Translation and international use of research data

To promote open science, the Finnish Social Science Data Archive also provides services to researchers and students who do not speak Finnish. In practice, this means that foreign language speakers are provided the opportunity to access quantitative research data in FSD holdings in English. In addition, we describe all data in English to allow the international audience to find them.

Translating variables, that is, the questions and response alternatives presented in a study, differs quite a lot from translating other types of texts. Translators of literature often have a great deal of linguistic and artistic freedom. The challenge of translating fiction usually lies in carrying the ambiguities of the source language over to the target language. Translating technical and informative texts requires clearer, more univocal interpretation, although depending on the text type a translator may have some creative licence.

Striving for fidelity and clarity

When survey questions are being translated, there's very little room for interpretation. In studying attitudes, how something is asked may have almost as big an effect on the frequency distributions as what is asked.¹ The users of translated data have to be able to trust that the translated questions correspond to the original ones as closely as possible in terms of both content and form. Researchers may draw far-reaching conclusions based on the translated questions, which is why it's essential that non-Finnish-speaking researchers gain an understanding of what the respondents were originally asked and how the questions were phrased.

One of the challenges of translating variables faithfully is posed by the errors and inconsistencies sometimes found in source texts. At its worst, an error in a Finnish question may affect how the respondents answer the question, and the non-Finnish user of the data will wonder what has caused the unexpected frequency distributions. When translators notice an error, they have to consider how significant it is and whether it has actually affected the responses. The translator also needs to decide how to communicate the error to data users. Replicating the error in the translation is rarely a good option because a user won't know whether the error originated from the source text or the translator.

As translators at FSD, we're lucky to have the possibility to provide further information on the translations to users through a separate file. These translation notes usually contain information on the word choices and translation strategies used.

Culture-specific expressions pose challenges

In general, a translator of research data should aim to create as neutral a translation as possible while avoiding interpreting variables too strictly from her or his own point of view. However, this does not mean that we translate data word for word; relaying the idea of the original text clearly and intelligibly is our priority.

Ultimately, the challenges of data translations are very similar to those encountered when translating other kinds of texts, such as how to express a concept that does not exist in another language and how to translate the nuances of the source language into the target language.

Questions presented in surveys often have elements or expressions that are specific to the Finnish culture and have no obvious counterparts in English. In these cases, the concepts need to be explained to foreign users. For example, in one dataset, the respondents were asked whether "red cottages and potato fields" are part of the Finnish rural landscape. This expression is similar to the "white picket fence" used in the United States referring to a sort of simple and idyllic life. However, because our users are from all over the world, domesticating the question for a culturally limited audience is not a good idea. For users who are not familiar with either the Finnish or the American culture, we would simply be replacing one culture-specific expression with another. In situations like this, the translation notes are of great help to us, as we are able to explain expressions that might otherwise be incomprehensible to some users.

The translation process at FSD usually takes a few weeks depending on the size of the data and the translators' workload. After the translation is complete, the data are available on our Aila Data Service to all users who need it in English. This is why we always put a lot of care and effort in translating all of the data instead of specific, individual variables needed by one user.

"Haste makes waste" also applies to questionnaire design and translation

Some might say that a data translation is never close enough to the source text to make research on translated variables feasible. It's been noted, for example, that some established measurements of well-being do in fact measure slightly different concepts in different languages.² Full equivalence can perhaps never be achieved, but this is rarely the goal of translation. It's good to remember that language is built on interpretations, and even speakers of the same language may understand the same question in different ways in attitudinal surveys.

Questionnaire design and translation do not always receive the attention they deserve, although well-formed questions are likely to lead to better validity. In international comparative surveys, it's particularly important that all respondents are asked the same things in the same way. If you plan to use questionnaires in more than one language, having questionnaires translated by a professional translator is something to consider when planning data collection. Surveys conducted in a single language also benefit from proofreading by a professional to prevent potential errors or inconsistencies from endangering the objective of the study.

Survey question form and wording require some effort, but there is no need to reinvent the wheel; inspiration for questions can be found, for example, in the UKDS Variable and question bank, the upcoming CESSDA Euro Question Bank and of course our question and variable search on Aila Data Service.

Henri Ala-Lahti
Research Officer
firstname.lastname [at] uta.fi

Niko Koski
Research Assistant
firstname.lastname [at] uta.fi

¹ Schuman, Howard & Stanley Presser (1996), Questions and Answers in Attitude Surveys: Experiments on Question Form, Wording, and Context. Thousand Oaks: SAGE
² Lolle, Henrik Lauridsen & Jørgen Goul Andersen (2015), "Measuring Happiness and Overall Life Satisfaction: A Danish Survey Experiment on the Impact of Language and Translation Problems", Journal of Happiness Studies 17 (4): 1337-1350.

This blog entry is also available in Finnish:
Suomenkielisten tutkimusaineistojen kääntämisestä ja kansainvälisestä jatkokäytöstä.