Pressroom
The latest news of everything memoQ. Browse this page to find press releases, news, articles, podcasts, and more.
Linguistic Validation
March 01, 2023
Mark Shriner
Mark Shriner is the strategic sales director for memoQ, leading the company’s market growth in the regulated industries. He has previously worked in several leadership roles in the localization industry including CEO Asia Pacific for CLS Communication.
Welcome back to The Lab, where we take a look at what’s cooking in life sciences localization. This month we are going to do a deep dive on the topic of linguistic validation (LV) to get an understanding of what it is, when and why it’s required, and what are some of the related challenges. We will also look at trends related to technology and economic imperatives that are introducing disruptions to the established LV process.
While Wikipedia’s definition of LV runs 601 words long, I’m going to summarize, to the best of my ability, and say that LV is a process used by organizations in the life sciences to ensure that the translation of certain content including patient-reported outcome (PRO) measures is linguistically accurate, readable, and understandable by the target audience. As an example, documents used for cognitive debriefing interviews would need to be accurately translated in a manner that laypersons in the target audience can understand. In short, the text needs to be both true to the source and easily understood by the target audience.
The most commonly used LV process includes two forward translations (FTs) consolidation or “harmonization” of the FTs, single or double back translations (BTs), and a review step. Each of these steps requires process and step-specific knowledge and skills and may be executed in a somewhat different manner depending on the specific project and content type to which it is being applied. For example, the processes for creating PRO questions might prioritize the readability of the questions for the target audience. However, the translation of content from clinician-reported outcomes (CROs) might favor a process that prioritizes the use of the exact medical terminology and descriptions used by the reporting clinicians.
Typically, there are two FTs. These are conducted by different translators and are supposed to prioritize readability of the translated content by the target audience. Literal translations or translations that use complex medical or scientific terminology are not typically acceptable for this step in the LV process. It’s important to understand the demographics of the target audience so that all relevant factors such as age and education level can be used to guide the readability of the translated content.
According to medical writer and the founder of Wordplay Translations, Eva Stabenow: “While precision is of the essence, ‘translating’ a source text word for word can easily lead to confusion at best and miscommunication at worst. Therefore, translators need to be keenly aware of how clinicians
and/or patients actually use and perceive their language.”
As previously mentioned, there are typically at two FTs required for each document. This is done to provide the reviewers of the translation more than one option from which to select the most appropriate translation. This selection process is commonly referred to as the consolidation or harmonization step where the best translations for each question or section of text are selected to create a single draft document.
The harmonization step can be challenging for a variety reasons including a lack of a readability assessment on the original source text, the use of complex terminology, the use of outdated terminology, and the application of machine translation (MT) for one of the FTs.
When asked about best practices for this stage of the LV process, Ana Sofia Correia, an English to Portuguese medical translator and writer, stated, “Best strategies include the use of two FTs produced by two different translators with experience in LV, doing a reconciliation based on a detailed list of concepts, and having a video call with the two forward translators to review the reconciled version.”
The consolidated or harmonized document is usually then forwarded to one or more translators to do back translation into the original source language. This translation is typically, but not always, supposed to be a literal word-for-word translation and not take into consideration the target audience. The back translated text is then reviewed by a subject matter expert to verify the trueness of the back translation with the meaning of the original source content.
Kristin Kamm, a freelance translator and editor specializing in clinical trial translation and owner of Turn of Phrase Translations, recommends: “Translators should be trained so that they don’t take the instruction to translate literally — well, too literally. For example, if a translation uses the culturally appropriate expression for an idiomatic expression like ‘raining cats and dogs,’ I’m of the opinion that the back translation should use the appropriate idiomatic expression as well. More concretely, if a French translation said ‘Il pleut comme vache qui pisse,’ I would back translate that as, ‘It’s raining cats and dogs,’ because that conveys the equivalent linguistic meaning.”
Kristin also pointed out the importance of managing the expectations of project managers and clients when comparing BTs with the original source.
“I think the most important thing for project managers to know, and for them to relay to clients, when it comes to BTs, in general but specifically in LV, is that BTs are a tool that are supposed to be used to point out potential errors or discrepancies from the source in a translation,” she said. “There should never be an expectation that the back translation is going to precisely mirror the source text because of the existence of synonyms in language. The important thing is that the exact same meaning is conveyed. That is the equivalency that needs to be achieved in any and all LV projects.”
In order for the methods used by linguists to be fully aligned with best practices and customer expectations, it seems that there is complete agreement across the industry that experience with clinical trials and ongoing training are both critically important. In fact, experience and training were the two key elements consistently mentioned as success factors by the experts I interviewed for this article.
Stephane Millet, director of strategic accounts for clinical and patient engagement solutions at Acolad Group stated, “Linguist training is crucial because translating reported outcomes is not the same as translating any other type of document. The translations have to convey the exact equivalent meaning, so it is important to provide a detailed training on the instrument that they are asked to translate, as well as detailed review of the concept definition.”
