Literary Theory & Criticism

Eight lines from 1794

William Blake etched these lines onto a copper plate in 1794 and printed them under a drawing of a rose eaten from the inside, a small worm curled at its heart. The poem is called The Sick Rose, and it stands near the front of Songs of Experience.

O Rose thou art sick.
The invisible worm,
That flies in the night
In the howling storm:

Has found out thy bed
Of crimson joy:
And his dark secret love
Does thy life destroy.

Thirty-four words. Here are three readings of them, each produced by a competent critic, each supported from the poem, and each incompatible with the other two about what the poem is even made of.

Reading one: a made object with a hinge in the middle

The first thing to notice is grammatical. The whole poem is a single sentence, and its subject and its verb are on opposite sides of the white space between the stanzas. You meet the worm in line two and then wait four lines, through the flying and the night and the howling storm, before you are told what it did. The delay is the poem. By the time you reach has found out, the finding has already happened and cannot be prevented.

Then look at what the words do to each other. Bed of crimson joy holds two registers at once: a flower bed and a marriage bed, red as petals and red as blood. Dark secret love puts the word love in the position of an agent that destroys, so the line does not oppose love to death, it makes love the instrument. Invisible worm is a small contradiction that the poem does not resolve, and the plate makes it stranger, because Blake drew the worm and you can see it.

Notice also what is never stated. The poem does not say what the sickness is, who the rose is, or what happened in the bed. Its entire content is the gap between an address (O Rose) and a diagnosis that never arrives. On this reading, that withholding is not a failure of information; it is the structure that makes the eight lines hold together, and paraphrasing them into a statement about disease or seduction would leave you holding the paraphrase and not the poem.

Reading two: what the poem is refusing to say

A second critic starts from the same words and treats them as symptoms. On this account, the interest of the poem is that it is not about a flower and never was, and yet it declines to say so.

The elements arrive already coded: a worm that flies at night and enters a bed, a redness the poem calls joy, a love that is dark and secret and lethal. Read as a dream is read, the poem performs two operations at once. It condenses, packing sexual act, infection and moral fall into one image of a worm in a rose. And it displaces, moving the whole scene into a garden where none of it has to be named. The critic's evidence is not the historical record but the pressure of the omission: a poem this short does not spend its second, third and fourth lines on weather unless the weather is doing the work of not saying something.

The poem's grammar helps this reading too. Every actor in it is displaced. The rose is addressed but never speaks. The worm is invisible. The love is secret. The only thing given plainly is the outcome, thy life destroy, and it is given in the last three words, the way a patient names the symptom before the cause. A reader trained in this method will say the poem knows more than it states, and that the reader's job is to reconstruct what the text is organised to avoid.

Reading three: the streets outside the plate

A third critic asks what was happening in London in 1794 and reads the poem as a document of it.

In the same collection, four plates away, Blake printed London, which ends with the youthful harlot's curse that, in Blake's spelling, blights with plagues the marriage hearse. Venereal disease in the London of the 1790s moved exactly as that line describes: from a paid encounter, to a husband, to a wife who had done nothing, to an infant born blind. There was no cure. Mercury was the treatment, and the treatment was itself a poison.

Read against that, the sick rose is not a symbol requiring decoding, it is a description. The invisible worm is an infection nobody could see; the night and the storm are when and where the encounter takes place; the bed of crimson joy is a specific bed; the dark secret love is secret because it is bought, and it destroys a life that was not party to the purchase. Songs of Experience is a book about institutions, the church, the charity school, the chartered street, the marriage contract, that turn what was free into property, and disease is what property does to bodies. This critic will add that the poem exists as an object because Blake and his wife Catherine printed and coloured every copy by hand, which is a fact about labour, not only about art.

What is actually in dispute

The three readings do not disagree about the words. They disagree about what a poem is, and it helps to see the disagreement laid out.

That last row is the one to keep. Every method in this course buys its power by refusing to look at something, and the refusal is not a flaw to be repaired by combining methods into a mush. It is the price of getting a clear answer to a definite question.

Key idea: A theory of literature is not a set of opinions about books. It is a set of commitments about where meaning is located, what counts as evidence for a claim about a text, and what a reading is supposed to produce. Change any one of those and you change what the poem is.

The reading that thinks it has no theory

Most readers, asked what theory they use, say none, and mean it as a defence of the poem against professional interference. Test the claim. Someone who says the poem is just about a rose dying, and that all this decoding is imposed, has already committed to three things: that the unit of meaning is the individual poem rather than the book or the century; that the right evidence is a plain reader's impression; and that the meaning of a poem is something that can be stated in other words, since they have just stated it.

Those are exactly the commitments the next two lessons will examine, and none of them is obvious. The second was denied by the New Critics as the affective fallacy. The third they called the heresy of paraphrase and treated as the central error of the age. Terry Eagleton put the general case at the start of his Literary Theory in 1983: "Hostility to theory usually means an opposition to other people's theories and an oblivion of one's own."

Jonathan Culler makes the sharper version of the point. Theory is not a body of knowledge you acquire before reading; it is what you are doing when you stop asking what a text means and start asking what makes meaning possible at all. The moment you say the poem is about sex because the worm enters a bed, you have assumed that images can stand for things they do not name, which is a theory of the sign, and you inherited it from somewhere.

Are all three readings equally good?

No, and the way they fail is instructive.

Within any method there are wrong answers. A reader who says the poem celebrates the worm's arrival has not produced a bold interpretation; the word destroy is in the last line. A reader who says the invisible worm is the East India Company has to show what in the poem, or in 1794, makes that specific rather than available. Each method comes with its own tests, and a claim that cannot fail one of them is not doing any work.

Between methods the comparison is harder, because they are answering different questions. What you can ask is what each one leaves unexplained. Reading one cannot say why this poem sits in a book called Experience, opposite a book called Innocence. Reading two struggles to be surprised: it finds concealment in every text, so the finding carries less information than it appears to. Reading three cannot explain why Blake wrote thirty-four ambiguous words rather than a clear pamphlet, which is to say it has trouble with the fact that this is a poem at all.

The point: The useful question about a method is not whether it is true but what it can find, what it cannot see, and whether the thing it finds could have turned out otherwise.

Common misconceptions

• Theory is what you do after you have understood the poem. Theory determines what counts as having understood it. Reading one and reading three do not disagree about a poem they both understand; they disagree about what understanding this poem would consist of.
• Every interpretation is as good as every other. No school in this course holds that. Even the most reader-centred position, which you will meet in the lesson on Stanley Fish, argues that readings are constrained, by the conventions of a community rather than by the text, but constrained.
• Theory is a French invention of the 1960s. The close reading practised in reading one was formalised in Moscow and Petrograd from 1915 and in England and the American South in the 1920s and 1930s. Aristotle's Poetics is a theory of literature, and so is Sidney's Defence of 1595.
• Using a method means applying a template. The templates in this course are real and you will run them, but a reading that could have been produced without opening the book is not a reading. The test is whether the text can still surprise the method.

Where this leaves us

You now have the shape of the whole course. Each lesson takes one method, states it in the terms its own defenders use, runs it on a text short enough to print, gives the objection that has cost it the most ground, and names what it cannot reach. The poem stays on the table throughout. When a method starts producing conclusions that no text could disturb, that is the moment to say so.

Worth holding on to: you cannot read without a theory, only without knowing which one you are using; and the most useful thing you can learn about any method is the list of questions it is built to ignore.

Sources

1. Blake, W. (1794). Songs of Innocence and of Experience. Project Gutenberg
2. Internet Encyclopedia of Philosophy. Literary theory. iep.utm.edu
3. Wikipedia. The Sick Rose. en.wikipedia.org
4. Culler, J. (2011). Literary Theory: A Very Short Introduction (2nd ed.). Oxford University Press.
5. Eagleton, T. (2008). Literary Theory: An Introduction (3rd ed.). Blackwell.

A twenty-year-old at the Stray Dog

In December 1913, in a basement cabaret in St Petersburg called the Stray Dog, a twenty-year-old student named Viktor Shklovsky gave a talk called The Resurrection of the Word. He printed it as a pamphlet the next year. Three years later, in 1917, he published the essay that is the reason his name survives outside Russia, Iskusstvo kak priem, which reaches English as Art as Device or Art as Technique.

The argument runs like this. Perception is expensive, so the mind economises. The tenth time you walk a road you no longer see the road; you recognise it and move on. Shklovsky calls this automatization, and he says it swallows everything it touches: objects, clothes, furniture, a wife, the fear of war. Life so lived, he writes, is as if it had never been.

Then the turn. If habit eats perception, and art is not decoration but an instrument, then art exists to give perception back. It does this by making things strange, ostranenie, usually translated as defamiliarization or estrangement, and by making the form difficult so that the act of perceiving takes longer. In the translation most English readers meet, the aim is "to make the stone stony." Difficulty is not a side effect of art. It is the mechanism.

The upshot: for Shklovsky the literary work is not a vessel carrying content. It is a set of devices whose function is to slow down and roughen perception, and the critic's job is to describe how those devices work, not what the work is finally about.

Tolstoy dusting a room

Shklovsky proves the negative half of his case with a diary entry. Tolstoy records dusting a room and coming to the sofa, unable to remember whether he had already dusted it. The movements were unconscious and therefore, Shklovsky says, unlived. If a whole life goes that way, it is as good as never lived at all.

He proves the positive half with Tolstoy too, and this is the part worth studying, because Shklovsky is not making a general point about vividness. He is naming a specific procedure: describe a thing without using its name, as though seeing it for the first time. In the story known in English as Strider, the narrator is a horse, and the horse cannot make sense of human property. People say my land, my horse, my woman, and yet many of them never touch the land, never ride the horse, do not live with the woman. The horse concludes that humans are governed not by deeds but by words. Nothing in that passage is untrue, and none of it can be seen while the word property is doing its usual work.

The same device is at work when Tolstoy describes a church service by naming only the physical actions performed with cloth and bread, or describes flogging as the practice of stripping people who have broken the law, throwing them down, and beating them on the backside with sticks. Shklovsky's point is that Tolstoy has not added moral commentary. He has withheld the name that lets us not look.

Try it as a procedure rather than a metaphor. Take a wedding ring: a small ring of the most durable metal available, worn permanently on one finger to signal that a contract exists. Take a lecture: a person reads aloud from notes while others make their own notes of what was read. The uncomfortable feeling those sentences produce is the effect Shklovsky is describing, and it is entirely a matter of technique.

Laying the device bare

Shklovsky's other favourite example is a novel from 1759. Here is how The Life and Opinions of Tristram Shandy, Gentleman begins:

I wish either my father or my mother, or indeed both of them, as they were in duty both equally bound to it, had minded what they were about when they begot me.

The narrator starts before his own conception and is not born until the third volume. His father's account of the accident to his nose takes precedence over the birth. A chapter is missing. One page is printed solid black, in mourning for Yorick. Another is marbled. A blank page is left for the reader to draw the Widow Wadman. The preface arrives in the third volume, and two chapters late in the ninth are left empty and then supplied several chapters later.

A reader in 1760 could call this a mess. Shklovsky calls it the most typical novel in world literature, and the claim is not perverse once you see the distinction it rests on. The events of a story in their natural order are the fabula. The order and manner in which the text delivers them is the syuzhet. Every narrative has both, and the two are never identical, because even the plainest chronological telling selects, compresses and delays.

Sterne's contribution is to make the gap between them impossible to ignore. Where an ordinary novel hides its arrangement, Sterne exhibits it, which the formalists called laying bare the device. That is why Shklovsky treats him as typical rather than eccentric: Tristram Shandy shows in the open what every novel is doing under cover.

In short: fabula and syuzhet give you a way to talk about form without talking about content. Ask what happened, then ask in what order and by what means you were told, and the difference between the answers is the work the text is doing.

What the group was trying to build

Shklovsky was not alone. The Moscow Linguistic Circle formed in 1915 around Roman Jakobson, and OPOJAZ, the Society for the Study of Poetic Language, formed in Petrograd in 1916 around Shklovsky, Boris Eikhenbaum and Yuri Tynyanov. What they wanted was a literary study that could be a discipline, with an object of its own.

Jakobson supplied the formula in 1921: the object of literary study is not literature but literariness, meaning whatever makes a given work a work of literature rather than a letter or a newspaper. That single move rules a great deal out. Biography goes, because it explains the author and not the text. Political history goes for the same reason. Psychology of the reader goes. What is left is the material and its organisation.

Eikhenbaum showed what the programme looked like in practice in 1919 with How Gogol's Overcoat Is Made. The traditional reading of Gogol's story treats it as a plea for a poor clerk, the humane classic of Russian literature. Eikhenbaum argues that the story is built out of skaz, the mimicry of a live speaking voice, with puns, mangled names, sudden shifts between comic patter and sentimental declamation. The famous humane passage, on this reading, is one more voice in the performance rather than the moral of the piece. You may not accept the conclusion. What matters is that the evidence is entirely in the sentence structure, and it is checkable.

Trotsky answers

In 1923 Leon Trotsky devoted a chapter of Literature and Revolution to the formalists, and it is the sharpest early statement of the objection this course will meet again in the Marxist lesson. He grants them something: the study of technique is real work, and the Marxists of his day had been careless about it.

Then he refuses the philosophy. The formalists, he writes, are followers of St John: for them, in the beginning was the Word. "But we believe that in the beginning was the deed. The word followed, as its phonetic shadow." Art, on Trotsky's account, is a function of the social process; it can be studied as a technique but cannot be explained as one, and a school that treats the poem as a self-sufficient system will end up describing the mechanism of a clock while refusing to ask who wound it or what time it keeps.

The state agreed, and with more force. By the late 1920s formalism was an accusation. In 1930 Shklovsky published a piece titled A Monument to a Scientific Error. Jakobson had already left for Czechoslovakia, where he helped found the Prague Linguistic Circle in 1926, and it is through Prague that formalist ideas reach the structuralism of the 1950s and 1960s. The line from a Petrograd society to Paris runs through an exile.

The problem the formalists found in their own theory

The most interesting objection was not Trotsky's. It came from inside. If a device works by making perception strange, then a device that succeeds gets imitated, and once imitated it becomes familiar, and once familiar it no longer defamiliarizes anything. Estrangement has a shelf life.

Tynyanov took this seriously and turned it into a theory of literary evolution. At any moment some element of the system is dominant and organises the rest; forms that have gone stale are pushed to the margins, and material from outside the recognised literature, letters, diaries, journalism, doggerel, gets promoted to fill the gap. Literary history stops being a parade of great men and becomes the study of a system reorganising itself under pressure. That is a formalist answer to a formalist problem, and it is why the school's late work is more interesting than the slogan it is remembered for.

The other objection, from Pavel Medvedev and Mikhail Bakhtin in 1928, holds that a device cannot be separated from the social evaluation carried in the words themselves, so the formalists' object of study was never as pure as they claimed.

Common misconceptions

• Formalism means the study of form as against content. Shklovsky's position is stronger and stranger: there is no content sitting behind the form to be extracted. What criticism can describe is what the devices do, and content is one of the effects they produce.
• Defamiliarization is a synonym for vivid description. It is a specific procedure with a testable signature: the customary name for the thing is withheld, and the thing is described as if by someone who does not know what it is for.
• The Russian formalists were apolitical aesthetes. Several were revolutionaries; Shklovsky drove an armoured car in 1917 and fled the country in 1922. The charge that stuck to them was philosophical, and then administrative.
• Fabula and syuzhet are just story and plot. The pair is closer to material and arrangement. Syuzhet includes every ordering choice a text makes, not only the causal shaping that the English word plot usually names.

What to carry forward

Formalism gives this course two permanent tools. The first is defamiliarization, both as a claim about what art is for and as a procedure you can run on any object by taking away its name. The second is the fabula and syuzhet distinction, which turns vague talk about structure into two questions with different answers.

It also sets up everything in Module 2. If a literary work is a system of devices, the obvious next question is whether the devices themselves form a system, a grammar shared across many works. Propp asked exactly that of the Russian folktale in 1928, and the answer he found is where structuralism begins.

Worth holding on to: the formalists won the argument about method, in that no one now writes criticism without attending to technique, and lost the argument about scope, in that almost no one now believes technique can be described without reference to anything outside the text.

Sources

1. Sterne, L. (1759-1767). The Life and Opinions of Tristram Shandy, Gentleman. Project Gutenberg
2. Trotsky, L. (1924). The formalist school of poetry and Marxism. In Literature and Revolution. marxists.org
3. Wikipedia. Russian formalism. en.wikipedia.org
4. Shklovsky, V. (1965). Art as technique. In L. T. Lemon and M. J. Reis (Eds. and Trans.), Russian Formalist Criticism: Four Essays (pp. 3-24). University of Nebraska Press. (Original work published 1917)
5. Erlich, V. (1981). Russian Formalism: History, Doctrine (3rd ed.). Yale University Press.

A summary that looks harmless

Wordsworth wrote these eight lines in Germany in 1799 and printed them in the second edition of Lyrical Ballads in 1800. They are the shortest of the poems now grouped as the Lucy poems.

A slumber did my spirit seal;
I had no human fears:
She seemed a thing that could not feel
The touch of earthly years.

No motion has she now, no force;
She neither hears nor sees;
Rolled round in earth's diurnal course,
With rocks, and stones, and trees.

Now the summary. A student writes: the speaker was in a kind of trance and did not worry about the girl, because she seemed beyond the reach of time. Then she died. Now she is inert and buried, turning with the earth among rocks and stones and trees.

Nothing there is false. Every sentence can be defended from the text, and the summary would receive a passing grade almost anywhere. This lesson is about where it fails, and the failures are specific enough to number.

Failure one: the word thing changes sides

In line three the speaker says she seemed a thing that could not feel the touch of earthly years. Read forward from line one, thing is a compliment. It means she was so alive, so far from decay, that mortality did not seem to apply to her; it is the sort of thing lovers believe.

By the second stanza she is a thing in the plainest sense: an object with no motion and no force, in a list with rocks and stones. The word has not changed; the world around it has. What the poem does, and what the summary cannot do, is make the reader complicit in the first meaning and then hold them there while the second arrives.

The student's paraphrase splits this into two propositions, one about a trance and one about a death. The poem is one word used twice.

Failure two: the death is not in the poem

Read the eight lines again and look for the event. There is no dying. There is no funeral, no illness, no news arriving, no name. Between the fourth line and the fifth there is nothing but white space, and on either side of that space the tense changes: seemed becomes has, and the verbs of the second stanza are all present.

