The translation of poetry is generally held to be the most difficult, demanding, and possibly rewarding form of translation. It has been a subject of a great deal of discussion, particularly within the field of literary translation, where far more has been written about the translation of poetry than about either prose or drama.
Translation of poetry is probably the subject in translation studies that triggers the strongest polemics. Even those not specialized in translation often have an opinion on the subject; consequently it is much platitudinized.
One of the most boring and useless debates concerns translatability and untranslatability of poetry. It is not worth while spending time on that, since there is a commercial and private production of translation of poetry, and thus a readers public ready to read such translated texts and to recognize, in a way, more or less perceivable traces of identity of this or that author.
Someone translates poetry and someone reads translated poetry, and that is more than enough.
Even for poetry, the translation dilemma is either creating a text enabling a reader to access the original, or creating a beautiful poetic text inspired by the original. Therefore, it is better make some distinctions on the aim pursued by translating poetry.
1. direct access to the original: probably the most common form of translation of poetry is metatextual, and consists in a critical apparatus prepared for a poem - in the same language of the poem or in another language - allowing people not particularly proficient in that language to access an interpretation of the text through a clarification of the semantic values of the original.
2. interlinear translation with parallel text: this is another form of direct access to the original, but in this case the aid is textual and not metatextual. Even if it is not always possible to call a parallel text "text". When the parallel verse is the reproduction, word for word, of the original verse, its only aim is to indicate the meaning (the one, among the many possible meanings, chosen by the translator) attributed to the individual words in the original, and seldom the whole result can be called "text" in the proper sense of the word, i.e. a consistent and coherent set of words.
3. philological translation: a translation that does not consider the readability of the text that is produced, only its philological adherence to the prototext. Aim of such a translation is to give access to the original for readers unable to access it through one of the previous strategies. Philological translation can be in prose or verse. When in verse, the verse of the metatext generally matches the verse of the prototext, but of course there are no rhymes (if not by chance), or pursued alliterations , and rhythm and other non-denotative aspects of the text are not considered. One of the most famous advocates of such a strategy is Vladimir Nabokov:
There is a certain small Malayan bird of the thrush family which is said to sing only when tormented in an unspeakable way by a specially trained child at the annual Feast of Flowers. There is Casanova making love to a harlot while looking from the window at the nameless tortures inflicted on Damiens. These are the visions that sicken me when I read the "poetical" translations from martyred Russian poets by some of my famous contemporaries. A tortured author and a deceived reader, this is the inevitable outcome of arty paraphrase. The only object and justification of translation is the conveying of the most exact information possible and this can be only achieved by a literal translation, with notes. (1973: 81)
4. single-dominant translation: usually the result of a poor and superficial analysis of the prototext, or of insufficient poetic competence, or of a low-profile publishing policy . One aspect of the original is found, the one most visible to the inexpert reader, like rhyme for example. In translation, the rhyme pattern is reproduced. Due to the anisomorphism of natural codes, pursuing the rhyme means obligatorily discounting the sense. For the dominant's sake, all the rest is lost, relegating the role of subdominant to the sense, when a part of it can be preserved. This kind of translation, especially when the rhyme is preserved and the measure of the verse is even, is also called "singsong" because of the effect similar to counting-out rhymes.
5. translation with a hierarchy of dominant and subdominants: this is the method that, while seeking an equilibrium between the opposite extremes of translatability ad untranslatability, takes for granted the impossibility to translate everything. It is a strategy deriving from Torop's total translation view. You first make a translation-oriented analysis of the prototext to identify the dominant elements in the source culture. Then such dominants are projected onto the receiving culture, and one must foresee the understandable elements, those textually incomprehensible and the partially understandable ones. Based on the model reader, the publishing strategy, the type of publication and, often, the translator's taste, one decides which important elements of the prototext can become dominants of the metatext, and which elements can be rendered only metatextually (through a critical apparatus)
Then a critical apparatus is made in which the metatext reader is told all that and a metatextual rendering of the translation residue (e. g. explaining the meter of the prototext that is not possible to reproduce in the metatext, or what connotative meaning a given poetic form in the source culture has).
When drafting the translated text, absolute precedence is given to the main dominant; once rendered, the translator tries to make room for the other dominants too, according to the hierarchy set during analysis.
The most important aspect of such an approach is absolute transparency of the decisions made by the translator (often by the publisher too) as concerns translation strategy. A translation of poetry that doesn't make clear what its carefully analyzed blind spots are, runs the risk of presenting itself as a "complete", "absolute" translation or, as some insist in saying, "faithful" translation of the original, a situation in which the reader comes out of feeling cheated, teased and/or manipulated.
6. cultural transposition: it is the strategy of people thinking of those who believe themselves able to find the cultural homologue of the poetic forms from a culture to the other. Let us see how David Connolly expresses the notion:
the sonnet form does not signify for the contemporary North American reader what it did for Petrarch's contemporaries in fourteenth-century Italy. Using the same form for a translation in a different age and a different culture may therefore carry quite a different meaning and produce the opposite of a faithful rendering. One solution is to look for a cultural equivalent (such as the English iambic pentameter for French Alexandrines) or a temporal equivalent (modern free verse for classical verse forms of the past) (1998: 174).
Putting aside the presumption implicit in the choice of a supposed "equivalent", cultural or temporal that it might be, since it is evident that such a choice is highly disputable anyway, such a strategy has a very low consideration of its model reader. It implies a person who isn't' open-minded enough to understand that a given form can have had a different meaning in another time or in another culture. This is what I have already written about rendering the reader responsible, and on esteem for the reader. With this kind of strategy one decides to underestimate her, to withdraw any responsibility she may have and, to top it all, to propose her a text that is very different from the original but that is presented as a "faithful translation".
7. poetic translation - author's translation: the translation is given a poet in the receiving culture. The result is often poetry, sometimes wonderful, sometimes better than the original. It is the best choice if one wants to produce poetic texts inspired by the original in another language, and if the philological interest is the last of the subdominants.