Structuralist Poetics

Another book I picked up years ago, Jonathan Culler's Structuralist Poetics, published in 1975 is still apparently worth reading. It's a bit of a snapshot I suppose, but from what I've read so far it does a decent job of summarising some of the theories and criticising them quite assertively. So I'll write it up here as I go through it, which means I have a bit of catching up to do.

The first chapter, "The Linguistic Foundation", is about the development of structuralism in linguistics, referencing Saussure of course, and the others who developed the system of signs. Culler is quite clear that structuralism and semiotics are the same thing. I don't think I've seen that identification made so baldly. He makes more reference to Chomsky than is usual. I think the point is to stress that linguistics and poetics (in a wide sense) aren't coterminous: there is an overlap but each activity has its own area of speciality.

This leads to the second chapter, "The Development of a Method: Two Examples", which looks in turn at Barthes' Systeme de la Mode and Lévi-Strauss's Mythologiques as attempts to apply a linguistic model to the understand of fashion any mythology respectively.

In Systeme de la Mode Barthes analyses a year's fashion press, trying to create from the captions of photos the code of what different types of clothes signify. For Culler, the big weakness is that the analysis is synchronic - ie it is based on only one year's fashion, whereas fashion is inherently diachronic; you can't ignore the difference between this year and previous years, which largely defines fashion.

With Levi-Strauss there's a different problem. He sought to identify certain features in world mythology that crop up across different cultures, but which have the same meaning. But the analysis doesn't have anything of the exactness of language.

The discussion of 'sun' and 'moon' is a case in point. Lévi-Strauss sees this opposition as a powerful mythological operator with great semantic potential: 'so long as it remains an opposition, the contrast between the sun and the moon can signify almost anything'.

So it is necessary to know something more than the structure to understand the myths. The chapter concludes that both these attempts have failed, but asks if literature might be more amenable to linguistic-based analysis, and so we move on to Roman "Jakobson's Poetic Analyses".

When I wrote about Jakobson earlier, I had doubts. Culler takes these further (showing why he's a professor and I'm not). He closely examines Jakobson's analysis of one of Baudelaire's "Spleen" poems. Jakobson tried to demonstrate the structure of the poem by looking for particular linguistic features, and showing that they formed a symmetry around the central stanza. Culler argues that by choosing different linguistic features you can show entirely different structures. It seems to me a complete demolition, not only of this analysis of this poem, but of this method of analysis entirely. The choice of linguistic features is arbitrary, with Jakobson's choice having no greater inherent worth than anyone else's. This seems close to Derrida's attack on Levi-Strauss: because the structuralist analysis doesn't have - can't have - a "centre" where the structure and the object of analysis coincide, any assumed starting point is as good as any other.

That might be the end of the book. Of course it isn't, and Culler suggests that Jakobson was looking for the wrong thing. Linguistic analysis precisely does not tell us what sentences mean - we already know that.

If one assumes that linguistics provides a method for the discovery of poetic patterns, then one is likely to blind oneself to the ways in which grammatical patterns actually operate in poetic texts, for the simple reason that poems contain, by virtue of the fact that they are read as poems, structures other than the grammatical, and the resulting interplay may give the grammatical structures a function which is not at all what the linguist expected. (p 73)

More to come.

Susan Sontag

First, a thought about the crazy economics of book publishing. Where the Stress Falls by Susan Sontag is a collection of essays and speeches from the last 20 or so years of her life. As I've noted earlier, it's a beautiful looking book, and the production inside is just as clean and stylish as the cover. But at about 350 pages, and with no new content, the cover price of £12 is ridiculous. Waterstone's has it in stock at full price, and not included in any 3 for 2 offer. Amazon has it for £8.60, which is more like it, but I bought it from a dealer in Amazon's marketplace for just £3.84 (plus p&p). To make any profit, the dealer must have got it for around £3.50. This is surely a mad, unsustainable business model.

Anyway, I bought the book largely because there's an essay on Roland Barthes. It praises him very highly, but largely despite his theoretical views. Earlier in this blog I looked at S/Z, where you can see Barthes succumbing to a classic narration, despite its lisibility. Sontag more or less argues that he was like that throughout his career; he was an old-fashioned practical criticismist in modernist clothing. I think there may be something in this. As with Stanley Fish, Barthes's analysis depends on many of the skills that people like Richards and Empson valued and developed.

A later piece, "On Being Translated", shows that Sontag doesn't hold with modernist denials of the primacy of the text. In a parenthesis she says:

You will have already noted that I am assuming that there is such a thing as an "original" text. Perhaps only now, when ideas utterly devoid of common sense or respect for the practice of writing have great currency in the academy, would this seem to need saying.

This is from a speech given at a conference on translation. It refers to Sontag's time in Sarajevo, working on a production of Waiting for Godot during the siege. Production was threatened because some people wanted a new translation of the play into Bosnian, to replace the existing Serbo-Croat one. But Bosnian is to all intents and purposes exactly the same language as Serbo-Croat. The call for a new translation was political.

