A translation memory (TM) is a database that stores “segments”, which can be sentences, paragraphs or sentence-like units (headings, titles or elements in a list) that have previously been translated, in order to aid human translators.

A translation memory is a database that stores sentences, paragraphs or segments of text that have been translated before.

The translation memory stores the source text and its corresponding translation in language pairs called “translation units”. Individual words are handled by terminology bases and are not within the domain of TM.

Software programs that use translation memories are sometimes known as translation memory managers (TMM) or translation memory systems (TM systems, not to be confused with a Translation management system (TMS), which is another type of software focused on managing process of translation).

Each entry, or segment, in the translation memory includes the original language, known as the “source” and its translation, known as the “target”. These pairs are called translation units, or “TUs”.

Translation memories (commonly known as a “TM”) are used with translation software, such as SDL Trados Studio, and automatically suggests stored identical or similar matches as new documents are translated. This means sentences, paragraphs or segments of text that have been previously translated never need to be translated again.

Translation memories support the localization process, dramatically improving the quality, speed, consistency and efficiency of every translation job.

How does a translation memory work?

Translation memories work as part of a computer-assisted translation (CAT) tool, such as SDL Trados Studio. When you open the document you want to translate (the source file), the translation memory checks to see whether any of the content has been translated before, and looks for any “100% matches” (identical matches) or “fuzzy matches” (similar, but not exact matches) that appear in the new source file. Any previously-saved translated text is instantly placed within the corresponding target segment within the Editor window of the CAT tool.

As you work your way through the source file, any suggestions from the translation memory can be either accepted or overridden with new alternatives. If a translation “unit” (a segment of source and target text) is manually updated, it can be saved to the translation memory for future use. In a similar way, all segments in the target file without a match would need to be translated manually. Once translated, these can then be added to the ever-growing translation memory.

When would I use a translation memory?

Translation memories are perfect for anyone who localizes content from one language into another. When a translator’s jobs regularly contain the same kinds of phrases and sentences, a translation memory will significantly increase the speed and quality of the translation.

Translation memories are very helpful when translating content out of context. An increasing number of organizations rely on Content Management Systems (CMS) to manage their information. A CMS allows individual blocks of text, rather than entire documents, to be created/edited and then published in a variety of different formats. A translation memory helps to make this process quicker and more consistent.

How does a translation memory differ from machine translation?

Translation memories are created by a professional translator and reuses content that has been previously translated to improve the speed, quality and consistency of future translations.

Machine translation (commonly known as “MT”) automatically translates without any human input. These kinds of tools are fast, but often produce poor quality translations as a machine is not fully able to understand the subtleties or contexts of language. As a result, quality and accuracy isn’t as good as a translation by a person. In addition, machine translation can only be used for a limited number of supported languages.

However, machine translation does have its uses and may be used in conjunction with a translation memory to increase the overall speed of a translation. An example would be when a translation memory does not have enough information to complete a segment of text, a machine translation provider can make a suggestion for the untranslated segment. The translator then post-edits the machine translation entry and saves the completed segment to the translation memory for re-use.

How does a translation memory differ from a termbase?

A translation memory stores segments of text as translation units. A segment can consist of a sentence or paragraph. The TM holds both the original and translated version of each segment for reuse.

Like a translation memory, a termbase is a searchable database. However a termbase contains a list of multilingual terms and rules regarding their usage. Terms can be anything specific to a company, such as the way that brand names or product names are translated, or indeed, should never be translated.

Translators typically use both a translation memory and a termbase within SDL Trados Studio.

Does SDL Trados Studio come with a ready-made translation memory?

Translation memories are created from the completed translations of a translator, and are not pre-packaged with SDL Trados Studio. By translating within Studio, or analysing previously translated content, a translation is built using a translator’s own work.

Can I make a translation memory using my previous translations?

If you already have documents that you have translated, you can make a translation memory from these files using a feature in SDL Trados Studio called Alignment. This is a simple process and allows you to leverage all of the work you’ve translated before, even if you have never used Studio or another CAT tool.