Nowadays, translation memories have become a key component of the quality translation production process. A translation memory is a data base containing original texts in one language aligned with their translations in another. The most important function of a translation memory is to provide total or partial suggested translations for a sentence and to standardise the terminology used. While translating, the database searches for MT segments in the language of origin. If the MT finds an exact match for the segment in the language of origin, this segment appears in the window, together with its translation, showing it is a 100% match. If the MT finds a segment which is not identical but similar to the original segement, this will also be shown together with its translation. These are known as "fuzzy match" segments. The cost per segment depends on the degree of repetition with a similar one. This means the rate per segment can vary between 20 and 100% of the rate for new words. This segment analysis is carried out using the translation programme that contains the memory.
The benefits it provides are:
- It guarantees terminological consistence: The programme automatically suggests the earlier translation, so the term is always translated in the same way.
- Shorter delivery time. Because the programme “remembers” previously translated material, it takes less time to translate the text.
- Cost reduction
- Alignment of previously translated texts: This allows texts to be aligned with their originals, so that subsequent translations can benefit from the advantage of using translation memories.