Stephane also suggested that translators would benefit from an understanding of how reported outcomes are used by sponsors.
Rafael “Rafa” Zaragoza, global director of quality, risk and compliance at Toppan Digital Language, agreed and added, “One of the challenging elements for LV is linguistic engagement and training, because finding linguists who are willing to undergo the necessary training required to successfully complete projects has proven to be difficult.”
One point of minor contention for this step in the LV process is whether back translators should be native speakers of the original source, or whether they should be native speakers of the target language of the forward translation.
Ana Sofia Correia believes that, “Without question, back translators must be native speakers of the original source languages, fluent with forward translation target language, extremely familiar with the target culture and be trained on back translation.” Kristin Kamm agrees and stated, “Translators should always be working into their native language, so for LV back translation projects, the back translator should certainly be a native speaker of the original source language.”
And, while Eva Stabenow confirmed that the usual practice is to use a native speaker of the source language, she doesn’t believe that is the key element.
She explained, “I find the native speaker vs. non-native speaker to be a bit of a red herring because a back translation (BT) needs to accurately reflect every last nuance of the forward translation (FT), and I have seen BTs by native speakers of the original source language that read wonderfully, but either misinterpret or do not faithfully render certain aspects of the FT.”
“Given that the sole purpose of the BT is to check the validity of the FT, I believe it is much more important that the person doing the BT has a truly excellent grasp of the FT target language and a good enough command of the original source language to produce a coherent text of good quality,” she added.
Technology currently facilitates LV workflows in a couple of key areas and is expected to become even more important as the nature of how trials are conducted and connected with other digital platforms. At this point, using a TMS that supports LV workflows can be a real game changer as it can automate tasks such as duel forward translations, provide support for the consolidation process, and then automatically prepare files for back translation and review.
This type of workflow support isn’t readily available in most TMS platforms and does require a great deal of LV process subject matter expertise. According to Richard Sikes, senior consultant for business services at memoQ, “There were several technical challenges that we solved to support LV workflows including supporting projects that might require dual forward translation for some files and single forward translations for others, consolidating the best translations and preparing them in a single document, support for ‘blind’ back translations of the consolidated document, and then finally delivering the completed back translation in a format that a reviewer who might not have any experience with a CAT tool or TMS can review the translation and provide feedback.”
While automating LV workflows is a huge benefit, the use of translation memories and glossaries receives mixed reviews. If forward translators leverage a TM or glossary, they run the risk of being led away from their readability requirements of the intended audience. Back translators using a TM might not get near enough to the literal translation that reviewers are expecting. However, as previously discussed, sometimes it behooves the translator to provide a less literal and more culturally relevant translation.
Much to chagrin of professional LV linguists, MT is increasingly being used for some steps in the LV process.
Ana Sofia Correia pointed out one potential issue of using MT for LV: “The Reconciliation step aims at creating a consensus between both forward translations, but MT typically provides a literal translation that doesn’t take cultural factors into account needed to present an idiomatic translation and it might provide a completely wrong translation.”
Ana Sofia points out two more potential issues of using MT in LV workflows: “If only one forward translation meets the established criteria, there is nothing to reconcile and, MT does not provide a gender-inclusive translation, which makes a significant difference in the linguistic validation for many gendered languages.”
Another type of technology that is impacting LV workflows and the tools used facilitate them is the wide-spread adoption of electronical clinical outcome assessments (eCOA) and electronic patient-reported outcome (ePRO) platforms.
Acolad’s Stephane Millet explained, “There has been a huge drive in recent years to move to eCOA platforms, and the Covid pandemic and movement towards conducting decentralized clinical trials has accelerated the use of eCOA and increased the amount of reported outcomes that are needed by the sponsors.”
While migration of the accepted content for questionnaires to an an eCOA system takes place after the LV process has been completed, there are steps that can be followed to optimize the migration processes. Likewise, the use of ePRO systems has effects on the LV process, and the use of both eCOA and ePRO systems are enhanced by integrations with any CAT, project management, business reporting, or translation management tools used during the LV process.
Ana Sofia Correia points out, “eCOA migration is a separate process. It takes place only after the linguistic validation of the questionnaire, which means that items and response options cannot be changed. You can only adapt instructions (like replacing ;circle the answer’). One of the main issues I encounter in my experience is the text expansion between languages. Screens are created based on the English version, but then you have languages for which you need the double or triple of that space. In this situation, decreasing the font size could solve the issue, but then it could render the text difficult/impossible for participants to read.”
The greater use of decentralized clinical trials, where technology is used to communicate with study participants and collect data (negating a need for all participants to meet in a central location), is also driving the adoption of eCOA and ePRO systems.
2023 • multilingual.com by Mark Shriner