I use translation memory because it's the best way to make an authentic
translation from an initial sentence or phrase. Because TM is based on an existing database
of manually entered translations and not on a computer calculation, each
subsequent translation is more authentic and more consistent than the last.
Also known as 'sentence memory' the process works on the basis of a Translator
making an initial, manual translation and entering it into the TM database. When
a document is being worked on it will parse the database as each sentence is
translated, looking for an exact match. As the database gets larger, each
translation increases the chance of finding a matching pair already located in
the database together with a percentage on the accuracy of the match, the
'fuzzy' factor. The final output of any TM translation will therefore be more
consistent.
The customer has the security of knowing that should he require amendments or
updates to already translated material the ensuing output will be a consistent
continuation of existing work.
Translation Memory is a form of 'Computer Aided Translation' (CAT), a general
term for technologies that aid a translator in his work with the aid of a
computer. As well as TM, the other most common form of CAT technology is 'Machine
Translation' or MT (not TM!). Most web based tools that you will encounter use
Machine Translation technologies that are based on a huge built-in dictionary or thesaurus of words and phrases, stored
on websites that are easily maintained but don't store any data that could be
useful to you the customer in the future. Although the technology 'seems'
terrific and you could get impressive results it is in fact not able to
make 'quality' translations or give consistent output because it will never be
based on anything attributed to you or previously translated work.
With technical translations and brochures containing repetitive information,
component numbers, indexes etc.. the Machine Translation process is particularly ineffective because each
translation needs to recalculate from scratch after each phrase that it encounters.
Interpretacao.com however uses Translation Memory (not MT) technology so you can
be rest assured that your translation data will be recorded from beginning to
end, resulting in consistency. There are a number of decent products around. I recommend
Wordfast because it's easily implemented into Microsoft Word, it's
fast, affordable, and one
of the most open ended and editable software products around.
* Note that the cost effectiveness of using Translation Memory for each project will be assessed
individually. Sometimes for small journalistic or literary articles it's quicker and
easier to translate manually using conventional tools and spell-checkers.
* It's worthwhile noting that although I don't recommend Machine Translation
software it is indeed a very valuable aid to the translator (in a similar
sense to using a conventional dictionary) and should not be disregarded.
