A translation memory is a repository for human-generated translations that enables the reuse of previously translated content. The text segments in the source text are usually stored in pairs in this database and can be retrieved via a search function if required. In this way, a professional translator can significantly improve the consistency and efficiency of their assignments.
Many translators use translation memory systems (TMS) in their daily work. The software evaluates to what extent elements (also: segments) of a text to be translated have already been translated and saved. If there are identical or similar segments, the system suggests the translation of the old text section to the translator. They then decide whether to accept, change or create a new translation. All other segments that are not yet stored in the TMS must be translated manually. Subsequently, they can also be added to the database.
Advantages
Translation memory systems support the translation process. By not having to translate segments of a text a second time, the translator can not only do their job faster, but also more accurately and efficiently. When several translators are used or when several documents are edited, the TMS ensures that all those involved work on the same database and that the translated texts are worded consistently.
Difference: translation memory and machine translation
Machine translation software automatically translates content, usually within a few seconds. However, this form of translation is prone to errors, as computers are not (yet) able to read “between the lines”. In certain cases, machine translation in conjunction with a translation memory is worthwhile in order to increase the speed of translation. The translation memory does not translate content on its own, but is merely a memory that enables the reuse of previously translated content.
Features
Translation memories are constantly growing and are available in different sizes. They have various functions that enable the analysis of existing translations. When searching for similar segments, algorithms are used that usually calculate a percentage similarity value. An exchange between different TM systems is possible in principle; the TMX format (Translation Memory eXchange) is used for this purpose.
These translation memory systems are available
With regard to the structure of translation memories, a basic distinction is made between two types of databases.
Data storage with context
On the one hand, there are systems in which reference texts are stored in pairs, separated according to source and target language. In this way, the context of the respective segments is maintained and enquiries can be made specifically about certain topic areas.
Data storage without context
On the other hand, there are database-oriented systems in which the segments are stored in isolation. They are mainly used when particularly large amounts of data have to be stored and have a positive effect on the runtime of the systems.
Load more
FAQ: More questions about translation memories
What is a translation memory system?
A translation memory system is a database in which source texts and their translations can be stored. The structured data facilitates the translation process.
How does a translation memory system work?
The software evaluates to what extent the text to be translated resembles already stored text segments. The translator then decides whether to use the suggested translation.
What are the advantages of translation memory software?
Translation memory systems allow translators to work much more productively. Avoiding multiple translations enables a fast, consistent and efficient translation process.
What are the best translation memory tools?
There are various translation memory providers whose software usually differs only slightly from each other. The most common systems are: MemoQ, Trados, WordFast, Across, Transit, Passolo and Déjà Vu.
This site is registered on wpml.org as a development site.
