I’ve been reading Rita Felski’s draft essay, Comparison, Translation, and Actor-Network Theory, which is about the implications of Bruno Latour’s Actor-Network Theory (ANT) for the study of comparative literature. She notes, quite accurately, that he has little to say about language. Noting that translation is a key notion for Latour – where it has to do with relations among all actors ¬– she goes on to point out that what ANT suggests about translation of texts from one language to another is quite different from translation’s standing in current literary theory (p. 5):
Translation thus becomes a key metaphor for thinking about relationality. It is not something imposed on the world—an act of aggressive encroachment on pristine otherness – but something that defines a world that is always already composed of acts of connection, negotiation, and transformation. Translations are the means by which paths are connected, actions are co-ordinated, and meanings transmitted.
I agree with her on both points and want to suggest what you might call a neo- or quasi-Latourian analysis of language and then situate linguistic translation within that.
Intermediaries and Mediators
In Reassembling the Social Latour distinguishes between intermediaries and mediators (p. 39):
An intermediary, in my vocabulary, is what transports meaning or force without transformation: defining its inputs is enough to define its outputs.[...] Mediators, on the other hand . . . transform, translate, distort, and modify meaning of the elements they are supposed to carry.
Taken as a whole, language is a system of mediators. In general individual statements, individual texts do not transport meaning of force without transformation. In any given case the transformation may be minor or extreme, but in any case we have no direct way of assessing this as we have no direct way of comparing what goes on in the speaker’s head (or the writer’s) with what goes in the in listener’s (or the reader’s).
But language, whether written or spoken, is a double system in which a string of signifiers carries or encodes a string of signifieds, to use Saussure’s terms. The signifiers are physical events in the public domain, vibrations in air, marks on a page, while the signifieds exist within the privacy of individual minds and brains.
The thing about signifiers, however, is that they are discrete, digital in today’s computer parlance. They are transmitted with high, if not always perfect, fidelity from one person to another. In signifier channel “defining its inputs is enough to define its outputs” – which is what Shanon’s information theory is about. The signifier system is thus, to a reasonable approximation, system of intermediaries. It is the system of signifieds that is less reliable (or more flexible). It is the system of signifieds that makes language a mediator rather than an intermediary.
In ordinary conversation back-and-forth interaction is able to resolve ambiguities and misunderstandings in a wide range of situations, though not always. In the case of written communication, such back-and-forth is somewhat more difficult or, in many cases, simply impossible. One cannot interrogate a dead writer on the meaning of an obscure passage.
The situation that interests Felski is that of translation of a text from one language to another. That is quite different from monolingual communication because the high-fidelity channel of signifiers now operates in a very different way. The reader of a translated text (in L2, language 2) isn’t reading the original string of signifieds (from L1). She’s reading a string of signifieds produced in a language that is not the original language.
The high-fidelity channel is “broken”, displaced, or dislocated and can no longer be considered an intermediary in Latour’s sense. It has become a mediator. More precisely, while the L2 channel that conveys the translated text to the reader is high-fidelity, an intermediary, it does not contain the origianal L1 string of signifiers.
In the case of face-to-face interaction where one or more translators are helping the two (or more) principals, we still have the possibility of resolving problems in real time. This, of course, disappears with written texts. In either case, what is preserved in a good translation? Meaning? And how do we determine that? That is, where do we stand to make that determination?
After all, as we’ve already noted, there is no way to make a direct comparison between the original meaning and the meaning understood. We can assert that, in some idea sense, both of those things exist. But we cannot measure or assess them. Perhaps we can drop them entirely?
Note that this problem does not disappear even in the case where the original writer is also the translator. For the problem arises within the language system. The set of relations between signifiers and signifieds is not the same in two different languages. That is the problem that must be dealt with and it doesn’t disappear even if it is the original writer who deals with both languages.
Some Exercises for the Reader
Exercise 1. I’ve spent a lot of time criticizing Dan Dennett’s account of memetics and, in particular, his account of language. On the one hand Dennett quite properly stresses the digital nature of phonology (the stream of signifiers). But he seems to think of the relationship between signifiers and signifieds as being a transparent one. And so he places memes, which in his account must bear the sematic information, inside people’s heads. What kind of account can he offer for translation between languages given that, in that situation, the high-fidelity channel is broken?
Exercise 2: Does this analysis of language translation also apply to music? I’m thinking of cases like the difference between Little Richard performing Tutti Frutti and Pat Boone performing it.