Main d'un robot et d'une personne sur un clavier

Pièges dans la post-édition de traductions juridiques automatiques (en anglais)

13.12.2023

Recent statistics show a significant increase in post-editing of machine-translated texts, with more than 50 per cent of language translation professionals using this technique in 2017. However, despite the efficiency and fluidity of neural machine translation (NMT), subtle errors can still creep in and go undetected.

The expertise of post-editors is therefore essential for ensuring the accuracy and reliability of NMT-generated output, especially in the field of legal translation, where precision is paramount.

When working on legal texts, post-editors should be aware of potential pitfalls in the following areas in particular:

Complex legal terminology

Legal documents often contain highly specialised and complex terminology. NMT tools can sometimes struggle to translate these terms accurately, resulting in errors that have to be identified and corrected by the post-editor themself.

Understanding of context

Legal language relies heavily on context, and subtle nuances and meanings are critical. NMT tools may not fully understand the context of legal texts, resulting in contextually inaccurate translations. The post-editor must therefore ensure that the translated text conveys the intended legal meaning.

Consistency and style

Legal documents require a high degree of consistency with regard to terminology and style. NMT systems can introduce inconsistencies or fail to maintain the specific legal writing style. The post-editor therefore has to ensure consistency throughout the text.

Ambiguities

Legal texts often contain ambiguities that human translators can interpret based on legal principles and context. NMT tools, however, may struggle to handle ambiguities appropriately. Post-editors must therefore resolve any uncertainties to ensure a legally accurate and reliable translation.

Cultural and legal nuances

Legal systems vary considerably from one jurisdiction to another, and certain legal concepts may not have a direct equivalent. NMT tools might not be sensitive to these jurisdictional nuances, which means post-editors have to adapt the translation to the relevant legal system.

Confidentiality and data security

Legal documents often contain sensitive and confidential information. The use of NMT tools can raise concerns about data security and confidentiality. Post-editors must ensure that the translation process complies with legal and ethical standards concerning the protection of sensitive information. Some NMT tools, such as DeepL Pro, ensure confidentiality by deleting all texts from the system and server as soon as the translation has been generated.

Orthography and grammar

NMT-generated translations may contain spelling errors that are not immediately apparent, especially ones relating to different variations of English (e.g. UK vs US). What’s more, in the case of texts that have been machine-translated from German to English, the grammar and sentence structures in the translated output are sometimes too close to the source, meaning the text can read more like a translation than an original piece of writing.

Typography and formatting

Another common challenge that post-editors encounter when working with NMT-translated texts is maintaining consistent typography and formatting. Neural machine translation can sometimes mishandle elements like line breaks, bullet points, bold text and italics. Post-editors must therefore meticulously review and correct these issues, as even small things like italics can sometimes make a big difference to the meaning.

Productivity and time constraints

Even with the assistance of NMT tools, post-editing can be time-consuming. Meeting tight deadlines while maintaining a high level of quality can be challenging for post-editors working on legal texts, especially if a lot of changes are needed.

Hallucinations

Some NMT systems will sometimes add content that doesn’t actually appear in the source text. Similarly, post-editors might find that important content is missing from the machine-generated translation. It is therefore crucial that they carefully check the output against the original text. With some NMT tools, the user can provide feedback directly to help improve output quality over time and improve translation accuracy.

In summary, German-English legal post-editors face a multitude of potential pitfalls when working with machine-translated texts. These difficulties underscore the essential role of qualified and experienced language professionals in ensuring that machine-generated translations are not only accurate but also reliable – especially in the field of legal translation, where precision is everything.

Machine translation post-editing (MTPE) has established itself in the language industry in recent years and that is unlikely to change anytime soon. Are you looking for a reliable partner who ensures your machine translated content is edited according to your needs and specifications? Then don’t hesitate to contact us, we’ll be happy to help.

WordPress Cookie Notice by Real Cookie Banner