My theory of (literary) language

Excuse the break from translation

I realise that I ought to write this down somewhere, and I have a blog!

My theory of language starts with JL Austin, who wrote How to Do Things With Words. The point of the book is that when you say a sentence (/utterance), you are performing an act in the world. That might be issuing a command, making a promise, requesting information, or providing information. The sentence or utterance does something.

The value of Austin’s insight is that it gives us a way to solve the problem of infinite polysemy. Words can mean anything, and in fact words do mean anything. The word horse can refer to the species equus caballus, or a particular animal of that species, or a statue of such an animal, or any one of a number of conventional metaphors relating to horses (dark horse, workhorse, etc.), or a new rhetorical twist that I expect you to understand in context… If you get down to it, each word in a sentence/utterance has an almost infinite meaning space, and therefore the meaning of each sentence becomes hopelessly underspecified. If any word in the sentence can mean anything, then how can we possibly know what the sentence means?

Austin’s understanding that sentences are not just strings of underdefined meanings, but attempts to do things, explains how real people solve this problem in real time. We judge from context what our interlocutor is doing, then work backwards to how the string of words they said might achieve this action. Sounds complicated, but people are really, really good at this, and it works out most of the time.

Vitally, this process has a well-defined success criterion: does the utterance succeed in doing the thing in the world that we both intuit the speaker wanted to do? So, when a barkeeper asks you what you want to drink, it doesn’t matter whether she says, “What’ll it be?” or “What can I get you?” or “Something to drink?” or “Any time you’re ready.” At the point when the customer understands the utterance well enough, the customer signals understanding (by ordering, or asking for more time, or asking for a recommendation, etc.) and the action of the utterance is successfully completed. Whatever understanding of the words the customer had is judged by both speaker (barkeeper) and listener (customer) to be good enough, and they move on.

(I call this kind of closure through conversational turntaking Trognon units, because I got the idea from a French researcher named Trognon. A Trognon unit consists of three turns: utterance, response, and confirmation. Once the confirmation has been given by the original speaker, the utterance is truly finished: the listener has responded, and the speaker has confirmed that the listener’s response was felicitous enough that the intended action has been completed. For example: “Anything to drink, ma’am?” “I’ll have a mineral water with ice.” “Would you like a snack with that?” The third utterance does not include any follow up questions or clarifications about the drink; this null response indicates that the customer’s reply was fully satisfactory, and the action of the initial question has been successfully completed.)

This kind of analysis also applies to written language. Most written language is also embedded in a turn-taking conversational structure that allows for interpretive closure. Text conversations are an obvious example. Letters/emails involve much longer conversational turns, but still fit the model. Even when a written turn is massive, like a government report, it still expects a response. The response of the reader and subsequent silence or clarifications by the authors of the report complete the Trognon unit and define the meaning of the words in the report as those meanings that the readers drew from their report in understanding the actions of the authors.

However, there is a form of linguistic communication that does not occur in a turn-taking process that allows for semantic closure, and that is artistic language: novels, poems, and essays. Art is one-way communication; no reply is expected from the reader; no reply would be felicitous. The action that these communications do in the world is just to be; the only response of the reader is to feel. And there is no single target reader: art is published for all to enjoy, meaning that there is no interlocutor to hear, understand, and stop the proliferation of meaning.

(There are exceptions to this rule, especially in the epistolary art of poem-letters in Tang poetry, which throw up very interesting confusions.)

This allows us to explain a couple of things. For example: Why is art so open to interpretation when other uses of language are not? It’s because the closure mechanism that usually nails down the meaning of ordinary utterances is not available to nail down the meaning of the words in art. Why does art need interpretation in the first place? Same: there is no target listener/reader who uniquely determines its meaning.

There also some implications for translation.

When translating non-artistic documents, you look for the meaning of the source text in the speech acts that it executes. Then you reproduce/mimic/describe those speech acts in your target translation (sometimes you can’t reproduce the acts, e.g. if they use source culture legal concepts, so you might mimic or describe instead).

When translating artistic documents, my goal is to open up and close off the same shape of potential meaning space as the source document. That is to say, I have to try to react to the source in as many different kinds of ways as possible; then to create an English target poem that suggests to the reader and allows the reader to react in that same set of ways. If a source poem has a dreamy quality, I want my target text to have a dreamy quality. If a source poem is about war, but seems to draw in themes of love, then I want my target poem to do the same thing.

It does mean that a translation of artistic work can never be right. The translation of a contract or a law or an email can be correct in a final sense: if it succeeds in doing the same speech act as the source text does. In my theory, artistic work doesn’t do a speech act, strictly speaking, so this success condition does not apply. All we can do with an artistic translation is follow the contours of the source text as closely as possible, always remembering the possibility that we’ve missed something or lost something.

There’s more to be said: I haven’t even touched on cross-cultural reception yet. And all this is far too abstract. I need to make it concrete with some nice examples. But I thought I’d get it down on paper just so that I have something to reference if/as necessary.