As said in Superman, “With great power comes great responsibility.” In studies collecting data in multiple countries, translations should protect the integrity of the questions. The current standard process is forward-translation followed by backward-translation (Larkin, Dierckx de Casterlé, & Schotsmans, 2007).
There are three steps: (1) translate the source version (e.g., English) into the target language (e.g., Korean); (2) translate the target version (i.e., Korean) back to the source language (i.e., English); and (3) review discrepancies (Llorente, Warren, Pérez, & Gleaves, 2013).
For our collaborative study of how service-learning influences youth purpose around the world, all questionnaires were originally in English. Some had already been translated into some of the partners’ languages. But most required translation into six languages: Korean, Simplified Chinese, Finnish, Brazilian Portuguese, Castillian Spanish and Catalonian Spanish.
Surveys were professionally translated in the US, then edited and approved by partner researchers as accurate to the understanding of concepts not just transliteration of words. For example, whereas “anxious” may have an appropriate translation in some languages, “jittery” is a more informal word that may not translate well.
The researchers chose not to remove any of the standard items on established questionnaires so that this study’s results may be compared to prior studies. Instead, they added the response options “not relevant in my country” and “other, please specify” to capture national differences on an item-by-item basis.
However, there is a trade-off with this method because of within-country variability. Study-abroad and immigrant students may interpret the question using meanings from their original country or the country in which they took the surveys. If they selected “not relevant in my country,” which country are they referring to? Thus, additional preliminary analyses were needed to clarify.
Plus, many languages have idioms and dialects that can cause miscommunication even among native born speakers (Llorente et al., 2013; Su & Parham, 2002). For example, two Spanish translations were made to be appropriate for both Catalonian and Castillian areas of Spain.
Finally, surveys included open-ended questions that could be qualitatively analyzed as well as the quantitative, standardized measures. This redundancy is recommended to clarify that the multiple-response questions are correctly interpreted by the respondents (Larkin, Dierckx de Casterlé, & Schotsmans, 2007).
By Cori A. Palermo, Clark University, USA
Read also Equivalence in Translation.
Larkin, P. J., Dierckx de Casterlé, B., & Schotsmans, P. (2007). Multilingual translation issues in qualitative research: Reflections on a metaphorical process. Qualitative Health Research, 17(4), 468-476. doi:10.1177/1049732307299258
Llorente, E., Warren, C. S., Pérez de Eulate, L., & Gleaves, D. H. (2013). A Spanish Version of the Sociocultural Attitudes Towards Appearance Questionnaire-3 (SATAQ-3): Translation and Psychometric Evaluation. Journal of Clinical Psychology, 69(3), 240-251. doi:10.1002/jclp.21944
Sinaiko, H. W., & Brislin, R. W. (1973). Evaluating language translations: Experiments on three assessment methods. Journal of Applied Psychology, 57(3), 328-334. doi:10.1037/h0034677
Su, C.-T., & Parham, L. D. (2002). Case Report – Generating a valid questionnaire translation for cross-cultural use. American Journal of Occupational Therapy, 56, 581-585. doi:10.5014/ajot.56.5.581