"That is totally me sausage" What's Holding Back Multilingual Business?

It’s hard to avoid the tired old metaphors of “explosion” and “flood” when speaking of the exponential expansion of Internet and other digital content. Consider this startling factoid: According to Google’s recent Think Quarterly on innovation, “In 2003, five exabytes of information existed. Now we generate that every two days.”

Three key factors drive the spike in inquiries Forrester has seen this year about translation technologies, practices, and service providers. The first is this inundation of content. Even if only a tiny percentage of it could be valuable in other languages, it overwhelms the capacity of human translators.

Second, the strongest growth in consumerism and purchasing power is happening in countries with young populations and expanding economies, such as the so-called BRICs and CIVETs. (That’s Brazil, Russia, India, and China plus the more obscure Colombia, Indonesia, Vietnam, Egypt, and Turkey.) Maintaining a consumer-oriented web presence in a single language today is akin to selling only to left-handed customers.

(And it’s futile anyway. Site visitors can use services like Google Translate to render your site in dozens of available languages, without your knowledge or any control over the quality of the translations.)

Finally, even if you ignore the consumer opportunities – and many companies still do – there’s the challenge of multilingual employee collaboration. Due to an overemphasis on the technology, many global enterprises seem surprised to learn that their slick new collaboration platform doesn’t do much to help the French collaborate with the Koreans or the Saudis if there is no common language.

Translation service providers have responded with new or expanded offerings, innovative business models, and user-friendly plug-ins. Last year, for example, SDL bought statistical machine translation vendor Language Weaver in order to extend and complement its human-based services. (Here's my blog on that buy.) Smartling, in contrast, positions itself as a software platform that facilitates and manages the creation of multilingual content, and provides no translation services of its own.

And yet, despite the titanic need and the ample solutions, both vendors and Forrester clients agree that enterprises are not rushing to embrace expanded use of translation. (“We’re calling to see what other companies are doing,” begins a typical inquiry about translation. And my answer typically begins, “Other companies are . . . calling to see what other companies are doing.”)

Conversations with vendors and end users reveal three main FUD-factors for this hesitancy:

·         It’s not secure enough. Whether machine- or human-based, the majority of translation processes require sending corporate content outside of the firewall. The growing use of cloud services exacerbates this anxiety.

·         It’s not good enough. “Das ist mir völlig Wurst” is an idiomatic German expression meaning “It’s all the same to me.” The free web-based Google Translator renders this as “That is totally me sausage,” which is rigorously literal and very funny, but useless as a unit of communication.

·         It’s not affordable enough. Fully human-based translation, at up to 30 cents per word, is a big-ticket item and restricts translation to very high-value content.  Despite sometimes dramatic decreases in costs due to machine translation technologies, the stigma of “too expensive” clings to content translation like grease to a sausage.

There are good (if not simple) responses to these fears, uncertainties, and doubts. I’ll discuss these and related challenges around content management in a multichannel, multilingual business environment with my colleague Stephen Powers at the Content and Collaboration Forum on September 23rd in Boston. In the meantime, I’d like to hear from you: What are your main drivers for increased translation of external or internal content? What questions and hesitations do you have about it? I would be very happy to use some of your business cases and experiences in our Forum presentation.


Localizing content + taxonomy

You bring up a great point, but shouldn't we clarify the difference between "translation" and "localization"? I see translation as the nearly-word-for-word transformation that leads to the hilarious example you give above. To me, "localization" is what's really needed, where meaning and context is provided, not just the machine-mind of exchanging bits for bits. For our company, we have to "localize" taxonomies for clients, because assigning meaning isn't a one-to-one task, especially for taxonomies, which are all about the tacit and explicit understanding of the many shades of a language.

While it's true that

While it's true that translation and localization are different but interrelated concepts, I think there is a mild confusion here. Localization deals with adapting a product or service —software game, movie, corporate tagline, etc.— to a locale or specific geography. An example would be adding the color red to corporate stationery for a subsidiary that will open in Shanghai, because red has the connotation of good luck in Chinese culture.

As such, localization is a category within translation. If the phrase “That is totally me sausage” is the output of MT or HT (machine translation, human translation, respectively), that is the result of poor, subpar translation. By definition, translation involves conversion, decoding, transcoding, recoding and transcreation of content, be it text, graphics, concepts or abstractions, tangible or intangible ones. The sausage example is just poor recoding, hilarious as it sounds. This reminds me a bit of Chomsky's example of generative grammar: a nonsensical but perfectly grammatical sentence, such as “Colorless green ideas sleep furiously.”

Same Drivers - Increased Scale - Shorter Cycles

While the drivers for translations remain the same as earlier, the need of the hour is the scale and the speed with which it needs to be done.

The typical drivers are

a) the aspiration to increase user engagement levels
b) entry into newer markets such as the BRICs and CIVETs and
c) new product launches or product enhancements

In quiet a few situations, there is a lack of a well oiled machinery to get the translations done by different external parties across the globe. Systems are being enhanced to enable this.

The need for a secure and optimized way of getting the translations done in a timely, high quality, high volume and cost effective manner is becoming increasingly evident.

It's Colombia, not Columbia

A provocative piece, made less relevant by an all-too-common inexcusable typo that mars a country's name. The name is Colombia, not Columbia.

My apologies for the

My apologies for the Colombian typo! And thanks for the other comments and many retweets. @Sam, I think you're right that the drivers are largely the same. However, precisely because machine translation dramatically increases speed, certain types of content, such as chat and customer support exchanges, can now be considered for translation for the first time.