Translation, Back-translation

When a survey instrument or questionnaire is developed for use across multiple cultures, a translation of the instrument may be necessary. Translation can be a complex process. Although creating a translation from one language to another that preserves the functional equivalence of the words is not too difficult, creating a translation that preserves the conceptual equivalence of whole sentences and paragraphs can be quite difficult. In addition, cultural differences may require changes to the instrument format or the interviewing procedure. Examples of issues in translation of survey instruments include:
  • Colloquialisms or slogans may translate badly from one language to another. For example, note the following message from a Copenhagen airline ticket office: "We take your bags and send them in all directions" [1]. Although this statement may be functionally equivalent to the original Danish, it would not be particularly reassuring to a native English speaker.

  • Words in one language may simply not translate to another language; either the concept conveyed by the word requires a multi-word explanation in the second language, or the concept conveyed by the word does not have an equivalent concept available in the second language.

    As an example of the former, English contains relatively few words to describe snow (i.e., snow, hail, sleet), but the language of the Central Alaskan Yupik Eskimos of North America contains many single words that more finely classify and define various types of snow (for example, "qengaruk" means snow bank, and "navcaq” means snow cornice, or snow about to collapse) [2].

    As an example of the latter, in the Sierra Leone languages Krio, Mende, and Koranko, the concepts of the threat of danger and the fear of danger are described with identical language, as one concept. In the Sierra Leone languages Kono, Temne, and Limba, however, it is possible to express the two different concepts.

One particular technique for making conceptual equivalence across languages more likely is back-translation. Back-translation is the translation of a survey instrument or questionnaire that has already been translated into a foreign language back to the original language. If at all possible, the back-translation should be done by a different translator than the one who did the forward-translation. After the back-translation, the original and back-translated instruments are compared and points of divergence are noted. The translation is then corrected to more accurately reflect the intent of the wording in the original language.

Alternatives to the use of back-translation include:

  • Multiple-forward translation. This is when two or more translators both translate the survey instrument from the original language to the new language, and the versions of the instrument in the new language are then compared.

  • Translation review by bilingual judges. This can be thought of as a variation of the back-translation procedure, but involves judges reviewing both the original-language version of the instrument and the new-language version of the instrument.

An ideal procedure for a multi-cultural survey will consider the multiple languages of the sample population from the conception of the questionnaire, and allow simultaneous drafting of the questionnaire in the multiple languages. It can be dangerous to draft a questionnaire in one language and then pass it over to translators, even with a back-translation check: the concepts contained in the survey might simply not exist in the surveyed population. That would be a cultural difference, not a matter of translation.

If it is impossible to develop simultaneous, multi-language questionnaires, the best procedure would be to use whatever translation techniques allow the instrument to collect comparable information across multiple languages and cultures. The use of translation and back-translation should be considered a minimum requirement towards assuring the quality of the survey data obtained [3].

1. From House of Horrors (19 December 2006).

2. From Counting Eskimo Words for Snow: A Citizen's Guide (12 September 2007).

3. Other sources for this encyclopedia entry include http://www.asiamarketresearch.com/ (19 December 2006) and Maxwell, B., “Translation and Cultural Adaptation of the Survey Instruments” in Martin, M.O., and Kelly, D.L. (eds.), Third International Mathematics and Science Study (TIMSS) Technical Report, Volume I: Design and Development, 1996, accessed from http://timss.bc.edu/timss1995i/TIMSSPDF/TRCHP8.PDF (19 December 2006).