Translation memory demystified

Translation memory is a type of software that stores a unit or “segment” of a source language together with its translated equivalent in the target language.

This is useful for several reasons. Firstly, it improves the consistency of translated documents, as once a segment of text is stored in the memory, the translator will be prompted to use it wherever it appears in a file. Secondly, where text is repeated in a document, the client does not need to pay for the sentence to be translated twice. Thirdly, it makes the process more efficient, so translations take less time.

Other useful feature common to most types of translation memory software are:

  • Ability to analyse many different file types to get a wordcount
  • Protection of code in web files such as HTML so that strings are not damaged in the translation process
  • Displays all text within a file, so that nothing is missed out of the translation
  • Integration with glossaries
  • Preservation of file format structure

About “matches”

Translation memory software categorises segments in several different ways:

  • 100% match – where the text is identical to a segment already stored in the translation memory
  • Repetition – where a segment is repeated with the file or batch of files
  • Fuzzy match – where the segment partially matches a segment in the translation memory, e.g.: if the segment “legal translation into French” was stored in the translation memory, “legal translation into German” would be a fuzzy match. Fuzzy matches vary between 75-99%, according to how similar they are to the stored segment.
  • No Match – this refers to any text that is neither repeated, nor occurs in the translation memory. These are sometimes referred to as “new words”.

Why do these matter?

When a file is analysed using translation memory, the software produces a breakdown of the repetitions, matches, etc in the file(s).

Translation providers will usually then discount the rate charged for fuzzy and 100% matches, as these have already been translated (or partially translated) and already exist in the TM. Repetitions are also discounted.

Benefits of using TM

  • Cost saving on texts that are identical or similar to other pieces you’ve had translated
  • Improved consistency across your multilingual content
  • Translation process is more efficient, and so can happen more quickly
  • Localisation of web content without the hassle caused by damaged code
  • You build a store or library of translated content that you can use and reuse

Next time you have a translation project, ask us about Translation Memory – we’ll be happy to explain the breakdown of your text and show you the savings you can make.