O Corpora, O Mores!
I have always considered corpora essential for the translation of various specialized texts. I actually believe that one cannot possibly begin to translate without consulting similar texts in the target language (the so-called parallel texts). It is highly essential to experience, to understand and ultimately to interiorize the textual techniques of your field of specialization within the target culture. Otherwise, your translations in the medical field, for instance, will not be in compliance with the standard drafting techniques and your (expert) target recipients will have a hard time understanding your translation.
Nonetheless, there is a risk of over-rating the importance of corpora. Due to their statistical significance and their vastness, we may be tempted to consider them the only norm for specialized or non specialized language. In other words, if a certain expression doesn’t exist in the corpus, we may believe that it doesn’t exist at all. Which is not necessarily true. We have to keep in mind that language and discourse are more than just a collection of words, sentences, paragraphs and texts. Besides, automated translation is greatly based on corpora, and as much progress as this technology has made in the past decade, it still fails to offer human-like results. And this is due to the fact that a machine does not understand the meaning of words, it cannot understand what a text is about. And we, as translators, do not translate words, collocations and sentences, we translate meaning. Or this is what we should do.
So, what is your opinion about corpora and how do you use them?
