Our clients vary quite a bit, as do their translation needs. If you have a high volume or content that’s repetitive in nature, you may already be well acquainted with translation memory, otherwise known as TM. But if not, you may wonder “what exactly is this and do I need it?” Let’s discuss!
What is Translation Memory (TM)? TM is a database that stores segments of translated text.
How does TM work? When you submit a file in an editable format (e.g., a Word document or InDesign file as opposed to a PDF), the content of the file is extracted into the TM software, then analyzed for repetition (exact matches) and fuzzy matches (similar but not exact matches). Before translation begins, matches are suggested by the TM software, which can then be accepted or overwritten by the linguist working on the text. As your TM expands, segments can be translated faster and at a lower cost.
Are repetitions free? No. Just because the segment is identical, doesn’t necessarily mean that the context in which it is used is the same. Linguists must edit these segments and occasionally retranslate them to ensure they flow correctly within the entire sentence. However, there are price breaks depending on whether the segment is a new, fuzzy, or repetitive match.
Who owns the TM from my projects? You own your TM; your translation provider maintains it.
Can I move TM from one company to another? You will have to verify with the translation vendor you are moving the TM to that the new vendor can utilize the TM, but yes, your previously created TM is portable to the agency of your choice.
Is TM software- or version-specific? There is an array of computer-assisted translation (CAT) tools out there; the most commonly used options are TRADOS, MemoQ, Déjà Vu, and Wordfast. Regardless of the technology your agency of choice is using, TM can be exported into a .TMX file, enabling the exchange of memories between specific tools.
How does TM differ from MT? Translation memory (TM) is a tool used by translators to store the text segments of human translations for reuse. The segments that are reused are specific to each client’s previous documents, establishing better context parallels and higher quality per segment. Machine translation (MT) is automated, nonhuman translation that isn’t client or subject matter specific. With MT, your files are uploaded into the software, which generates an output into whatever language you choose. MT lacks context and quality controls.
How can I learn more? Visit our website to read more about our translation memory capabilities, or ask us directly at info@mcelroytranslation.com.