ABBYY Aligner 2.0
Translation Memory is a database containing a set of texts translated into different languages and divided into fragments which are juxtaposed according to their content. That means that fragments of the source text are juxtaposed to the fragments with the similar content translated into another language.
Such texts translated into different languages are called parallel texts (or bilingual texts).
Fragments in which each text is divided are called segments. Segments are usually made of one or more sentences, parts of sentences or word groups, and, in some cases, words.
Juxtaposed parallel texts are called aligned parallel texts.
Thereby the parallel text alignment (or bilingual text alignment) is a juxtaposition of the source and translated texts by segments. Translators usually call this process translation alignment. A Translation Memory database consists of such aligned translations.
TM bases are used by translators when working in various computer-aided translation tools (CAT-tools) which allow executing TM base search, finding repeated segments in new texts and replacing them with translations taken from previously translated texts.
In other words, translators, when working with CAT-tools, use them to divide new texts into segments and check if these segments match those stored in translation memories. When a given match percentage is detected, translation variants are displayed. If a source segment fully matches a translated segment in the database (translation memory), the latter can automatically replace the former. New segments can be slightly different from those stored in translation memories. In this case, it is also possible to insert the stored segment into the text, but a translator will have to make necessary changes. Thus, a translator either accepts the suggested variant, or changes and accepts it, or translates the segment as a new one.