A VLTM, otherwise known as a “Very Large Translation Memory” is normally server-based translation memory that contains a very large collection of previous translations. These are sometimes utilized by translators to perform their translation more rapidly. They are based on a server because they would work too slowly on a remote computer due to the massive amount of information. Many automated online translation platforms used by agencies use such tools. Please read my posting on translation technology or translation memories if you are new to this topic.
Application programming interfaces, also known as APIs, can also be used to interface with various servers and even online translating systems so that the results appear right in the translator’s CAT tool. It “pre-translates” material that the translator can look at and “fix up”. It is believed that by doing this, the translation process is made quicker. This is also known as machine translation.
Ferris Translations does not particularly agree with these methods. Although it may be quicker to have sentences pre-translated and less typing might be involved, unless you are talking about a TM that is dead-on with regard to the subject matter, it may also be more work for a good translator to use this method of translating instead of just translating it outright. Any project manager knows that rework is always more costly and time-consuming than just doing something correctly from the start and, in this case, from scratch. From a quality perspective, the latter is much more desirable.
Concerning translation memories in general, I do not believe that it is worth the effort of saving translation memories if the content is not interrelated. Translating a bunch of articles based on different things and saving a translation memory generated is quite useless, because there should not be any matches. The same goes for the translation of any literary material. It is not necessarily considered good style to repeat sentences again and again in the same way, even if the same thought is there. On the other hand, for technical manuals, it would be not only necessary but quite desirable to use a translation memory to ensure translation continuity.
© 2014 – 2016, Ferris Translations e.U. All rights reserved.