In an increasingly globalized world, researchers often want to conduct cross-cultural surveys to examine some phenomenon in different cultural contexts. These studies often necessitate translating a survey from one language to one or more other languages. When translating surveys, researchers should aim for having multiple versions of a survey instrument that are as similar as possible in their meaning and interpretation despite language differences. The translation process, though, raises a number of issues that complicate that goal.
How to Translate
It may be tempting to use a machine translator such as Google Translate to convert a survey, but these tools have substantial limitations in their ability to translate accurately. Language is usually too complex for most automated translators to convert correctly, taking into account the context of phrases and the variations on language that can express the same sentiment. A simple way to illustrate this problem is to enter a phrase into Google Translate in English, convert it to some other language, and then translate it back to English. You likely will not get the same sentence you started with when you back-translate the phrase.
For the highest quality translations, human translators are best to ensure the correct meaning of a question is conveyed. It is essential, however, that the person tasked with the translation is equally comfortable in both English and the target language. If possible, involve multiple translators in the process so that errors in translation can be reduced.
The following example, found online, is the verbatim text of the back of a bootleg Chinese DVD:
Vincent LaMarca job is to arrest kieelers, but this job is different. The suspect he is tracking is his own son. He a cop, LaMarca must tring the acclised to Justice. As a Pather, he must find wining actoce Robert De Niro, Frances Mc Dernad and James Franco. De Nice memorably plays LaMarca, burdened by tragedy and his failings as a father…and now putting his life on the line to do by his family and profession. “Do Nino has been sawing us greet stutt for30 years bou. But in this movie he shows us something nes.” The San Fancisce Chroniole Mick La’s Mick LaSalle wrote. Put another way: a greet actor rederines his greatness in City by the Sea.
This is clearly a phonetic translation. The description of the movie as a whole conveys the sense of the plot, but the clear problems are the spelling, grammar, and sentence structure. This is an excellent example of the problems and complexity involved in translations, and a reminder that producing poor translations is dangerous to survey quality. It is essential that the person(s) tasked with the translations be genuine experts in the target language and that their work, when possible, should be independently verified.
Finding a qualified translator can be difficult, though, depending on the subject matter of the survey. If the content is related to a specialized field, it is critical that the translator possess some basic knowledge of the survey subject matter. Using a translator without content-specific knowledge increases the probability of translation errors substantially. Indeed, if the survey is about technical or scientific content, most fluent translator may not be qualified.
Even for single language surveys ambiguity can be a major concern, for multiple language surveys ambiguity can be one of the biggest obstacles to translation. To minimize the problem of ambiguity the source survey needs to be made as clear as possible. Ambiguity is not always obvious and can sometimes only come to light on a second or third reading, or by two people interpreting the same question differently. Some ambiguity can go unnoticed and can lead to respondents answering a completely different question to that intended by the survey’s author, in which case the results will be fundamentally flawed.
There are many regional differences even within the same language. In English, for example, British English and American English often differ in their vocabulary. The word “pants” in British English refers to what Americans would call underwear, but the same word to Americans refers to what the British would call trousers. The same is true of other languages. Vocabulary and even sentence construction are often not the same in the French that people in France speak, for example, as compared to the French that people in Quebec speak.
If a survey is to be published across two or more dialects of the same language, all differences need to be addressed. When having the survey professionally translated into a non-English language, the person responsible for the translation must know the dialect to which they are translating. If it is to be accepted that there are significant and potently problematic differences between established dialects of the same language, take time to consider the potential extent of fundamental differences between two distinctly different languages.
The conventions of writing dates often differ across cultures. Americans tend to put the month before the day (e.g. January 11, 2000; 1/11/2000). The British, for example, tend to place the day before the month (e.g. 11 January, 2000; 11/1/2000). Be aware of the standard when adapting a survey across languages or dialects of the same language
Author: Patrick R. Miller, DISM Survey Research Associate