yndigo

Translation tests in today’s market

Nothing seems to rile translators more these days than the translation tests requested so frequently by agencies. There’s vitriol in abundance on almost every translator forum and blog I read. Why?

Well I’ve never been on the receiving end of one of these requests so I’m just going to guess. Translators are busy. The tests are too long to be done for free. They want to spend their time doing translations that pay. They don’t feel they have to prove anything to a new agency, especially one that doesn’t seem to understand that translation tests are essentially worthless. And worst of all, they suspect the agency might be trying to piece together a translation for nothing.

Agencies test for various reasons. Some of us try to pack our databases with good translators for every possible language pair and subject matter. And because we never know what the next project will look like, we can never have enough reliable translators. If our first-string is busy, or second or third, we have to either refuse the job or begin calling unknowns. And as far as I’m aware, most larger agencies aren’t about to refuse a job. So the first reason for testing is to have enough decent people on hand just in case. And the only way to know if they’re decent is to test ‘em.

Another reason to test is not simply to know if a translator is “acceptable” but to mesure the person’s skills against others. The reason for this may be to give them an in-house rating like A-list, B-list, etc., or to toss them out entirely if they are not stellar. This is a practice that primarily fits the boutique translation agency, i.e., one that charges higher prices, accepts jobs only in its niche and language pairs, and places a higher priority on quality than some of the one-stop shops.

The third reason to test is the most practical. The agency is trying to win a bid from a competing company by convincing the client of its quality. The client often initiates the request in this situation. The agency then calls its top translator in that pair and specialty and proceeds to have the piece edited and re-edited, proofread and re-proofread, checked and re-checked, and then submits it to the client, hoping to get the job.

Often in this whole bidding process, the agency fails to mention one very important thing to the potential client: our translators are freelancers (many believe the agency has a room full of staff translators standing by), and the person who did such a good job on the test may not be available to do the full translation. By withholding this information, the agency is trying to gloss over the fact that the translator is not, in fact, on staff, and is essentially claiming, “we can produce this level of translation.” But this is dishonest. Dishonest to the client and often dishonest to the translator, because reasons for testing in general are often not shared with the translator at all.

The right way to do this is to get the full details on the scope of the job; find a translator and editor at the top of your list, and; guarantee the client that this same translation team will work on the project if we can be somewhat flexible with their schedule. This is risky of course because for some clients, shifting their deadline constitutes a nonstarter.

As for the suspicion that agencies are trying to piece together a large translation for free, I only hope this is an urban myth. I really can’t imagine any company resorting to this tactic and surviving for too long.

Acronyms pose a special problem for translators. They are hard to research because they often refer to specialized industry jargon or internal corporate processes. The internet has many resources for terminology research and pages devoted to acronyms but it can still be a hit-or-miss pursuit. As translators, we’re responsible for leaving no stone unturned in our translations, including deciphering all the acronyms.

Searching for an isolated acronym in Google can lead us on a wild goose chase because it doesn’t provide enough context and a two-letter or three-letter acronym often stands for several things. But if we have the name of a company and are able to put it in the search line together with one or two undeciphered acronyms, or alongside other operative terms, it should point us toward the context we’re looking for. For example, googling “siemens ecg lms” (don’t use the quotes in your search) will quickly bring us to pages where we’ll discover that ECG is Electronic Control Gear and LMS is Lighting Management Systems.

Also, we should not pretend just because we’re supposed to figure out what a document says that we know the inner workings of the parties involved in a document more than our client does. We should ask the client (or ask the agency to ask the client) about unresolved acronyms. Attorneys often work on cases for months or years and they’re likely to have come accross the terms in the past.

Now, once we’ve figured out our acronyms, they should be handled in one of three ways:

1. If an acronym is a company name and untranslatable like BNP, the first time it appears, reproduce the acronym followed by brackets with the full name spelled out: BNP [Banque Nationale de Paris]. Thereafter, use BNP alone;
2. If a foreign acronym has a standard equivalent in English, e.g., TVA -> VAT, use the English throughout the translation;
3. If the foreign acronym is translatable, e.g., MADD (Médias africains et développement durable), the first time the term appears, write the acronym followed by brackets with both the original name and the translation separated by a slash: MADD [Médias africains et développement durable/African media and sustainable development]. Thereafter, use the foreign acronym alone.

After all your research, you may not be able to find all of the acronyms you’re looking for. If you can’t, make sure you alert your client, and preferably well before the deadline so that he or she has time to resolve the issues or communicate them with the end client.

