Assuring quality in all its aspects —semantics, grammar, stylistic and cultural accuracy, optimal format, correct terminology— is one of the pillars of localization workflows. In this regard, quality is not ensured by applying one single action. On the contrary, many factors contribute to quality, such as linguist selection, efficient use of tools and different processes of quality control.
There are certain textual materials that are highly sensitive and have a major impact on the life of people related to them. This is the case, for instance, of pharmaceutical and clinical protocols, clinical research, financial or medical reports. When working with these documents, clients and/or Language Service Providers (LSPs) can opt for adding another quality assurance step, called back translation. This step can be useful in the following scenarios:
- The content needs rigorous control, since any semantic error can have practical consequences.
- Sometimes clients don’t speak nor read in the target language and want to check the accuracy of the translation by themselves.
- When reviewing transcreated text, clients want to assess how the linguistic team translated a slogan or other marketing copy, and what meaning it conveys in the target language.
The three-column text
The aim of back translation is to have another instance of control of the target text’s quality; mostly in terms of terminology and semantic accuracy. It consists of performing another translation, but this time of the target text into the source text language.
For example, if we translated a clinical protocol from English into Spanish, we would apply another process of translation, but from the Spanish target into English again. The resulting new English text will never be an outbound translation. The new English translation is not intended for publishing nor distribution. The outcome of a back translation process is only for internal use in the localization process. The linguistic team performs it only for quality assurance purposes.
Moreover, we can point out the following characteristics for back translation outcomes:
- They are literal.
- They will never result in a version exactly the same as the original source; it helps to identify discrepancies or errors in the translation.
- A different linguist not involved in the original translation workflow works on it. Sometimes, they don’t even have access to the first source text (blind translation).
After the above-mentioned step, the same linguist or a different one will read and compare the source text against this new outcome. If no critical difference in meaning or mistranslations are flagged, the translation is considered accurate, and the process finishes. On the other hand, if the reviewer spots discrepancies, he or she will mark them up for further review, in which the team fixes the translation. Sometimes, they fill reports with the errors found and corrected in this step.
The process of back translation is very thorough and implies an exhausting revision of the translation that goes beyond editing or proofreading. However, it demands a well-planned and wide deadline, and the use of several professionals. Because of this, LSPs and/or clients should consider if back translation is appropriate for a certain project considering time, resources and budget. If not, there may be other ways of ensuring quality too, through one of the multiple means the localization industry has come to develop.