Do you think - once the translator has had some small communication with the author about which bits are important and which not - that a mix of both the approaches you outline might work? Especially for work like Sebastian Barry, or Cormac McCarthy? Both of who play with words and structure so much?
I actually think that all real translation is a hybrid between these approaches; that's because you
can't be 100 % faithful in another language, and you'll always stick to the book (or it wouldn't be a translation). I don't set out translating with a conscious desire to be faithful to the original at the expense of smooth readability. It's really just an extension of my general personality (e.g. I try to be as verbatim as possible when quoting others, and I fret about misrepresenting them even so, by selective quoting...).
It's when discussing translation problems that you realise which position you tend towards:
For example, in a poetry translation workshop (English --> German), the word "bingo" came up in a british poem. I translated the game as "Bingo", only changing capitalisation (in accordance with German capitalisation). Other people used different words, I forget which, but I think "Tombola" was one of them.
Now, there are basic questions you could ask:
- Does a native speaker of German understand a reference to "Bingo"? (I do think so; a couple of years later national television started a Bingo Show on TV, so now there's really no question of "Bingo" going mainstream.)
- Does the context make clear that we're talking about a game, or do you have to know what "Bingo" is, to get the game reference?
- Is the exact manner of the game important to the poem? Are slight substitutions acceptable?
etc.
Whatever you do, you're taking risks. Use "Bingo" and you're risking to kick people out of the poem. Use "Tombola" and you risk "false" associations (not to mention that - at least in Austria, where I live - there is an ambiguity in usage, where Tombola could reference the Italian game, or simply a charitable lottery with donated prizes...).
You don't always pick the "faithful" version, and you don't always pick the "smooth" version, as risks vary in likelihood and severity of catastrophe. (For example, likelihood of people not understanding "bingo" has decreased with a national telivision game show airing, but the misunderstanding, should it occur, would still be as severe as ever.) Finally, if the text contains other culture-specific words, some of which are
more familiar than others, you run into the problem of consistency. If, say, the word "bingo" would occur in a very specific cultural context, it's harder to justify substitutions, than when it occurs in a metaphoric subtext.
You really shouldn't use your translation philosophy as a prison. But when you're looking at what you think is a "bad" translation, it helps to realise that other people have other values. And values change with time.
Take, for example, what German television did to
Cheers. (Click on more infos, and read.) They gave the cast Germanised names; e.g. Norm = Helmut.
This was controversial back then (especially since they made changes to the characters, too!), but it's unthinkable in present day Germany. The re-runs of
Cheers do not use these old dubs, either. They couldn't. A modern audience would be baffled. This reflects a change in priorities, which in turn reflects higher global mobility (of both people and information).
What the distinction means, in the end, is that: if you want a cultural conversion (for whatever reason - e.g., there are dialectal versions of the Australian movie
Babe [the talking piglet], in both Austrian and Swiss dialects), I'm probably not the translator you want. I
could probably do it to some extent, but someone else would deliver better quality at a higher speed. It doesn't mean that I pick my philosophy and stick with it come what may.
And again, this is all theoretical, since I'm no professional translator (even though my only publications to date are two translated poems, one of which must have been decent as it has been re-printed).