Translation Memories
What is Translation Memory?
We have been using translation memory software (in one form or another) since we started as a translation company (in 1986).
The system has recently become an important commercial entity with translation memory systems now fully integrated in our workflows.
Not to be confused with machine translation (which is completed solely by a computer), a translation memory electronically stores and reuses previously translated material.
As we translate, the translation memory captures and remembers the translations for future reuse. Applying sophisticated linguistic algorithms, the translation memory segments source and target content into units of reusable content in the form of database records. As new content is ready for translation, translation memory searches sentence-by-sentence for pre-existing translations and identifies all existing perfect and close matches.

Why Use Translation Memory?
We use translation memory technologies for the following reasons.
- Economy. It saves us and therefore our client's money by not having to re-translate everything again. This is particularly useful if the text in question is part of an update or similar project to one that we have previously translated.
- Quality. It improves consistency. Once the translated terms are validated they are stored within our translation memory system which will make sure that every future release of your product will contain the previously validated terms.
- Speed. By re-using previously translated text, translation memory speeds our workflow up breaking through the standard limit of 2,000 words per day per translator. Also by using technology to help us we can place tens of translators to work simultaneously on one project. We are doing this at the moment to help the London Organising Committee of the Olympic Games to deliver over 1 million words in less than one month.
- Compatibility. We can import translation memories that your company may have previously prepared. We can even import previously translated documents into our system to create memories. So even if you haven’t got any memories (or your translation company pretends that they can’t export them), we’ll be able to generate them.

An Example of Translation Memory Workflow
Imagine a 50-page user manual, which needs to be translated before being shipped with a specific piece of equipment.
The initial translation is carried out with the use of a translation memory system, saving you time and money at an early stage. All of the segments of text (a segment can be anything from a title to a full paragraph) are saved in our system along with the validated translation.
A few months later, your technical authors have to make amendments to the user manual. You send the file to us for analysis (this can be in any of the popular authoring systems, eg InDesign, FrameMaker, XML). Our system indicates the segments of text that have changed and the new segments for translation.
The amended text is translated (as opposed to the whole document) and presented to your in-country team for validation (as pdf). New terms are amended (if needed) and the final document processed and prepared for print/distribution.
The more translation that we carry out for a particular organisation the larger the translation memories will be, and, therefore, quicker turnarounds and lower costs (per unit word) will be experienced.
Supported file formats
Translation memories can be used with many file formats. A few examples are listed below. Do not hesitate to contact us should your required format not be there:
- Word, Excel
- htm, html, jsp, xml
- DTP formats: FrameMaker, QuarkXPress, InDesign
- Properties file
- Resource files
Supported languages
All European (Western and Eastern) languages, Chinese, Arabic, etc.
Any Questions?
We understand that this is a complicated topic (especially to people outside of the translation industry). So if you have any questions please do not hesitiate to contact us on 01908 572600, email info@k-international.com or via our contact us page on this website.