Double Source Double Translation Design For A Leadership Assessment Questionnaire

The Halin Prémont Enneagram Indicator, developed by Belgian researchers Philippe Halin and Jacques Prémont, is a scientifically validated questionnaire designed to determine the respondent’s Enneagram type. It is used for leadership coaching and has 52 questions used to attribute a score out of 100 for each of the 9 personality types in the Enneagram.

Project Scope

cApStAn was responsible for upgrading the existing French to English translation to source version status and then using the French and English source versions to produce German, Spanish, Swedish, Turkish, Italian, Hebrew and Arabic versions, applying a double source double translation design. The design envisaged a number of different steps:

• Producing a second English translation
• A reconciliation between this version and the previous one
• An adjudication procedure to finalize the text, with the support of subject matter experts
• A pilot phase during which cognitive labs were held
• Using the French and English source versions produce German, Spanish, Swedish, Turkish, Italian, Hebrew and Arabic versions, applying a double source double translation design

Once the new English version was deemed fully equivalent to the French, and scientifically validated, the two sources (French and English) could, at least theoretically, be interchangeable. In other words, in order to create a new language version, one could either translate from the French source version, or the English source version. Other major projects where the double translation double source design has been used include OECD PISA — for which cApStAn has assured the language component since its inception, in 2000 — in which the double translation double source design was used for questionnaire items and cognitive test items.

Challenges

Using two source versions (English and French) replete with idiomatic expressions to produce quality translations in the required languages was a challenge, as it required a complex procedure to avoid loss of meaning, maintain comparability across and achieve fluidity in all the languages.

Project execution process

We will take the case of the translation to German, as an example.

To start with, the questionnaire was translated from both the English and the French source versions. Then a senior German linguist merged Translation 1 and Translation 2, selecting the best elements from each for the reconciled version. The linguist also earmarked any potentially controversial issues and any choices that could have been difficult to make for a discussion in an adjudication meeting (where Translator 1, Translator 2 and the Reconciler/Adjudicator would come together).

In any adjudication meeting, once the different people involved come to a decision, there can still be residual errors and inconsistencies, so to address that generally there is a proofreading phase, conducted by an external professional proofreader. The proofreader’s brief is not to look at equivalence any more, but to make the final merged version error free.

The table below illustrates how the double source double translation design played out for a sample questionnaire item of the HPEI being translated into English and then German.

• Translation and adaptation guidelines were prepared for the French source version and adapted for the English source version, where needed.
• Translator 1 translated the questionnaire item from French to German; Translator 2 translated the questionnaire item from English to German.
• The item analyzed in the table contains the French expression “prendre du recul”, which was translated into English as “stepping back and taking stock”.
• Each translator documented their difficulties and their doubts (see T1 and T2 comments in the table).
• During the reconciliation process the comments were discussed and, during the adjudication phase (see table 2 below), where a cApStAn moderator was present, it was found that it was not possible to follow both sources. In other words it was not possible for the German translation to follow perfectly the French version, or to follow perfectly the English source. In this particular case the decision was made to follow the English, acknowledging that there was a slight shift in meaning between French and English, and thus between German and French (the English version introduced an additional element, “taking stock” that was not inherent in the French).

Advantages of the double source/double translation design

• A comparison between the English and French versions can serve to give a good measure of the freedom of translation that is deemed “acceptable”. It is not, of course, a literal translation, but a translation that “functions” in the same way. In other words there may have been some intentional deviations from the initial French to reach something that sounded more “natural” in English. This measure of translation leeway is quite precious, especially if there is an adjudicator that masters both languages.

• There is also the strength of striving for semantic equivalence via triangulation to consider. So, for example, if a translator looks at the French and does not see exactly what the best way would be to translate from the French into German, he/she can always check the English source version. If it is worded slightly differently and it strikes chords, he/she can decide that this would work better. Then he/she can go back to see if the translation from English into German still works, compared to the French source. The triangulation enriches the process.

• In multilingual projects if one always uses a single reference language (which in international projects would most likely be English) one could put undue weight and importance to certain formal characteristics of that language. For example, if you have a language that is built with subject, verb and object, and then you have another language that is built differently, there is a risk that you could try to fit the formal characteristics of the source version into the target version. It isalways good to have a different reference point at the start.

Disadvantages of the double source/double translation design

• One of the main disadvantages of the process is that it is highly dependent on the language skills of the reconciler/adjudicator, who has to be trilingual. In addition to mastering his own native language he needs to have excellent skills not in one but in two different foreign languages, and understand the nuances and subtleties of both.

• There may also be a language barrier during the adjudication if the two translators are “only” bilingual and only speak one of the target versions and one of the sources. In this case the adjudication process will prove to be difficult.

• Finally, even if it is enriching to have two source versions, this also brings some uncertainties. For example, how should one proceed if there is a slight discrepancy between the two sources, if there is a nuance that is introduced in one compared to the other? Which source version should be given precedence?

Conclusions and lessons learned

The combination of a robust process design and cApStAn’s efficient project management contributed to the successful execution of the double source double translation approach required for this project. The questionnaires were rolled out on schedule and the client was very happy with the outcome in terms of quality, cost and timelines committed by cApStAn.

As mentioned above there are pros and cons in this design. One of the disadvantages is that it adds a further layer of decision making, making things more difficult.

As a recommendation, if two source versions are available and validated, it may be preferable to translate from the main source, or one of the sources, to produce both Translation 1 and Translation 2 — from the French in this particular case — and use the other source only for cross-checks, rather than having two different starting points. his way one would be able to maintain the advantages of having a double source, i.e. the measure of translation leeway that is deemed acceptable, and a benchmark to rely on, but with a single starting point, which makes things easier.

Do you want to translate your assessments into multiple languages?

Select some sample items and request a free pilot at hermes@capstan.be to discuss your requirements.