The database that is stored the already translated sentences or paragraphs in the form of segments for helping the human translator, is known as the Translation Memory (TM). The Source text and its respective translation are stored in the form of language pairs in the translation memory, which are known as the “Translation Units”. However, the individual words are not in the TM domain as they are handled on the basis of terminology. The software that is used TM is usually referred as translation memory managers (TMM). TM is used with a CAT tool, terminology management systems, word processing program, multilingual dictionary or simple MT output.
How the Translation Memory works?
The TM program actually breaks the text that needs to be translated into small segments, it then searches for the matches between the segments and source of the previously translated text that is stored in the TM, it then presents the matching pairs as the candidate of translation. The candidates act as an option for the translator and he can replace the translation with the candidate or modify it. The modified translation or the fresh translation is also stored in the database of the TM.
There are some memory systems that search only for matches that are at 100% while some systems retrieve from the similar segments with the differences flags. However, the ideal TM system searches only for the text in the source segments. The segment that does not match is manually entered by the translators, which is, later on, stored in the database on the TM system for future assistance.
Benefits of using a Translation Memory System
The TM systems are very suitable for specialized vocabulary documents and technical documents, other benefits of translation memory include;
Some Obstacles in using the Translation Memory System
The main problems in using a translation memory include;