The speech also covers some general points about translations. Unsurprisingly, given the above quotation, Sontag says the translation must serve the original text, but accepts the spread of means in which this can be attempted.

Phèdre in Yorkshire

Today's audience, it seems, consumes Racine in a purely anthological way: in Phèdre, it is Phèdre the character they come to see; more than that they come to see the actress herself, to see how she will play the role. [...] The text itself is seen as an assortment of materials, from which to choose one's pleasure. Memorable lines and famous tirades stand out from a background of boredom and obscurity. It is for these lines, these tirades, this actress that one goes to the theatre. One puts up with the rest in the name of culture, in the name of the past, in the name of a poetic flavour patiently awaited because centuries of the myth of Racine have localised it. Performances of Racine are a mixture of boredom and celebration, this is to say essentially a fragmented spectacle.

That is the start of Roland Barthes' 1958 essay "Dire Racine", collected in the book Sur Racine. It's my translation. I could translate it further. The first two words really ought now to be “the audience of 1958”. Or, to make the underlying point clearer, I could recast it as Shakespeare, and substitute Hamlet or King Lear for Phèdre. I think that could work pretty well (once I had also swapped in soliloquies for tirades). The point is that familiarity, particularly a culturally shared and shaped familiarity, risks changing the way a text is received. (I'd also change the title from Reading Racine to Speaking Shakespeare. Equally stupid sounding, after all.)

In the rest of the essay Barthes looks at what actors can do to speak the text in those circumstances. Surprisingly, in an essay in French about plays written in French and performed in French for a French-speaking audience, it sparks some interesting ideas about translation, which I'll be considering in the light of Ted Hughes' translation of Phèdre which is currently on stage at the National Theatre.

Barthes argues that the actor in Racine typically interprets the text by indicating certain key words and phrases. He uses two culinary metaphors: first the actor “chews” (mâche) the the text; he also says the actor acts like a parent, cutting up the food for a child. The outcome of this process is to enable the anthological approach, which Barthes identifies as a wholly bourgeois approach to art. He compares this to the use of rubato in music, which is the imposition of the artist's view of the piece. J S Bach springs to my mind here: the best performances, for me, are those in which the player most scrupulously follows the written notation. If Bach wanted rubato he'd have written it. Bach's music, like Racine's text, has effects written in it, that are produced simply by playing it. On the other hand, it's foolish to assume that there can be any pure, uninterpreted reading of a text.

Barthes writes that the actor works on a false belief that the words translate the thought (“les mots traduisent la pensée”). I've quoted the original because “mots” much more closely refers to words as individual items, rather than as a stream. He talks about actors speaking the text as a didactic exercise, not aesthetic. Their purpose is to make key words stand out, and those key words are the ones that express a fairly literal meaning.

At the same time, actors seem to try to disguise the alexandrine structure; instead they use pacing and rhythm that try to approximate to normal spoken language. He says the alexandrine contains its own musicality; there is no need for the actor to sing, to add a “secret music”.

English audiences for Racine don't have the background anthology of Racine in their head. They're not looking out for the memorable lines. So actually that familiarity isn't a problem. English audiences come to a performance almost naively. The actors don't need to cut up the meat for them. Maybe the text can speak directly through its own resources.

In the programme for the National's production, Blake Morrison writes about the translation. He's impressed by the fact that it sounds like Yorkshire, but then he would be. He quotes three lines of Racine in the “standard blank verse translation” and Hughes's version, which I may as well requote:

Athens revealed to me my haughty foe.
As I beheld, I reddened, I turned pale
A tempest raged in my distracted mind.

Hughes offers:

Suddenly he was there
Standing in front of me,
He had simply appeared -
Staring at me,
The man created
To destroy me.
Before I could grasp what I'd seen
I felt my face flame crimson – then go numb.
My whole body scorched – then icy sweat.

The original is:

Athènes me montra mon superbe ennemi.
Je le vis, je rougis, je pâlis à sa vue ;
Un trouble s'éleva dans mon âme éperdue ;

Obviously, the original is best. The simplicity of that second line is unparalleled in either of the Englishings. The staccato impact works so well among the longer more formal words, and the stress pattern almost demands a syncopation. But what's so notable about Hughes's version is that it adds so much meaning, letting the words translate the effect of the music of the lines. The short anglo-saxon based words – grasp, scorched, sweat – don't shock because they live in a context of short punchy words. And there's no music left.

Morrison clearly sees Hughes's version as an improvement on the standard blank verse translation. But I think that's precisely because it adds something – something Yorkshire – to the meaning. As foreseen, Hughes's lexicon is much bigger – not just here but throughout. A lot of the power of Racine comes from that very sparseness of words – that sense that all the characters are debating what these concepts – gloire, coeur, honneur – mean. It's claustrophobic, matching the closedness of the setting.