The paraphrase inserts the sentence then she died, which is exactly the sentence Wordsworth declined to write. Its absence is not coyness. Sealed in the first stanza and the silence in the middle are the same gesture: the speaker's slumber is what let him not think about mortality, and the poem's own silence about the death repeats the evasion at the level of form.

What matters here: the summary supplies information the poem withheld and calls it clarification. Whatever the poem was doing with that withholding is gone from the summary, and it cannot be got back by adding a note.

Failure three: the last line does two opposite things at once

Rolled round in earth's diurnal course, with rocks, and stones, and trees. Diurnal is Latinate, technical, astronomical, the vocabulary of a planet in orbit. It grants the dead girl a place in a magnificent motion, and if you read the line for its music it is the grandest thing in the poem.

Then the last line is seven words, all but one of them a single syllable, joined by two ands: rocks, and stones, and trees. Rocks and stones are the same substance twice, which is the tell. The list has stopped meaning anything by the time it reaches trees; it flattens. The girl who could not feel the touch of earthly years is now being carried around the axis of the earth with the gravel.

Which is it, the consolation of the grand orbit or the horror of the gravel? The poem does not choose, and the last line's power comes from holding both. A paraphrase must choose, because prose cannot mean two incompatible things in one clause. That is the whole argument of this lesson in one line of verse.

Brooks names the error

In 1947 Cleanth Brooks ended The Well Wrought Urn with a chapter called The Heresy of Paraphrase. His claim is not that summaries are useless. It is that the paraphrase is a scaffolding, convenient for talking about the poem, and that criticism goes wrong the moment it treats the scaffolding as the building.

What replaces the paraphrase as an account of what a poem is? Brooks says structure, and he means something specific by it: not a logical argument or a narrative, but a pattern of resolved stresses, an equilibrium of attitudes held in balance. The working vocabulary of the school follows from that. Irony is the pressure that context puts on a statement, so that a word means differently here than it would in a dictionary. Paradox is a statement that must be false to be true. Tension, in Allen Tate's coinage, is what a poem holds between the abstract meaning and the concrete image.

Run those on our eight lines and the account is quick. The irony is the word thing. The paradox is that the sealed spirit was right, she cannot feel earthly years now. The tension is diurnal against stones. The poem is not a statement about grief plus some decoration; the arrangement is the meaning, and the arrangement cannot be restated.

Two fallacies that clear the ground

For that programme to work, two familiar moves have to be ruled out, and in two essays in the Sewanee Review W. K. Wimsatt and Monroe Beardsley ruled them out.

The intentional fallacy, in 1946: "the design or intention of the author is neither available nor desirable as a standard for judging the success of a work of literary art." The argument is practical before it is philosophical. If the poem succeeds, the intention is in it and you do not need the letter; if the poem fails, no letter can rescue it. Wordsworth's note that the poem was written in Germany tells you nothing about whether the last line consoles.

The affective fallacy, in 1949: the confusion between the poem and its results, between what it is and what it does to a reader. The objection is that reports of feeling are reports about readers, and readers vary; a criticism built on them ends in impressionism, where the poem disappears and only the critic's nervous system remains.

Between them the two essays leave the poem alone on the desk, which was the point. Note what has been shut out: biography, letters, the author's stated aims, the reader's tears, and the history in which the poem was written.

The dispute this poem actually caused

In 1950 F. W. Bateson read the same eight lines and argued that they end in something like pantheistic magnificence: the girl is taken up into the motion of the earth, and the last stanza is a consolation rather than a grief. Brooks read them as shock and inertness, the beloved reduced to an object among objects. Two competent critics, one text, opposite conclusions about its dominant feeling.

In 1967 E. D. Hirsch made this disagreement the hinge of Validity in Interpretation. Both readings, he pointed out, are consistent with the words on the page; the words underdetermine the choice. So if criticism is to be knowledge rather than preference, something outside the words has to break the tie, and the only candidate is what Wordsworth probably meant, reconstructed from the rest of his work and his period. Hirsch draws his famous distinction here: meaning, which is fixed by the author and does not change, and significance, which is what the text comes to mean to later readers and changes constantly.

That is the intentional fallacy answered on its own ground, and it is why the argument did not end in 1946. If you rule out intention, you may be left with two readings and no way to choose. If you rule it in, you have to defend a claim about a dead man's mind on the evidence of documents he did not write for you.

Where the method came from, and what it was for

The technique is older than the school. In Cambridge in the 1920s I. A. Richards handed his students poems with the authors' names removed and collected their written responses. Practical Criticism, published in 1929, prints the results, and they are unflattering: readers of the best university in the country misread stress, missed syntax, and graded poems by whether the sentiments matched their own. His student William Empson published Seven Types of Ambiguity in 1930, at twenty-four, and made the multiplicity of meaning a technical subject.

The American branch had an institutional motive. John Crowe Ransom argued in Criticism, Inc. in 1937 that literary study had to become the work of professionals with a method of their own, rather than a mixture of philology, moralising and appreciation. Brooks and Robert Penn Warren then wrote the textbook that carried it into every classroom, Understanding Poetry, in 1938. It arrived just before the postwar expansion of American universities, and it had one unbeatable practical virtue: the entire evidence for a reading fits on the page in front of the student, so a class of five hundred with no library and no Greek can all do the work.

What it cost

Three bills came due. The first is the canon. A method built on irony, paradox and compression rewards the short metaphysical lyric and penalises the long, the plain and the polemical; Donne rose and Shelley fell, and the poem as such came to mean a page of dense verse.

The second is history, which the method did not so much refute as decline to admit. That is the opening Marxist and historicist criticism will take, and Module 5 is where the argument gets settled or at least stated properly.

The third is the claim to neutrality. Ransom and Tate had published in 1930 in I'll Take My Stand, a manifesto for the agrarian South against industrialism, and later critics have argued that the ideal of the poem as a self-sufficient organic whole was not a view from nowhere. Whether that undermines the method or merely explains its appeal is a live question; what is not in doubt is that the men who called for a criticism free of politics had a politics.

Common misconceptions

• The heresy of paraphrase forbids summarising a poem. Brooks summarises constantly. The heresy is treating the summary as the poem's meaning, so that everything the arrangement was doing is written off as ornament.
• The intentional fallacy says authors do not know what they are doing. It says that a statement of intention is not evidence for the value or meaning of the finished work, because a successful poem carries its design inside it and a failed one is not saved by a letter.
• The New Criticism was one school with one doctrine. Richards was a psychologist of meaning whose position the affective fallacy essay rejects; Empson kept using biography and intention all his life; Ransom disagreed with Brooks about structure. The label was applied loosely and then hardened.
• Close reading and New Criticism are the same thing. Close reading is a practice you will use in every later lesson, including the Marxist and postcolonial ones. New Criticism is a set of claims about what may count as evidence, which most later critics reject while keeping the practice.

Pulling it together

The chain of this lesson: a paraphrase of eight lines fails in three identifiable places, the failures share a cause, Brooks names that cause, and two essays remove the escape routes that would let a reader avoid facing it. What you are left with is a discipline that can be taught, examined and argued about, which is more than criticism had before 1929.

You are also left with a poem whose dominant feeling two serious critics could not agree on, and a method that forbids the appeal that would settle it. Hold on to that. Every school after this one is in some sense an answer to the question of what to do when the words on the page underdetermine the reading.

Bottom line: the New Critics were right that the arrangement is the meaning and cannot be restated, and their opponents were right that the arrangement alone will not always tell you which of two readings to believe.

Sources

1. Wordsworth, W. (1800). A slumber did my spirit seal. In The Poetical Works of William Wordsworth (Vol. 2). Project Gutenberg
2. Wikipedia. Intentional fallacy. en.wikipedia.org
3. Wikipedia. The Well Wrought Urn. en.wikipedia.org
4. Wimsatt, W. K., and Beardsley, M. C. (1946). The intentional fallacy. Sewanee Review, 54(3), 468-488.
5. Hirsch, E. D. (1967). Validity in Interpretation. Yale University Press.

A card index in Leningrad

In 1928 Vladimir Propp published a short book on a hundred Russian wonder tales taken from Afanasyev's collection. He had done something nobody had thought worth doing: instead of sorting the tales by theme or origin, he wrote out what happened in each one as a sequence of actions, and compared the sequences.

What he found is the reason Morphology of the Folktale is still in print. The characters vary endlessly, but the actions do not. There are thirty-one of them, he claimed, they never occur in a different order, and no tale uses all of them. He called them functions, and defined a function as an act considered from the point of view of its significance for the course of the action. Who performs it is a separate question; the seven roles that do the performing, villain, donor, helper, princess and her father, dispatcher, hero, false hero, are also fixed.

That is a strong claim about literature: that beneath a hundred different stories there is one story, and it can be written down. This lesson is about the tradition that took the claim seriously, and it proceeds by running its procedures rather than describing them.

Saussure, and why difference comes first

The theory underneath the procedures comes from a linguist who did not publish it. Ferdinand de Saussure died in 1913, and in 1916 two colleagues assembled the Course in General Linguistics from his students' lecture notes. Four of its claims do the work.

• The sign has two faces: a signifier, the sound or mark, and a signified, the concept. Neither is the thing in the world.
• Their relation is arbitrary. Nothing about a tree requires the sounds of arbre or Baum, which is why languages differ and why usage can change.
• Value comes from difference. In the system, cat means what it means because it is not cut, not cap, not dog. Saussure's formulation is that in language there are only differences, without positive terms.
• Langue, the shared system, is separable from parole, the individual utterance, and the system can be described as it stands, synchronically, without tracing its history.

Transferred to literature, those four claims give you a programme. A poem or a tale is a parole; the business of criticism is the langue behind it, the system of conventions that makes it readable. Meaning is not a matter of what a word points at but of its position in a network of contrasts. And the system can be studied as it is, without the history that produced it.

Run the procedure: Little Red-Cap, first pass

Take the Grimms' version, Little Red-Cap, printed in the Household Tales. The mother's instruction is worth having exactly: 'walk nicely and quietly and do not run off the path, or you may fall and break the bottle, and then your grandmother will get nothing.'

Now the sequence, with Propp's names.

The procedure has produced something checkable. Every row can be argued with by pointing at the text, which is more than most interpretation offers.

Second pass: change one input

The Grimms' tale does not stop at the punishment. A second wolf appears; Red-Cap is on the same errand; and this time she does not leave the path. She goes straight to the house, tells her grandmother what she saw, and the two of them keep the door shut. The grandmother has her pour the water from cooking sausages into the trough outside; the smell draws the wolf onto the roof, he slips, falls into the trough and drowns.

Compare the two moves. The interdiction is the same. The violation does not occur. Reconnaissance and delivery do not occur, because the girl says nothing to the wolf. And with no complicity there is no villainy, so there is no need for a rescue and no huntsman: the donor role passes to the grandmother, who supplies the means, and the victim defeats the villain herself.

That is the payoff of running a procedure rather than admiring one. The second episode is not a different story with a different moral, it is the same function set with one input flipped, and Propp's grammar predicts precisely which downstream functions disappear.

So what?: a structural description is worth having when it tells you what would have to change for the story to come out differently. If a method cannot answer that, it is a filing system.

What the table cannot see

It cannot see the red cap. The velvet, the colour, the fact that the girl is named after a garment, none of that is a function, and Propp's method is indifferent to it. It cannot see that Perrault's 1697 version has no huntsman, no rescue and no second wolf: the girl is eaten and the tale ends with a verse warning about the wolves who follow young ladies into their houses. It cannot see that the Grimms were editing an oral tradition into a book for bourgeois households in 1812. It cannot tell you why anyone has told this story for three hundred years.

Claude Levi-Strauss made a version of this objection in 1960. Propp, he argued, had separated form from content so cleanly that he was left with an empty grammar into which any content at all could be poured, and the price of the abstraction was that the tales stopped being about anything. His own method looks for bundles of relations, oppositions such as raw and cooked or nature and culture, that recur across a body of myth. Propp's reply was that the objection asked him to answer a question he had not asked.

Barthes: five codes on one sentence

In 1970 Roland Barthes published S/Z, which cuts a thirty-page Balzac novella into 561 fragments and reads them one at a time through five codes. The codes are not a hierarchy and they are not exhaustive; they are five kinds of question you can ask at any point in a text.

Take the first sentence of the Grimm tale: 'Once upon a time there was a dear little girl who was loved by everyone who looked at her, but most of all by her grandmother, and there was nothing that she would not have given to the child.'

• Hermeneutic, the code of the enigma: the sentence opens a question and does not answer it, and the last clause promises a gift, so a reader is now waiting for something.
• Semic, the code of connotation: dear, little, loved by everyone. These are not information, they are attributes, and they build a character out of adjectives before anything happens.
• Symbolic, the code of antitheses: village and wood, child and grandmother, inside and outside. The tale's oppositions are already being set.
• Proairetic, the code of actions: TO GIVE has begun, and it will run through the cap, the cake and the bottle of wine.
• Cultural, the code of received knowledge: Once upon a time signals the whole convention of the folktale, and grandmothers dote on grandchildren is offered as something everybody knows.

Notice that the cultural code is where ideology hides. The sentence presents family sentiment as a fact of nature and not as a claim, and it is precisely because it is not argued that it is hard to argue with. Barthes's word for a text that keeps these codes quiet and reads smoothly is readerly; a writerly text makes the reader do the work of production.

The author dies, and something is gained

Three years before S/Z, Barthes published The Death of the Author. The argument is that a text is not a line of words releasing a single meaning from an Author-God, but a tissue of quotations drawn from innumerable centres of culture, and that to assign an Author to a text is to close it, to furnish it with a final signified. His last sentence is a slogan and was meant to be: "the birth of the reader must be at the cost of the death of the Author."

Two years later Michel Foucault answered it with What Is an Author, which does not restore the author but describes the author-function: the historically specific machinery by which a name is attached to some discourses and not others, groups them, gives them status, and makes it possible to prosecute someone. Statements about a text's meaning cannot be settled by the name, but the name is doing work that a theory has to account for.

Genette: five questions for any narration

Structuralism's most durable export is the toolkit that Gerard Genette assembled in Narrative Discourse in 1972. Take the opening of Great Expectations: 'My father's family name being Pirrip, and my Christian name Philip, my infant tongue could make of both names nothing longer or more explicit than Pip. So, I called myself Pip, and came to be called Pip.'

The gap between mood and voice is the whole comedy of the sentence. Infant tongue is not a phrase an infant could produce, so the adult Pip is standing behind the child, affectionate and slightly amused. Ask the five questions of any narrative and you will find things you would otherwise have felt without being able to name.

Common misconceptions

• Structuralism means finding hidden symbols. It is the opposite. The claim is that meaning comes from position in a system of differences, not from a secret content buried in an image.
• Propp proved all stories are the same. He described a specific corpus, the Russian wonder tale, and said so. Applying the thirty-one functions to a modern novel is an analogy, and the strain shows quickly.
• The death of the author means biography is banned and anything goes. Barthes is making a claim about where meaning is produced, not licensing arbitrary readings; S/Z is one of the most disciplined close readings ever published.
• Narratology is jargon for things everyone already knows. Try describing the difference between an iterative and a singulative telling, or between focalization and voice, without the terms, and the sentence gets long and wrong.

The short version

Structuralism replaces the question what does this text mean with the question how does this text mean, and answers it by describing the system that makes meaning possible: Saussure's differences, Propp's functions, Barthes's codes, Genette's five categories. Its discipline is real, its results are checkable against the page, and its vocabulary is now the common property of criticism, used every day by people who would not call themselves structuralists.

Its weakness is the one Levi-Strauss named and the next lesson will press much harder. A system of differences with no positive terms is stable only if the differences hold still. If the oppositions the system rests on can be shown to leak, the whole procedure changes character, and that is the argument Derrida brought to Baltimore in 1966.

The core of it: a structural reading earns its keep when it can say what would have to change for the text to come out differently, and it costs you everything that the text has that its system does not code.

Sources

1. Grimm, J., and Grimm, W. (1812). Little Red-Cap. In Grimms' Fairy Tales. Project Gutenberg
2. Wikipedia. Morphology of the Folktale. en.wikipedia.org
3. Propp, V. (1968). Morphology of the Folktale (2nd ed., L. Scott, Trans.). University of Texas Press. (Original work published 1928)
4. Barthes, R. (1974). S/Z (R. Miller, Trans.). Hill and Wang. (Original work published 1970)
5. Genette, G. (1980). Narrative Discourse: An Essay in Method (J. E. Lewin, Trans.). Cornell University Press. (Original work published 1972)

Baltimore, October 1966

The Ford Foundation paid for a symposium at Johns Hopkins that October called The Languages of Criticism and the Sciences of Man. Its purpose was to introduce structuralism to the United States, and the guest list was the movement: Lacan, Barthes, Todorov, Georges Poulet, Lucien Goldmann, Jean Hyppolite. Levi-Strauss was invited and did not come.

On the third day a thirty-six-year-old philosopher from the Ecole Normale Superieure read a paper called Structure, Sign, and Play in the Discourse of the Human Sciences. Jacques Derrida had published almost nothing in English. His paper argued that the science the conference had assembled to celebrate rested on a move it could not justify.

The claim: every structure has had a centre, and the centre cheats

Derrida's opening is a piece of intellectual history. The concept of structure, he says, is as old as thought, and it has always been given a centre: a point of origin or reference that organises the whole and limits what he calls the play of its elements. The names of this centre change, and he lists them, essence, origin, end, substance, truth, consciousness, God, man, but its function does not.

The trick is in the position. The centre governs the structure, so it must belong to it. But if it were merely one element among others it could be moved and substituted like any other, so it must escape the structure it governs. The centre is therefore inside and outside at once, which is, in Derrida's phrase, contradictorily coherent.

He then shows the problem at work in Levi-Strauss. Structural anthropology rests on the opposition between nature, what is universal and spontaneous, and culture, what varies by rule and prohibition. The incest prohibition wrecks it: universal, therefore natural, and a prohibition, therefore cultural. Levi-Strauss calls it a scandal. Derrida's point is not that the anthropologist blundered. It is that the scandal is exactly what you should expect if the opposition was never a fact about the world but a tool of description that the description then forgets it is using.

Why this matters: the argument is not that oppositions are useless. It is that they are hierarchical, one term always privileged and the other defined as its lack, and that the privileged term turns out to depend on what it excludes.

