XTM Xplains
Yes, a translation management system (TMS) will typically support both machine translation (MT) and translation memory (TM) functionalities. MT and a TM are two fundamental features of any modern translation workflow, and therefore normally well supported by TMS systems.
A translation memory (TM) is the core basic feature of a TMS. A TM is a repository that stores previously translated segments. When a translator comes across a segment that is exact or similar to a previously translated one, the TM suggests the stored translation so that the linguist can reuse if exact it or slightly edit it if necessary. A TM is essential to maintain consistency in translations and also reduces unnecessary and redundant translation tasks. It’s also a tool that saves companies money, since they do not have to pay again for repeated translations. A study by Masaro Yamada from Rikkyo University in Japan indicates that translation memories can save companies up to 30% in localization spend and increase productivity by up to 70%.
A machine translation (MT) engine is software that can translate content from one language to another using algorithms. A TMS normally supports a wide range of MT engines, and how many depend on the provider. Before purchasing any TMS, it’s important to verify with the provider which MT engines it can support and whether it includes your MT engine of preference. The use of MT can also help with localizing content and publishing it more quickly. Once it’s published, it can always be reviewed and re-published if any edits need to be made. MT translates text from one language to another. It provides quick and rough translations and is particularly useful for large volumes of content where speed is prioritized over absolute accuracy, such as instruction manuals.
By combining TMs and MT, a TMS makes the translation process much more efficient. A TM suggests pre-existing translations that are exact or similar to the one being translated, helping linguists translate more quickly. An MT engine will generate translations automatically, so linguists don’t have to translate from scratch and only need to do some post-editing on them if needed. These two features are key contributors to improving translation quality, consistency, and cost-effectiveness.
Assess the suitability of machine translation engines for compatibility with the TMS.
Provide training to translators and linguists on the optimal utilization of TMs within the TMS, with a focus on promoting consistency and leveraging pre-existing translations.
Foster collaboration between human translators and machine translation output to obtain optimal outcomes, where human post-editing can enhance and refine the quality of machine-generated translations.
Update and maintain the translation memory database regularly to guarantee the accuracy and relevance of stored translations.
A good TMS should integrate both machine translation and translation memory capabilities. This integration allows for the effective utilization of existing translations and language assets (glossary, term bases, etc.) and the automated generation of new ones. By leveraging TMs and MT, translation quality, consistency, and cost-effectiveness are greatly enhanced. When choosing a TMS, it is important to ensure compatibility with preferred machine translation engines and prioritize the efficient use of translation memory to improve productivity and streamline workflows.
The first part of a translation process is content being sent to a TMS for translation. This can be done via an integration from a content-authoring platform, like a CMS, or by uploading files like a Word document or an Excel spreadsheet.
The content is then ready for translation in the TMS, separated into segments to make it easier for the translator. The translator then translates the content or it’s done automatically via machine translation.
Once all the content is translated, the linguist submits it to the next workflow step, which could be a review step, for example. By submitting it, the translated content gets stored in the translation memory.
In future projects, if a translator comes across exact or similar segments to those in previous projects, the translation memory will suggest a translation to the linguist. If it’s an exact segment, it will suggest it as a 100% match. If it’s similar but not quite the same, it will suggest it as what’s known as a “fuzzy match” which is a number between 50 and 99%, depending on how similar it is. This way, a linguist doesn’t have to translate the segment from scratch and only has to do minimal post-editing. Thanks to a TM, stored translations can be fully or partially reused, which saves time for the translator.
We’ll be delighted to help! Get in touch to discuss more about how TMs and MT can lower your localization costs.
