Translating Dostoevsky: What Actually Works
Crime and Punishment is one of the most translated novels in Russian literature, which means there are dozens of English versions sitting on bookshelves right now. Most of them are fine for casual reading. None of them capture everything the original does. I learned this the hard way after spending three weeks on a translation exercise that fell apart because I wasn't paying attention to register shifts. The core problem with translating this novel isn't vocabulary. It's voice. Raskolnikov speaks differently depending on who he's talking to, what state he's in, and whether he's actually making sense or just performing coherence for someone else. Dostoevsky writes these shifts deliberately, and a translator who flattens them has already lost the book. There's no single best translation. There are only translations that prioritize different things. Pevear and Volokhonsky are known for being ruthlessly literal. They preserve sentence structure, repetition, and the jagged rhythm of Russian syntax even when it feels awkward in English. Constance Garnett, the old standard, smooths everything out into something readable and mid-nineteenth-century British sounding. Garnett's version is easier to read but completely changes the tone. She renders Katerina Ivanovna's dramatic speeches as if they belong in a Victorian melodrama rather than a feverish Russian apartment.
I once tried working from a Pevear-Volokhonsky passage and a Garnett passage side by side and realized Garnett had essentially rewritten an entire internal monologue. Not paraphrased. Rewritten. She changed a character's thoughts into narration. That's not a translation choice. That's a creative intervention that most translators would consider crossing a line.
What Most People Get Wrong About This Process
Beginners tend to focus on accuracy at the word level. They look up every unfamiliar term, consult a dictionary, and try to map Russian words to English equivalents one by one. This approach produces text that is technically correct and completely lifeless. Dostoevsky's sentences do not work the same way in English as they do in Russian. Period. Trying to force structural fidelity is how you end up with prose that reads like a legal document written by someone who studied Russian in college. What actually matters is register matching and emotional continuity. You need to understand what the narrator is doing in a given scene before you translate a single sentence. Is the narration unreliable? Is it close third person filtered through Raskolnikov's deteriorating mental state? Or is it an omniscient voice stepping back? The answer changes how you handle every line that follows. Here's a specific example. In the opening chapters, there's a passage where Raskolnikov is walking through St. Petersburg and the city itself feels hostile. The Russian uses a particular kind of compressed syntax, short clauses stacked without much connector. A naive translator might add conjunctions to make it flow. That's wrong. The compression is the point. The reader should feel cramped. The workaround I use is to read the passage aloud in Russian first, notice where my breath naturally catches, and then reconstruct those same chokepoints in English. It takes longer initially but prevents you from accidentally smoothing out the texture of the original.
Get the Full Details
Technical Considerations That Matter More Than You Think
Russian has grammatical gender, aspectual verb pairs, and a case system that encodes information English doesn't have. When Raskolnikov describes himself, the Russian can shift between reflexive and non-reflexive verb forms to signal his self-alienation. English doesn't have this tool. You have to recreate the effect through word choice, punctuation, or sentence structure instead. There is no perfect solution. There are only compromises you accept consciously rather than accidentally. Another issue is the philosophical passages. Dostoevsky inserts essays within the narrative. These are sometimes written in a different register from the rest of the novel. If you translate the essay passages in the same voice as the scenes, you lose the deliberate tonal break. I've seen versions where Raskolnikov's theoretical justification for murder reads like casual conversation because the translator never adjusted the register. That's a serious problem because the whole novel depends on the gap between his intellectual claims and his emotional collapse. If you're doing this work seriously, I'd recommend having at least two other translations open while you work. Not to copy from them, but to see where different translators made different choices and why. It trains your eye to recognize the decisions that matter rather than the ones that don't. Most of the differences between translations are trivial. The ones that aren't are usually invisible to anyone who hasn't read both versions carefully.
Where This Approach Fails Completely
Machine translation will not work for this text. Not even close. Neural MT systems handle straightforward prose reasonably well. Crime and Punishment is not straightforward prose. The stream-of-consciousness sections, the fever dreams, the deliberate syntactic fragmentation, the ideological debates embedded in dialogue. These are the exact types of text where MT breaks down most visibly. I ran a test passage through a major translation engine once just to see. The output was grammatically coherent but semantically hollow. It preserved the facts of the scene but lost every emotional cue. The result was the literary equivalent of a spreadsheet. Even human translators using reference materials can get this wrong if they treat it like a technical translation task. There's no terminology glossary that will save you from a misread character motivation. The book requires literary sensitivity, not computational precision. If you're looking for a version to read, pick based on what you want. Garnett for historical context and ease. Pevear and Volokhonsky for structural fidelity and roughness. Oliver Ready's newer Penguin edition is worth considering too. Each makes different sacrifices. The text itself never changes, but your experience of it absolutely does.