Not a mood, a procedure

A year later Derrida published three books, including Of Grammatology, which does the same operation on the opposition between speech and writing. Western philosophy from Plato to Saussure treats speech as the natural, living, present form of language, and writing as a derived, secondary technology for recording it. Derrida calls that ranking logocentrism, and he attacks it not by asserting the opposite but by reading Rousseau closely.

Rousseau calls writing a supplement to speech. Watch what the word has to do. A supplement is an addition to something already whole; but you supplement what is lacking, and you do not add to what is complete. So the very word Rousseau uses to keep writing in its subordinate place concedes that speech had a gap in it, and needed help. Rousseau uses the same word for other substitutes he condemns, and the logic runs the same way each time.

That is the procedure, and it has four steps you can run on any text: find the opposition and the hierarchy; find the passage where the privileged term needs the subordinate one to be described at all; invert the hierarchy; then refuse to settle in the inversion, because a simple reversal keeps the same structure. What survives is not a new hierarchy but the demonstration that this one was constructed and is unstable.

Derrida coined a word to hold the point. In French, difference and the spelling he invents for it, with an a in place of the second e, sound exactly alike; the difference exists only in writing, which is the joke and the argument at once. His term names both differing, since meaning depends on contrast, and deferring, since the contrast never terminates in a final presence. Saussure said value comes from difference. Derrida's coinage says that if that is true, then no signified is ever fully present, because every term carries the traces of the terms it is not.

One line of Yeats, two readings that cannot both be right

Paul de Man gave the literary version its cleanest demonstration in an essay of 1973. Yeats's poem Among School Children, from the 1928 collection The Tower, ends:

O body swayed to music, O brightening glance,
How can we know the dancer from the dance?

Read it as a rhetorical question, which is how it is nearly always read and how it is printed on posters: we cannot tell them apart, form and meaning are one, the dancer is the dance. Read it as a literal question and it says something else entirely: by what means could we possibly distinguish the two? That is a request, and it comes from someone who badly needs the distinction. In the poem, which has spent seven stanzas on ageing, labour and the difference between the image and the thing, the literal reading is at least as well supported as the rhetorical one.

De Man's point is precise. The two readings are grammatically identical: nothing in the syntax marks the difference, and no amount of closer attention to the words will settle it, because the words are the same words. The choice has to come from somewhere else, and there is nowhere else that is not another reading. He makes the same point with a joke from a television comedy, where a character is asked whether he wants his bowling shoes laced over or laced under and answers by asking what the difference is. His wife explains the difference. He is furious, because he was asking her not to bother him.

Remember: deconstruction's literary claim is not that a text means nothing. It is that a text can generate two readings that are both grounded in it and cannot both hold, and that grammar has no resource for choosing.

The Yale school

De Man was at Yale, with J. Hillis Miller, Geoffrey Hartman and, in a looser association, Harold Bloom; the group published Deconstruction and Criticism in 1979 and gave American graduate schools a method. Miller applied it to the poem you closed-read two lessons ago. Where Brooks and Bateson disagreed about whether the last stanza of the Lucy poem consoles or appals, Miller's answer is that the poem produces both and offers no ground for choosing, so the critical deadlock is not a failure of the critics but a property of the text.

Notice how that reframes the whole New Critical enterprise. The New Critics prized ambiguity and paradox because a good poem resolves them into a unity. De Man's version keeps the ambiguity and removes the resolution. Same close attention, opposite conclusion about what close attention finds.

Abrams answers

In the spring 1977 issue of Critical Inquiry, M. H. Abrams published The Deconstructive Angel, with Miller's reply printed alongside it. Abrams is not a naive opponent; he grants that language is a system of differences and that meaning is never total.

His objection is about practice. Writers and readers, he says, operate a shared linguistic and literary institution whose norms are learned and largely determinate; that is why a scholar can establish what a Romantic writer meant by a word in 1798 and be corrected if wrong. If that were not so, the deconstructive critics could not conduct their own arguments, quote their opponents accurately, or expect to be understood when they say that texts cannot be determinately understood. The performance refutes the doctrine.

Miller's reply, The Critic as Host, works on Abrams's own metaphor: he traces the words parasite and host to a common root and shows that each contains the other, guest and enemy, stranger and host, so that the metaphor Abrams used to dismiss deconstruction as a parasite on real criticism cannot be kept stable. Whether that answers Abrams or changes the subject is exactly the question the dispute turns on.

Searle answers

In 1977 John Searle replied in the journal Glyph to Derrida's essay on Austin's speech act theory. Derrida had argued that Austin's exclusion of jokes, plays and quotations as parasitic uses of language cannot be sustained, because the possibility of being repeated in another context, iterability, is a condition of any sign functioning at all.

Searle's counter is that Austin's distinction is not a metaphysical hierarchy but a methodological simplification, of the kind every science makes, and that Derrida has confused the general iterability of linguistic forms with the specific question of what a speaker does on an occasion. Derrida answered at length, and with mockery, in Limited Inc. Searle's later verdict, in print, was that the writing was obscurantist and that when the arguments were made clear they were either false or trivial.

That charge acquired an institutional form in May 1992, when Cambridge proposed an honorary doctorate for Derrida and nineteen philosophers, among them W. V. Quine and Ruth Barcan Marcus, wrote to The Times to oppose it, saying his work did not meet accepted standards of clarity and rigour. The university put it to a vote, which is unusual, and awarded the degree 336 to 204.

1987, and an archive in Belgium

In 1987 a Belgian graduate student, Ortwin de Graef, found nearly two hundred articles that de Man had written for the collaborationist Brussels newspaper Le Soir during the German occupation, including a piece of March 1941 on Jews in contemporary literature that entertained the idea of a separate Jewish colony without visible discomfort. De Man had died in 1983 and had told no one.

The discovery does not refute a theory of reading; an argument is not answered by its author's life. But it changed how the theory was heard, because de Man's later work insisted that texts undo their own claims and that confession and apology are rhetorically unreliable, and readers now had a reason to hear those claims as autobiography. Derrida's response, a long essay in Critical Inquiry in 1988, was itself attacked for reading the wartime articles too generously. What the episode really tests is a question this course keeps returning to: how much of a method's authority rests on the person who taught it.

What would settle the dispute

Both sides have something they could be shown. Abrams and Searle would have to concede ground if deconstructive readings turned out to be reliably reproducible: if trained readers, working independently, found the same instabilities at the same points in a text, the way independent scholars converge on a date or a source. Derrida and de Man would lose ground if the celebrated undecidables dissolved under better historical and linguistic information, so that the Yeats line, say, could be settled by evidence about Yeats's usage and the poem's manuscripts.

Neither test has been run to anyone's satisfaction, which is one reason the argument went quiet rather than concluding. What is not in doubt is the influence: the demonstration that oppositions are constructed and unstable was picked up almost immediately by critics who wanted to work on the oppositions that organise sex, race and empire. Gayatri Spivak translated Of Grammatology into English in 1976 and wrote a preface that is itself a landmark; you will meet her again in the postcolonial lesson, and Butler's account of gender in the queer theory lesson runs the four-step procedure on a hierarchy Derrida never discussed.

Common misconceptions

• There is nothing outside the text means reality does not exist. The French sentence is better rendered as there is no outside-text, and Derrida later glossed it as there is nothing outside context. The claim is about the impossibility of a meaning fixed from outside all context, not a denial of the world.
• Deconstruction means any reading is as good as any other. Derrida and de Man both insist that a deconstructive reading must first produce the most rigorous available reading of what the text says, since the instability is shown by the text's own terms and not imported.
• Deconstruction is destruction. The word translates Heidegger's Destruktion, a taking apart to see how something was built. The four-step procedure ends by refusing the simple inversion, which is why it is not a swap of one hierarchy for another.
• It is a way of reading literature. Derrida's targets were Plato, Rousseau, Husserl, Saussure and Austin. Literature departments adopted it; philosophy departments, in the English-speaking world, largely did not, and that division is most of what the 1992 Cambridge quarrel was about.

Looking back

Structuralism proposed that meaning comes from a system of differences. Deconstruction accepts that premise and refuses the stability that structuralism assumed came with it: if terms mean by differing, then meaning is never present in one place, and the oppositions that organise a system will be found leaning on what they exclude.

The gain is a procedure that reliably finds the load-bearing opposition in a text and the passage where it fails. The cost is that the procedure is difficult to falsify, produces a similar-shaped result whatever it is applied to, and gave a generation of critics a vocabulary that was easy to imitate badly. Both halves of that sentence are worth carrying into the rest of the course.

The takeaway: when you meet a confident opposition in a piece of criticism, ordinary against literary language, centre against margin, form against content, ask which term is doing the privileging and whether it can be described without borrowing from the term it excludes. That question is Derrida's, and it survives every objection in this lesson.

Sources

1. Lawlor, L. (2023). Jacques Derrida. Stanford Encyclopedia of Philosophy. plato.stanford.edu
2. Internet Encyclopedia of Philosophy. Derrida, Jacques. iep.utm.edu
3. Wikipedia. Deconstruction. en.wikipedia.org
4. Abrams, M. H. (1977). The deconstructive angel. Critical Inquiry, 3(3), 425-438.
5. de Man, P. (1979). Allegories of Reading: Figural Language in Rousseau, Nietzsche, Rilke, and Proust. Yale University Press.

A question nobody has settled since 1898

Henry James published The Turn of the Screw in Collier's Weekly over twelve weeks in 1898. A young woman with no money and no experience takes a post as governess to two children at a country house in Essex. Her employer, whom she meets once and finds handsome, imposes one condition: he is never to be troubled about them. At Bly she comes to believe that two dead servants, Peter Quint and Miss Jessel, are appearing to the children and corrupting them. By the last page one child has fled and the other is dead in her arms.

Here is the problem. Nobody else in the book ever sees a ghost. The housekeeper, Mrs Grose, cannot read and takes the governess's word for everything. The children never once confirm an apparition. So either the ghosts are real and the governess is the only person with the courage to see them, or there are no ghosts and a young woman's mind has destroyed two children. The text supports both, and it has supported both for more than a century.

This lesson works that problem, because the psychoanalytic tradition in criticism is best understood through the argument it started and could not finish.

Wilson's answer, 1934

In 1934 Edmund Wilson published an essay arguing that the ghosts are hallucinations produced by the governess's repressed desire for her employer. His evidence is worth listing, because it is better than its reputation.

• The apparitions arrive at moments when she is thinking about the master, and the first arrives at the exact instant she is imagining a man appearing to admire her.
• She never gets independent confirmation. Mrs Grose identifies Quint from a description the governess supplies, and identifications made that way are worthless.
• The two ghosts are a manservant and a governess who were lovers, that is, a version of the cross-class romance the governess has been fantasising about.
• Her behaviour intensifies as she is left alone with the children, and the boy dies in her grip during an interrogation she is conducting.

Robert Heilman answered in 1947 with the other case. James called the story a fairy tale and an amusette in his preface. The governess produces a detailed description of a man she has never seen or heard of, and the description fits a real dead servant. Miles is expelled from school for something never named. A reading that dissolves the ghosts must explain away the plot's own machinery, and it usually does so by treating every piece of counter-evidence as further proof of the governess's derangement.

Key idea: the deadlock is not caused by insufficient close reading. Both sides read closely, and the closer they read, the more each finds. That is the fact the rest of this lesson has to explain.

What Freud actually gives a critic

Freud published Das Unheimliche in 1919, and it is the single most useful essay he wrote for literary study, because it is a piece of criticism rather than a case history.

He starts with a word. Heimlich in German means homely, familiar, belonging to the house; it also means concealed, kept from sight. So the word slides into its own opposite, and unheimlich, the uncanny, is not simply the unfamiliar. Freud's conclusion, quoting Schelling, is that the uncanny is what ought to have remained hidden and has come to light. The frightening thing is the familiar returning in a form we do not recognise.

Then he reads a text. In Hoffmann's The Sandman, of 1816, a student is destroyed by his childhood terror of a figure who throws sand in children's eyes so that the eyes jump out. Ernst Jentsch had explained the story's effect by intellectual uncertainty, chiefly over whether the beautiful Olimpia is a living woman or an automaton. Freud disagrees flatly, and his argument is textual: the doll is a subplot, while the threat to the eyes recurs at every turn of the story and organises its structure. Fear for the eyes, he claims, is a substitute for the fear of castration, and the story terrifies because a repressed childhood dread returns.

You may reject the conclusion. Look at the method: identify what recurs, notice what the text keeps returning to rather than what it announces, and treat the point of maximum obscurity as the point of maximum interest. That is a reading procedure, and it works on texts whose authors never heard of Freud. Applied to Hamlet in The Interpretation of Dreams in 1900, it produces the claim that Hamlet cannot kill Claudius because Claudius has done what he himself wished as a child, which is either a brilliant solution to a four-century puzzle or a demonstration that the method can generate an answer to anything.

Lacan moves the object

Jacques Lacan changed what psychoanalytic criticism is about, and the change is what makes it usable now.

Start with the paper he delivered in Zurich in 1949 on the mirror stage. An infant between six and eighteen months, still unable to control its own limbs, sees itself in a mirror and greets the image with delight. What it recognises is a unity it does not possess: the image is whole, coordinated and bounded, while the body it belongs to is not. The self, on this account, begins as an identification with an image that is a misrecognition, and it stays that way. The I is not the foundation of experience but a picture we have adopted.

From there the famous formula: the unconscious is structured like a language. Lacan takes Freud's dream mechanisms and maps them onto rhetoric, condensation onto metaphor and displacement onto metonymy, borrowing the pair from Jakobson. The consequence for criticism is direct. If the unconscious works like language, the critic does not need access to a mind, and the fantasy of putting a dead author or a fictional character on a couch can be dropped. What is analysable is the text's own language: its substitutions, its slides, its gaps.

Lacan demonstrated this in a seminar on Poe's story of a stolen letter, whose contents are never revealed. His reading tracks how each character's position is determined by their relation to the letter rather than by their psychology, and he concludes that a letter always arrives at its destination. Derrida attacked the reading in 1975 for smuggling a guaranteed truth back in, and Barbara Johnson then published an essay showing that Derrida's critique repeats the very move it accuses Lacan of making. Three readings, each catching the previous one in the trap it described.

Felman turns the question around

Which brings us back to Bly. In 1977 Shoshana Felman published a long study of The Turn of the Screw that refuses both available answers.

Her observation is that the critical literature has the same shape as the story. The governess believes the children are concealing something and reads their innocence as evidence of it. The Freudian critics believe the text is concealing something and read its apparent supernaturalism as evidence of it. The anti-Freudian critics believe the Freudians are deceived, exactly as the governess believes Mrs Grose is deceived. Each party occupies a position the novel has already written, and the accusation of blindness is the story's own mechanism, not a discovery about it.

The wider claim follows: literature is not the raw material on which psychoanalysis operates. Applying a theory to a text assumes the theory is the knowing party and the text the object; Felman argues that James's story knows something about interpretation that the interpreters keep proving by their behaviour. The relation is mutual implication, and the critic who forgets it becomes a character.

Working the passage

Here is the moment where all of this can be tested. The governess is walking at dusk and thinking about her employer:

Someone would appear there at the turn of a path and would stand before me and smile and approve.

Two sentences later, a man is standing at the top of the tower:

What arrested me on the spot ... was the sense that my imagination had, in a flash, turned real. He did stand there!

Run the three methods. Wilson's reading takes the sequence as the plainest possible evidence: the wish is stated, and then the wish is fulfilled in a form the wisher cannot own. The Lacanian reading attends to the language rather than the diagnosis: the sentence my imagination had turned real does not describe a perception, it describes an identification, and the figure on the tower occupies the place the governess has just prepared for a man who would look at her and approve. Felman's reading points at the phrase turned real and asks what it does to a reader who is being asked to decide whether the figure is real. The novel supplies the doubt as part of the apparition.

Notice that none of the three requires a claim about Henry James's private life, which is what psychoanalytic criticism looked like before Lacan and why it earned its reputation for vulgarity.

Three objections worth taking seriously

The first is falsifiability. Karl Popper used psychoanalysis as his standard example of a theory that cannot fail: evidence confirms it, and denial of the evidence confirms it as resistance. Applied to criticism, this means a psychoanalytic reading can be produced for any text whatever, and a reading available in advance is not evidence about the particular text.

The second is clinical. Freud's case histories have been reexamined for a century, and the Dora case in particular is now read by many as a record of a young woman's account being overridden. A method whose founding evidence is contested does not thereby become false, but it cannot be cited as settled science in a literary argument.

The third comes from inside literary studies and is the sharpest. Wilson's diagnosis of a sexually frustrated spinster whose fantasies kill a child is itself a period document: an unmarried woman without money, employed by a man who will not speak to her, is exactly the figure Victorian medicine liked to call hysterical. Feminist critics have argued that the Freudian reading reproduces the story's own class and gender arrangements while claiming to expose them, and that the more interesting hidden content of the novel is the terror of a woman with total responsibility and no authority, which the text states in plain view.

Common misconceptions

• Psychoanalytic criticism means decoding symbols. Towers and lakes as body parts is the parlour game version. The method that survives is procedural: attend to repetition, to what a text will not say, and to the point where its language slips.
• It analyses the author. Early practitioners did, and the results were poor. After Lacan the object is the text's language, which is publicly available and can be argued about.
• The uncanny means anything creepy. Freud's definition is specific: the familiar, repressed and now returning in unrecognised form. A monster that was never familiar is not uncanny in his sense.
• Felman's reading is a refusal to decide. She makes a strong claim, that the text produces the deadlock deliberately and that the critics' certainty is part of its design. That claim can be argued with; it is not a shrug.

Putting it together

The tradition offers three distinct things, and it is worth keeping them apart. From Freud, a reading procedure that treats repetition and evasion as evidence and has produced real discoveries about texts. From Lacan, a redirection from the psyche to the language, which is what makes the method arguable in public. From Felman, a warning that a method applied to a text may find only what it brought.

What it does not offer is a way to settle the question about the governess. After ninety years the argument stands where Wilson and Heilman left it, and the best explanation of that stalemate is Felman's: the book was built to produce it.

Bottom line: use the psychoanalytic method for what it is good at, finding the place where a text's language exceeds what its narrator can admit, and be suspicious of any reading of this kind that could have been written before the text was opened.

Sources

1. James, H. (1898). The Turn of the Screw. Project Gutenberg
2. Freud, S. (1919). Das Unheimliche. Project Gutenberg
3. Wikipedia. Uncanny. en.wikipedia.org
4. Felman, S. (1977). Turning the screw of interpretation. Yale French Studies, 55/56, 94-207.
5. Heilman, R. B. (1947). The Freudian reading of The Turn of the Screw. Modern Language Notes, 62(7), 433-445.

