A translation memory is a type of software application that allows translators to reuse existing translations. In essence, a translation memory — or TM — is a database of text segments in one language together with their corresponding translations.
When the translator starts work on a new segment, the TM checks whether that sentence has already been translated. If a match is found, it is presented to the translator, avoiding the need to translate the sentence from scratch. Even when there is no exact match, the program can find similar sentences the translator has used before. These so-called fuzzy matches give the translator a starting point and, once again, save them from having to begin from nothing.
Segmentation
Translation memories work at the sentence level. When you use a TM, the source document is broken down into the sentences or segments that compose it. The segment is the smallest unit of text that can be reused when working with a TM.
Smaller units of text — such as individual words — cannot be reused because they may appear in different contexts and therefore require different translations. A word-for-word translation does not produce usable results.
Repetitions, 100% matches and fuzzy matches
Each translation segment is compared with what is already stored in the TM, and the results are split automatically.
A segment that is identical to one already in the TM is considered a 100% match. At some point in the past, the program encountered this exact segment and the translator supplied a translation.
When there is no exact match but the TM contains segments similar to the one being translated, those are shown as fuzzy matches. The match percentage can range from 0% to 99%; the higher the percentage, the closer it is to the new translation:
- A 99% match may differ in only a single letter or punctuation mark.
- A 75% match might differ in several words.
- Matches below 70% are generally not useful.
When a document contains several identical segments that are not yet in the TM, these segments are called repetitions. Most TM tools can identify repetitions before the translation begins. The advantage of repetitions is that, once the first occurrence has been translated, all the others become 100% matches.
File analysis and pricing
Before translation starts, the file is analysed against the TM. This analysis produces statistics that show:
- The total word count.
- The number of words that are repetitions.
- The 100% segment matches.
- The fuzzy matches.
This breakdown — combined with the type of text — gives a close approximation of the final cost of the translation. The most common categories used to break down a translation are:
- Repetitions
- 100% match
- 95%–99% match
- 85%–94% match
- 75%–84% match
- No match / unique words (74% or less)
The higher the match level, the less effort the translator needs to put in. Each category can be assigned a different price or discount. Any percentage below 75% is charged at the full per-word rate.
Benefits of using a translation memory
The benefits of using a translation memory are clear:
- Lower translation costs.
- Shorter turnaround times.
- Greater terminological consistency.
Because less work is required from the translator, using a TM means the translator can deliver sooner. Finally, the segments suggested by a translation memory help the translator stay consistent with previous translations, raising the quality and coherence of the final text.
Got a translation project?
At Textualia we are sworn translators accredited by the Spanish MAEC. If your project needs a sworn translation from English or French into Spanish, or you have questions about any related service, write to us at info@textualia.com or via WhatsApp on +34 646 07 68 54 and we will put together a tailored quote.