The following was written by me in another context, as part of a conversation. I'm reproducing it here in case it is of general interest:Literal. Flowing. The whole question of what translation actually *is* and what counts as good or bad translation, is a huge one, of course. It's been theorized and discussed in a voluminous secondary literature. I've only dipped into this literature, but I found Lawrence Venuti's book, The Translator's Invisibility enormously helpful, and a real eye-opener. One of Venuti's big empirical discoveries in this book (from doing statistics on book reviews) is that "good" is a term that means very different things to different national audiences. There's no absolute goodness, where translations are concerned. In the Anglophone world, a translation that is generally judged "good" by reviewers is one that reads as if the original author wrote in English. In the German-speaking world, readers typically feel translations naturalized in this way to be deceptive or ethically false. Germans want their translations to feel foreign, to reproduce certain syntactic awkwardnesses of the source language. French and Italians are different again. Venuti's book is very enlightening, I thought.
My personal position, as a translator into English, is that the deeper my understanding of the Sanskrit sentence I'm working on is, the more natural and even idiomatic becomes my English translation. For me, translation is a sentence-by-sentence process of internalization. I first have to really, fully GET the sentence in Sanskrit, without translation. I often achieve this by repeating it again and again, silently, until I truly hear and understand it in Sanskrit, not in English. Then, I write out what it means in English. I will quite happily invert the syntax of a Sanskrit sentence, move the subject or verb from the end to the beginning, re-order it's clauses, paraphrase past gerunds with a temporal sequence (not "having eaten, he left" but "he ate and then left"), use different English words for one Sanskrit word in different contexts, or break a Sanskrit sentence into several separate English ones.
Some of my German-speaking colleagues find this hard to swallow. They sometimes feel I am betraying the Sanskrit, that I'm not accurate. In my view, my translations are MORE accurate, because they (try to) reproduce more accurately what happens in the mind of a fluent Sanskrit reader, say a pandit. I'm not attempting to teach English readers how Sanskrit syntax works. I'm not attempting to protect myself against critical colleagues by producing "defensible" translations that transparently reveal the underlying Sanskrit syntax, and thereby demonstrate that I know Sanskrit grammar. What I am trying to achieve is to give my readers an experience that is as close as possible to the experience that a Sanskrit reader of the past would have had. I presume that they were comfortable with the language, and that as they read, their attention was engaged with the actual meanings of sentence- and paragraph-length units.
However, I hasten to add, that my own translations are by a native English speaker and for an English audience. If I ever had to translate into German, I would have to rethink my style considerably, and adapt it to the expectations of a German readership, perhaps abandoning some of my cherished stylistic preferences.