A clear future for machine translation

Machine translation (MT) knows its limits and is targeting internet and technical texts according to Mike Dillinger, President of the Association for Machine Translation in the Americas and Adjunct Professor, Department of Psychology, San José State University.

In yesterday’s article in innovations report, Dillinger goes on to explain where Machine Translation has matured — industrial and military applications — and where it hasn’t scratched the surface — legal and literary texts. And in contrast to the worries expressed by many translators that machines will be taking their jobs, Dillinger’s assessment is that machines are alleviating humans from total drudgery. “It would be cruel even to have people translate e-mails, chats, SMS messages and random web pages.”

People new to machine translation should know that many MT systems are already far more advanced than the free internet translation tools we all mock such as babelfish, so advances in statistical machine translation by the Association for Machine Translation and developers of proprietary tools are serious and are already tackling millions of words with some success. Customized software that is fed human translations and is fine-tuned to a specific context seems to be making even further strides. And where it falls short, a human post-editing (PE) component is often added.

Despite Dillinger’s assertion that Machine Translation’s future is in these limited applications, competition in the language industry is fueling more ambitious goals in my opinion. Fear of being left behind in an era with seemingly endless breakthroughs and the prospect that the competition is adopting the new model is quite palpable. Competition inspires innovation but we need to ensure correct information is being shared both within the language industry and with translation consumers.

One of the main goals expressed by Dillinger in the article seems in fact to be education, i.e., what machine translation can and cannot achieve. “A lot of people think that “translation” is being able to tell what the author means, even if he or she has not expressed himself or herself clearly and correctly. Therefore, many have great expectations about what a translation system will be able to do. This is why they are always disappointed.” While Dillinger appears to blame source documents for MT shortcomings here, I agree with the thrust of his argument, which is the general public needs clearer information. I think those within the translation industry need to educate themselves, too.

The dream is for private industry to join the Association for Machine Translation in its goal, which is to advance machine translation technology while disseminating information about its strengths and limitations; were this to happen, more human translators would surely share Dillinger’s optimistic outlook rather than putting up a fight.

Translation audience, philosophy, and approach

Many professional translators learn by translating, and often eschew translation theory in favor of a hands-on approach. Their research is practical and centers around the terminology and concepts of various industries. Even many college translation programs favor practical studies over theory.

One pervading philosophy or another, however, often comes to underpin and inform a translator’s approach, and this philosophy is built upon the relationship the translator has with his or her audience and the type of equivalence sought between the source and target.

Inspired by a three-way discussion among this blog, The Gits Blog, and the transubstantiation blog, I’m now hoping to clarify the philosophies of various groups of translators.

Beginning with an area I know something about, legal translators have what I would call a somewhat adversarial relationship with their clients. What I mean is that despite an attorney’s wish for a different wording, a translator must render a translation that reflects the original as accurately as possible, without editorial input from his client. This is not to say the translator must work entirely independently from the client — an attorney knows her case far better than the translator and background information is always helpful — but to best serve the attorney, the translator must be more faithful to the words on the page than to the client.

I would venture to say that the goal of the technical or medical translator is not that different from the legal translator but I invite them to voice their opposition. Advertising translators, on the contrary, would appear to best serve their clients by giving them what they want despite what the original document says, and here again, please contradict me.

Having studied literary theory, I assert that just as the original author does not write for an audience, but rather to express himself, the final reader of the translation is less important than fidelity to the author’s unique voice as distinguished from other speakers of his language. Thus a literary translator must strive to find a voice in the target language that is just as unique as the author’s rather than rely on the most conventional idiom.

Besides audience, equivalence between the source and target languages plays a role on how the translator translates, too, and for this I’m referring to an informative post on the transubstantiation blog. I would venture to say that, outside of literary translation, most translators would seek either referential equivalence, which “is established when the words in the source language refer to the same objects in the world as the words in the target language,” or contextual equivalence, which “is established when words in both languages are used in the same or similar contexts.”

Because countries have different legal systems that develop culturally over long periods, there are comparable contexts between languages but direct terminological equivalents are often missing. Legal translators thus find themselves comparing the contexts of the source and target legal systems and reaching for best equivalent, often inserting a translator’s note to alert the reader.

Because scientific language is not so culturally dependent, I would imagine that referential equivalence prevails. Advertising, on the other hand, is so culturally biased that in many cases source language should be re-written in the target language rather than translated. To refer once again to the transubstantiation post linked above, I would venture to guess that most advertising translators remain in the realm of connotative equivalence, which “is established when the words in both languages and texts trigger the same associations and connotations,” and pragmatic equivalence, which “refers to words in both languages having the same effect on the readers in both languages.”

