Translation, interpretation, and outright errors

I’m happy to agree with Russell Scott Valentino that literature is a “narrow and hyper-conventionalized form of communication,” though I’d insist that the actual use of non-literary language is also governed by conventions that we internalize and forget, that linguistic conventions arise in every repeated communicative situation. You can find my earlier thoughts here and here. Rightly or wrongly, I’m inclined to see human language as something special, and non-linguistic symbolic systems as qualitatively different, while RSV seems to lump non-literary language together with forms of communication like “how close you stand to someone” and put literature in a special category.

I don’t want to keep going around on that subject, but there’s a different tension in RSV’s latest post that I find interesting because I agree with two parts of his argument that seem contradictory:

If we were discussing misrepresentations or outright semantic errors, then I would be on board. But that has not been the issue in these posts. Interpretive differences, like those that often surface in comparisons among multiple published translations of the same text, are not misrepresentations, they are interpretive differences. […] The original book, original poem, original story, or play, or literary essay is not a unitary thing.

[…]

To be clear, I am excluding outright errors (where the translator has clearly not understood the source language) and deliberate manipulation (where the translator has willfully made the text into something else, e.g., an updated version of The Inferno replete with contemporary figures in their appropriate circles of hell). […] As a result, no one is being called in to tell us what is the “true” or “most accurate” or “authentic” or “ideal,” because this monolithic, unitary, most accurate version does not exist anywhere in reality. […] Thinking that it does exist would be like thinking that there is one correct interpretation of any artistic work and that you could write that version down somehow, capturing the entirety of it in other words than those in which it was first expressed.

Theory and practice seem to be pointing in different directions. What I’m calling “theory” is the idea that a text can be interpreted many ways, and it’s impossible to reduce even one sentence from a literary work to a single correct interpretation. The disclaimers about errors and misrepresentation seem to come from “practice.” We all know that sometimes a translator just plain gets something wrong — I know I have. Strong claims that it’s impossible to judge how faithful a translator is to the original work fall pretty fast to John Cowan’s reductio ad absurdum: if no one can say what any part of Anna Karenina means for sure, and one interpretation is as good as another, why not translate the first sentence as “it was the best of times, it was the worst of times”?

Both these things are obviously true, aren’t they? Literary works can be interpreted in many ways, and a translator can be guilty of “outright errors” or “misrepresentation.”

But they can’t both be true, at least not the way I understand these ideas in RSV’s telling.

Here’s how I understand RSV’s position: We have a field of possible interpretations, where if Constance Garnett translates something one way and Isabel Hapgood translates it another way, it’s an interpretive difference. A Russian-speaking reviewer shouldn’t come along and say that Garnett’s reading is better than Hapgood’s because Garnett makes an ambiguous word in chapter 41 harmonize with a use of the same word in chapter 62, and Hapgood doesn’t. Then we have a space outside that field, where the translations are different because Hapgood has committed an outright error (or deliberate misrepresentation) and Garnett hasn’t. Bilingual readers know which cases are errors, which are misrepresentations, and which are interpretive differences.

I don’t think evaluating translations can work this way, with a bright line between error and permitted interpretation. I believe in the errors, and I believe in polyvalence and ambiguity and their wonderfulness in literature, but I don’t believe in the bright line. We’re not always sure, and when we are, we’re one piece of information away from changing our minds. What seemed like a clear error (or non-error) on the part of a translator may no longer seem so after we recognize an allusion or learn an obsolete, technical meaning of a word we know in another sense.

I guess I see the fidelity of translations like this: there’s a continuous spectrum from persuasive interpretation, to possible interpretation, to an interpretation that’s a stretch, to a potential error, to a clear minor error, to a howler. A reader with specialized knowledge, such as knowledge of the source language or experience as a literary translator, is well placed to make an argument about where on this spectrum a concrete choice by a translator falls. It’s fine to argue that one error-free, defensible interpretation is less faithful to the original than another, and this doesn’t require us to believe that the original text has one fixed meaning: the original (and the translations that come from it) can have multiple but not infinite interpretations, which are plausible but not interchangeably, equally plausible.