FSD Bulletin

Issue 31 (3/2010)
10.12.2010

ISSN 1457-7682

Front page
Previous issues
Editorial staff

» latest issue

FSD Bulletin is the electronic newsletter of the Finnish Social Science Data Archive. The Bulletin provides information and news related to the data archive and social science research.


FSD

Finnish Social Science Data Archive
Tel: +358 40 190 1432
Fax: +358 3 343 9088
E-mail: fsd@uta.fi


Team-based Approach Best for Questionnaire Translations

Taina Jääskeläinen

Translating questionnaires from one language into another is challenging. Unsuccessful translations may make the question inapplicable for cross-national comparisons. The two main targets of survey translation are: a) questions should measure the same construct across cultures/languages (construct equivalence), and b) questions should be understandable to the respondents and function independently in the target culture (=instrumental translation).

Translation process

DictionariesThe traditional method of one translator translating the questionnaire and someone else back-translating it for quality review has received critique over the recent years. Nowadays several-stage TRAPD model involving a team is recommended: Translation, Review, Adjudication, Pretesting and Documentation. The process starts with two or more independent translations, which the team reviews and then decides upon the version, while adjudication means deciding on the final version, possibly after expert comments. At the last stage, the translated questionnaire should be read carefully without looking at the original, as it should function independently as a survey instrument.

Translation team members need to have knowledge of the study, instrument design and the subject matter, as well as cultural and linguistic knowledge required to produce appropriate questionnaires in the target language. One person may have more than one expertise area mentioned above. It is crucial that the team has knowledge of what exact concept each question is meant to measure.

Linguistic challenges

The general recommendation for questionnaire translation is that the translated question should correspond as much as possible to the original. However, translation teams are in perpetuity torn between the requirement of formal, semantic equivalence and the requirement of comparable constructs. Literal translation may lead to too formal, complicated or awkward text, or the unidiomatic or improper use of the target language.

Close translations are often not even possible since translation is by nature interpretative. Does the word 'government' appearing in a question refer to the cabinet (prime minister and ministers), both the cabinet and Parliament, or state administration in general (including the above plus the ministries and other government agencies) or public administration in general (adding local government to the above)? Since most languages do not have a term equivalent of the widely scoped 'government' the exact meaning of the word must always be interpreted in translation.

Across languages, concepts have areas of shared (etic) meaning but also facets of meaning specific to each culture (emic meaning). The challenge in cross-cultural surveys is to disentangle emic and etic meanings in apparently similar concepts. Concepts may have similar facets in different cultures/languages but some facets may be present in one culture while missing in another. Or one facet may be emphasized in one culture while ignored in another.

For example, a question "How often do you hold some sorts of services or rituals in the memory of your direct ancestors?" could be easily translated into Finnish by using the dictionary equivalent rituaali for the English word ritual. But the Finnish word rituaali has an emic connotation of being something very ceremonious and solemn. The translation team would need to know whether also more 'low-key' activities are included in the concept. Most Finnish ways of remembering their ancestors (like taking candles or flowers to the grave) tend to be too low-key to be conceived as rituaali, so if such behaviour is to be included, it is better to phrase the question differently in order to make respondents include all types of remembrance.

System-related questions

The team approach and focusing on constructs are particularly important when translating system-related questions such as questions on health care, social security, legal or education systems. Health care systems, for instance, differ greatly between countries in terms of who pays for health care and who organizes it. Therefore even two countries sharing the same language have their own terminology for their systems. It is notoriously difficult to develop valid measures covering systems. Even with carefully constructed questions some adaptations are needed in the wording of the translated question to make it understandable to respondents.

Quality assurance of translations

The traditional method of checking questionnaire translation quality is back-translation. The target-language questionnaire is translated back into the source language by someone else than the original translator. Research indicates that this method does not necessarily detect translation errors. For instance, too literally translated idiomatic expressions (make ends meet, get their just desserts) back-translate beautifully even when incomprehensible or absurd in translation. Paradoxically, the more stilted or unnatural the translation is, the more easily it often back-translates.

Back-translation will not reveal etic and emic meaning differences discussed above although they may have a clear impact on comparability. In the case of multiple equivalence, the source language word can only be represented by two or more words in the target language. If the translated text has used only one word in the target language, giving the concept a narrower scope than in the source text, back-translation does not give any warning of this difference in scope. All these problems are more likely to become apparent in a translation team review. Furthermore, knowledge that back-translation will be used may cause the translator to favour overly close/ literal translation.

The team approach and various pre-testing methods are more useful in ensuring translation quality. If the source questionnaire exists in two languages (e.g. in French and English), a double translation design may be used. Two independent translations are produced, one of each source language questionnaire and these two translations are then reconciliated by a third person or a team.

Documentation of the translation process

It is useful to document the following for the benefit of future users of the data:

  1. Any adaptations made in translation of the original source questions, such as modification of content or format of the questions, with clarification of the reason for the adaptation.
  2. Changes across translations of the same question in different survey rounds, with clarification of the reason.
  3. Notes of questions that are for some reason inapplicable or problematic for a country, and have the risk of producing a biased measures.

All this information is important to researchers doing cross-national comparisons or longitudinal research. Documentation should be done at the translation stage since memory is short and the changes made and the reasons for them are difficult to remember afterwards.

 

In addition to her own experiences, the writer has used as sources:
» Cross-Cultural Survey Guidelines (CCSG), developed as part of the Comparative Survey Design and Implementation (CSDI) Guidelines Initiative. Web publication: http://ccsg.isr.umich.edu/index.cfm [retrieved 3 August 2010]
» Harkness, J (2008). Round 4 ESS Translation Strategies and Procedures. Web publication: http://www.europeansocialsurvey.org [retrieved 3 August 2010].
» Harkness, J (2003). Questionnaire translation, in Fons J R, Mohler, Peter Ph (ed.) Cross-Cultural Survey Methods, Hoboken, N.J. John Wiley & Sons, Inc, 35-56.
»Harkness J, Pennell BE, Schoua-Glusberg A. (2004). Survey Questionnaire Translation and Assessment, in S, Rothgeb J, Couper M et al. Methods for Testing and Evaluating Survey Questionnaires. Hoboken, N.J.:John Wiley & Sons, Inc., 453-473.
» Usunier, J-C (1999). The Use of Language in Investing Conceptual Equivalence in Cross-Cultural Research. Conference paper: Seventh Cross-Cultural Consumer and Business Studies Research Conference. Web publication: http://marketing.byu.edu/htmlpages/ccrs/proceedings99/usunier.htm [retrieved3 August 2010].