A sonnet written on the roof of a coach

On the morning of 31 July 1802, William and Dorothy Wordsworth sat on top of the Dover coach as it crossed Westminster Bridge. They were on their way to Calais so that William could see Annette Vallon and their nine-year-old daughter Caroline before he married Mary Hutchinson. Dorothy's journal records the crossing, and the sonnet that came out of it carries a date in its title that does not match.

Composed upon Westminster Bridge, September 3, 1802

Earth has not any thing to show more fair:
Dull would he be of soul who could pass by
A sight so touching in its majesty:
This City now doth, like a garment, wear
The beauty of the morning; silent, bare,
Ships, towers, domes, theatres, and temples lie
Open unto the fields, and to the sky;
All bright and glittering in the smokeless air.
Never did sun more beautifully steep
In his first splendour, valley, rock, or hill;
Ne'er saw I, never felt, a calm so deep!
The river glideth at his own sweet will:
Dear God! the very houses seem asleep;
And all that mighty heart is lying still!

Two critics read it. Both are good at their jobs. They agree on every word and reach verdicts that do not fit together.

The first reading: what the sonnet does

It is an Italian sonnet, so the argument turns at line nine, and the turn is from the city to the sun, that is, from the made to the given. The octave holds the city; the sestet measures it against valley, rock and hill and finds the city not inferior. For a poet who had built a career on the claim that nature teaches what cities cannot, that is a reversal, and the form stages it.

The governing figure arrives in line four: the city wears the morning like a garment. A garment is put on and taken off, so the beauty is explicitly temporary and explicitly not the city's own. Then the list, ships, towers, domes, theatres, and temples, five nouns in one line with no adjectives at all, plain to the point of inventory.

Notice how much of the poem is negative. Earth has not any thing, never did sun, ne'er saw I, never felt. The praise is built out of denials of comparison, which is a way of praising without describing. And the poem ends on a word that will not resolve: still means calm, and it means motionless, and applied to a mighty heart it carries the possibility of arrest. Smokeless is the same kind of word, beautiful and abnormal at once.

That is a strong reading. It stays inside the fourteen lines, it accounts for the syntax and the turn, and it finds a real ambiguity at the close.

The second reading: what the sonnet leaves out

The second critic starts with a question the first cannot ask. Why is the air smokeless?

Because it is about five in the morning and the coal fires are not lit. London in 1802 had upwards of nine hundred thousand people, and its air was famous for being unbreathable. The poem catches the city in the one hour when the labour that produces the smoke has not begun. Its beauty is not a property of the city; it is a property of the city with the work stopped.

Then the list. Ships, towers, domes, theatres, temples: a fleet, the offices of finance, the church, the entertainment of the rich. The ships on that river carried sugar and cotton, and Parliament would not abolish the slave trade for another five years. In the sonnet these are scenery, arranged for the eye of a man on a coach roof, and the people who built them, sailed them, cleaned them and were traded through them are not merely absent. Their absence is the condition of the view.

Finally the figure at the close. A mighty heart makes the city an organism, something that beats by nature. A city is not an organism. It is a set of relations among people, and calling it a heart converts a human arrangement into a fact of biology, which is precisely the operation that Georg Lukacs called reification: a relation between people takes on the character of a thing.

In short: the second critic is not accusing Wordsworth of anything. The claim is about what the poem's form accomplishes, and the claim is checkable against the poem, which is what makes it criticism rather than complaint.

Where the dispute actually is

Lukacs, and the argument about which form tells the truth

Lukacs made reification the centre of History and Class Consciousness in 1923, and then spent decades applying it to the novel. His literary position, stated most sharply in 1957, is that nineteenth-century realism, Balzac, Scott, Tolstoy, gives a truer account of the world than modernism, because the realist novel shows individuals as products of historical forces, and its typical characters embody tendencies of a whole society. Modernism, on his account, presents the human being as solitary, ahistorical and fixed, which mistakes the experience produced by capitalism for the nature of humanity.

Brecht answered that Lukacs had turned realism into a formalism of its own, prescribing the shape of the 1840s novel as an eternal standard and mistaking a set of techniques for reality itself. Realism, in Brecht's version, means whatever lays bare the causal network of society, and that may require montage, interruption and estrangement rather than a well-made plot. The quarrel is not a footnote. It is the clearest case of two Marxists deriving opposite aesthetic verdicts from the same politics, which tells you the connection between politics and form is not a deduction.

Althusser and Macherey: reading the silences

Louis Althusser redefined ideology as the imaginary relationship of individuals to their real conditions of existence, lived rather than believed, and reproduced through institutions such as schools, churches and families. That removes the crude picture of writers consciously serving a class interest and replaces it with something harder to see and harder to escape.

Pierre Macherey drew the critical conclusion in 1966. A text is not a unified whole with a hidden message; it is structured by what it cannot say, and its silences are not gaps in a full statement but the shape ideology leaves in it. The critic's job is not to restate the work more elegantly, which is what appreciation does, but to produce knowledge of it by locating the places where it fails to be seamless. Notice that this is an exact answer to the New Critical ideal of organic unity: for Macherey the unity is the illusion, and the seams are the evidence.

Williams: the base was never a floor

Raymond Williams spent his career dismantling the model of a base that determines a superstructure, on the ground that it treats culture as a reflection of something more real, when culture is one of the ways the social order is produced in the first place. His alternative, cultural materialism, treats writing as a material practice among others.

Two of his terms are worth having. A structure of feeling is the lived quality of a period as it is being lived, before it has hardened into doctrine: what people in 1802 felt about cities, in the form of unease and enthusiasm rather than argument. And any moment contains dominant, residual and emergent elements at once, which is why a period never has a single ideology.

His most useful book for this lesson is The Country and the City, of 1973, which reads the English country-house poem from Jonson onward and notices that the estate is always presented as a place where food arrives of itself, fruit offering itself to the hand, and the labourers who grew it are missing. That is precisely the operation the second critic found in the Westminster Bridge sonnet, and Williams demonstrates it across two centuries of poems, which turns a suspicion into a finding.

Jameson: form is sedimented content

The Political Unconscious opens in 1981 with two words: "Always historicize!" Fredric Jameson then makes a claim that goes beyond content. Genres and forms are not neutral containers; they carry the traces of the social situations in which they were developed, so a form is itself an ideological act. Narrative, on his account, works as a socially symbolic act, offering imaginary resolutions to contradictions that cannot be resolved in life.

Applied to our sonnet: the octave and sestet structure exists to stage an argument and resolve it, and a form built for resolution, applied to a city built on trade and slavery, will produce a resolution whether or not one is available. The form is doing ideological work before the poet chooses a word.

The formalist reply, at full strength

Three objections, in ascending order of force.

First, portability. Almost everything in the second reading could have been written from a history of London without opening the poem. A reading that does not need its text is a lecture with a quotation attached.

Second, over-prediction. If every text produced under capitalism conceals labour, then finding concealed labour in one text tells you nothing about that text. Terry Eagleton concedes the force of this in Criticism and Ideology, which is why he goes to Althusser and Macherey for a model in which the text works on ideology rather than reflecting it.

Third, the record. Deriving politics from form has failed repeatedly, and the Lukacs and Brecht quarrel is the demonstration: two committed Marxists, opposite verdicts on modernism, and the twentieth century did not vindicate either prediction.

What would settle it

Parts of this are testable. The claim that the poem's beauty depends on the absence of work can be checked by reading the poems where the work is present, and Williams's book does exactly this at scale. If country-house poems and city sonnets turned out to include labourers as often as they exclude them, the claim would be in trouble.

What cannot be settled by evidence is the prior question of what criticism is for. If it exists to describe how a made thing works, the first reading is complete and the second is background. If it exists to explain what literature does in the world, the first reading is a description of the mechanism with the question of its purpose ruled out. Both critics can be right about the poem and still disagree about everything.

Common misconceptions

• Marxist criticism judges writers by their politics. Lukacs's favourite novelist, Balzac, was a royalist. His argument is that Balzac's realism records the society accurately in spite of his opinions, which is a claim about form, not allegiance.
• It means literature reflects the economy. That reflection model is the one Althusser, Macherey, Williams and Jameson all reject. The text works on ideology, or produces it, or fails to say something; none of that is a mirror.
• Ideology means false beliefs people could correct. In Althusser's version it is a lived relation, reproduced through institutions, and not primarily a matter of what anyone consciously thinks.
• Cultural materialism is just historical background. Williams's argument is that culture is not the decoration on top of material life but part of how material life is made, which is why he objected to the base and superstructure picture in the first place.

Where this leaves us

You now have both readings of one sonnet and the machinery behind each. The formalist can tell you how the poem is put together and cannot tell you why it is put together that way. The Marxist can tell you what the poem does in the world and is always at risk of not needing the poem at all.

The best work in this tradition, Williams on the country house, Jameson on genre, is the work that pays the formalist's price: it stays with the details of particular texts long enough that the historical claim could have come out otherwise.

Why this matters: a historical reading earns its authority the same way a formal reading does, by being answerable to something that could have turned out differently. Ask of any political reading what the text would have to look like for the reading to fail.

Sources

1. Wordsworth, W. (1807). Composed upon Westminster Bridge. In The Poetical Works of William Wordsworth (Vol. 2). Project Gutenberg
2. Lukacs, G. (1923). Reification and the consciousness of the proletariat. In History and Class Consciousness. marxists.org
3. Wikipedia. Structure of feeling. en.wikipedia.org
4. Williams, R. (1973). The Country and the City. Oxford University Press.
5. Jameson, F. (1981). The Political Unconscious: Narrative as a Socially Symbolic Act. Cornell University Press.

Five names on a blackboard

In the summer of 1971 Stanley Fish was teaching two courses in the same room at a summer linguistics institute in Buffalo. The morning class was linguistics, and at the end of it he had written a list of names on the board, an assignment of authors his students were to read:

Jacobs-Rosenbaum
Levin
Thorne
Hayes
Ohman

The afternoon class was on English religious poetry of the seventeenth century. Fish drew a frame around the list, added a page reference to make it look like an assignment, told the students it was a religious poem of the kind they had been studying, and asked them to interpret it.

They did. Rosenbaum contains rose and, in German, tree, so the rose tree, so the Virgin. Jacobs gives Jacob's ladder, the ascent from earth to heaven. Thorne gives the crown of thorns. Ohman, read aloud, is an address: O man. Within minutes the class had produced a reading of the ascent to God through Christ's suffering, complete with observations about the poem's shape on the page, the ladder-like descent of the single words, and the way the final line turns outward to the reader.

These were good students working honestly. Fish's conclusion, published in Is There a Text in This Class in 1980, is the one that makes this lesson necessary: the students did not find those features in the poem. Their interpretive strategies made both the features and the poem, because they were asked to be readers of religious poetry, and readers of religious poetry are people for whom a rose is a Marian symbol before they have decided anything.

Where the argument started: Milton's traps

Fish had arrived at this by way of Milton. In Surprised by Sin, of 1967, he argued that Paradise Lost is designed to catch its reader out. Here is Satan in the first book:

Here at least
We shall be free; th' Almighty hath not built
Here for his envy, will not drive us hence:
Here we may reign secure; and, in my choice,
To reign is worth ambition, though in Hell:
Better to reign in Hell than serve in Heaven.

Read it and you will feel the pull. The rhythm is magnificent, the sentiment is defiant liberty, and the last line is one of the few in English poetry that people who have not read the poem can quote. Then the narrator's voice returns to describe the speaker as vaunting aloud, though racked with deep despair, and the reader who was moved is caught.

Fish's claim was that this is the poem's method throughout: it seduces and corrects, so that the experience of reading it re-enacts the fall. The important move for theory is the one underneath. The right question about a line of poetry, Fish said, is not what does this mean but what does this do, in time, to a reader moving left to right through it. He called it affective stylistics.

The point: if meaning is an event that happens to a reader rather than a content stored in a text, then the sequence of a reader's responses is not evidence about the poem, it is the poem.

The fallacy, revived on purpose

You met the objection two lessons ago. Wimsatt and Beardsley had named this the affective fallacy in 1949 and treated it as the road to impressionism: readers vary, so a criticism built on response has no object.

Fish's answer is that the New Critics' alternative was never available. There is no way to describe a poem's formal features without already reading, and reading is something a trained person does in a particular way. The blackboard experiment is the demonstration: the students' account of the poem's shape was not false to what was on the board, and there was no poem on the board.

Iser: the reader completes what the text leaves open

A gentler version of the same insight had been developed at Constance. Wolfgang Iser, building on Roman Ingarden, observed that no text specifies everything. A novel tells you a woman entered a room; it does not tell you her gait, the light, whether she had decided anything on the stairs. Ingarden called these places of indeterminacy, and said the reader concretises them.

Iser makes the gaps constitutive rather than incidental. A narrative is a set of views arranged in sequence with blanks between them, and the reader's wandering viewpoint builds connections across the blanks, revising as it goes. What a text controls is not the reader's conclusions but the pattern of invitation and frustration that produces them. His implied reader is not a person but a position the text constructs: the set of responses the text is built to elicit.

The difference from Fish is exactly here, and their public argument in 1981 turned on it. For Iser, the gaps are in the text, so the text retains authority and the reader's freedom is real but bounded. Fish's reply was that Iser cannot have this: if the determinate part of the text is only determinate to a reader who already knows how to read it, then the line between what is given and what is supplied cannot be drawn, and Iser has kept a text-in-itself he is not entitled to.

Jauss: what a period was ready to read

The third position changes the timescale. Hans Robert Jauss, in his Constance inaugural lecture of 1967, asked why literary history had become a list of authors and dates that nobody read for pleasure or instruction, and answered that it had left out the readers.

His central concept is the horizon of expectations: the assumptions about genre, form, subject and morality that a work meets on arrival, which are historically specific and reconstructable from reviews, sales, imitations and quarrels. A work can confirm that horizon, in which case it is enjoyed and forgotten, or it can violate it, in which case it produces what Jauss calls aesthetic distance, and the gap between the work and what readers expected is the measure of its novelty.

His demonstration is a comparison. Madame Bovary and Ernest Feydeau's Fanny were published in the same period and treated similar material, and Fanny was by far the more successful with contemporary readers, while Flaubert's novel was prosecuted and then, over decades, became the model of the modern novel. Nothing in either book changed. The horizon moved, and Flaubert's impersonal narration, which readers in 1857 found cold and immoral, became the thing a novel was supposed to do.

That gives literary history a subject matter again: not a sequence of works but a sequence of receptions, in which the meaning of a work includes the history of the questions readers have brought to it.

If the reader makes the meaning, what stops anything going?

This is the objection everyone raises, and Fish's answer is the most interesting thing he wrote. You are never free, he says, because you never read as an individual. You read as a member of an interpretive community, a group that shares strategies for constituting texts. The community produces the reader and the text together, which is why the students in Buffalo all produced roughly the same poem rather than twenty different ones.

So the constraint is real, but it is not the text and not the individual. It is a set of professional and cultural habits that make some readings obvious, some available with argument, and some literally unthinkable. On this account a reading is wrong not because it violates the words but because no community would accept it, and Fish is candid that this makes correctness a social fact.

Three objections

First, change. If communities determine what their members can see, how does anyone ever produce a reading their community does not already hold? Fish answers that communities are not monolithic and that readers belong to several at once, which many critics regard as a description of the problem rather than a solution.

Second, surprise. Readers report being brought up short by texts, corrected by them, made to abandon a reading they arrived with. Iser's model explains this easily, since the text sets the pattern of frustration. Fish must treat the experience of being corrected by a text as itself an artefact of community norms, which is possible but strained.

Third, practice. Fish taught, examined and marked students, and marking assumes some answers are better than others in a way that survives disagreement. His reply is that professional consensus is exactly what marking measures, which is consistent, and which concedes that the theory cannot distinguish between a reading being right and a reading being accepted.

Common misconceptions

• Reader response means every interpretation is valid. Fish's whole point about interpretive communities is that constraint is total: you cannot read otherwise than as your training permits. What he denies is that the constraint comes from the text.
• Iser and Fish hold the same position. Iser keeps a determinate text whose gaps the reader fills; Fish denies that the determinate part can be isolated. Their exchange in 1981 was pointed for this reason.
• Reception study is just tracking a book's popularity. Jauss reconstructs the horizon a work met, so that the gap between expectation and work becomes measurable evidence about the work, not only about its audience.
• The school abolished close reading. Fish's Milton readings are line-by-line and slow, and Iser's analyses of eighteenth-century novels are minutely textual. The change is in what the description is a description of.

What you now know

Three positions, in order of how much authority they leave the text. Iser: the text sets the blanks, the reader fills them, and the interplay is the work. Jauss: the work's meaning includes the history of its receptions, and novelty is measured against a reconstructable horizon. Fish: the strategies precede the text, and the text is what a community's strategies make.

The lasting effect is that no one now writes as though a poem simply contains its meaning, waiting. Even critics who reject Fish accept that a reading is performed by someone trained to perform it, which is why the next module can ask a question that would have looked improper in 1949: what happens to the reading when the person doing it is not the reader the text expected?

Worth holding on to: the useful version of this school is not that meaning is subjective. It is that description is never innocent, so a critic who reports what a text obviously does should be able to say who the obviousness belongs to.

Sources

1. Milton, J. (1667). Paradise Lost. Project Gutenberg
2. Wikipedia. Reader-response criticism. en.wikipedia.org
3. Wikipedia. Reception theory. en.wikipedia.org
4. Fish, S. (1980). Is There a Text in This Class? The Authority of Interpretive Communities. Harvard University Press.
5. Iser, W. (1978). The Act of Reading: A Theory of Aesthetic Response. Johns Hopkins University Press.

A story an editor did not want to print

Charlotte Perkins Gilman sent The Yellow Wallpaper to Horace Scudder at the Atlantic Monthly, and he returned it. By her account he wrote that he could not forgive himself if he made others as miserable as he had made himself reading it. It appeared instead in The New England Magazine in January 1892.

Gilman had material. After the birth of her daughter she had been treated by the neurologist S. Weir Mitchell with his rest cure, and she says his instruction was to live as domestic a life as possible and never to touch pen, brush or pencil again. She wrote the story, she said in 1913, not to drive people crazy but to save people from being driven crazy.

