Why a 100% translation memory match is not 100% free
Computer-assisted translation (CAT) tools, particularly the translation memory (TM) feature, are one of the best technologies invented for the translation industry. As the name implies, CAT tools assist translators, but they do not replace them.
TM stores every pair of units (words, phrases or sentences) translation memory so that they can be re-used in the future – either again in the same file where text repeats itself, or in future files with the same text. It acts like a big glossary (of words, phrases or sentences) that helps maintain consistency in style, terminology and minimizes the amount of repetitive work that a translator does. These reductions in repetitive work reduce time for the translator and costs for the client, but while the translation is automatically retrieved from the TM, it still needs to be checked based on the context of the new text and file. This is because different languages work differently based on context. Therefore, many times, the translator needs to make adjustments to the text according to the rules and usage of the target language.
Here are some simple English-to-Spanish examples:
ENGLISH | SPANISH | Notes |
nurse | enfermero (male) | Translation changes depending on the gender of the nurse. |
enfermera (female) | ||
the first patient | el primer paciente (male) | Even though the translation for “patient” is the same, the modifiers change depending on the gender of the patient. |
la primera paciente (female). | ||
You must read this form | Debe leer este formulario (adult) | Translation of the verb changes depending on the age of the audience or the number of people addressed. |
Debes leer este formulario (child) | ||
Deben leer este formulario (group) |
So even though the translation in the TM may be correct, it may not fit the specific context or new audience.
Here is a more complex example where a change in the source text causes a change in the target text, but on a different unit. Version 1 of a client’s document is below:
Tell the study doctor about any changes in your health
ENGLISH | SPANISH |
If you participate in this study, you must do the following: | Si participa en este estudio, debe hacer lo siguiente: |
|
|
|
|
In version 2, the client updated the source text. The first line changed from “you must do the following:” to “you will be expected to:”, and so the translation of the bullet points after the changed unit will need to change, even though the English text for the bullet points did not change:
ENGLISH | SPANISH |
If you participate in this study, you must do the following you will be expected to: | Si participa en este estudio, se esperará que usted: |
|
|
|
|
Here is another complex example. The translation memory has the following data:
ENGLISH | SPANISH |
Invoice | Factura |
Date | Fecha |
Number | Número |
Invoice
Date |
Invoice
Number |
The source text is below:
However, the TM translation is not correct and would need to be edited:
ENGLISH | SPANISH |
Invoice | Factura Fecha de |
Date | Fecha factura |
Invoice | Factura Número de |
Number | Número factura |
In brief, translations is the interpretation of meaning and not words. While a CAT tool can suggest a 100% correct word match, a human needs to check if it is actually 100% correct in meaning. BURG Translations respects the inherent difference between languages and thus does not assume by default that a 100% TM match in words is a 100% match in meaning.