I'm enjoying and admiring TBK without being able to figure out why, 30 years ago, I thought it was the best novel I'd ever read. My 25-year-old self was different from the current me, but not by any means prone to unaccountable literary judgments.
When I'm done reading the much-acclaimed Pevear/Volokhonsky translation, I may go back and read some key passages in the much-maligned Constance Garnett version. I can tell that P & V are more accurate: Garnett sometimes makes silly-sounding choices, right from the first page, where P & V describe daddy Karamazov as "the type of man who is not only worthless and depraved but muddle-headed as well," and Garnett says he is "a type abject and vicious and at the same time senseless." I remember thinking in my first reading that "senseless" sounded like a poor choice, and wondering whether it was Dostoevsky's fault or Garnett's. The dialogue sometimes sounds less stilted in P & V as well.
And yet, and yet . . . I wonder whether Garnett may have captured the "flow" of the novel, the emotional ups and downs, better than P & V. The novel seemed much more powerful when I read it thirty years ago--the psychology of the tragedy had a Shakespearean intensity, although Garnett's diction was anything but Shakespearean. It would be interesting if Garnett were more accurate overall, despite P & V being more accurate word by word. In translations, momentum can be as important as word choice. Some translations are faithful to the original at the cost of missing the essence--in being a bit too precise, they fail to carry the emotional weight of the original into the new language, whereas a freer translation better conveys why the poet bothered to write the poem.
One of the critiques I've dipped into encourages me in these speculations. Victor Terras's book "Reading Dostoevsky" has an interesting appendix called "How Much Does Dostoevsky Lose in English Translation?" Terras compares Garnett and P / V to the Dostoevskian original and finds that both have to sacrifice many untranslatable nuances. He concludes by saying, "Pevear's translation serves the scholarly reader better, as it brings him or her closer to Dostoevsky's craftsmanship. Garnett's somewhat old-fashioned English has great charm and is close to the ethos of Dostoevsky's Victorian narrator. It is not quite Dostoevsky, falling short of the prodigious energy of his dialogue, but the general reader may find it preferable to Pevear's." Terras seems to be responding to the same differences that I'm talking about.
Some old French guy famously says that translations are like women: "lorsqu'elles sont belles, elles ne sont pas fideles, et lorsqu' elles sont fideles, elles ne sont pas belles." ("When they are beautiful, they are not faithful; when they are faithful, they are not beautiful.") That's unfair to women but just right about translations. When we find a translation that is both belle and fidele, we know we are in the presence of genius (Richard Wilbur's Moliere comes to mind). Dostoevsky may still be awaiting his genius.
Of course, there are other translations of TBK (David McDuff's and Andrew McAndrew's are easily available). I guess I'll never know whether there's a better version than Garnett and P & V, because I'd have to read the whole thing each time to really test them--after all, word for word, P & V are clearly better than Garnett. It's no big deal to compare a half dozen translations of a sonnet; an 800-page novel is something else again. (And, of course, not knowing any Russian is a bit of a handicap. . .)
P.S.: I googled the belle/fidele quotation, but the results were inconclusive: depending on which unreliable source you believe, it's by Voltaire, or Thomas Mann, or a contemporary Moroccan writer named Tahar Ben Jelloun.