Machine Translation: Electric Car of the Language Services Industry?

The Obama Administration recently challenged the language services industry to achieve “automatic, highly accurate and real-time translation between the major languages of the world.”

It was a bold but reasonable statement, and one that directly challenges human translators and multilingual content professionals, whose work millions of people depend on to eliminate global language barriers. It has significant implications for CMS and content managers, too, as CMS tools and professionals increasingly need to efficiently manage multilingual web properties.

I immediately wondered how the industry would respond. I didn’t have to wait long to find out.

The next day Dr. Jiri Stejskal, president of the American Translators Association (ATA), published a response letter to President Obama. The letter urged Obama “to take a long-term approach to language security by investing in human skills and promoting greater awareness of and expertise in foreign languages.” (The text was actually bolded in the letter.)

The next sentence in the ATA’s letter reveals an attitude similar to that displayed by the automotive industry in recent decades on the issue of electric cars:

Are we against technology? Certainly not – in fact, most professional translators already use computer tools to speed up their work.

With all due respect, this is about as cogent a statement as a car industry exec downplaying electric cars by stating that gas-powered vehicles already incorporate electrical systems. I don’t want to carry the comparison too far, but I can’t help but see a legitimate parallel.

When California created more stringent fuel efficiency guidelines in the early 1990s, automakers quickly responded, in part, by rolling out electric vehicles to comply. However, the automakers were widely accused of deliberate self-sabotage, failing to adequately promote their electric vehicles in order to create the false impression that consumers were not interested in electric cars, while fighting against the [California] mandate using lobbyists and lawsuits. (Wikipedia)

From the ATA letter to President Obama:

Despite all the changes wrought in our lives by technological advances, no computer can match the language skills of a five-year-old child. The reason is simple: Computers cannot translate effectively – that is, they cannot entirely convey meaning from one language to another…

Let me be clear about my stance on this: I’ve delivered numerous presentations on what a disaster it can be to rely on machine translation alone. Or to use it in the wrong situations.

At its current level of sophistication and quality of output, machine translation should be avoided for creative copy: branding, marketing, advertising, literature, poetry and so forth. In this type of writing, language is often manipulated, plays on words made and other devices employed to purposely use language creatively and outside the way it has traditionally been interpreted. By definition, this creates a serious hurdle for accurate machine translation.

Sensitive, highly nuanced documents related to diplomacy, law, and medicine, among others, should also not typically be entrusted to machine translation.

But even in some of these cases, a hybrid of human and machine translation is a perfectly viable option. By employing a solution that is customized for the domain or profession, output is improved through limiting the scope of allowable substitutions. This technique is particularly effective in domains where formal or formulaic language is used. It follows then that machine translation of government and legal documents more readily produces usable output than conversational or less standardized text. (Wikipedia)

No one is saying that quality human translation is going away any time soon. In fact, I speculate we’ll need huge numbers of skilled human translators for decades to come. But the misgivings and fear – bordering at times on enmity – that currently exist within the professional translation associations toward those toiling away on improving machine translation needs to give way to acceptance and greater collaboration between them – especially as the volume of web content continues to grow, and the opportunities to create truly global communications expands relentlessly.

Just as the planet cannot sustain an ever-increasing number of petroleum-fueled, inefficient cars on the road, global communication, education and cooperation can no longer be constrained by the bottleneck of, in the ATA’s words, qualified human translators. As I said in my last article here, the time for machine translation has come.

About the Author

Steve has spent more than a decade building global brands and helping clients succeed online in the U.S. and abroad. A seasoned online business strategist and marketer, Steve has significant experience in the technology, entertainment, health care, publishing and travel & tourism industries, among others. In addition to writing magazine articles and white papers about international marketing and technology trends, Steve addresses audiences around the country, from the Japan Society of New York to the Software Association of Oregon. Steve is a citizen of the world, with a focus on Asia, and currently resides in Portland, Oregon where he works for Connective DX, a global agency serving clients from offices in Portland and Boston.

More articles from Steve Kemper


4 responses… read them below or add one.

  1. M.D. says:

    “I speculate we’ll need huge numbers of skilled human translators for decades to come”
    If that happens, the professionals who work today, considering basic laws of economics, will earn even less for a highly specialized task. The associations should, as it’s clear, defend the point of view of professionals not the perspective of the very entities who exploit them. There is only one thing professional translators can do considering the current situation: to refuse to review projects that use MT. If MT is that good, let it review its own work.

  2. Steve Kemper says:

    Thanks for your comment, M.D. As someone who has done professional translation and has many friends still doing it, I hear you. I wasn’t blowing smoke when I said I have slides in recent presentations cautioning against machine translation for most purposes.

    That said, the landscape is changing and it’s not an option to fail to acknowledge that. I understand the frustration, but refusing to review machine translation output seems like a viable option only until it’s not anymore. Granted, that may still be a few years out. And, as I said, there will likely always be plenty of content for specialists who don’t want to be in the mix with machine translation at all. But that doesn’t negate the growing need for it.

    Thanks again!

  3. Eve Bodeux says:

    I love this comment: “If MT is that good, let it review its own work.” I don’t think you (Steve) really make the point clear enough that, in the future (and even now), there will always be need for human translators and there will also be need of MT due to huge volumes of translation needed. MT is suited to a certain type of translation (as you note). I would say, the US administration shows some naivety about how languages and technology work together. Sure, they can achieve their goal of “accurate” and “real-time”, but only in controlled situations – not what they mean, I am sure. People who do not work with languages always think it is easier than it is…

  4. Mike Unwalla says:

    @M.D.: The associations should, as it’s clear, defend the point of view of professionals not the perspective of the very entities who exploit them.

    A professional person has a duty to his or her client. If the best option for the client is MT, a professional person should tell the client. To be credible, a professional association must do more than only defend the point of view of professionals. The clients are important, too.

Leave a Reply
  1. Fields marked with * are required.
  2. We will not publish your email.