The narrator of the story is a woman confined to an upstairs room for a rest cure by her husband, who is a physician. Early on she writes: 'John is a physician, and perhaps ... perhaps that is one reason I do not get well faster. You see, he does not believe I am sick!' By the end she is on the floor, having torn the paper off the wall, and her husband has fainted across the doorway: 'I have got out at last, in spite of you and Jane! And I have pulled off most of the paper, so you cannot put me back!'

Three different feminist projects have read that story. They are not three flavours of one method. They ask different questions, use different evidence, and produce claims that could contradict each other.

The three projects

Kate Millett's book opens by reading sex scenes from Henry Miller, Norman Mailer and D. H. Lawrence as scenes of power, and it treats the reader's discomfort as evidence rather than as bad manners. Judith Fetterley pushed this further in 1978: the woman reading the American canon is asked to identify against herself, and the first act of a feminist critic is to become a resisting reader.

Elaine Showalter named the second project gynocriticism and defined it as the study of women as writers, with its own subjects: the history, styles, genres and careers of women's writing, and the tradition running through them. She proposed a three-phase chronology for the English novel: an imitative phase, a phase of protest, and a phase of self-discovery. The work it produced is mostly archival, and its results are physical: books back in print.

Adrienne Rich gave the third project its name in a 1972 lecture. Re-vision, in her definition, is the act of looking back, of seeing with fresh eyes, of entering an old text from a new critical direction, and she treats it as an act of survival rather than of scholarship.

Key idea: critique reads the texts we have, gynocriticism finds the texts we did not have, and re-vision reads a text we have as if from the position it was not written for. Confusing them produces most of the bad arguments in this area.

Reading the table on one story

Now run each project over Gilman's story.

Critique takes John. He is a physician, and the story gives him the whole apparatus of authority: diagnosis, prescription, the choice of room, the power to decide what his wife may write and whom she may see. He calls her blessed little goose. He laughs at her, which she notes, and immediately adds that one expects that in marriage. The critique reading shows how a discourse that presents itself as care operates as control, and it does not need to establish anything about Gilman's opinions to do it: the evidence is the distribution of speech and decision in the text.

Gynocriticism takes the story's own history. It was published, anthologised once as a horror tale, and then effectively disappeared. It came back into general circulation in 1973, when the Feminist Press reissued it with an afterword by Elaine Hedges. What gynocriticism found was not a hidden meaning but a missing book, and its characteristic argument is that the shape of the canon is a fact about publishing, reviewing and syllabus-making rather than about quality.

Re-vision takes the ending. Read as a case history, the last page is a woman's final collapse into psychosis. Read from a different position, it is the only successful escape in the story: she has got out, she has destroyed the surface that held the trapped woman, and the man who confined her is on the floor. The line to hold is in spite of you and Jane, since Jane may be the narrator's own name, in which case what she has escaped is the self that agreed to be managed.

The madwoman, and the anxiety of authorship

The Madwoman in the Attic, published in 1979 by Sandra Gilbert and Susan Gubar, is the book that made the third and second projects into one argument.

Their claim starts from Harold Bloom, who had described literary history as a struggle in which a poet must overcome a precursor: the anxiety of influence. Gilbert and Gubar answer that a nineteenth-century woman writer had a different problem. Not the fear that her predecessors were too strong, but the fear that writing at all was improper, since the available images of women in literature were the angel, selfless and silent, and the monster, appetite and voice. They call this the anxiety of authorship.

The consequence they trace through Austen, the Brontes, Eliot and Dickinson is a doubled text: a proper surface and a rebellious figure who acts out what the heroine cannot. Bertha Mason, locked upstairs in Jane Eyre, burning the house and tearing the veil, is the type case, which is where the book gets its title. In Gilman's story the doubling is on the wall, in the figure of the woman behind the pattern who is trying to get through, and the narrator ends by joining her.

The thesis is elegant and the evidence is enormous, six hundred pages of it. Both facts matter to the objections.

Woolf, and the material argument

Behind all three projects stands A Room of One's Own, published in 1929 from lectures given at two Cambridge women's colleges. Virginia Woolf's method is to invent Shakespeare's equally gifted sister Judith and follow what would have happened to her: no grammar school, betrothal at her father's choice, laughter at the stage door, pregnancy, suicide. The point is not that women could not write but that writing requires conditions, and Woolf names them with a number: five hundred pounds a year and a room with a lock.

That argument is materialist in the way Module 3 would recognise, and it is why the feminist tradition has been comfortable with historical criticism and uncomfortable with the New Critical rule that the conditions of production are inadmissible.

Three objections

The first is theoretical. Toril Moi argued in 1985 that Anglo-American feminist criticism had stayed naively realist: it judged texts by whether they gave true or positive images of women, and it assumed a unified female self doing the writing. The French tradition, Cixous, Irigaray and Kristeva, locates the trouble in language and the subject rather than in representation, and asks what a writing that does not obey the existing order of the sentence would look like. Whether that is a deeper analysis or an evasion of politics is still argued.

The second is the sharpest, and it is about the madwoman. Bertha Mason is a Creole woman from Jamaica whose family money is what makes Rochester's position possible. Reading her as the white heroine's psychological double turns a colonised woman into a symbol of someone else's repression. Gayatri Spivak made this argument in 1985, insisting that the novel cannot be read without the imperial history that supplies its plot, and Hazel Carby made a parallel case about the exclusions built into the category of womanhood the tradition was defending. Barbara Smith had already argued in 1977 that a criticism which treats black women's writing as a special case is not a general criticism at all. Jean Rhys had done it in fiction in 1966 by giving Bertha a childhood, a name and a first-person voice.

The third is the essentialism problem, and it is the hinge into the next lesson. Gynocriticism needs the category women's writing to be stable enough to have a history. If the category is itself produced by the arrangements the criticism opposes, the ground moves, and the next lesson is about the theorists who moved it.

Common misconceptions

• Feminist criticism means looking for strong female characters. None of the three projects does that. Critique reads power in the text's distribution of speech, gynocriticism works in archives, and re-vision changes the reader's position rather than the text's cast.
• It applies only to writing by women. Millett's book is almost entirely about male novelists, and Fetterley's method is designed for the male-authored canon.
• The Madwoman in the Attic argues that these writers were secretly angry women. The claim is about a structural problem in the position of the woman writer, and the evidence is a repeated textual pattern of doubling, not biography.
• The postcolonial objection is that Gilbert and Gubar were careless. Spivak's argument is stronger: the reading works, and what makes it work is a use of Bertha that repeats the novel's own imperial logic.

Summing up

Three projects, one story, three findings that do not collapse into each other. The critique reading shows a discourse of care operating as control. The gynocritical reading explains why you had probably not read the story before about 1975. The re-visionary reading turns the last page from a collapse into an escape.

The tradition's strength is that its claims are checkable: publishing records, syllabi, the distribution of speech in a text, the fate of a story between 1892 and 1973. Its exposure is that the category it organises itself around has been contested from inside, first by black feminist critics, then by the theorists of the next lesson, and the contest is not finished.

The upshot: ask of any feminist reading which of the three projects it belongs to, because the objection that defeats one of them frequently does not touch the others.

Sources

1. Gilman, C. P. (1892). The Yellow Wallpaper. Project Gutenberg
2. Wikipedia. Gynocriticism. en.wikipedia.org
3. Wikipedia. The Madwoman in the Attic. en.wikipedia.org
4. Rich, A. (1972). When we dead awaken: Writing as re-vision. College English, 34(1), 18-30.
5. Gilbert, S. M., and Gubar, S. (1979). The Madwoman in the Attic: The Woman Writer and the Nineteenth-Century Literary Imagination. Yale University Press.

Fourteen lines addressed to a man

Shakespeare's sonnets were printed in 1609. The first hundred and twenty-six are addressed to a young man. Here is number twenty, in the old spelling of the quarto lightly modernised:

A woman's face with nature's own hand painted,
Hast thou, the master mistress of my passion;
A woman's gentle heart, but not acquainted
With shifting change, as is false women's fashion:
An eye more bright than theirs, less false in rolling,
Gilding the object whereupon it gazeth;
A man in hue all hues in his controlling,
Which steals men's eyes and women's souls amazeth.
And for a woman wert thou first created;
Till Nature, as she wrought thee, fell a-doting,
And by addition me of thee defeated,
By adding one thing to my purpose nothing.
But since she prick'd thee out for women's pleasure,
Mine be thy love and thy love's use their treasure.

This lesson runs a procedure over that sonnet and its neighbours, in four steps, and then stops to ask whether the procedure has any business being there at all. That last step is not politeness. It is the strongest objection in the field, and it comes from inside.

Before the procedure: what the sonnet does

Start with a formal fact anyone can check. Every one of these fourteen lines ends on an unstressed extra syllable: painted, acquainted, passion, fashion, rolling, controlling, gazeth, amazeth, created, defeated, doting, nothing, pleasure, treasure. No other sonnet in the sequence is entirely in feminine rhymes. A poem about a man with a woman's face is written in the rhyme that criticism of the period called feminine, all the way through.

Then the argument. The young man has a woman's face and a woman's heart without what the speaker considers women's faults. Nature made him as a woman, fell in love with her own work, and added one thing, which defeats the speaker. The couplet then divides the man in two: the speaker keeps his love, and women may have the use of his body.

That division is doing a great deal of work, and it is where every reading of the sonnet begins, because you can take it as a disavowal of physical desire or as a joke that only makes sense if the desire is on the table.

Step one: find the triangle

The procedure comes from Eve Kosofsky Sedgwick's Between Men, of 1985. She takes from Rene Girard the observation that literary love triangles are not what they appear: the bond between the two rivals is at least as intense, and as determining, as either one's bond to the beloved. Girard treated this as a fact about desire in general. Sedgwick asks what happens when you notice that in most of these triangles the two rivals are men and the third party is a woman.

So step one is mechanical. Find a triangle in the text. In the sonnets you do not have to look far. In numbers forty to forty-two, and again in the hundred-and-thirties, the young man and the woman the speaker calls his mistress are involved with each other. Sonnet forty-two states the arrangement precisely: That thou hast her it is not all my grief, and yet it may be said I loved her dearly; that she hath thee is of my wailing chief, a loss in love that touches me more nearly.

Read that slowly. He has lost both of them, and the greater grief is the loss of the man.

Step two: ask what the third party is for

Step two asks what role the woman plays in the structure. Sedgwick borrows here from Gayle Rubin's account of the traffic in women, in which exchanges of women among men are how alliances between men are made and ranked. In the sonnets the mistress is the medium through which the speaker and the young man are connected, blamed, forgiven and connected again. Sonnet forty-two ends by concluding that since the friend and the speaker are one, the woman loves the speaker alone, which is a piece of arithmetic that only works if the man matters more than the woman.

Key idea: the triangle is not a love story with a rival in it. It is a relationship between two men that requires a woman in order to be conducted, and the method's first payoff is that it makes the woman's structural position visible rather than treating her as a plot device.

Step three: locate the discontinuity

Step three is Sedgwick's own contribution and the one that makes the method more than a diagram. She proposes the term homosocial for the whole range of same-sex bonds, from friendship to mentorship to rivalry, and then observes an asymmetry. For women in modern Western culture, the range from friendship to sexual love has been treated as a continuum: nothing catastrophic happens when a woman's affection for a woman is described in warm terms. For men it is not a continuum. It is broken by a sharp and policed line, and the policing is homophobia, which functions to regulate all relations between men and not only the sexual ones.

Apply that to the couplet. The joke about Nature adding one thing performs the break in a single line: it names the body, declares it irrelevant to the speaker's claim, and assigns it to women. That is exactly the manoeuvre Sedgwick's model predicts in a culture where intense male bonds are both required and dangerous, and it is why the sonnet needs a joke at that point rather than a statement.

Step four: refuse to rename, and historicise

Now the step that stops the procedure from becoming a machine for outing dead poets.

Michel Foucault argued in 1976 that sexuality is not a natural fact that societies repress but a discourse that produces the categories through which people understand themselves. His famous claim is that the sodomite had been a temporary aberration, whereas the homosexual became a species, and he dates the transformation to the psychiatric writing of the later nineteenth century. Alan Bray's Homosexuality in Renaissance England, in 1982, established the historical version of the same point: Elizabethan England had a fierce legal and theological category of sodomy, applied to acts, and no notion of a person defined by a stable orientation. Men shared beds, wrote love letters and called each other bedfellow without any of it meaning what a modern reader assumes.

So the question is the sonnet gay is malformed. The productive question is what arrangements of desire, and what limits, the poem is working inside, and what it has to joke about in order to say what it says. That is why the field settled on the word queer, reclaimed as a term of analysis around 1990: it names a relation to norms rather than a type of person.

So what?: the procedure is only as good as step four. Steps one to three describe a structure; without the historical question they describe it in the wrong vocabulary and call the anachronism a discovery.

The closet as a structure of knowledge

Sedgwick's Epistemology of the Closet, in 1990, generalises the argument. Her claim is that an unresolved crisis over homosexual and heterosexual definition, dating from the end of the nineteenth century, structures a whole range of modern oppositions that appear to have nothing to do with sexuality: knowledge and ignorance, private and public, natural and artificial, health and illness, innocence and initiation.

She sorts the available positions into two. A minoritizing view treats the question as one about a distinct minority of people. A universalizing view treats it as an issue for everyone, since the definitional crisis shapes everyone's identity and knowledge. Both, she argues, are in active use, often by the same people in the same argument, and the incoherence is not an error to be tidied up but the shape of the field.

Notice the epistemology in the title. The closet is not primarily a metaphor for concealment; it is a structure in which knowing and not knowing are distributed and enforced, and it can be read in texts by tracking who is permitted to know what.

Butler: performativity, and what it is not

Judith Butler's Gender Trouble appeared the same year. Its central move takes the deconstructive procedure from Module 2 and runs it on the pair sex and gender. If gender is the cultural meaning given to a natural sex, then sex is the ground and gender the interpretation. Butler argues that the ground is itself produced: what counts as a natural body, and which features of it matter, is already an effect of the system that the distinction was supposed to explain.

From there, performativity. Gender is not something one is, expressed through acts; it is constituted by the repeated acts themselves, and the appearance of an inner essence is the effect of the repetition rather than its cause. There is no doer behind the deed. Drag interests her not because gender is a costume but because a good impersonation exposes that the original was also an imitation, of an ideal nobody embodies.

The misreading was immediate and Butler answered it in 1993: performativity is not performance, and nobody chooses their gender in the morning the way one chooses a shirt. The repetition is compelled, sanctioned and largely unconscious, and its constraint is the point.

Objections

Three, of unequal force. The anachronism objection is the serious one, and step four exists to answer it; critics still disagree about whether the answer is sufficient when the vocabulary of analysis remains modern.

The second is the charge of over-application. A method that finds a broken homosocial continuum in every male friendship in literature has stopped being informative, and the test is the one this course keeps applying: could the reading have been written without the text.

The third comes from inside the field. Cathy Cohen argued in 1997 that queer theory's central opposition, queer against straight, obscures how race and class distribute normativity, since a poor black single mother is punished by the same norms while being nobody's idea of queer. Sedgwick herself, in the same period, began to doubt the exposure model of criticism, and that doubt is where this course goes in the lesson on post-critique.

Common misconceptions

• Queer theory claims historical figures were gay. The Foucault and Bray argument runs the other way: the modern category did not exist, so the interesting question is what arrangements a text works inside, not what label its author would take.
• Homosocial is a polite word for homosexual. Sedgwick coined it precisely to keep them apart, so that the relation between them, continuous for women, broken for men, becomes the object of study.
• Performativity means gender is a free choice. Butler's clarification is explicit: the acts are compelled and sanctioned, and treating performativity as voluntarist gets the argument backwards.
• Sonnet 20's couplet settles the question of the speaker's desire. The couplet divides love from bodily use, which can be read as a disavowal or as a joke that presupposes what it disavows. The poem does not settle it, and criticism that says otherwise has usually decided first.

Pulling it together

You now have a procedure with four steps: find the triangle, ask what the third party is for, locate the point where the continuum is broken, and historicise the vocabulary before naming anything. Run on the sonnets, it produces a reading in which the mistress is a medium, the greater grief is the loss of the man, and the couplet's joke is doing structural work.

You also have two larger claims to argue with: Sedgwick's, that a crisis of sexual definition organises oppositions that look unrelated to sex, and Butler's, that the natural ground beneath a cultural category is produced by the same system that the category was invented to explain. Neither is provable in the way a dating of a manuscript is provable. Both are testable in the weaker sense that matters here, which is that they predict where to look and can be wrong about what is found.

Worth holding on to: the useful queer reading is not the one that finds desire where nobody had noticed it. It is the one that shows what a text has to do, formally, in order to keep a relation sayable.

Sources

1. Shakespeare, W. (1609). Shakespeare's Sonnets. Project Gutenberg
2. Wikipedia. Queer theory. en.wikipedia.org
3. Wikipedia. Epistemology of the Closet. en.wikipedia.org
4. Sedgwick, E. K. (1985). Between Men: English Literature and Male Homosocial Desire. Columbia University Press.
5. Butler, J. (1990). Gender Trouble: Feminism and the Subversion of Identity. Routledge.

A speech in the Commons, June 1910

Arthur Balfour rose in the House of Commons to explain why Britain was in Egypt. His argument was not about strategy or trade. It was about knowledge: the British, he said, knew Egyptian civilisation better than they knew any other, better, by implication, than Egyptians knew it themselves, and that knowledge was itself the qualification for governing. Two years earlier, Lord Cromer, who had run Egypt for a quarter of a century, published a book explaining that the mind of the Oriental was picturesque but inaccurate, and that his reasoning was of the most slipshod description.

These are not obscure documents. They were written by a former prime minister and a proconsul, for public consumption, and they treat a claim about how a whole population thinks as a settled matter requiring no evidence.

Edward Said put them at the front of a book in 1978 and asked what kind of statement they are.

How the book came to be written

Said was born in Jerusalem in 1935, raised in Cairo, and educated at an English-language school there before being sent to the United States at fifteen. He took a doctorate at Harvard on Joseph Conrad and joined the Columbia English department in 1963 as a comparatist of the entirely conventional kind. By his own account, the war of 1967 changed what he could ignore: he found that the region he came from existed in American public discourse only as a set of stock figures, and that the figures had a long and respectable literary pedigree.

Orientalism, published in 1978, is the result. Its thesis is stated early and is worth having exactly: Orientalism is a Western style for dominating, restructuring and having authority over the Orient. Three things in that formulation do the work. It is a style, so it is formal and describable. It is Western, so it tells you about Europe rather than about Asia. And it has authority as its purpose, which is a claim about function, not about sincerity.

