Timeo Danaos Autem Machina Translationem Ferentes (I Fear Greeks Even When They Are Bringing Machine Translation)

According to estimates found mostly only on Internet, although I did hear it mentioned by an Indian journalist yesterday on Le Kiosque on French TV, a weekly program in which foreign journalists are freely discussing in a non-native language current topics, something that American viewers are not likely to see any time soon on their alphabet “news channels” as I wrote in this post, 100,000 to 200,000 Indian farmers committed suicide since 1997 as suicide was the only way for them to escape debt.

A major reason for these suicides is that seeds saved from crops in previous years, which used to be free to these farmers, were replaced by patented, genetically modified seeds that the farmers have to purchase now every planting season from giant corporations such as Monsanto. Monsanto seed, which was supposed to multiply the crop of poor farmers and thus make them prosperous, is now making them dead instead of rich.

Europeans and especially the French are very distrustful of genetically engineered food, which is supposedly much cheaper to produce and in some respects better than traditionally grown types of food. This is why this type of food must be clearly labeled as such in EC countries so that the customer would have a choice. In the United States, however, the law specifically prohibits labeling that would indicate that what the customer is about to devour is genetically modified “frankenfood”, precisely so that the customer would not have a choice. To me this is another indication that just like in India, corporations reign supreme in the United States as well.

What does all of this have to do with machine translation? Well, perhaps nothing. But there are certain comparisons that can be made.

When computer-assisted translation (CAT) tools such as Trados were first promoted to translators, the narrative used to get them hooked was that these tools would make it possible for translators to multiply their daily output and thus make much more money. Although the output of words per day was probably increased for translation of some documents such as highly repetitive manuals, the contracts that translation agencies now send to translators to sign typically include a clause specifying partial payment, a fraction of what translators used to make per word, for what is now referred to as “full matches” and “fuzzy matches”.

It would seem to me that the Monsanto seed model has been planted in the translation industry and is now used to aggressively redirect profit away from translators into somebody else’s pocket.

According to a new model that some MT enthusiasts are promoting now for early adoption by translators as I wrote for example in this post about an article in the ATA (American Translators Association) Chronicle, translators either need to integrate machine translation in their work, or risk becoming modern day luddites.

On the other hand, the contracts that some translation agencies are now sending to translators for signing prohibit the use of machine translation by human translators in their translations (see a sample sentence from such a contract in the Comments section below). There are two reasons for this:

1. Concerns about confidentiality, which may be compromised by using MT online, and

2. Concerns about quality, since when you run text in a foreign language through machine translation software and then edit the result so that it would look like human translation, the most likely result is inferior quality and mistranslations.

But some translation businesses are using the opposite approach: they are looking for translators who would be editing MT, either at a low hourly or a low per word rate, which could then be sold to clients at a hefty surcharge as the result of human translation, or at least as translation with human input.

I would be rather skeptical about the viability and profitability of this business model, but what do I know. I am just a translator. There must be many enterprising geniuses out there who are busy perfecting the business aspects of this new business model for translation as I am writing these words.

At the same time, a strong argument in favor of a translation model built on automated processing of texts through MT software, which would be later improved by a limited intervention of humans, is what one could call herd mentality of translators, especially on the part of the hungrier and less experienced ones, who years ago decided that in order to survive they must spend hundreds of dollars for “licenses” for Trados, which I understand cost 800 dollars and must be frequently renewed. As comments on my blog indicate, translators may hate Trados, but they are buying it because everybody else is doing it.

It will be interesting to see which approach to MT – incorporation or prohibition of MT in human translation – will prevail in the coming years in the translation business, specifically:

1. The approach that prohibits the incorporation and direct importation of MT into human translation, which is what some translation agencies are now including in their contracts, or

2. The approach that says that MT should be used because only luddites would try to resist integration of translation with human translation, while the incorporation is improved with subsequent post-editing.

Both approaches can of course also coexist at the same time depending on the type of the translation segment that we are talking about.

Some translators, including this mad patent translator, never bothered to purchase a CAT tool and probably never will because they see them as being largely useless for their purposes. Similarly, I use MT all the time, but only as a dictionary, which is free as well as omnipresent these days. I would never use MT “post-edited” in my own translations.

It is a very powerful argument when something is available free.

But it is well worth remembering that the lovely parting gift that the cunning Greeks left at the gates of Troy, which later became known as the Trojan horse, was also free.