ABSTRACT
Translation memory systems of this type can be particularly useful if the source-language text is an updated version of a document which has been translated previously and then stored together with its translation (for example a computer manual). When starting to translate the new text using the translation editor, the system automatically performs a segmentation of the source text and looks up the segments in the translation memory database. If a segment is found, the translation stored with this seg ment is offered as a possible equivalent. It can be adopted by the translator as it is, amended, or rejected. As soon as the translator has fin ished the translation of this segment, the new source and target segments are again stored together in the translation memory.