Language as a weapon

Posted on September 8, 2010 by Donna Parrish

A whistleblower tells ABC News (US) that many interpreters hired to work for the US army in Afghanistan cannot speak the local language very well. Story here. This is probably no huge surprise, and definitely not a surprise to the armed forces personnel who are trying to “win the hearts and minds” of the Afghan people.

What is additionally distressing is that, according to Paul Funk, the whistleblower, language proficiency tests were falsified by Mission Essential Personnel in order to get more contractors working, earning MEP the accompanying fees. Doing this with one interpreter — when you consider the possible consequences — is heinous. But, according to the report, 25% of the interpreters did not pass their test.

Marc Peltier, MEP’s COO, said that an internal company survey that showed 82 percent of its customers were satisfied with the performance of its translators. When I was in school, 82% was a solid “C” grade. Now, I am sure there are some excellent interpreters in place. But to have an overall satisfaction rate of 82%, that means there are some interpreters endangering US and Afghan people (even more than they are already), with consequences I shudder to think about.

According to ABC, some Afghan veterans have rated the value of a skilled interpreter as equal to that of a working weapon or sturdy body armor.

This is not news to us in the language industry. But evidently that importance has not been brought home to the contractors or the agencies that hire them.

Post to Twitter

Posted in Uncategorized | 1 Comment

Measuring Information Quality: What’s Missing?

Posted on August 18, 2010 by ultan

A lot written about information quality recently. But how do we measure it?

There is a perception out there that information quality generally is very poor. But how do we know? All that digital content out there, but most of it is never read, and only a fraction of that is ever translated (and translators are often the only readers of that content). Just because a user guide isn’t read doesn’t mean the information itself within is poor quality. Perhaps, the accompanying software’s usability is spectacularly good and so it isn’t needed? Who knows?

In the GILT industry, information quality is all too often only assessed in terms of the cost, time, and effort to produce and then translate content (and usually one function is measured in isolation of the other). We have all kinds of metrics about “spend” published, time-to-market statistics analyzed, the opinions of professional linguists and terminologists debated, complicated mathematical formulae promulgated (trust me, if you reach that level you’ve clearly no real work to do), QA checklists written, certifications from standards bodies waved under our noses, and all the rest, in an attempt to define information quality. All good, though how efficient or applicable to the real world some of these things are is debatable.

However, often there’s a decider of information quality that is missing from these methodologies: the user of the information.

We need to move the key determinant of information quality to the user community: engaging users, and research, analyzing and understanding how users search for, retrieve, and use information to solve problems. For example, what search keywords and phrases do they use? Which pages do they read the most? Which parts of those pages are read and how? And so on.

The tools and opportunities for this are everywhere. Ever studied web server logs? Done any eye tracking studies (see image below) before and after an information quality project? Conducted comprehension studies on the outputs? Observed how real users consume information? Found out what terminology they actually use when helping each other, on support forums, and when they customize and extend software?


So, let’s look at costs and translatability issues, post-editing metrics, number of flagged errors in QA, and so on, sure. But let’s connect it to the user experience too, regardless of language, and give the user the final say. Make users the arbiters of information quality.

Otherwise, we’re really just talking to ourselves.

Post to Twitter

Posted in Language in the News, Personalization and Design | Tagged , , , , , , , , , | 3 Comments

Natural User Interface Gestures. Are They Global?

Posted on August 18, 2010 by ultan

I was reading an interesting BBC article about how language changes are being driven by the internet. Examples are cited from English, Ukrainian, and more. In its own right, this is an interesting subject, reminding us that such changes are global.

How such changes can be reflected in our information quality efforts, and how applicable they are to everyday ‘plain’ use as opposed to specialized, domain-specific needs is a subject I’ll return to.

My eye was drawn to the following in the article:

The dreaded force-quit process of pressing ‘Control, Alt, Delete’ is known as Дуля (dulya).