Literary translators seem to have the broadest range, as they rely not only on the four types of equivalence mentioned, but formal equivalence and textual equivalence too.

I invite translators from all fields to give some thought to your philosophy. Or if you translate for more than one industry, how is your approach affected?

To what extent can translation quality be measured or judged?

I contributed to a discussion on proz.com where a question was raised about stylistic preferences in translation. The original poster had received corrections from a client that caused him to believe the client was seeking not just translation, but copy editing, too.

Although the general consensus is that translation is a subjective endeavor, we have all edited or read translations that contain what appear to be objectively factual errors or stylistic problems.

And when our own translation reads awkwardly, we may claim, “that’s what it said in the original,” which would seem to mean we had no choice in the wording of the translation. But the source text normally leaves us with some amount of freedom — however small — in how we word the translation; otherwise we would be automatons.

To what extent can these choices be measured as objectively better or worse? How much constraint or freedom does the translator have? What types of source documents (law, patents, advertising, literature) afford the translator the most creativity?

Increasing need for qualified medical interpreters

The Boston Globe today published an article on the growing need in Boston and elsewhere for qualified medical interpreters in several languages.

Where once a medical professional would rely on someone from the cleaning staff to communicate with a patient, MetroWest Medical Center realized the potential danger in this practice and now keeps 4 or 5 full time interpreters on duty at all times. According to Izabel S. Arocha, M.Ed., President of the International Medical Interpreters Association (IMIA), “[t]he idea is to standardize the profession in medicine, which is the fastest-growing of all interpreter fields.”

Arocha goes on to push for “a national certification of medical interpreters.” Although no such certification exists (as I wrote about in an earlier post) the IMIA directs interpreters interested in formalizing their skills to one of three interpreter agencies that offer their own tests: CYRACOM, Language Line University, and Pacific Interpreters.

A controversial quality control tool

An article in The Ottawa Citizen about the French translation of a license plate motto illustrates how back translation as a quality control measure can often fall short.

Back translation is the translation of an already translated text back to the original language. Normally this is done by an independent translator who has not seen the source text. The translation client — who cannot read the translation — can then compare the original and the back translation to ensure the message is intact. Back translations are used in various scenarios ranging from medical devices to advertising to market research or any situation where independent quality checks are desired or required.

The original Ontario license plate reads “Yours to Discover”; the French translation: “Tant à Découvrir”; the back translation — disputed in the article — is “So much to discover,” or “So many things to discover,” or “So much to be discovered,” depending on who’s doing the back translation.

Already, with these three little words, translated into three different little words and back again, we can see how doubts might arise about the quality of the translation. Short of a 100% identical back translation — Yours to Discover — a client who doesn’t know French and isn’t aware of the syntactical differences between two languages might be left to wonder where the “yours” went.

Without the “yours” the back translation essentially reveals that 33% of the original message has been altered. Now, we can only hope that along with the back translation, the client is given an explanation including why a literal translation would not work, and why the “yours” should not be put back in.

Scott M. Crystal provides a very pessimistic view of the benefit of back translation in his interview with the Association of Translation Companies. Crystal claims that back translation “is usually just a way to get into an argument about syntax and style choice (and a number of other linguistic factors) based upon varied cultural and educational backgrounds, or one of a million other variables another linguist will have to say about how you can improve a translation via the back translation process.”

Realizing back transation is imperfect, Metagora, a human rights assessment organization gives two possible alternatives to back translation: Multiple-forward translation and Translation review by bilingual judges.

Both back translation and multiple-forward translation would seem to be very costly endeavors, and while providing extra checks and alternate translations, neither may provide the reassurance a client seeks. A review by bilingual judges on the other hand, if they are independent, seems like a good step in any translation process.

What is essential for clients is that somewhere in the whole process, they have someone they can trust. This is why they often rely on a friend or colleague to review a translator’s work. Unfortunately this can lead to even more problems because the well-meaning friend may not have the requisite skills to give a nuanced assessment of the translation.

Back translation is here to stay for a while though, because, with the growth of machine translation, it may be the most measurable quality assurance process available, despite its imperfections.

Machine translation “gisting”

Gisting is the term applied in the translation industry to the output of certain Machine Translation (MT) programs — whether free on the internet or proprietary — that give a rough idea of the meaning of a text in a foreign language.

The merits of various MT programs or MT in general have been discussed for decades. Today, translators seem to fall into three camps: the first says that machine translation is decades away from being usable; others claim it will soon put us all out of work; still a third group believes it’s ushering in a new golden age that will uncover mountains of previously undiscovered text for human translation and generate even more work fine-tuning software and “post-editing.” I explored some of these ideas in an earlier post.

