Getting Lost in Translation

Sometimes a simple word can make all the difference. Consider a recent project to develop a brand strategy platform. My task: to test a series of positioning statements across the globe. Our client: one of the world’s largest technology companies. The claims: close customer collaboration, the ability to deliver seamless solutions and the belief that people can do amazing things. The result? Customers in France had an immediate and visceral reaction to the phrase “natural collaborators,” which conjured images of Vichy France. The Brazilians had no problem with the word “collaborators,” but took issue with the word “natural” (translated as “naturais,” but interpreted as “average” or “normal”). Living in a country with the world’s largest Roman Catholic population, the Brazilians were downright annoyed when they read the word “creed/crença” (translation of “belief”), arguing that discussing religion in the context of IT was wildly inappropriate.

These types of missteps are all too common and can be expected when the translation process isn’t given its due.

Global research has high stakes for brands—requiring more money, more time, more complexity, more participants and more choreography. It goes without saying that when you’re conducting research in multiple languages and regions, translation is critical. But rarely is it straightforward.

You might imagine that translation of research materials—the questionnaire, moderator guide, positioning statements—is a fairly straightforward process; in fact, deceptively simple. I’ve even had clients ask if they could do their translations using Babel Fish or Google Translate. Others have suggested that they dig up a colleague who is fluent in the target language. This is unadvisable for several reasons. Often, the layperson will transliterate—a literal, word-for-word translation, which results in a stiff and awkward read. What you need instead is not just fluency in the target language, but a much deeper understanding of how to use certain idioms, tone and syntax (rules or principles that govern sentence structure).

So, let’s talk about best practices.

Translation should be a three-step process:

Step 1: Professional translation
This means specialists—in-country, native speakers. There are significant differences between Québécois French and Parisian French. You don’t want to uncover those discrepancies when you’re knee-deep in data—cursing the fact that you didn’t find someone fluent in the target dialect.

Step 2: Client review
This step is critical for two reasons. First, it provides a second filter that the translator doesn’t have—subject-matter expertise. For instance, I would not expect a translator to know industry jargon (think pharmaceutical, high-tech, finance). When it comes to this specialized vocabulary, the client should adjudicate. As important is the role of the in-country client team—which is why I often ask for direction from the local offices as opposed to headquarters. These in-country employees can alert you to potential landmines—commonly, issues that have absolutely nothing to do with translation (e.g., labor protests, industry norms, competitive dynamics, cultural issues, topics in the news). The client team located at headquarters can be blissfully unaware of these issues, which can result in a slew of unanticipated responses from respondents.

Step 3: Back- or reverse-translation
As the name implies, this simply means translating the first translation back into English (or whatever language the researcher and client team are using). This is definitely a situation where “checks and balances” prevail. The back-translation should be done by a second translator—not the individual who did the first translation. I would recommend making this the last step (as opposed to the client review). Back-translation is your chance to validate the translator’s work, as well as any subsequent edits from the client team. Avoid the sin of transliteration here, too. The word-for-word translation can be poles apart from the author’s original intent.

Says Richard S. Paegelow, from the American Translators Association Chronicle (August 2008):

“For example, the Spanish word prima has more than 15 possible translations, including “the cost of an insurance policy” and “female cousin.” An incorrect back translation leads to a false conclusion about the accuracy of the original translation.”

In addition to the three-step process, other issues can surface. For instance, there are times when problems are revealed in the source text phrasing. This occurred recently when I was testing positioning statements in the U.S., India, U.K., Brazil, China and France. Although grammatically correct, several of the phrases were far too wordy for the respondents (English- as well as non-English speakers) to fully comprehend. In such a case, the translation team should not be shy in raising the issue with the original author. One phrase in particular—“the next generation of technologies that are no longer disconnected and unaware”—proved to be particularly confusing for Portuguese speakers. Because the direct translation (“inconscientes”) is reserved for people and not things, the translation of “unaware” became “alienated/alienadas.”

This type of rigor is not reserved for foreign languages. The same holds true for different dialects. In a Johannesburg study, we quickly uncovered some problems with the English ad copy. “Coffers” was used in one ad headline, but was seen by English-speakers as old-fashioned and very close to the Afrikaans word for suitcase (“koffer”).

Acronyms can prove particularly challenging—in English and any other language. In another study for a large credit card company, one advertisement used the acronym ACH (Automated Clearing House). But the British respondents we interviewed were unfamiliar with that term, and instead used BACS (Bankers Automated Clearing System). Meanwhile, the South Africans use another acronym altogether—ACB (Automated Clearing Bureau).

A rigorous translation process requires time. That time will depend upon the translation vendor you’re working with, the amount of copy that must be translated and the speed with which the client team can turn around reviews and approvals. As a rule of thumb, we schedule 1½–2 weeks for the entire translation process. This includes the original translation, the client review and the back-translation. But there are ways to save time. If you have the appetite for rush fees, you can expedite the process with most vendors. If the translated text will appear online, some vendors can save time by working directly with HTML/XML files because they don’t have to copy and paste translations from Word or Excel. Also, some vendors will provide online portals to help accelerate the client review cycle. On the flip side, additional time is needed if you’re translating visual elements (e.g., text on notional applications such as packaging, business cards or advertisements). This may require specification of non-Latin scripts, such as Simplified Chinese, Arabic or Devanagari.

Budgeting varies, depending upon the word count, number and type of languages. Generally, we assume $2,500 per language for translation of a 20-minute questionnaire. If you’re translating an extensive volume of text, some vendors will provide discounted rates based on the percentage of repeat text.

In summary, the translation process may seem like a long and tortuous journey, but it is one where shortcuts should be avoided. If you follow the three steps, you’ll avoid last-minute surprises that can derail your research project.

View PDF

Register now to comment

Read Our Newsletter