SimulTrans Localization Blog: SimulTips

Why Create a Translation Memory?

[fa icon="calendar"] September 14, 2016 / by Eli Karpodini

Translation Memory.pngImage: Paul Yoakum

In today's profit–and time-driven economy, efficiency is of the utmost importance. The translation and localization industry, of course, is no different. Neither clients nor LSPs (language service providers) like spending time doing the same thing twice. This means that once a portion of content is translated, the translator should never have to translate it again.

In that respect, a common practice followed by numerous LSPs and their clients is the use of Translation Memories, or TMs as they are known in the industry.


What is a TM?

A TM is in essence a database of existing translations. In other words, a repository of source text, for instance English, paired to its target text in another language, for instance French.

So, every time a new string of text is translated, it is stored in a database along with its matching translation, forming a translation unit (TU). Then, when the translator works through a translation memory system, the tool searches through the TM for a match of each segment.

If that segment, or a similar one, is available in this database, then the system retrieves one or more TUs from the TM and presents them as results to the translator. These results are accompanied by a "match indicator" which is basically the degree of similarity between the new segment for translation and the one available in the database.

What are Translation Memory matches?

If the segments are identical, the match indicator will come up as a 100% match or a context match (CM), when the segment before the given segment is also identical. If there are some differences between them this "match indicator" will appear reduced and this is considered a fuzzy match.

In our industry, we consider an useful match to be any segment in the 75-100% range. Anything under this threshold is considered “noise” as editing those matches might be counter-productive and translating from scratch is usually preferred in these cases. Usually the translators adjust their translation tool settings to ignore any TM result with a “match indicator” under 75%.

Benefits of a TM
As already mentioned, the benefits of the TM are linked to some of the main parameters of every project: time, cost, but also quality.

It goes without saying, that by recycling already existing translations, a project’s turnaround time is considerably reduced. By leveraging their own or other linguists’ work, translators are more productive as they can increase their daily word throughput and they can also dedicate more time to the truly challenging parts of translating.

Another project parameter that has been greatly influenced by TMs is project cost. Cost and TM leverage decrease and increase respectively, on diametrically opposite directions. The higher the TM leverage, the lower the translation cost. As translators have to spend less time on the segments that come from the TM, this allows for price reduction, usually based on a word rate.

For example, the standard price for a new word is €0.10 per word. By using a TM, a lower rate for fuzzy matches can be obtained. Fuzzy matches between 75-90% can be charge at a % of the new word rate, for instance €0.08. 100% or CMs are offered for an even lower rate or percentage or in certain cases free of charge if a review is not required.

Last but not least, the use of a TM ensures consistency in terminology and style. Even when multiple translators work simultaneously on a translation project or when a new translator is involved, the use of a TM ensures consistency and preserves the client's terminology.

On top of that, the TUs bear some metadata that allow the linguist to make informed decisions based on the origin of the TM result. So, for each TU, the linguist can know when and by whom it was created, and decide accordingly which one to use.

For instance, they might select the latest or the most reliable translation (coming from the most seasoned linguists) or they might decide to create a new translation if they consider the ones retrieved from the TM to be outdated or not suitable.


In my opinion, TMs are undeniably one of the most–if not the most–useful tools that translators have at their disposal nowadays, and also one of the most valuable assets of any company when it comes to translating their content.

Want to learn more?
Watch  Translation Memory Video

Topics: Localization Technology, Translation Services

Eli Karpodini

Written by Eli Karpodini

Elisavet Karpodini is a member of the locsolutions team at SimulTrans. Elisavet has studied Translation, International Relations and European Studies, and holds an MSc in Translation Technology.