Machine translation from humans?

A lot of attention is being spent on errors created by machine translation due to a misinterpretation of meaning. Oddly enough this same type of error can be caused by poorly trained linguists.

A large amount of blog posts and news stories about translation are geared toward the efficacy of machine translation and tools like Google Translate. The errors you most often read about are due to a misinterpretation of the meaning of words or phrases in the source text. We recently reviewed a client project where this occurred due to an overly ambitious employee that had some education in language during college and figured that there could be no harm in modifying some previously approved ad copy with the assistance of an online dictionary.

The new ad copy used the word “save” (as in save money on this offer) and that was the only new word required to change a previously translated phrase. The employee decided to use the Italian word “Salvare”. When the ad was reviewed in country it was flat out rejected because of the headline. Salvare is indeed a valid entry in a definition of save however it means rescue or save someone or something from a catastrophe. The proper term would have been “risparmiare”.

This type of error is known as a false cognate. Machine translation tools and people with no formal translation training or education often make this type of error.One of my personal favorites in Italian is the word for Factory. Fattoria which sounds much like factory is actually the proper word for a farm. A factory is a fabrica.

This is a simple example of how having a good resource like a dictionary and some previously approved text does not make you a qualified translator. The mistake outlined above could have easily been avoided by having the change reviewed by a qualified translation provider.