As part of the new issue of Words Without Borders, translator Michael Emmerich has a very interesting essay on translation.
In order for "translation" to have any meaning at all, it must be translatable into other languages; but the moment it is translated, it is swept up in a system of differentiations different from the one in which it is enmeshed in English—indeed, it doesn't even have to be translated, because the word itself implies its own connectedness to these other systems of differentiation. Translation must be viewed as a node within which all the ideas of translation in all the languages there ever have been or could ever be might potentially congregate, intersect, mingle. Or we could say that the word "translation" is haunted by all the concepts it might translate, the words with which it may be translated. A word like "dog" can be understood, if only provisionally, in terms of its difference from an (indefinite) string of other terms in English; "translation" is made doubly provisional by its inevitable connection to other, non-English ideas of translation that could, at any moment, be brought to bear on the English word, just as the English word can be brought to bear, through a subtle process of productive mistranslation, on the Japanese word hon'yaku. "Translation" is, that is to say, always waiting to be redefined, not through its difference but through its similarity to other terms in other languages.
If this is not as obvious as it probably should be, it is because too frequently we consider translation from a perspective that has nothing to do with translation. We focus almost exclusively on translations. There are originals and translations, source texts and languages and target texts and languages, domestic and foreign, those who commission translations and those who consume them—everything but translators engaged in the act of translation. There are several reasons for this. The most important, perhaps, is that it is difficult to get a handle on what exactly a translator is doing when she translates. Consider this description of the process by Donald Philippi:<
Whatever happens after a translator sits down at the computer, it isn't anything material. What realm do we enter when we boot up our computer, attune our mental faculties to that odd wavelength of ours, and ascend into the ethereal realm of the translator's daily praxis? The translator's consciousness is not focused on any object, but is rather liberated from the world of material objects. The translator's realm is on a highly abstract plane, rather like that of a mathematician, grammarian or logician. The material objects are distanced. The domain of consciousness in which the translator operates is detached from the whole natural world. Abstracted from reality, the translator operates outside the spatio-temporal system in the world of pure consciousness. As Edmund Husserl would say: "Between the meanings of consciousness and reality yawns a veritable abyss" (Ideas, p. 138).
Later on, Emmerich puts it well when discussing the translator's role:
Rather than imagine the translator as someone who stands between languages, cultures, and nations, we would do better to cultivate an image of him as a ghost who haunts languages, cultures, and nations, existing in two worlds at once but belonging fully to neither.
I also try to balance myself somewhere between Japanese and Englishor rather to keep myself as much as possible in both languages at once, like a ghost that isn’t entirely here but isn’t entirely gone, either. To produce a work that feels right to me both as English prose and as a translation from Japanese. This isn’t a very good answer to your question, but all I can say is that each author and each work and each sentence in each work requires that one perform a different sort of balancing act. Usually several different balancing acts at once, in fact.