Human Touch: Should Brands Avoid Machine Translation?

Imagine yourself in a foreign country where you don’t know the language. Even simple tasks like ordering food and checking into a hotel feel difficult. Now, multiply that frustration by a factor of several thousand and you’ll understand the challenge facing global marketers.

Given the strict deadlines, tight budgets, and million-word workloads, efficiency is a top priority. But with so many horror stories surrounding translations gone wrong, can marketers really trust automated solutions to deliver professional results?

Man vs. Machine

As artificial intelligence continues to disrupt unexpected sectors of the economy, the language services industry has recently been cast as a key bellwether. If machine translation tools can match or replace professional translators, what other essential human skills could software soon automate?

Those who have been following the debate closely, however, can tell you that the answer is not as ominous as headlines often suggest. In fact, there’s a pretty clear gap between what humans and machines do well in the language services space.

Machine translation has no rivals when it comes to cost or speed. Powerful, automated engines will help you process thousands of words in minutes — and only charge you a few pennies for the privilege. Conversely, it would take several days and several hundred dollars for a professional human translator to complete the same task.

Humans have an insurmountable edge when it comes to quality and originality, however. Even as technology matures and machine learning adapts, man repeatedly trumps machine in translation quality tests. And when machines misinterpret your brand message, the costs of correcting that communication may be more than merely financial.

Man + Machine

Savvy global marketers don’t sacrifice. Instead, they deploy each type of translation resource in a context that plays to its strengths.

Content that demands uncommon originality or uncompromising quality should still be assigned to human translators. Marketing slogans, conversion copy, and checkout flows are all examples of text most brands would deem too sensitive to trust to a machine. In fact, multiple layers of human editing may be justified as well.

Not every character of content holds the same commercial value, however. That’s why it absolutely is safe to consider using machine translation for high-volume, less-essential text that needs to be translated quickly.

Take customer reviews, for instance. A translation error in that context may be regrettable, but it is far from fatal to the business. And the risk is hardly a reason to rationalize spending several weeks and several thousand dollars professionally translating user-generated content.

But even if you do ultimately decide that none of your content is fit for machine translation, there’s still one automated advantage that’s worth your attention.

Translation management software can radically accelerate the work surrounding linguistics. From instantly gathering your initial collection to automatically pushing projects down the approval chain, machine intelligence can absorb a number of tedious tasks and free humans to pursue the critical and creative thinking they do best.

Learn More

For a closer look at how global brands get the best out of both man and machine, explore the potential of adaptive machine translation in our Smartling + Lilt Overview.

[Editor’s Note: This post was originally published in November 2015 and has recently been updated for accuracy and clarity.]