So Hughes's version (as the programme scrupulously calls it) adds explicit meaning to the text, making it much more lisible, and completely changing the experience. Later (later in my life, I mean, not later here – I've nearly finished) I'll consider if a translation ought to preserve the experience.

The birth of the author

More thoughts about Roland Barthes, who's probably best known for the concept of the 'death of the author'. It's linked to the lisible/scriptible comparison seen in S/Z, in that the author of a modern work isn't the same controlling presence that Balzac, or closer to home Dickens, was. It's more than that, though, it suggests that the text is produced by language through the author, and then assembled by the reader.

Foucault asked "What is an author?", and described an author function, which means that Shakespeare the author is not the same as Shakespeare the man. So that it's logically possible to ask "Was Bacon Shakespeare?". Shakespeare the author is largely defined by the work, so that William Shakespeare's will, for example, or T S Eliot's note to the milkman, are not the work of Shakespeare or Eliot the author. It's kind of obvious once you think about it. But I think there's a historical shift, and the concept of an author has grown. Largely, perhaps, because of increasing protection of intellectual rights. When a novel was published as being by "A Lady", the readers had no idea of an author, but probably had some idea of what to expect, in the same way that one knows what to expect from a "Mills and Boon". This affects the way writers write. They are conscious that a poem, for example, is part of an oeuvre. It means that individual poems can rely on the reader's knowledge of the work, and can therefore be harder to understand.

James Joyce had to turn up in this blog sooner or later, and his stories and novels build on those that have gone before. Ulysses is never going to be easy to understand, but it's easier if you've read Portrait.

The Cantos is an extreme example. You need to know all Ez's work to understand any of it.

Anyway, this is a preliminary note. It's something I'll be coming back to.

Ez said (Canto XXXVIII)

Canto XXXVIII begins with a quotation from Dante about counterfeiting money. For once the quotation is referenced, suggesting it's important. The rest of the canto is about economics and the arms trade. The thing that strikes me about many of these passages is just how bound they are to their time. Which is weird, because Ez's effort is clearly to yoke different eras and civilisations together in his analysis. Maybe it's just like the way past visions of the future are so often so quickly dated. Focussing on an imagined world, the writer doesn't notice the assumptions implicit in the text, which stand out glaringly to later readers.

Writers/readers. Guess, from the title of this post, what I've been reading. S/Z by Roland Barthes, that's what. I don't think Barthes had much to say about Ez specifically, but in S/Z he looks at the difference between conventional 'realist' writing and modernist writing; he was more concerned with the nouveau roman but I'm sure it can apply at least as well to modernist poetry.

He describes two kinds of text: lisible and scriptible in French. Literally, they mean readable and writeable, although they are unhelpfully translated, normally, as readerly and writerly. I'm reading the book en français, naturellement, so will use the French terms.

A lisible text is one in which the meaning and connotations are closely controlled by the text. Although there may be a certain pluriel or plurality of meaning, the apparent transparency of the text is an illusion. Barthes demonstrates this by an incredibly detailed analysis of 'Sarrasine', a story by Balzac, looking at how it employs various codes to build the meaning, and also showing how it is deeply intertextual - defining beauty, for example, in terms of earlier texts (including artworks). It's a tour de force, even if now it's suffering a bit from the blindness to unconscious assumptions that infects visions of the future. He says that some of the meaning depends upon some shared understandings. Now, about 50 years later, it's becoming clear that at least some of those understandings are time-bound.

By contrast, a writeable text is one that doesn't really exist until a reader constructs it out of the loosely linked raw materials. (This is why I think 'writerly' is a bad translation - it appears to put the writer in the foreground, which is precisely wrong.)

It's a simplistic contrast, as Barthes accepts, and his use of a particularly good text shows it up. A lot of the analysis shows that the reader has a large input into the construction of the meaning, or at least in choosing from the plural understandings that it allows. A work can't last if it allows no scope for the reader to be involved in its construction. I'd suggest there's also a case to be made that a purely writeable text can't last, either.

Anyway, the book has given me some things to think about. Here's one extract (my poor translation) following an unattributed fragment of speech:

It's impossible, here, to attribute to the comment an origin, a point of view. Now this impossibility is one of the means by which the plurality of text grows. The more the origin of a comment is unclear, the more the text is plural. In a modern text, the voices are denied any source: the speech (discours) or better still the language (langage) speaks, and that's all. In a classic text, on the other hand, most statements are sourced; one can identify their father and their owner. Sometimes it's a consciousness (a character or the author), sometimes it's a culture.

Clear relevance here to the unattributed quotations and comments throughout the Cantos. But again, as I've said above, I think there are cultural codes or assumptions embodied in the text, which are now dated and have to be re-learnt.

But is too much plurality a bad thing, Roland, hein?