In short: Said's argument is not that European writers said unkind things about the East. It is that a body of scholarship, fiction, travel writing and administration produced a coherent object, the Orient, that could then be known, taught and governed, and that the object was largely an artefact of the producing.

The archive

The book's evidence is a two-hundred-year run of texts treated as one field. Silvestre de Sacy organising Arabic into a curriculum in Napoleonic Paris. Edward William Lane living in Cairo in the 1830s and publishing an account of the manners and customs of the modern Egyptians that remained a reference work for a century. Ernest Renan on the Semitic languages, deriving a racial psychology from a grammar. Then the administrators, Cromer and Balfour, using the scholarship as a warrant.

Said distinguishes what he calls manifest and latent Orientalism. The manifest layer is the stated content, which changes with fashion and evidence: what a given writer claims about Egyptian irrigation or Arabic poetry. The latent layer is the set of assumptions that does not change across two centuries: that the Orient is unchanging, sensual, despotic, feminine and incapable of representing itself. Writers who disagree violently at the manifest level share the latent one, and that stability across disagreement is his central piece of evidence.

Foucault plus Gramsci, and the trouble it makes

Said names his tools. From Foucault he takes discourse: a regulated body of statements that produces its object and the positions from which one may speak about it. From Gramsci he takes hegemony: rule that works by consent and cultural leadership rather than by force alone.

These do not fit together comfortably, and his critics noticed at once. Foucault's discourses have no authors and no outside, while Said keeps naming individual writers, praising some for their honesty and blaming others, which requires a humanist standard of truth that Foucault's framework denies him. James Clifford put this in a review in 1980: to say that the Orient is a fiction produced by the West seems to require knowing what the real Orient is, and Said explicitly refuses that knowledge.

Said's answer, roughly, is that the contradiction is real and productive, and that a criticism which cannot say some representations are more answerable to their subjects than others is useless. Whether that is an answer or an admission is still argued.

Contrapuntal reading: a dead silence at Mansfield Park

In Culture and Imperialism, in 1993, Said proposed a method for reading canonical works, which he called contrapuntal after the musical term: hold the metropolitan narrative and the imperial history it depends on in the ear at once, neither cancelling the other.

His example is Mansfield Park. Sir Thomas Bertram's household in Northamptonshire is funded by his estate in Antigua, which in 1814 means sugar and slave labour. He leaves for Antigua because things there are going badly, and the novel's moral crisis at home happens in his absence. When he returns, Fanny Price asks him about the slave trade, and Austen records the response in five words: there was such a dead silence.

Said's reading does not accuse Austen of anything, and it does not claim the novel is secretly about Antigua. It claims that the domestic order whose fine discriminations the novel exists to examine is materially produced elsewhere, that Austen knows this and puts it in the book, and that a criticism which treats Antigua as background has decided in advance which parts of a novel are its subject.

You can test the method by asking what it would take to refute it: show that the Antigua estate is incidental to the plot, or that the silence carries no weight in a novel otherwise minutely attentive to what people say.

The objections to Said

Bernard Lewis replied in 1982 that the book was full of factual errors, that it ignored the German scholarly tradition, which had no colonies to serve, and that treating a whole discipline as an ideological function insults the many philologists whose work was simply good. Robert Irwin extended the case in 2006 in book length.

The sharper objection came from the left. Aijaz Ahmad argued in 1992 that Said's method traps him: by reading imperialism out of a canon of European literary texts, he makes literature the primary site of colonial power and leaves out the economic history, the anticolonial movements and the actual resistance, so the colonised appear only as an effect of European discourse. Ahmad also pointed out that a book claiming Europe silences the Orient is itself built almost entirely from European sources.

Spivak: who can be heard

Gayatri Chakravorty Spivak, who had translated Derrida's Of Grammatology in 1976, took up exactly that gap. Her essay Can the Subaltern Speak works through the British prohibition of widow immolation in India, abolished in 1829.

Two accounts of that episode were available then and remain available now. The colonial account: white men are saving brown women from brown men. The nativist account: the women themselves wanted to die. Spivak's point is that both are constructed by men in positions of authority, and that between them there is no place from which the widow's own account could have been recorded, because the archives that survive are the archives of the two contending patriarchies.

Her case study is her own great-aunt. In 1926 Bhubaneswari Bhaduri, a young woman in Calcutta, hanged herself. She waited until she was menstruating, so that the obvious reading, an illicit pregnancy, would be unavailable. Decades later Spivak learned that she had belonged to an armed group in the independence struggle and had been unable to carry out an assigned killing. The suicide was composed as a message, and the family read it as a love affair anyway. That, Spivak says, is what she means by the subaltern cannot speak: not that such people are silent, but that the speech act does not complete, because there is no listening position that will receive it as what it is.

She also gave the field a term she later regretted, strategic essentialism: the temporary acceptance of an essentialist identity for a political purpose, which she came to think was being used as a licence for exactly the essentialism it was supposed to suspend.

Bhabha: Macaulay's class of interpreters

In 1835 Thomas Babington Macaulay wrote a minute on Indian education, and the government acted on it. He argued that a single shelf of a good European library was worth the whole native literature of India and Arabia, and proposed to form a class of persons Indian in blood and colour, but English in taste, in opinions, in morals and in intellect, who would interpret between the government and the millions it governed. The English Education Act followed.

Homi Bhabha reads that sentence for what it cannot control. The colonial project needs the mimic man, and it needs him to remain distinguishable: reformed and recognisable, almost the same, but not quite. That gap is not a failure of the policy, it is the policy, and it makes colonial authority permanently ambivalent, because the mimicry that secures it also mocks it. A perfectly successful imitation would dissolve the difference the whole arrangement rests on.

From this Bhabha builds hybridity: colonial cultures do not meet as two intact blocks, one imposing on the other, but produce a third space in which the authoritative version is repeated with a difference and cannot be repeated intact. The objection, made most sharply by Arif Dirlik in 1994, is that this description flatters the metropolitan intellectual: hybridity is a comfortable condition in a university in New York and a much less comfortable one in a refugee camp, and a theory that finds subversion in every act of imitation risks not noticing which side won.

Common misconceptions

• Orientalism argues that Western scholarship on Asia is all worthless. Said repeatedly praises individual scholars. The claim is about a discourse whose latent assumptions survive across two centuries of disagreement at the manifest level.
• Postcolonial criticism means studying literature from former colonies. Said's most influential readings are of Austen, Kipling and Conrad. The method is about the metropolitan canon as much as anything else.
• Spivak claims subaltern people are literally silent. Her claim is about reception: the speech act fails to complete because no available listening position will take it as what it is, which is why her example is a woman who arranged her own death as a legible message.
• Hybridity means cultural mixing is a happy outcome. Bhabha's term names a structural instability in colonial authority, not a celebration of multiculturalism, and the objection that it is often read as celebration is a fair one.

What to remember

Three moves, in order of what they ask of a reader. Said: treat a body of writing as a discourse and look for the assumptions that survive across disagreement. Spivak: before asking what a text says about the colonised, ask whether the archive it draws on could have recorded them at all. Bhabha: look for the point where colonial authority requires an imitation that undoes it.

The next lesson takes a single novel, Conrad's Heart of Darkness, and reads it twice, once with Chinua Achebe and once with the critics who answered him, because the disagreement between them is the clearest test case this field has produced.

The core of it: the durable contribution here is not a verdict on any particular book. It is the demand that a reading say where its evidence came from, and who was in a position to produce it.

Sources

1. Wikipedia. Orientalism (book). en.wikipedia.org
2. Wikipedia. Can the Subaltern Speak. en.wikipedia.org
3. Wikipedia. English Education Act 1835. en.wikipedia.org
4. Said, E. W. (1978). Orientalism. Pantheon Books.
5. Ahmad, A. (1992). In Theory: Classes, Nations, Literatures. Verso.

The defence, stated as an argument

Here is the answer most students give, and most syllabuses imply, when the question of racism in Heart of Darkness comes up. It has three steps.

1. The novella is an attack on European imperialism, and a fierce one for 1899.
2. The offensive descriptions belong to Marlow, a character inside a frame narrative, not to Conrad.
3. Therefore the charge of racism is a category error: the book is on the other side.

This lesson takes that argument seriously enough to test each step, because it is not a stupid argument. Two of its three premises are true. The conclusion does not follow, and finding the exact point of failure is more useful than either endorsing or denouncing the book.

What Achebe put on the table

On 18 February 1975, Chinua Achebe gave a lecture at the University of Massachusetts Amherst titled An Image of Africa. He was already the author of Things Fall Apart. His conclusion was that Conrad was a thoroughgoing racist and that the novella should not be called a great work of art.

The evidence is textual and specific. First, the Africans on the riverbank are given no language. What Marlow reports is a burst of yells, a whirl of black limbs, a mass of hands clapping, of feet stamping, of bodies swaying, of eyes rolling, and then the steamer moving on the edge of a black and incomprehensible frenzy.

Second, the two Africans in the book who do get direct speech are given four sentences between them. The headman of the crew asks Marlow for the prisoners: Catch im. Give im to us. Asked what he would do with them, he answers Eat im. The other is the manager's servant, who announces Kurtz's death with contempt: Mistah Kurtz ... he dead. Achebe's point is not that these lines are unrealistic. It is that in a novel of a hundred pages this is the whole of African speech, and both instances are grotesque.

Third, and this is the argument that has been hardest to answer, Africa in the book functions as a setting for a European drama. The question the novella asks is what the wilderness does to Kurtz and to Marlow. The people who live there are the medium in which that question is posed, and the book's moral crisis does not require them to be anything more.

Where the defence holds

Step one of the defence is simply true, and it is worth knowing how true. The Congo Free State was the personal property of Leopold II of Belgium, and the rubber quota system there was enforced by hostage-taking, mutilation and killing on a scale that historians still argue about in millions. Conrad went up that river in 1890. Roger Casement's official report appeared in 1904 and E. D. Morel's campaign followed; Conrad's novella came first, in 1899, and Casement and Morel both knew it.

And the book does not flinch. In the grove of death Marlow finds men who have crawled away to die: They were not enemies, they were not criminals, they were nothing earthly now ... nothing but black shadows of disease and starvation, lying confusedly in the greenish gloom. He notes that they had been brought there in all the legality of time contracts. That sentence is doing exactly what a critique of imperialism should do: it names the paperwork.

Step two is also partly true. There is a frame: an unnamed narrator on a yawl in the Thames estuary reports Marlow's story, so the novella has at least two narrating levels, and Marlow's judgements are visibly the judgements of a man who has been damaged.

Where the reasoning starts to slip

The frame does less work than the defence needs.

It insulates opinions, not descriptions. When Marlow delivers a view about restraint or about work, the frame lets a reader hold it at arm's length. But the passage about the whirl of black limbs is not an opinion, it is the narration of what is there, and the frame narrator never offers a competing description. Nothing in the book sees the riverbank differently.

Then there is the passage Achebe treats as the centre of his case. Marlow says the men on the bank were not inhuman, and then adds that this was the worst of it, this suspicion of their not being inhuman. The humanity of the people on the bank arrives in the novella as a horror, and it arrives as a suspicion rather than a fact. Whatever the frame is doing, it is not putting distance between the book and that sentence, which is the moral hinge of the passage.

What matters here: a frame narrative distances a reader from a character's claims. It does not distance a reader from the terms in which the world of the book is rendered, because those terms are what the reader has to see with.

The load-bearing error

Now the actual failure, and it is in step three.

The argument treats anti-imperialist and non-racist as the same proposition. They are not. It is entirely possible, and in 1899 it was common, to attack a colonial regime for its cruelty, its greed and its corruption of Europeans, while regarding the colonised as scenery. The eighteenth and nineteenth-century record is full of humanitarian campaigners whose case rested on the suffering of people they never troubled to imagine as agents.

Once the two propositions are separated, the dispute becomes tractable. Conrad's book is a genuine indictment of the Congo enterprise and a book in which Africans appear almost exclusively as bodies, noise and menace. Both can be established from the text, and neither cancels the other. What Achebe demonstrated is that the second was invisible to a critical tradition that had been discussing the first for seventy years, and the invisibility is a fact about the criticism as much as about the novel.

What the defence gets right that Achebe's case understates

Cedric Watts replied to Achebe in 1983 and made two points worth keeping. The first is comparative: measured against Conrad's contemporaries writing about Africa, the novella is markedly less confident about European superiority, and Marlow's opening speech that conquest is mostly the taking away of things from people with different noses is not a sentiment you find in the adventure fiction of the period.

The second is about Conrad's position. He was a Pole from a family broken by the Russian empire, writing in his third language, and his other fiction is unusually alert to the machinery of imperial administration. Watts also objected that Achebe quotes selectively, which is true and is also true of Watts.

Achebe's answer, restated in interviews for the rest of his life, was that being ahead of one's contemporaries is a historical fact and not an exemption, and that a work still taught as one of the central books in English carries a present responsibility that a period defence does not touch.

The two readings side by side

Said's third position

Edward Said, who wrote his doctorate on Conrad and returned to him for thirty years, refused both verdicts. His account in Culture and Imperialism is that Conrad saw further than almost any of his contemporaries into the fraudulence of the imperial mission, and could not imagine a world in which the colonised had a history of their own after it ended. The novella therefore contains a devastating critique and an absolute limit, and the limit is not a flaw to be excused but a structural feature of writing from inside an imperial culture at its height.

That position has the advantage of predicting what a reader will find: a book that keeps opening the question of European legitimacy and keeps closing the question of African agency. It is also the position that makes Achebe's Things Fall Apart legible as a reply rather than a rebuttal, since it supplies precisely what Conrad could not.

What would settle it, and what would not

Evidence about Conrad's private opinions would settle almost nothing. The dispute is about what the novella does, and a letter cannot repair a page.

What would move it is textual. If a reader could show a place where the book renders an African consciousness, or where the narration offers a competing description of the riverbank, Achebe's central claim would be weakened at its strongest point. Conversely, if the defenders' claim that Marlow's incomprehension is exposed by the novel is right, there should be a moment where the text marks his failure to understand as his failure rather than as the country's strangeness. Find that moment and the argument shifts. Both sides have been looking for a century.

Common misconceptions

• Achebe wanted the book banned. He wanted it read differently and taught with its problem visible, and he said so repeatedly. His target was as much the criticism that had not noticed as the novel.
• The frame narrative settles the question. It distances Marlow's judgements. It does not distance the terms in which the riverbank is described, and no other description is offered.
• Conrad was typical of his time, so the charge is anachronistic. Casement and Morel were his contemporaries, and so were African and African American writers who described Africans as people. Being of one's time is a description of a range of options, not of a single one.
• You must choose between the two readings. The most defensible position, and Said's, is that the novella is both a serious critique of empire and a book whose imaginative resources stop at the water's edge.

Recap

The defence fails at one identifiable step: it treats anti-imperialism and non-racism as the same claim. Once they are separated, the evidence on both sides can be held at the same time, and the dispute stops being a referendum on whether a book is good.

Keep the debugging habit. When two competent readers disagree about a text for fifty years, the productive move is rarely to find more evidence for one side. It is to write out the argument in steps and look for the premise that has been carrying two propositions at once.

The point: a critique of a system and a failure of imagination about the people inside it are different things, and a book can do both at full strength on the same page.

Sources

1. Conrad, J. (1899). Heart of Darkness. Project Gutenberg
2. Wikipedia. An Image of Africa. en.wikipedia.org
3. Wikipedia. Congo Free State. en.wikipedia.org
4. Achebe, C. (1977). An image of Africa: Racism in Conrad's Heart of Darkness. Massachusetts Review, 18(4), 782-794.
5. Watts, C. (1983). A bloody racist: About Achebe's view of Conrad. Yearbook of English Studies, 13, 196-209.

A pamphlet printed in 1588

Thomas Harriot was a mathematician sent by Walter Raleigh to the colony on Roanoke Island in 1585. He came back after a year, and in 1588 published A Briefe and True Report of the New Found Land of Virginia, a document written to keep investors interested. It divides into three parts, and the first is a list of what he calls commodities already found or to be raised, which will enrich the providers and profit the country.

Deep in the section on the inhabitants, Harriot reports what happened when the English left a village. People died in numbers, of diseases the Algonquians had no experience of, and only in the towns the English had visited. He records their explanation: that men were coming behind the English who were invisible and without bodies, and who at the English request killed the people by shooting inuisible bullets into them.

This lesson runs a procedure on that pamphlet. The procedure is New Historicism, and it has five steps.

The five steps

1. Start from an anecdote or an odd document rather than from a literary work.
2. Put a literary and a non-literary text of the same moment side by side, with neither ranked as the background of the other.
3. Ask what is being exchanged, negotiated or borrowed between them.
4. Ask whether the trouble the texts produce is contained by the forms that produce it.
5. State the result in both directions: the text has a history, and the history reaches you as text.

Steps one and two: Harriot beside Prince Hal

Stephen Greenblatt did exactly this in an essay of the early 1980s, and the pairing is Harriot's report with Shakespeare's second tetralogy.

Start with the danger in Harriot's document. There was a hypothesis in circulation in the sixteenth century, associated with Machiavelli and dangerous to hold, that religion is an invention of rulers for keeping subjects obedient. A colonial administrator's report is an odd place to find it tested, but that is what Harriot's account amounts to: he describes using the Bible, and English instruments, clocks, guns, magnets, writing, to produce in the Algonquians the belief that the English had divine backing, and he records the results with the detachment of a man running an experiment.

Now put beside it Prince Hal in Henry IV Part 1, alone on stage in the second scene, having just agreed to a robbery: I know you all, and will awhile uphold the unyoked humour of your idleness. He is telling the audience that his companions are material, that his disgrace is a strategy, and that the reformation he plans will look better against the darkness he is deliberately building.

Both texts do the same thing. Each produces a genuinely subversive perception, that authority is manufactured, and each places that perception inside a structure that makes it serve authority. Harriot's report of doubt about religion is written by a man who is recording it in order to govern better. Hal's cynicism is the mechanism by which a legitimate king is produced.

Step three: what is being exchanged

Greenblatt's term for the third step, developed in Shakespearean Negotiations in 1988, is the circulation of social energy. A play does not reflect its society and does not merely draw on it for material. It acquires the intensity of things that already have force in the culture, exorcism, treason trials, royal entries, punishment, and gives them back changed and repeatable.

This is what replaces base and superstructure. There is no direction of determination to establish, because the traffic runs both ways: the theatre borrows the ceremony of power, and power borrows the techniques of the theatre.

