What Is a Translation Memory and How Does It Work?

Saving time and money on translation projects, increasing the quality of the output text or allowing several translators to work on a project simultaneously are just some of the great advantages of using a translation memory (TM).

This memory is truly an ally and an essential tool for both translators and client companies. And the more it grows, the greater the benefits.

What is a translation memory and what is it used for?

A translation memory is created using computer-assisted translation (CAT) software. It can be defined as a linguistic database in which translation segments (words, expressions, phrases, etc.) are continuously recorded for subsequent reuse as a reference when translating the same or similar text. 

CAT software facilitates the translation process and has the ability to break the input text into segments called translation units (TUs), in order to record the equivalents of these segments in other languages in a file or database. This database is known as a translation memory or TM.

In short, translation memories are used to speed up the search for reference terms in the translation and to facilitate the use of recurring phrases. In addition, they can be passed on to new translators or clients in order to maintain style and terminology.

CAT software is a strong trend in the translation world for 2023.

Find out more:  

Top Machine Translation Software Trends for 2023

How does a translation memory work? 

As we have seen, translation memories are composed of sections or fragments of texts that have been previously translated and approved. Thus, when the translator works on a new project, the CAT tool automatically divides the source text into segments or sentences, lists, or tables.

For each new fragment processed by the translator, the translation memory searches for matches between that new fragment and the sentences it has previously recorded. 

If it detects matches between the new segments and those already stored, the system displays them on screen in pairs (source language and target language), together with the match percentage. The TM source segment will also indicate the elements that have been changed, added, or removed in comparison to the current source segment.

Consequently, the translator can analyze and accept the suggestion based on the percentages, or they can modify it and replace it with another completely new translation, but which will also be stored for future translation.

The match percentage found by translation memory management can be classified as follows:

• 0 to 74 % match: this means that the translation memory has no matches or that the matches it has are too low to be useful. 

   

• 75 % to 99 % match: this is a partial match. That is, the compared segment has some similarity to the segments stored in the translation memory. 

• 100 % or exact match: this means that the analyzed segment is identical to the one stored in the translation memory. 

The differences between a translation memory and machine translation 

It is essential to differentiate the management of a translation memory from the work performed by machine translation. 

Firstly, machine translation is the type of translation performed by a machine without human intervention, while translation memory automatically compares a source text with segments of previously approved translations. 

Therefore, using translation memories is not synonymous with machine or automatic translation. It only means that comparisons with segments of previous translations are made automatically before the translator validates or edits them. 

Read more:

Everything you need to know about machine translation

When should a translation memory be used?

Translation memories are used to speed up the translation process and obtain higher quality content, especially when:

• Documents include technical content from specialized sectors, such as scientific or legal texts. 

• Documents have a large number of repeated words or phrases, such as technical manuals.

• Documents that need to be updated, such as user manuals, are reviewed and translated.

• A large volume of content needs to be translated and the work must be distributed among a team of translators. 

• Entire localization projects need to be processed.

• Greater consistency and speed are sought in the process performed by a CMS content manager.

The advantages of using translation memories for your business 

At Pangeanic, we use translation memories to speed up translations and tailor our work to the schedule you and your company need, to produce more consistent texts, to reduce costs and to help you build your international brand and image.

Our expert translators rely on translation memory management to ensure consistent use of terminology for each project. They identify key terms, develop multilingual glossaries and create style guides for each client, according to the target region. By doing so, we ensure that your translation project communicates the essence and style of your brand. 

For large projects, our teams of translators and linguists have access to a translation memory that allows them to work simultaneously, while ensuring consistency in terminology and style of content. 

At Pangeanic, we use translation memory technology to provide your company with more cost-effective, higher quality translations in a shorter time and with the guarantee that it will build and protect your brand globally. 

Tell us about your project and, at Pangeanic, we'll help you optimize your translations while maintaining your company's identity.