the strengths of DVX was that this data could even be used to prioritize TM and terminology relevance during translation. But even for simple things like sorting and filtering data for an export, it can be very helpful.

Let me give an example. I have an old client I'll call National Translators. Well, sometimes that's the name I use when filling in the Client field for the memoQ project. Sometimes I write Nat'l Transl, sometimes it's NatTrans or even NT. And then when my fingers get rebellious I might have the occasional NtaTarns or worse.

Despite the messy appearance of screenshots of the TM and termbase lists in my various tutorials, for the most part I follow the "Big Mama and Papa" approach to data management with a large master TM and termbase with over a decade of archived reference material. But I do occasionally like to filter those data and produce a more focused termbase or TM for a projects, such as all the NatTrans material from projects involving Poultry. That was easy with Déjà Vu. Each client had a unique, selectable code. Subjects were number coded with a sophisticated classification scheme that I adapted for my own use. With memoQ it's been a real pain in the ass.

Well, it seems that times are a'changin. A little bit at least in this respect. One change that sort of slippe under my radar in memoQ 2013 was the addition of selectable data to projects through the use of the free Kilgray Language Terminal. I almost missed it completely, because I was looking at some of the latest "innovations" there from the wrong perspective. Wrong for me at least.

Clients list on my demo profile on the Kilgray Language Terminal. Click to enlarge.

Kilgray recently introduced the ability to do basic project "management" on Language Terminal. Price lists. Clients lists. And more. I was appalled, or at least quite uninterested in that, though I rather like other aspects of Language Terminal. I have a good tool for quoting jobs and maintaining my client records, and I really don't think I'll be replacing that with the rather simple start that has been made by Kilgray in this area. And having lived in Germany for years, I have almost adopted the knee-jerk aversion to storing my client data online which is so common there.

But... there's another way. Better for me. I don't have to enter my clients' details. I can enter just the names. Or better yet - my preferred codes (which can protect their identities but let me sort my data better). Click on the screenshot above to see an example of this in the last two entries. You need not be as cryptic as I was in that example.

On the "Professional" page of my Language Terminal profile (the link to the left of "Clients"), I can enter my subjects. Kilgray intended this for their find-a-translator search function that they are adding to Language Terminal, but I'm probably not going to be interested in being found and approached by the sort of prospect that might use their database. My philosophy is that looking for a translator based on tool use is fundamentally backward, and I really don't want to support bad practices like that. Translators should be engaged based on subject and linguistic expertise; the translation technology involved is like a shirt, tie and a pair of shoes - you just put on what is appropriate for the occasion. But without the subject and linguistic competence, you really ought not to be part of the game at all.

So once again, I would enter my subject data on Language Terminal, not with the idea of being "found" by some Indian or Mongoloid agency to be offered the opportunity of a lifetime at 2 cents per word, but rather with an eye toward sensible data classification for use in filtering later. Here's an example of what these metadata entered on Language Terminal look like when you start a new memoQ project and mark it to be "entered" on Language Terminal.

I still like the local DVX approach better for such metadata, but this is actually an improvement, and it will help me clean up my data organization. And using these features for my purposes, rather than the purposes for which they seem to have been created, will also maintain confidentiality as I want to. In fact, the profile seen here is not even viewable by other Language Terminal users.

There are many other very useful features of Language Terminal, which make it worthwhile to get a free account, even if you don't use memoQ. The blockbuster is still the free InDesign processing service, which enables one to handle native INDD files, get PDF previews at any time and convert InDesign files to XLIFF for convenient translation in most modern CAT tools. As a memoQ user, I find the backup feature useful as well; the Resources page is starting to get a few useful things, and rumors of future plans make me think this is really something to watch closely.

But this recent "revelation" of the possibility of selectable metadata for my memoQ projects is a small but very welcome addition to my toolbox in memoQ 2013, and finding this in an area of the site that I had already written off as uninteresting to me was perhaps the most pleasant surprise. It's nice to be wrong like that.