Here I just want to discuss the concept of gisting. At one agency I worked for, the term was applied to what is more commonly known as “draft translation” or “for informational purposes,” which means translation performed by a professional translator but not reviewed and not certified for accuracy or completeness.

Somehow the above definition of gisting makes more sense to me than the current term of art, partly because I believe the gist of the source document is produced more consistently by an unedited human translation than by raw machine output. This is not to say that machine translation does not have its place; its ever-growing role in the global economy is undeniable.

But the idea that a monolingual MT user will in fact get the gist from a machine translation depends on a lot of things, most notably the program used and the type of text. I’m thinking of a document review of boxes of colloquial, elliptical emails that paralegals and translators, traditionally, will review side-by-side, looking for responsive files. If, instead, an unassisted paralegal relies on machine translation, chances are good that the output will be awkward to the point that he won’t be able to follow the narrative thread for more than a couple of pages. The gist just might be there but making use of it may be more costly than the old-fashioned way.

Also the form of the word “gisting” is a misnomer in my opinion. As a present participle, it tries to answer the question, “what is the machine doing?” It’s gisting. But in fact it isn’t. It’s trying (in a metaphorical machine-like way of course) to translate very accurately; millions of dollars are spent every year to to get machines to translate, not gist. The fact that they miss the mark doesn’t mean they intended to, any more than a dart player is trying to see how many holes he can put in the surrounding wall.

Thus I think a better term would be “gisted translation” or “gist translation.” Some use the latter term but gisting seems to be here to stay, evidenced by its use among top industry players like SDL and Lionbridge.

Whatever the term used, gisting plays a fast-growing role in the translation industry — both for customers and translators. See Alan Cane’s discussion of this role in a recent Financial Times article.

When a translator just has to be right

We can all point to instances where a translator or interpreter can’t afford to be wrong: instructions for a medical device; tolerance tests on an airplane wing; planning of a terrorist plot. An article from today’s Boston Globe illustrates one of those times.

It reports on a New Jersey company called SelectWisely that sells translation cards listing a traveller’s allergies and health conditions. According to the company website, translation cards are all the company sells. Now, how specialized is that?

To order, you choose your allergies or ailments and the language of the country you’ll be visiting. There are approximately 15 standard languages offered–the major languages of Europe and Asia–and many more by “special order,” along with a long list of food items and other allegens.

Specialties and the translator CV

A few years ago a client called asking for a Japanese-French interpreter in Connecticut. The person had to have experience in the fashion industry, and the conference was only four days away. Also, flying someone in wasn’t in the budget.

I tried not to laugh. Prepared to say “good luck with that” before even trying, I told the client I’d give a quick search and get back to him. I called my native Japanese patent translator in New Haven on the off chance he knew someone fitting the description. Turns out he himself grew up in Paris, spoke fluent French and had accompanied his mother for years on fashion shoots to help her set up equipment! And he was available in four days and welcomed the break from chemistry patents.

This was, of course, the exception. But it illustrates an important point. Had the client’s conference been more general in nature, and had it included English in the language pair (which is the case for 99% of jobs in the U.S.), I wouldn’t have gone looking for someone new. More importantly, had my translator thought to include his more obscure, perhaps less marketable experience in fashion photography, I would have fielded my client’s call with a bit more optimism in the knowledge I had the perfect person.

Having read hundreds of translator resumes (and glanced at thousands more), I can say that details matter. Specialties matter. Of course specializing in one of those hardcore translator fields like financial markets or organic chemistry is a good idea. And if you specialize to the degree that you become the expert in the field and all referalls tend to point back to you, all the better. But if you’re a medical (general) translator, to borrow ATA’s moniker, and your hobby is orchids or horses, or your background includes submarine technology or rare stamps, include it.

Also, if you’ve learned a new subject, update your resume and send it again. Despite claiming a database of thousands of translators, agencies are always on the lookout for competent, reliable and specialized translators. Project managers simply do not have the time to use more than 30 or 40 people on a regular basis, thus many translator profiles and resumes lay dormant, often forever, and a freshly sent resume often gets a look before older ones, especially if it fits an immediate need.

I also like to see details of past jobs. Interpreters seem to include these more than translators — perhaps because the number of jobs is fewer and they can be summarized on a couple of pages. But translators could learn from this. If the subject matter of certain past jobs stands out from the plain vanilla contracts or licence agreements, give some details on your resume (without mentioning the parties obviously).

Include all of the competencies you’ve acquired either on the job or through past occupations or pastimes that may put you in a better position to translate a specific document than someone else. These definitely get noticed.