Translation Memory: Overview, Benefits, and Common Questions

Translation Memory: Overview

A translation memory (TM) is a bilingual database or software system that stores translated content and automatically suggests these stored translations when used with computer-assisted translation software (aka CAT tools). The entries in the database consists segments of text, such as words, sentences, and paragraphs[1][2]. For example, consider the description below this famous painting by René Magritte:

Photo credit: http://museumtv.fr/portfolio/rene-magritte/

Note: This is not a pipe. It is a wonderful piece of art!

If this phrase were to be translated into French for France, then the TM would look something like this:

Later on in the same file or on a future project, the TM will provide translation suggestions if the source text is exactly the same or very similar. Very similar text is referred to as a “fuzzy match”, meaning that it is not an exact match, but it contains a certain percentage of the same text so that the previous translation could be useful. After reviewing the suggestion, it is up to the translator to make any necessary corrections to the translation before confirming the exact or fuzzy match and entering it into the TM. If the segment contains entirely new text, then the translator will translate from scratch with the option to search the TM for key words or fragments.

• Example: If the new text to be translated was “Note: This is not a painting.”
  • The TM would automatically suggest “Remarque:” as an exact match since “Note:” stayed the same. Often, exact matches do not need to be edited, though it is important to check for a change in context. If this meant “note” as in a musical sound, then it would need to be edited to “Note” in French and would add another entry to the TM.
  • In the next segment, “pipe” changed to “painting”. The TM would suggest the previous translation of “This is not a pipe.”, but flag it as a fuzzy match so that the translator knows to go in and make edits to the translated text. After the translator confirmed the edits made in the CAT tool, a new segment would be added to the TM as shown below:

The Benefits of Using a TM

Lowered costs and shortened timelines: Utilize existing translated material which provides a more cost-efficient translation process and faster delivery times.

One benefit to using a TM is that project costs and turnaround times will become more efficient as more projects are completed since the more content a TM contains, the more possible suggestions it can offer. It takes less time and energy for a translator to confirm exact matches and edit suggestions for fuzzy matches than to translate from scratch for two reasons: (1) review and edits take less time than original translations and (2) the translator doesn’t have to manually look up each entry to ensure that the content matches previous translations.

Improved quality: Ensure no content is missed and ensure consistent and accurate treatment of content.

Another benefit is that use of a TM will improve quality by making sure nothing is missed and improving consistency. CAT tools take the content from source documents and place them in a bilingual table for the translator to work with. This makes it easier to ensure that all segments are accounted for in the translation process.

Using a TM within the CAT tool significantly improves consistency within a project and across many projects. Without a TM, the translator would have to take the time to manually look up each segment or rely on memory to ensure consistency throughout a project and with past translations. This can be incredibly time-consuming and prone to error. The TM automates this process through providing suggestions for exact and fuzzy matches, and can also be searched for key terms and phrases[3][4].

Common questions:

How is a TM different from a glossary?

A glossary is a collection of preferred terms, abbreviations, and short expressions. These terms might be found within phrases of the translation memory, but are usually put into a bilingual glossary to ensure consistency of key terminology throughout the segments of translated content before finalizing the project and importing the text into the TM[5][6]. Glossaries might also contain a column of notes explaining certain terms in case there are translation differences based on context (see example of the two translations of “note” in the table below).

An example of what a glossary might look like:

Having a glossary as well as a translation memory increases the CAT tool’s ability to automatically suggest terms since you can have both a TM and a glossary providing suggestions while translating. You can search a translation memory by individual word, but the glossary helps automate that process. There are also automated quality checks within CAT tools that verify the use of glossary terms, which can be really beneficial to reduce confusion to the target reader and maintain a consistent message in marketing materials.

How is a TM different from a machine translation? 

In short: a translation memory is the result of hard work put in by a native translator with experience in the field, while a machine translation is done entirely by a computer with no human involvement[7].

A machine translation is essentially the computer software that completes translations using previously translated content, rules and/or algorithms. A machine translation uses and builds a TM much like a translator would, but still requires that extra time be spent post-editing by a professional translator to ensure accuracy and consistency. At this point in the technology’s evolution, we recommend that machine translation be used by translators as a supplementary tool instead of replacing the initial translation step for the reasons outlined in this article https://cpgcanhelp.com/machine-translation-what-is-ii-and-how-should-it-be-used/

[1] What is Translation Memory? SDL Trados, https://www.sdltrados.com/solutions/translation-memory/

[2] Computer-Assisted Translation Tools: A Digest, Paul Froese, https://atasavvynewcomer.org/tag/cat-tools/

[3] Top 3 Translation Memory Benefits and Why You Need to Build One Now, Chelsea Ramage, https://www.interproinc.com/blog/translation-memory-benefits

[4] How Translation Memory Can Save You Money and Time, Errol Cleary, https://www.vengaglobal.com/blog/translation-memory-can-save-money-time/

[5] How does Translation Memory work? EVS Translations, https://www.evs-translations.com/blog/how-translation-memory-works/

[6] What’s the Difference Between Translation Memory (TM) and A Glossary? Language Scientific, http://www.languagescientific.com/whats-the-difference-between-translation-memory-tm-and-a-glossary/

[7] An Introduction to Translation Memory, Emma Goldsmith, https://www.sdltrados.com/download/an-introduction-to-translation-memory/101902/