Read my review of ‘DTP for Translators’, hosted by the North West Translator’s Network in October 2022.
Hi everyone! Last month, I attended a webinar of desktop publishing (DTP) for translators, hosted by the North West Translator’s Network (NWTN). After a call for volunteers, I was then asked to write a short review of the session for the NWTN website. Click the link below to read my review!
Read my latest WMTS blog post on the perils of the 101 match, CAT tool errors, and the importance of human proofreading.
Today, computer-aided translation (CAT) tools are an integral part of a translator’s arsenal: increasing output, improving terminology accuracy, and streamlining the project management process.
This said, a CAT tool is only ever as good as the person using it. All the fancy bells and whistles that many tools offer are ultimately superfluous if thorough and careful proofreading falls by the wayside.
I was recently asked to proofread a translation of a clinical trial application form for a study taking place in France. A well-known CAT tool had been used to complete the task, but the translator’s apparent faith in the technology resulted in a fundamental mistranslation of the source text.
When authorising clinical trials to be conducted in France, the French health authorities make use two methods of authorisation when responding to applicants: expressed or implicit.
Expressed authorisation means written confirmation is sent to the applicant confirming the status of a clinical trial application, whereas written confirmation is not sent under the implicit regime. Indeed, implicit authorisation equates to the expression: “If you don’t hear from us, assume everything’s fine.”
In the translation I was asked to review, a ‘101’ or ‘perfect’ match appeared in a segment referring to the aforementioned authorisation methods. A perfect match is a segment of previously translated text stored in a translation memory (TM). When the corresponding, identical source text for this translation is recognised by the CAT tool as appearing in a subsequent task, the software auto-generates the previous ‘perfect’ translation of the source text into your translation. Unfortunately, the source text in the segment explaining implicit authorisation had been fundamentally mistranslated, stating that a lack of response indicated rejection of the application, rather than authorisation thereof.
This highlights a key CAT tool issue: The peril for the perfect match, and the intellectual malaise it induces. As a translator, it’s nice to see a green 101 symbol next to a given segment or segments of text; however, as any responsible translator knows, you can’t simply regard such segments as ‘translated’ and move on. All segments, including 101 matches, must be carefully reviewed for contextual accuracy regardless of match accuracy scores. Additionally, as a method of best practice, each client should have their own dedicated TM, as opposed to simply maintaining a generic ‘catch-all’ TM, which reduces the risk of contextual mistranslations.
This is where careful, human proofreading is essential. As discussed in my previous post, human proofreading should be a deliberate and careful process and is essential to catching such CAT tool errors and providing sufficient quality assurance for our clients.
Find out more about my editing and proofreading services, and SUBSCRIBE to the WMTS blog by completing the form at the bottom of this page.
Read the latest WMTS blog post on editing back translations for clinical trials and best practice in proofreading.
Revising a fellow linguists work can be fraught with complexity. I was recently asked to review four proofread, back-translated informed consent forms (ICFs) as the end-client raised concerns about the sheer number of changes made to the back translation (BT). This particular job was a laborious and difficult one, but, ultimately, it was also a rewarding experience, as I now use it as a case study for considering: When editing/proofreading, how much change is too much?
BT is, in general, a complex undertaking, and differences between the BT and the original translation (OT) often cause headaches for end-clients and language professionals alike: highlighting issues with the OT and inadvertently revealing the translators’ differing biases and linguistic preferences. This being the case, a conservative approach to proofreading BTs stands us in good stead, as we’ll examine below.
I took this job with certain expectations, as it’s not very often I get asked to review a proofreader’s work. In accordance with ISO 17100, industry best practice states the translation should be undertaken by one linguist, and another linguist proofreads it before being sent for client approval. For there now to be a further step in the process, I knew the issues had to be fundamental, but the full extent of the problems only became clear once I started looking at the files.
Regardless of whether you’re translating or proofreading, with ICFs, it’s best to start with the main patient ICF, as large swathes of other related ICFs, such as pregnant subject and pregnant partner ICFs, are often drafted—or copied and pasted by the authoring physicians, as the case often is—using text from the longer, more comprehensive main subject ICF. Doing so will help build in consistency across the entire project.
From the first two pages alone, the issues were clearly wide-ranging and significantly impacted all four ICFs. As you can imagine, the task of reviewing the reviewer took much longer than originally anticipated and, after all four files had been assessed, reviewer X had made several basic errors:
To aid the translation and review process, the end-client had supplied their own terminology glossary and style guide. However, throughout the four ICFs, reviewer X deliberately chose to swap out terminology employed by the translator and specified in said glossary for preferential, interchangeable nouns. As a linguist, you should always yield to the expressed will of the client where it exists, and where there’s doubt, discuss any concerns with them ‘cum exsurgunt’. X also made various syntactic and stylistic changes that were non-essential when compared to the original BT. Unless such changes are necessary to correct mistranslations or more accurately reflect the meaning of the source text, further edits are not generally required. In the present case, X’s syntactic choices took the meaning of the BT further away from that of the OT, causing confusion for the end-client when they compared the BT with the original source text.
Perhaps most damning of all, X even introduced their own typos and errors into the BT while editing and, rather inexplicably, deleted sections of text which clearly appeared in the OT. As a linguist, such sloppiness is unforgivable, especially given that the sections of translated text were error-free to begin with. It laid to bear X’s poor quality assurance practices and their lack of attention to detail. As the old adage says: “if it ain’t broke, don’t fix it”.
As stated earlier, all four ICFs were riddled with similar errors and, as a seasoned linguist, the slap-dash execution gave me the very distinct impression of someone under pressure to get this job done. Little care and attention were paid in this piece of proofreading and, ultimately, the project—and the client—suffered because of it.
So, to sum up:
Find out more about my editing and proofreading services, and SUBSCRIBE to the WMTS blog by completing the form at the bottom of this page.