“A dulya is an old-fashioned Ukrainian gesture using two fingers and a thumb – something similar to giving a finger in Anglo-Saxon cultures,” she said.

“And you need three fingers to press the buttons. So it’s like telling somebody (a computer in this case) to get lost.”

This made me consider a section in the 150-page Apple iPad User Guide (PDF), listing out the natural user interface, finger-based, gestures for that device. For example:

  • Two-finger flick up: Read all, from the top of the screen.
  • Three-finger flick up or down: Scroll one page at a time

I wonder if some of these might be ‘tricky’, internationally? Certainly the use of hand and finger gestures (or, indeed, of other body parts) has long been regarded as a potential international no-no when designing graphics. I’d love to hear your opinion.

For some, it would seem that some of these gestures are even positively disgusting!

Post to Twitter

Posted in Language in the News, Personalization and Design | Tagged , , , , , | 2 Comments

¿Socially-Enabled What?

Posted on July 28, 2010 by ultan

The mystery of Galvão birds. Twitter use in Japan and Brazil. Different trending topics in the US depending on whether you are black or white. The “wisdom of the flock.” How the new media doesn’t help us. “Imaginary cosmopolitanism.” How “we think we’re getting a wide view of the globe.” Except we’re not. We’re in social media “filter bubbles.” We don’t see the wider picture.

There are lots of people who don’t get the notion of true global connectivity, replacing the essence of global relationships between real people with a ’Globish‘ Anglo-American-centric debate about digital infrastructure instead. But, this guy coming up does.

If you want to move on past that tired social media guru “we’re all connected now through Facebook and Twitter” stuff (actually the second easiest job in the industry,) then he’s for you.

So, I urge you to view the very refreshing and insightful TED talk called The Need for Cultural Translation with Community Media by Ethan Zuckerman of Global Voices. PBS tells us his talk is about exploding “the big myth that the web is bringing us closer to other cultures or countries – when we’re on the web, we’re basically in our own small islands of our social networks.”

Ethan says we need “DJs … skilled human curators” who “speak the language of the West and of other cultures at the same time.” We need more “xenophiles,” those “bridge figures” who can and do cross cultural divides, rewiring information systems across language and culture so we get that wider picture.

The talk has subtitles in English and Portuguese different languages.

We need more Ethans too.

Post to Twitter

Posted in Language in the News | Tagged , , , , , , | Leave a comment

Far Too Easy: Knocking Machine Translation

Posted on July 20, 2010 by ultan

The easiest job in the translation industry has to be the one disparaging machine translation (MT). As a former boss of mine would say when he saw such e-mails: “Those guys are not busy.”

A day doesn’t pass without some link being circulated my way, either on Twitter or e-mail, highlighting poor applications of machine translation. Some dismally poor translated output is shown (almost always the result of the use of a free online translation service) along with some subjective declaration about how using Google Translate, Bing Translator, and others, is dangerous to your brand’s reputation, will drive away your customers, and so on (translation: “I haven’t a clue about the potential or proper application of MT really, but I think my job is under threat or you must pay me to do it instead”).

Of course feeding text randomly into a free online MT engine will result in a less than perfect translation, at times by a very long shot indeed. Except, you can get equally dreadful results by giving poor quality information, with no context or glossary, to the wrong type of human translator too (and still pay for the privilege).

Such tiresome “MT be bad” examples have brought nothing new to the translation debate in two decades. And then, of course, there are times when these free online services do a very good job indeed. But those examples are rarely declared. Natch.

For those who do understand the potential and application of MT, they must counter all this stuff by the mass circulation of correct examples of MT output. Match every lazy, bad example with a better, applicable one. The role of information quality, of customization, and post-editing needs to be explained more. And, critically, the role of MT in alleviating information poverty too must be brought to the fore.

How many of the MT knockers out there are permanently offering translation services for free for high value, life-saving information in Asia or Africa?

Post to Twitter

Posted in Translation Technology | Tagged , , | 7 Comments