Excerpted (muchly!) from Robert Bly’s The Eight Stages of Translation, Rowan Tree Press, 1983. It’s now out of print and difficult to locate. In the book he goes through the translation process with a sonnet from Rilke’s “Sonnets to Orphesus.” I won’t type out his various versions, but “seeing” how he makes choices and changes them is fascinating. Below I’m outlining the steps, giving his descriptions, advice, and small bits of his commentary.

From the introduction:

In this essay I will not deal with the theory of translation but I will try to answer the question: What is it like to translate a poem? We’ll look mainly at the difficulties. The difficulties are all one difficulty, something immense, confusing, all of a piece. One translates a poem in fits and starts, getting a half line here, weeks later the other half, but one senses a process. I’m going to simplify the process into eight stages. I mean by that the stages one goes through from the first meeting with a poem to its recreation, when one says goodbye to it. As I’ve mentioned above, the stages will often collapse into each other, or a single line will suddenly go through all eight stages in a flash, while the other lines lie about looking even more resistant than before. What I will do then is to pretend that all goes in order; but this is an ancient ploy. When one makes a map, one pretends the earth can be laid out flat. But a map helps us to visualize the territory.

Stage 1:

During the first stage we set down a literal version, we don’t worry about nuances—English phrases that are flat, prosaic, dumpy are fine. We only want the thrust. […] As we read the literal, our first reaction is: What happened to the poem? Where did it go? So we read the original again and it’s still marvelous; so evidently something has been left out—probably the meaning. Before we go further with a translation then, we have to deal with the issue: What does the poem mean?

Stage 2:

         To find that is what I call the second stage. Some translators just print the literal version; they turn away from this stage. If we enter it, we will need everything we have learned in literature courses, or from our own writing, and all the (language of the original poem) we can scrape up in order to penetrate the “problems.” Often friends are helpful at this stage, to bring up quirky details that we haven’t noticed.

         [when wrestling with the deepest levels of the poem] we feel ourselves drawn here into areas we do not feel confident in, even to ideas we cannot accept. If we cannot accept them, we will resist them as a translator and do a poor job translating the poem. During this stage, then, we test how far we are willing to go. It’s clear that the poet is ahead of us, otherwise the poem would not be worth translating; it would have nothing “to say.” […] In a good poem, which violates certain secret assumptions, this second stage may take several hours. I spent a long time on this stage alone, some of it arguing over the text with other students and translators.

         […] The more one talks, the more clear Rilke’s beliefs become, and so his meaning. He is certain of it, and so the German has a lovely enthusiasm, expressed in lifting, joyful rhythms. If that store of feeling is beyond the translator, he or she should leave them poem be. At the end of this stage, the translator should ask himself whether the feelings as well as the concepts are within his world. If they are not, he should stop. […] In the second stage, we decide whether to turn back or go on.

Stage 3:

         If we decide to go on, we return to our literal version and see where it lost the meanings just found. We redo the literal and try to get it into English this time. We think of the genius of the English language, what its nature is. I’ll call that the third stage.

         During this stage, we use all we know about the structure of the English language. During the composition of the literal version we followed the word order of the original German, and by doing that found ourselves drawn in the whirlpool of the delayed verb. German gains energy at times by delaying the verb, and even the main noun, so it appears late in the sentence. English gains energy the opposite way, by embarking the main noun immediately and the verb soon after. Most sentences in English that begin with prepositions, with “into” or “upon the” or “for the” tend to be weak in practice; this is not a doctrine but something we observe in reading or writing English.

         […] Leaving the word order of the original poem behind is often painful; beginning translators especially resist it. They feel disloyal if they move the verb, but each language evolves in a different way and we cannot cancel a thousand years of language evolution by our will. Moreover, if we are disloyal to German, we are at the same moment loyal to English. The word order of Spanish is closer to that of English, and this stage is usually less painful when translating from Spanish. So then, after redoing later lines, thinking solely in this stage of the sentence and clause structures natural to English, we would arrive at a new draft. We ignore the sentence structure of the German original, and try to move all sentences bodily into the genius of English.