Translation will often leave translator and reader alike dissatisfied. 'Translation of poetry into poetry is difficult, because the translator must give a brilliancy to his language without that warmth of original conception from vvhich such brilliancy would follow of its own accord,' as Coleridge put it in the preface to his version of Schiller's Wallenstein. Most translators of poetry, if they have any ambition to achieve such 'brilliancy' at all, will start jettisoning features of the original in an attempt, well-conceived or not, to give the translated poet an authentic-seeming voice in the target language. 'I have made little attempt to mimic the formal or linguistic characteristics of the originals,' notes David Hinton on his versions of Tu Fu, and we applaud, all things considered. Hinton has aimed 'to create reciprocal configurations', he says (a strange phrase); 'to re-create [ … ] the poems [Tu Fu] might have written had he been writing in today's English.' Self-evidently, this is where the problems of question-begging begin with a vengeance.

As readers, we want to know whether Hinton's strategy is vindicated by the final product. The reader's dissatisfaction with translations is different from the translator's, since it originates in the result, not the process. I have come to think that I shall never know that Pasternak was a great poet unless I learn Russian; no English versions that I have seen convince me, pleasant though they are, so I shall have to take Pasternak's greatness on trust. Similarly, an ...