Why this matters: the method's real claim is that a literary text is one practice among the practices that make a society, and therefore that a play can be evidence about a culture in the same way a court record can, without either being the cause of the other.

Step four: is the subversion contained?

Greenblatt's answer for the Elizabethan material is that subversion is produced in quantity and that the containment is not a lid put on afterwards; the same operation produces both. Authority in these texts requires the thing that threatens it, the way Hal requires Falstaff.

British critics working at the same time on the same plays disagreed. Jonathan Dollimore and Alan Sinfield, in the collection Political Shakespeare in 1985, argued that the plays contain real dissidence that was not contained: theatres were closed, plays were censored, Richard II was performed on the eve of the Essex rebellion in 1601, and authorities behaved like people who thought something was at stake. They called their approach cultural materialism, after Raymond Williams, and its quarrel with New Historicism is political: if power always produces its own opposition and always absorbs it, criticism becomes a description of a closed system, and the critic has nothing to do but admire the machinery.

This is the sharpest division in the field, and it turns on evidence: whether the record shows dissidence that had effects.

Step five: the chiasmus

Louis Montrose gave the position its slogan in 1989, and it is worth learning as a pair of tasks rather than a phrase: the historicity of texts and the textuality of history.

The first half says every text is embedded in a specific moment, made by someone with a position in it. The second half says the past is not available except through what survives, and what survives is documents, which have rhetoric, gaps, interests and audiences of their own. An archive is not the world; it is a set of texts that require reading exactly as a poem does.

That second half is what separates New Historicism from the older historical scholarship it displaced, which treated background as solid ground on which literature could be stood.

Where the method went

Catherine Gallagher, who wrote Practicing New Historicism with Greenblatt in 2000, defended the anecdote directly. Its function, she argues, is counterhistorical: an odd detail that does not fit the established narrative of a period opens a hole in it, and the point of starting from the strange rather than the typical is to avoid confirming the story you already had. Her own work on Victorian fiction and political economy shows the method outside the Renaissance, reading novels and economic pamphlets as participants in one argument about labour and value.

Four objections

• Arbitrary connectedness. Why this document beside this play? Walter Cohen's objection is that the method licenses a striking juxtaposition without a rule for what counts as a relevant one, so any two texts of a period can be made to illuminate each other.
• The anecdote's representativeness. A single strange case is offered as a way into a culture. Gallagher answers that the strangeness is the point. The objection stands anyway: nothing in the method tells you when a detail is a symptom and when it is a curiosity.
• Unfalsifiable containment. If evidence of dissent shows subversion, and evidence of order shows containment, the thesis cannot lose. Cultural materialists press exactly here.
• Form goes missing. Reading a play and a pamphlet as equally texts is the method's strength, and it means that what makes the play a play, its verse, its structure, its theatre, can drop out. That objection is one of the reasons the return to form in Module 6 happened.

The same document, read for place

Now change one thing. Take Harriot's pamphlet again and ask, not about power, but about land.

Read that way it is an inventory. Its first and longest part is a list: grasses, silk, flax, pitch, tar, furs, deerskins, sassafras, cedar, wine, oil, sugar cane, iron, copper, pearls, and the maize and beans grown by people whose farming methods Harriot copies down carefully because they work. The document's organising idea is that a place is a set of extractable commodities and that the people already living there are, in his phrase, not to be feared.

Ecocriticism, which acquired a name and a scholarly association in the early 1990s, is defined in Cheryll Glotfelty's 1996 reader as the study of the relationship between literature and the physical environment. Its first wave was largely about nature writing and Romanticism: Lawrence Buell's The Environmental Imagination, of 1995, proposed criteria for an environmentally oriented work, chiefly that the nonhuman is present as more than a frame for human history and that human accountability to the environment is part of the text's ethical orientation. Jonathan Bate had already argued in 1991 that Wordsworth's poetry is a serious ecological argument rather than a retreat from politics.

The second wave complicated the founding term. William Cronon argued in 1995 that wilderness is a cultural invention with a history, produced partly by removing the people who lived in it. Timothy Morton went further in 2007, arguing that the concept of Nature, held at an aesthetic distance and admired, is itself an obstacle to ecological thinking. Environmental justice criticism then shifted the field's attention from unpeopled landscapes to the places where damage is actually distributed.

Notice what the two readings of Harriot share. Both refuse the boundary of the literary text, and both treat a document as a participant in the world rather than a report on it. They differ in what they hold constant: one keeps power in view and the other keeps the land in view, and the same seventy pages support both.

Common misconceptions

• New Historicism is just history of ideas with better anecdotes. The textuality of history is the difference: the archive is not solid ground beneath the literature but another set of texts requiring reading.
• Containment means nothing ever changes. Greenblatt's claim is local, about a particular set of Elizabethan texts and institutions. Whether it generalises is exactly what the cultural materialists deny.
• Ecocriticism means writing about nature poetry. Since the mid-1990s the field has worked on cities, waste, disaster, agriculture and colonial resource inventories, and Morton's argument is that the category of Nature is part of the problem.
• These approaches abandon close reading. Greenblatt's essays are minute readers of single sentences. The change is what a sentence is taken to be evidence about.

The takeaway

The procedure, once more: start from a document, pair it with a literary text without ranking them, ask what circulates between them, ask whether the trouble is contained, and remember that your evidence about the past is itself made of texts.

Run honestly it produces readings no other method reaches, and this lesson's pairing is the demonstration: a colonial pamphlet and a history play turn out to be doing the same thing with authority. Run carelessly it produces a striking juxtaposition and a conclusion that was available before the archive was opened.

In short: the test for a historicist reading is whether the archival material could have surprised the argument. If the pamphlet is only there to illustrate a thesis about power, you have written the thesis, not the reading.

Sources

1. Harriot, T. (1588). A Briefe and True Report of the New Found Land of Virginia. Project Gutenberg
2. Wikipedia. New Historicism. en.wikipedia.org
3. Wikipedia. Ecocriticism. en.wikipedia.org
4. Greenblatt, S. (1988). Shakespearean Negotiations: The Circulation of Social Energy in Renaissance England. University of California Press.
5. Buell, L. (1995). The Environmental Imagination: Thoreau, Nature Writing, and the Formation of American Culture. Harvard University Press.

A poem that has already done the critique

Emily Dickinson wrote this in about 1863. The version below follows the first printed edition of 1890, with her dash restored in the seventh line.

Much madness is divinest sense
To a discerning eye;
Much sense the starkest madness.
'T is the majority
In this, as all, prevails.
Assent, and you are sane;
Demur - you're straightway dangerous,
And handled with a chain.

Now apply the method this course has spent thirteen lessons building. The poem shows that sanity is not a medical fact but the verdict of a majority; that the category polices dissent; that the chain is where classification ends. It exposes the operation of power in a discourse of health.

That reading is correct. It is also entirely contained in the poem, which said all of it in eight lines and said it first. And here is the question that a group of critics began asking around 2009: if the text already knows, what exactly is the critic doing?

The charge

Rita Felski made the case at length in The Limits of Critique in 2015. Her claim is not that critique is wrong. It is that critique is a mood and a genre with a recognisable style, and that it has been mistaken for rigour itself.

She takes from Paul Ricoeur the phrase hermeneutics of suspicion, coined to describe Marx, Nietzsche and Freud, who share the assumption that consciousness is not what it takes itself to be and that the analyst's task is to get behind it. Felski's contribution is to describe what that assumption does to critical prose. It produces two spatial habits: digging down, to a depth where the real content lies, and standing back, to a distance from which the text's complicity becomes visible. It casts the critic as detective and the text as a suspect with something to hide. And it produces a distinctive affect, a vigilance in which being caught out is the worst thing that can happen to a reader.

The result, she argues, is a criticism that can only ever find one kind of thing, and that has no vocabulary for what draws people to books in the first place. Her Uses of Literature, in 2008, proposed four modes of engagement, recognition, enchantment, knowledge and shock, precisely because the professional vocabulary had no room for them.

Her positive proposal borrows from Bruno Latour, whose essay of 2004 asked whether critique had run out of steam as it watched its own techniques of doubt turned against climate science. Latour's alternative is to treat objects as actors: to ask what a thing does, what it connects, what it makes possible. Applied to literature, that means asking what a text does in the world, what attachments it forms, what it enables its readers to do, rather than what it conceals.

The surface, and why 2009

Stephen Best and Sharon Marcus introduced a special issue of the journal Representations in 2009 with a piece called Surface Reading. Their target is symptomatic reading, which you met in Module 3: the model from Althusser and Macherey, carried into literary study by Jameson, in which the text's silences and contradictions are the way to its unspoken ideological content.

Their proposal is to attend instead to what is evident, perceptible and apprehensible in a text, and they list several versions of surface: the material surface of the book, the verbal structure of the sentence, patterns visible only to a machine, and an ethical attentiveness that Marcus calls just reading, meaning reading that lets a text say what it says.

The date matters, and they say so. The generation that built symptomatic reading assumed that power conceals itself and that exposure is therefore a political act. They wrote after a decade in which torture was authorised in published memoranda and photographed by its perpetrators. When domination is conducted in the open, the critic who announces that something hidden has been uncovered is not obviously performing a service.

Key idea: the post-critique argument is historical before it is philosophical. It claims that a method built for a world where power hides loses its edge in a world where power does not bother.

Sedgwick, who got there first

The essay everyone in this dispute cites was written in 1997 by Eve Sedgwick, whom you met in the queer theory lesson. She distinguishes paranoid from reparative reading, and she is careful to say the paranoid position is not a mistake: it is a strong theory, anticipatory, mimetic of what it exposes, and it earned its authority in a decade when official denial about an epidemic was killing people.

Her doubt is about its faith in exposure. Paranoid reading assumes that demonstrating the operation of an oppressive system will weaken it. Her observation from the AIDS years is that this is often simply false: the workings were exposed, repeatedly and well, and the exposure changed very little because the knowledge was never what was missing.

The reparative position is not optimism. It is the practice of assembling, out of a culture that was not built for you, resources adequate to living in it. Sedgwick's examples are the uses queer readers have made of texts that were hostile to them, and the point is that this is a form of skilled reading with its own difficulty, not a retreat into liking things.

Affect, briefly

Affect theory supplied the vocabulary. Sedgwick took from the psychologist Silvan Tomkins a model of discrete affects, shame in particular, that are not reducible to ideology. Brian Massumi drew a line between affect, understood as a bodily intensity that precedes the person, and emotion, which is that intensity named and made social. Lauren Berlant gave the field its most portable concept in 2011: cruel optimism, an attachment to an object whose pursuit actively obstructs the flourishing it promises. That is a description of how people stay attached to bad arrangements which does not require them to be deceived, and it is exactly the sort of account critique's exposure model cannot easily produce.

The reply

Critique did not concede.

The first reply is that the target is a caricature. Bruce Robbins put it most bluntly: nobody who actually practises symptomatic reading believes that texts are simple containers of hidden ideology or that critics are detectives, and Felski's composite portrait is assembled from the weakest practitioners. Jameson's own readings are attentive to form to a degree that the description of digging beneath the surface does not capture.

The second reply is that surface reading is not available. Ellen Rooney argued in 2010 that deciding what counts as a surface is itself an interpretive act, and that just reading conceals its own selections behind the appearance of mere description. Reading the sentence rather than the ideology is a choice about relevance, and choices about relevance are what interpretation is.

The third is political. Carolyn Lesjak and others argue that dropping suspicion at a moment of widening inequality is a strange response, and that the mood the new critics recommend, attention, attachment, description, is the mood of accommodation. If exposure has not worked, the argument goes, that is a reason to ask why, not to stop.

Felski's answer to all three is that she is arguing against a monopoly, not for a prohibition: critique remains one available mood among several, and the pathology she names is a discipline in which no other mood counts as serious work.

The poem, read the other way

Back to Dickinson, and this is where the dispute becomes concrete rather than a quarrel about tone.

A surface reading attends to what the eight lines do. The poem is built on a chiasmus in its first three lines, madness to sense, sense to madness, which enacts the reversal it describes. It moves from an abstract vocabulary of Latin origin, madness, divinest, sense, majority, to a final line of plain monosyllables ending on chain. The syntax narrows: the first sentence has a subordinate clause and a qualification, the last has none. And the poem places the reader in the position it describes, since to agree with it is to assent and to disagree is to demur, which are the two options it has just named.

A reparative reading asks a different question: what does a person do with this poem? It has been carried by readers who were themselves being classified, and its usefulness is not that it informs them that classification is social, which they knew, but that it supplies a form in which that knowledge can be held without dissolving.

Neither of those readings competes with the critique reading. All three are true. What they show is that the critique reading, offered as the finished job, leaves most of the poem on the table.

Common misconceptions

• Post-critique means reading uncritically. Felski's case is against a monopoly of one mood, and her own book is a sustained critical analysis of a scholarly genre.
• Surface reading means ignoring context. One of Best and Marcus's surfaces is the pattern a machine finds across thousands of books, which is as contextual as criticism gets. The claim is about depth as a metaphor, not about scope.
• Reparative reading means liking things. Sedgwick's reparative reader is assembling resources from a hostile culture, which is difficult work and often does not end well.
• The debate settled in favour of post-critique. It did not. Both positions are actively practised, and the interesting recent work tends to run both, with the second turn of this lesson as its model.

Where this leaves us

The dispute comes down to a question about literary knowledge. If a reading is valuable because it tells you something the text was hiding, then critique is the whole job and everything else is appreciation. If a reading is valuable because it makes something available that was not available before, then exposure is one route among several, and description, attention and use are others.

What would move the argument is the thing that moves every argument in this course: results. If suspicious readings keep finding what they went in for, and surface readings keep finding features nobody had described, that is evidence about both. The next lesson turns to a group of critics who took Best and Marcus's third kind of surface, the pattern visible only to a machine, and made a research programme out of it.

Bottom line: ask of your own reading whether the text could have told you what you found. If it could, the interesting question is not what you exposed but what the text is doing with the knowledge it already has.

Sources

1. Dickinson, E. (1890). Much madness is divinest sense. In Poems by Emily Dickinson. Project Gutenberg
2. Wikipedia. Postcritique. en.wikipedia.org
3. Dauenhauer, B., and Pellauer, D. (2022). Paul Ricoeur. Stanford Encyclopedia of Philosophy. plato.stanford.edu
4. Best, S., and Marcus, S. (2009). Surface reading: An introduction. Representations, 108(1), 1-21.
5. Felski, R. (2015). The Limits of Critique. University of Chicago Press.

Twelve disputed papers

The eighty-five essays of The Federalist were published in New York newspapers in 1787 and 1788 under the name Publius. Three men wrote them. Most attributions are secure, but twelve papers were claimed by both Alexander Hamilton and James Madison, and the dispute ran for a century and a half on internal evidence and political sympathy.

In 1963 two statisticians, Frederick Mosteller and David Wallace, settled it well enough that the finding has never been seriously overturned. Their procedure has five steps, and every quantitative literary study since has the same skeleton.

1. Turn the question into something a count can answer. Not who wrote better, but which of two known writers is more likely to have produced this text.
2. Choose features the writer does not control. Not subject matter, imagery or argument, which are shared by design, but function words: upon, whilst, while, on, there, by.
3. Build a comparison set. The undisputed papers, plus other writing by both men, so the rates can be estimated from texts of known authorship.
4. Fit, then validate on data you did not fit. Hold out papers whose authorship is known and check that the method assigns them correctly.
5. Report the uncertainty. Not a verdict but odds, and the sensitivity of those odds to the choices made in steps two and three.

The single most quoted detail is the word upon. Hamilton uses it at roughly three occurrences per thousand words. Madison hardly uses it at all. Neither man decided this; nobody chooses a rate for upon. That is exactly why it works, and it is the whole logic of stylometry: the evidence is the part of writing that is beneath the writer's attention.

The conclusion, on the numbers, was Madison for all twelve, most with very long odds.

What matters here: the study is convincing because the question was narrow, the ground truth was real, and the method could have failed the validation step. Hold those three conditions in mind, because the argument in the rest of this lesson is about what happens when one of them is missing.

What Moretti proposed

In 2000 Franco Moretti published an essay on world literature containing a coinage. If the object of study is a world literature of thousands of traditions, no scholar can read it, so knowledge will have to come from a different relation to texts: distant reading, in which distance is not an obstacle but a condition of knowledge, and the unit of analysis is smaller or larger than the text, a device, a theme, a genre.

His companion argument is about the archive. The nineteenth-century British novel canon is a couple of hundred books. Somewhere over twenty thousand were published. Criticism has been describing a fraction of one per cent and calling it the history of the form. What happened to the rest is a question no amount of close reading can answer, because the rest cannot be read.

The Stanford Literary Lab, founded in 2010, published its findings as pamphlets. The first, Quantitative Formalism, asked whether a machine given nothing but word frequencies could recover the genres of a set of novels. It largely could: the clusters the computation produced corresponded to genres critics had assigned by reading. The finding is not that computers understand novels. It is that genre leaves a trace in the frequencies of ordinary words, which means genre is a formal fact and not only a marketing category. The second pamphlet turned plots into networks, with characters as nodes and speech between them as edges, and showed what happens to Hamlet's structure when you delete a single character.

Change one input: from authorship to value

Now take the same machinery and point it at a different question. Ted Underwood and Jordan Sellers asked whether a model trained on word frequencies could distinguish volumes of poetry reviewed in prominent journals between 1820 and 1919 from a random sample of volumes that were not.

It could, with accuracy well above chance, and the interesting part is what happened across time. The model trained on one decade kept working on later ones, and the misclassifications were not random: books the model treated as more literary than their reception tended to be ahead of the trend rather than off it. The picture that emerges is of a slow, directional drift in what counted as poetic distinction, running for a century in a consistent direction, rather than the sequence of revolutions and reactions that literary history usually narrates.

Now compare the two studies at each of the five steps. Steps one, two and three are as sound in the second as in the first. The break is at step four. The Federalist study validates against a fact: someone wrote those essays. The prestige study validates against reviewing, which is a social outcome produced by editors, networks and class. The model is an excellent predictor of what got reviewed. Whether that licenses a claim about literary standards is a question the numbers cannot answer, and Underwood is careful about it in a way his headlines often are not.

The same authors, with David Bamman, produced a finding that shows the method at its strongest: measuring the proportion of English-language fiction written by women across a century and a half, they found it fell from roughly half in the mid-nineteenth century to about a quarter by the middle of the twentieth, then rose again. That is a claim about the archive, not about value, and it is the kind of thing no amount of close reading could have established.

