I have been using oDesk.com lately to have survey text translated into a number of different languages. For my Facebook Privacy Attitudes survey in 2012, the total was 13 different languages.
Some oDesk contractors are confused as to why I would get their work double-checked. I have even been told by one ‘experienced’ contractor that double-checking their work is a ‘waste of time’. To me, a reaction like this actually shows a lack of experience. That is, any experienced translator, especially one whose work includes translation of survey texts (survey instruments), will know the importance of having one’s work looked over by a ‘fresh’ set of eyes. No matter how experienced the translator is, this will always be the case.
So, here is an explanation as to why I get at least one independent translation check for every original translation:
Even if I had the original survey translation done by a career veteran translator of 30 plus years, academic cross-cultural research methodology would require that I have an independent check. Why?
Look at it this way: If a chemist – a physical scientist – wants to measure some kind of chemical process, they would construct some kind of physical instrument to measure the process. They would use high quality glass, sensors, specific electronics, etc. Once they had constructed their instrument, they would test the instrument to make sure that there were no inconsistencies in the construction of the instrument. Even if the most experienced scientist created an instrument themselves, they would test and retest the instrument, and get others to look over their work and test it, to make sure it is accurate. If the instrument is not accurate, then any measurements they take with the instrument will be useless, and totally unreliable; the scientist will not be able to make any conclusions from the data.
It is the same in the academic field of social and psychological science. In this field, we create ‘instruments’ for measuring cultural differences, psychological processes etc. These instruments consist of ‘textual scales’. No matter how experienced the person was who created the scales, an experienced social scientist would, in the name of scale validity (that is, instrument accuracy), have someone look over their work; ‘test’ the instrument, so to speak. This is even more so when translation is added to the mix. Even very small errors in translation can render data gathered from survey instruments useless.
In this way, the fact that I want to have my “translated instruments” checked is not because I doubt contractors’ abilities. It is because that is what science requires.
Now, this does not mean that translations should be literal translations. An important concept in survey text is ecological validity. That is, if the survey text does not seem natural to a reader in the target language, then there is a possibility that the survey participant will give up half-way through the survey, or, worse still, finish the survey but be so distracted by the strange looking text that they will not answer in a natural way. Therefore, in a translation, as much meaning as possible should be retained, with as much ‘naturalness’ in the target language as possible.
So there you have it. That’s why I request independent translation checks. It’s not because I doubt contractors’ (your) abilities. Independent translation checks are just part of the process of creating an accurate instrument; the construction of which you are an integral part.
For an even more in-depth, systematic discussion on survey text translation, see this article: http://www.thesportjournal.org/article/cross-cultural-research-and-back-translation