A back-translation (BT) is when a translation or target text (TT) is translated back into its original source language (SL). When the back-translation is completed, it is then compared to the original source text (ST) in an attempt to gauge the quality of the translation, which I believe is incredibly stupid. While it's true that gauging the quality of translations is very difficult, I don't believe for one second that BTs are a viable way to do it.
The idea of BTs is based on a premise similar to that of mathematics, but is foolishly founded on the concept of a number of assumed exact equivalents between languages, be they lexical, syntactic, semantic, etc.
In maths, the equals sign indicates that the equations on either side of the symbol are the same. However, even this logic seems to be flawed, since if you imagine that numbers are words, you could easily say 7 = 7, but also that 7 = 6 + 1 or 7 = 5 + 2, etc.
If you can express sums in multiple ways, you can certainly express almost any sentiment in plenty of different ways. This is the problem I have with back-translations, as they assume that there is only ever one "correct" way to translate any given phrase and translating it back should yield exactly the same results.
Another issue I have with BTs is the assumption that discrepancies between the BT and the ST are due to mistakes by the first translator and issues in the TT. Comparing the ST and BT to one another completely ignores the TT and therefore the entire work of the first translator, as the BT is simply a derivative work of the BT.
|Valet parking wasn't really necessary 100 years ago.|
Imagine you give your car to a valet to park it while you have a meal. Later, when you collect your car, a different valet is driving your car and the car has a huge scratch down the side of it. Who scratched your car?
If you use the logic of BTs, the first valet definitely damaged your car as it has changed since you left it. However, without seeing the car's journey to and from the parking space, there's no way of telling who caused the damage.
What do you think of BTs? Is there a better way to gauge translations? Tell us your opinion in the comments below!