What a count can and cannot establish

Da's objection

In 2019 Nan Z. Da published an article in Critical Inquiry that made the strongest available case against the field. Her method was to take published computational literary studies and attempt to reproduce them.

Her charge comes in two halves, and it is important that they are different. Some papers, she argues, contain statistical errors: they report patterns that do not survive correction for multiple comparisons, or they use techniques that assume conditions the texts do not meet. Others are statistically sound and produce results that were never in doubt, so the machinery has confirmed what any reader knew. Between the two, she claims, very little is left.

The replies, published as a forum by the same journal, ran along three lines: that Da's own reproductions contain errors; that she selects weak papers and generalises; and that confirming what readers thought they knew is a real result, since a great deal of what readers think they know turns out on measurement to be false. That third reply is the strongest, and it makes the field's own standard clear: a quantitative study earns its place when it could have contradicted the received account.

Bode's objection

Katherine Bode's argument, made at book length in 2018, is aimed at step three rather than step four, and it is the one working practitioners have found hardest to answer.

A corpus is not the literature of a period. It is what a digitisation programme happened to scan, which is a function of what libraries collected, which is a function of what was reprinted, reviewed and thought worth keeping. Studies built on such corpora then discover patterns that may be artefacts of the selection. Her proposal is that literary data should be prepared the way a scholarly edition is prepared, with its history of transmission documented, rather than treated as a transparent window onto the past.

Notice that this is Montrose's second task from the previous lesson, restated for a new medium: the textuality of history applies to a database as much as to an archive box.

Common misconceptions

• Distant reading means never reading closely. Moretti's own pamphlets move constantly between scale and passage, and the Literary Lab's genre finding was interesting only because critics had already classified the novels by reading them.
• The numbers are objective and the interpretation is added afterwards. Every choice in steps one to three is interpretive: which texts count, which features are measured, what the categories are.
• A statistically significant result is a literary finding. Significance says a pattern is unlikely to be chance in that data. Whether it is about literature depends on the argument connecting the feature to the claim, which is where most disputes actually sit.
• Da's article refuted the field. It forced a standard of reproducibility that the field mostly lacked, and the better recent work publishes code and data. Whether the results survive is being settled paper by paper.

Putting it together

The five steps are the whole method: state a question a count can answer, choose features the writer does not control, build a comparison set and be honest about how it was assembled, validate against something outside the model, and report uncertainty. A study that skips step four is a description, which may still be interesting; a study that hides step three is not evidence at all.

What the field has genuinely added is a class of claims nobody could make before: about archives rather than about masterpieces, about slow change rather than about ruptures, about what most books were doing while the canon was doing something else. That is a real extension of what criticism can know, and it is the reason this lesson is in the course even though the count will never tell you what a sentence is doing.

The upshot: ask of any quantitative literary claim what the ground truth was, how the corpus was assembled, and whether the result could have come out the other way. Those three questions dispose of most of the bad work and identify the good.

Sources

1. Stanford Literary Lab. Pamphlets. litlab.stanford.edu
2. Wikipedia. Distant reading. en.wikipedia.org
3. Wikipedia. Stylometry. en.wikipedia.org
4. Da, N. Z. (2019). The computational case against computational literary studies. Critical Inquiry, 45(3), 601-639.
5. Underwood, T. (2019). Distant Horizons: Digital Evidence and Literary Change. University of Chicago Press.

Sixteen lines, and four things critics have meant by form

Blake printed this in 1794, four plates after the poem this course opened with.

I wander through each chartered street,
Near where the chartered Thames does flow,
A mark in every face I meet,
Marks of weakness, marks of woe.

In every cry of every man,
In every infant's cry of fear,
In every voice, in every ban,
The mind-forged manacles I hear:

How the chimney-sweeper's cry
Every blackening church appals,
And the hapless soldier's sigh
Runs in blood down palace-walls.

But most, through midnight streets I hear
How the youthful harlot's curse
Blasts the new-born infant's tear,
And blights with plagues the marriage hearse.

Everyone in this course would agree that this poem has a form. Almost nobody would mean the same thing by it.

Four conceptions, side by side

Run the first three on Blake. A New Critical reading finds the poem's unity in the word chartered, applied first to a street and then, absurdly, to a river, so that the idea of legal ownership is extended to something that cannot be owned; and in mind-forged manacles, where a metaphor of restraint is relocated inside the head. A structuralist reading places the poem against the ballad quatrain and the street cry, and notes that its four-beat line and its repetitions belong to a public, oral, common form rather than to a learned one. A historicist reading supplies the Corporation of London's charters, the 1792 proclamation against seditious writings, the Chimney Sweepers Act of 1788 that was not enforced, and the venereal disease that carries the last stanza.

Each is good. Notice that none of them can say much about how the poem's shape and London's shape are related, except by treating one as a reflection of the other.

Levine's move: forms have affordances

Caroline Levine's Forms, published in 2015, takes a word from design. An affordance is what a material or a shape makes possible: glass affords transparency and shattering, a handle affords grasping. Levine's claim is that formal patterns work the same way. A bounded whole affords enclosure and exclusion. A rhythm affords repetition, synchronisation and endurance. A hierarchy affords ranking and subordination. A network affords connection across distance.

Two consequences follow, and they are the argument.

The first is portability. A rhythm is a rhythm whether it is a metrical line, a factory shift, a liturgical calendar or a school bell, and what it affords is the same in each. This licenses a critic to move between a poem's form and an institution's form without claiming that one caused or reflected the other.

The second is collision. Any actual situation contains many forms at once, and they do not nest neatly. The prison is a bounded whole; the working day is a rhythm; the family is a hierarchy; the epidemic is a network. Politics, on Levine's account, happens at the points where such forms cut across one another, and literature is unusually good at showing those points because it arranges forms deliberately.

Key idea: the new formalism does not treat form as the aesthetic part of a text that historical criticism ignored. It treats form as the thing literature and society have in common, which is what allows a formal reading to be a political one without becoming a reflection theory.

Blake through the four forms

Whole. The poem's first word after the verb is chartered, and a charter is an instrument of enclosure: it makes a bounded thing with an inside and an outside. Applied to the Thames, the absurdity is exactly the affordance of the form being exceeded. The poem's own quatrains are bounded wholes too, four of them, each closed by a full rhyme, and the boundaries hold even as the misery accumulates.

Rhythm. Four beats a line, almost without variation, and the word every seven times in sixteen lines. What a rhythm affords is continuation, and this one affords no exit; the reader walks the streets at the speaker's pace. Set that beside the rhythms the poem names, the sweep's morning cry, the soldier's watch, the night streets, and the metrical pattern and the working day become the same kind of thing rather than a metaphor for one another.

Hierarchy. The third stanza is built from two rankings: sweep beneath church, soldier beneath palace. Both are stated as vertical relations, and both are described by something moving the wrong way up them. The cry appals the church; the sigh runs down the palace wall.

Network. The last stanza is a chain of transmission with four links: the harlot's curse reaches the newborn's tear and blights the marriage hearse. Infection is a network, and networks do not respect enclosures.

Now the collision, which is the payoff. The third stanza's hierarchies are meant to hold people in place. The fourth stanza's network runs straight through them: disease travels from the street to the marriage bed and back, indifferent to charter, rank or parish boundary. The poem's political force is not a statement it makes but the fact that its last stanza's form cannot be contained by its third stanza's form.

That reading is not available to any of the other three conceptions, and it is checkable against the words on the page, which is the test this course has applied throughout.

Why this had to be a fight

Susan Wolfson published Formal Charges in 1997, in the middle of the historicist decades. Her target was an assumption rather than an argument: that attention to poetic form was a retreat from politics, a relic of the New Criticism, and that a properly historical critic reads form as the disguise ideology wears.

Her evidence is that the Romantic poets themselves were arguing about form constantly, in prefaces, letters and the poems, and that treating their formal choices as unconscious symptoms means ignoring what they actually said. Wordsworth's decision about blank verse, Keats's quarrel with the sonnet's rhyme scheme, Byron's stanza: these are arguments conducted in form, and a criticism that cannot read them has lost access to what the writers were doing.

Marjorie Levinson sorted the results in 2007 and found two different projects sharing a name. One, which she calls activist, folds form back into historical reading and treats formal analysis as a way of doing history better; Wolfson and Levine belong here. The other wants a return to close reading as a corrective to historicism, and is closer in spirit to a revival than to a new method. The two are often confused, including by their practitioners.

The objection

Jonathan Kramnick and Anahid Nersessian put the strongest case against the expansive version in 2017. If form names the sonnet, the prison, the working day, the network, the hierarchy and the whole, then form has stopped picking anything out, and a term that applies to everything explains nothing.

Their positive point is sharper than a complaint about vagueness. Explanations in criticism, they argue, are local: the reason a sonnet's volta works is not the same kind of reason as the reason a prison's walls work, and collapsing them into one vocabulary of affordance loses precisely the specificity that made formal analysis valuable. A critic who says that a poem and a factory share a rhythm has made an analogy, and analogies need an argument to become explanations.

Levine's answer, in outline, is that the shared vocabulary is a hypothesis to be tested case by case, and that the test is whether it lets you see a collision you would otherwise have missed. Whether the Blake reading above meets that standard is a fair question to put to it.

Common misconceptions

• The new formalism is a return to the New Criticism. The activist version is the opposite: it treats form as what literature shares with institutions, which is exactly the connection the New Critics ruled out.
• Affordance means a form determines its use. It means a form makes some uses easy and others hard. A network affords connection; whether that connection spreads a disease or a strike depends on what else is in the situation.
• Formalism ignores history. Wolfson's argument is historical throughout: her evidence is what Romantic writers said and did about form in their own moment.
• Form and genre are the same thing. Genre is one kind of bounded whole with a history. Rhythm and network are not genres at all, and part of Levine's point is that the four do not reduce to each other.

What to carry forward

The return to form did not overturn the historicist decades; it made a claim about how to do them. If form is portable, then reading a poem's shape carefully is a way of studying the shapes a society is made of, and the old choice between attending to the words and attending to the world stops being a choice.

Keep the objection with the method. The vocabulary of affordance is only worth its cost when it produces something the older conceptions could not: a collision, a mismatch, a place where two patterns cannot both be satisfied. If a formal reading ends in a statement of correspondence, a poem that mirrors its society, it has produced an analogy and called it a finding.

In short: ask what a form makes possible and what it makes hard, then find the other form it cannot accommodate. That question is the whole of the method, and Blake's last two stanzas are the demonstration.

Sources

1. Blake, W. (1794). London. In Songs of Innocence and of Experience. Project Gutenberg
2. Wikipedia. Affordance. en.wikipedia.org
3. Wikipedia. Formalism (literature). en.wikipedia.org
4. Levine, C. (2015). Forms: Whole, Rhythm, Hierarchy, Network. Princeton University Press.
5. Wolfson, S. J. (1997). Formal Charges: The Shaping of Poetry in British Romanticism. Stanford University Press.

The decision

Kate Chopin published Desiree's Baby in Vogue in January 1893. It runs to about two thousand words and is set on a Louisiana plantation called L'Abri in the years before the Civil War.

Desiree was found as a foundling asleep by a stone gatepost and raised by the Valmondes. Armand Aubigny falls in love with her, marries her without caring that her parentage is unknown, and they have a child. Over three months the household changes: the neighbours whisper, Armand's manner hardens, and Desiree finally sees what everyone else has seen. She asks him what it means. It means, he answered lightly, that the child is not white; it means that you are not white. He tells her to go. She walks into the bayou with the baby and does not come back.

Weeks later Armand burns her things in a bonfire, and among them is an old letter from his own mother to his father. The story's last words are that letter: night and day, I thank the good God for having so arranged our lives that our dear Armand will never know that his mother, who adores him, belongs to the race that is cursed with the brand of slavery. Then the story stops. No narrator comments. Armand's reaction is not described.

You have now met nine ways of reading. The exercise of this lesson, and of any actual critical work, is not to admire all nine. It is to choose one, defend the choice against a named rival, and say what it costs. This lesson is a procedure for doing that.

Three bad ways to choose

By preference. You liked the deconstruction lesson, so you deconstruct. This produces readings that are about the reader's education rather than about the text, and it is detectable: the argument would have been the same with a different story.

By sampler. A paragraph of each: here is the formalist reading, here is the feminist reading, here is what a Marxist would say. This is a demonstration that you attended a course. It produces no claim, because nine partial claims that never meet cannot contradict each other and so cannot be wrong.

By identity. The author is a woman, so the method is feminist; the setting is colonial, so the method is postcolonial. Subject matter is not method. A feminist reading of this story might be about narrative voice, and a formalist reading might turn out to be about property law.

Four questions that decide it

1. What is the puzzle? Not the theme, the puzzle: something in the text that is strange and calls for explanation. If you cannot name one in a sentence, you are not ready to choose.
2. Where would the evidence be? In the words, in an archive, in the reader's experience, in a corpus of similar texts, in what the text does not say.
3. What does the method rule out? Every method buys its clarity by refusing to look somewhere. Name the refusal before you start.
4. What would show the reading is wrong? If nothing could, you have a position rather than a claim.

The core of it: a method is chosen by a puzzle, not by a preference, and the sign that you have chosen well is that the text can still embarrass you.

The candidates, on this story

Working one choice all the way

The puzzle. The story ends inside a document, and stops before anyone reacts to it. The final knowledge belongs to a reader and, presumably, to Armand, and to no one else. Desiree died not knowing.

That is a question about how narration distributes knowledge, so it starts as a formal question. But the reason the distribution matters is that in Louisiana descent was a legal fact with consequences for inheritance, marriage and status, and the story's plot turns on which parent's descent is known. So the choice is a formal analysis whose payoff is historical, which is roughly what Module 5 and Module 6 were arguing about.

The reading. Notice first that the narration is external and sparing throughout: it reports what people say and do and almost never enters a head. The exception is the middle section, where the text follows Desiree's dawning realisation closely enough that the reader arrives at the knowledge with her rather than before her. That is a choice, and it produces the story's cruelty: we are placed inside the person the plot destroys, and outside the person the plot exposes.

Second, the letter. It is not a revelation the narrator makes; it is a document Armand happens to be burning, and it is quoted rather than summarised, so the reader receives it as evidence rather than as a report. What it establishes is not that Armand is wrong about Desiree, which the story never resolves, but that his certainty was never grounded in anything except her being the one whose parentage was unknown.

Third, the timing. The story withholds the letter until after Desiree is gone, so that the last paragraph converts every earlier scene into a different scene. On a second reading, Armand's love for a woman with no known family is not generosity but the only marriage available to a man who cannot afford scrutiny.

The claim. The story's form is an argument about how the law of descent works: it makes the woman whose ancestry is undocumented carry the consequence, while the man whose ancestry is documented and false is not even required to know. The final silence is the point. Nobody in the story is obliged to say anything, because the arrangement functions perfectly well on ignorance.

What would falsify it. If the narration entered Armand's mind as often as Desiree's, the asymmetry claim would fail. If the letter came earlier, the rereading effect would not exist. Both are checkable in ten minutes.

Naming the cost

Every reading owes a sentence in this shape: this reading buys X by ruling out Y, and it would be wrong if Z.

For the reading above: it buys an account of how the story's structure enacts a legal asymmetry, by ruling out the plantation's economy, the actual statutes, Chopin's magazine readership and everything a corpus could tell us about how common such plots were, and it would be wrong if the narration's asymmetry is not there on the page.

Notice that naming the cost is not an apology. It is what makes the reading arguable, because it tells a rival critic where to attack.

Defending the choice against a rival

The strongest rival here is the historicist reading, and a defence has to state it fairly first: the story's meaning depends on legal and social facts about descent in Louisiana that the text assumes and never explains, so a reading that stays inside the words is reading a document in a language it has not learned.

The answer is not that history is irrelevant. It is that the historicist reading, on its own, cannot explain the one thing that makes this a story rather than a case: the placement of the letter. Statutes do not have last paragraphs. The formal reading needs the history to say why the asymmetry matters, and the historical reading needs the form to say why this text is worth reading rather than the statute. That is a real answer, and it concedes something, which is what an answer should do.

What is not a defence: saying that all readings are partial, that methods can be combined, or that the reader should decide. Those sentences are true and they settle nothing.

Common misconceptions

• Combining methods is the sophisticated option. Combining is only sophisticated when the methods are doing different jobs in one argument, as above. A list of perspectives is not a combination.
• Naming the cost weakens the reading. It is the difference between an argument and an assertion. A reading with no stated limits cannot be engaged, only agreed with or ignored.
• The best method is the newest one. The formalist question that opened the reading above is the one Brooks would have asked in 1947. What has changed is what it is allowed to lead to.
• Theory is what you apply after understanding the text. The first lesson of this course made the opposite case, and this last one is the demonstration: the choice of method decides what will count as having understood Desiree's Baby.

Looking back

Seventeen lessons, and one sentence has been true of every one of them: a method buys its power by refusing to look somewhere, and the refusal is the price of a clear answer rather than a flaw to be corrected.

The New Critics refused history and produced a discipline that could be taught. The structuralists refused the individual work and found the system that makes it readable. Deconstruction refused the stability of the system and found the oppositions leaning on what they exclude. Psychoanalysis refused the surface. Marxist criticism refused the boundary of the page. Reader response refused the text-in-itself. Feminist, queer and postcolonial criticism refused the neutrality of the reader and asked who the text was written for. New Historicism refused the hierarchy between literature and document. Post-critique refused the monopoly of suspicion. Distant reading refused the canon. The new formalism refused the choice between the words and the world.

You do not have to adopt any of them permanently. You do have to be able to say which one you are running, what it is refusing, and what would show it wrong. That is the whole of what this course was for.

Bottom line: take a text you care about, name the puzzle in one sentence, choose the method whose evidence is where the puzzle is, write the reading, and finish with the sentence that names the cost. Then hand it to someone who prefers a different method and see whether they can find the seam.

Sources

1. Chopin, K. (1893). Desiree's baby. In The Awakening and Selected Short Stories. Project Gutenberg
2. Wikipedia. Literary theory. en.wikipedia.org
3. Leitch, V. B. (Ed.). (2018). The Norton Anthology of Theory and Criticism (3rd ed.). W. W. Norton.
4. Barry, P. (2017). Beginning Theory: An Introduction to Literary and Cultural Theory (4th ed.). Manchester University Press.
5. Culler, J. (2011). Literary Theory: A Very Short Introduction (2nd ed.). Oxford University Press.