What Is Translation Memory?

Translation Memories (TM’s) are libraries of historic translated strings, stored by language, that we build up in our translation workbench tools every time we perform a piece of translation work for a client.

In general, TM’s are built up over time as we do continued volumes of work for our clients. As such, the efficiencies from leveraging TM’s can drive down the effective cost of translation work over time – across updates and new releases of software and across different media (multiple UI platforms, Online Help, Docs, Technical Product Specifications, Marketing and Web Collateral, etc.)

At Iota, we ensure that the TM’s we help develop for our clients remain the property of the client. These can, over time, form the basis of a significant asset to our clients, given the cost efficiencies they can deliver as their businesses grow and expand into new international markets.

When analyzing any new content for translation, it is first run through a tool which compares this new content against the TM. Any matches that this analysis throws up fall into a number of categories:

[a] New Words refers to source language content which, when compared to existing TM libraries, is found to have either (i) no pre-existing legacy translated text from previous projects or (ii) cannot be
matched to more than 85% of the segment of text concerned. New Word content requires full translation and is generally charged at an agreed rate per word.

[b] Fuzzy/Partial Matches (85%-99%) refers to source language content which, when compared to existing TM libraries, is found to have legacy translated text from previous projects which can be partially matched to more than 85% but less than 100% of the segment of text concerned. Fuzzy or Partially Matched content requires review and in-context translation of the unmatched element of each such segment of text only, rather than of the entire segment. Hence, it costs a little less – 66% of the cost of New Words.

[c] Repetitions refers to segments of source language content which, when cross-referenced within the entire source language content provided for translation are found to be repeated, often in a range of contexts, throughout. Repetitions require proofreading and review of each occurrence of the agreed, in-context translation of repeated segments of text. This is a little less expensive again – 33% of the cost of New Words.

[d] 100% Matches refers to segments of source language content which, when compared to existing TM libraries, are found to have legacy translated text from previous projects which can be fully (100%) matched to the source language content provided for translation. Generally, our clients place full reliance on the integrity of the TM software and no review of the automated output from the TM is requested for 100% Matches. In the event that a review of 100% Matches is requested by the client, charges are applied at rates similar to those for Repetitions.

Any questions? Get in touch…