9. Worked Analyses

Eight close readings and a revision case. The apparatus is the same throughout: identify the primary domain (§4), audit the two channels (§6), and apply the tests that bear (§7).

9.1 Austen, Pride and Prejudice (1813) — the Hunsford proposal

Darcy arrives at the Collinses’ parsonage and, after a silence, declares himself:

In vain I have struggled. It will not do. My feelings will not be repressed. You must allow me to tell you how ardently I admire and love you.

Four sentences, and every one of them is about him. The declaration proper occupies the last; the first three are an account of an internal contest he has lost. Elizabeth is the occasion of the sentence, not its subject. Austen then compounds this by summarising what follows rather than quoting it — Darcy’s dwelling on her family’s inferiority, on the degradation of the connection, on the obstacles his judgement had opposed to his inclination — and by noting that the warmth with which he pressed these points seemed due to the rank he was wounding and was unlikely to recommend his suit.

The technique to notice is the switch to report. The most damaging material in the proposal is not quoted. Austen gives the reader Darcy’s opening in direct speech, then withdraws to indirect summary for the offence, then returns to direct speech for the confrontation. This is the nineteenth-century speech/report boundary (§3.5) used with complete deliberation: direct speech for what must be experienced, report for what would be tedious to endure at length and is more damning in summary.

Elizabeth’s refusal answers a declaration of feeling with a point of form — the established mode on such occasions being to express a sense of obligation, however unequally the sentiments are returned. The move is the relational one described at §4.3: she declines the frame before declining the offer. He has spoken as a man conferring; she replies as a woman receiving a social communication, and the mismatch strips the proposal of its assumed standing.

Where it fails a test, and why that matters. The exchange fails test 6 (implicature) rather spectacularly: both parties say what they mean, fully and at length. This is the on-the-nose exchange as a deliberate structure. Its force comes from the surrounding novel, in which nobody ever says what they mean, and the Hunsford scene is the first place where the pressure becomes too great for the machinery of manners to contain. The gap-closing move (§6.5) is being used here in the middle of the book, by two people simultaneously, and it costs both of them.

9.2 Austen, Pride and Prejudice (1813) — the Pemberley exchanges

The Pemberley scenes are the structural answer to Hunsford, and they are conducted almost entirely in ritual formulae.

Elizabeth, having toured the house believing Darcy absent, meets him in the grounds. What follows is a sequence of exchanges about how long she has been in Derbyshire, when she left Longbourn, whether the Gardiners are well, and whether he may be permitted to introduce his sister. Nothing of consequence is said. Both parties are visibly unable to complete their sentences; Austen notes their confusion, their repetitions, their asking of questions already answered.

This is the ritual formula (§6.4, mechanism 9) at maximum load. The content of the exchange is null by design: what is being transacted is entirely in the second channel, and the reader computes it from three signals — the disproportion between the occasion and the parties’ agitation, the repetitions (which flout the maxim of manner), and the single substantive act buried in the phatic material, which is Darcy’s request to introduce Georgiana. In a novel governed by the rules of introduction, that request is a declaration; nothing else in the scene needs to be.

Together, the Hunsford and Pemberley exchanges form the chapter’s clearest available demonstration of the two-channel principle:

The lesson for a writer is that the second scene is only legible because of the first. Austen has trained the reader across four hundred pages to read civility as a code, and at Pemberley she can therefore write an entire reconciliation out of enquiries about the weather in Derbyshire.

9.3 Dostoevsky, The Brothers Karamazov (1880; trans. Constance Garnett) — the Grand Inquisitor

Ivan recites to Alyosha a “poem” he has composed: Christ returns to Seville during the Inquisition, is recognised, is arrested, and is visited in his cell by the Grand Inquisitor, who explains at length why the Church has corrected his work — that men cannot bear freedom, that they will exchange it for bread, mystery and authority, and that Christ’s refusal to give them these was an act of contempt for human weakness. Christ says nothing throughout. At the end he kisses the old man; the Inquisitor releases him and tells him never to return.

This passage tests the chapter’s model harder than any other passage discussed here, and it is worth being explicit about where the model bends.

It is a speech, not an exchange, by the definition at §2.1. The Inquisitor’s addressee never speaks. By test 2, half the want-test fails; by test 3, no status moves within the speech; by test 10, there are no interruptions. By §8.11 it is the failure pattern of the speech answering a question nobody asked, at a length no workshop would tolerate.

And it works, for four reasons the model can articulate.

1. It is embedded in an exchange that is running. The Inquisitor speaks to Christ; Ivan speaks to Alyosha, and that outer exchange has two engines. Ivan wants his brother’s agreement, or his brother’s horror; Alyosha wants not to lose his brother. Everything the model asks for is present at the outer level, and the inner speech is the instrument by which the outer want is pursued. The set speech is a move in a conversation happening one frame out.
2. The listener’s silence is the content. Christ’s not answering is the passage’s argument. A reply would concede that the Inquisitor’s terms were the terms; silence refuses the frame, and the kiss is a response in a register the speech cannot reach. This is the silence domain (§4.6) carrying the entire thematic burden of a chapter in which one character does all the talking.
3. The dialectic is left open. The Inquisitor’s case is made in its strongest form — this is the requirement at §4.4 — and the narration does not arbitrate. Bakhtin’s account of polyphony (§3.4) is built substantially on this chapter, and the standard critical observation is that Dostoevsky, an Orthodox Christian, wrote the most powerful case against his own position in the language. Alyosha’s objection, when it comes, is not a refutation; it is the claim that the poem praises what it means to blame.
4. The two channels are present, but relocated. There is no subtext in the ordinary sense: the Inquisitor states his position completely. The gap is between what Ivan says and what Ivan is — a man constructing an unanswerable argument in order to find out whether he can live inside it, and using his brother as the test. This is the Dostoevskian two-channel structure identified at §8.3, and it is why the chapter is character work of the first order despite containing no characterisation in the usual sense.

What the model learns from the case. The requirement of two engines holds; it simply does not have to be satisfied at the level of the quoted speech. A sustained monologue is an exchange whenever the frame around it is contested — which is the same principle that licenses Nelly Dean (§9.5) and the naive-listener convention (§8.1), arriving from the opposite direction.

9.4 Hurston, Their Eyes Were Watching God (1937) — vernacular, free indirect, and the right to speak

Two moments carry the novel’s dialogue argument.

The first: at the lighting of the street lamp in Eatonville, the crowd calls for a speech from Janie, and Joe Starks refuses on her behalf — his wife knows nothing about speech-making, he never married her for anything like that, her place is in the home. The exchange is four lines long and settles the terms of a marriage.

The second, years later, in the store: Jody mocks Janie’s body in public, and Janie answers in public, telling him that when he pulls down his britches he looks like the change of life. The store falls silent. He never recovers.

Why the second lands is entirely a matter of the first. The novel has established the porch as a space Janie is excluded from — the silence domain (§4.6) operating structurally rather than locally, across two hundred pages. The fact about the retort that matters most is not its content but its venue. It is spoken where she has not been permitted to speak, in the register of the place, and it is unanswerable in that register. This is the gap-closing move (§6.5) executed in the relational domain rather than the emotional one: what closes is not the distance between saying and meaning but the distance between Janie and the floor.

On the phonetic rendering. §8.9 argues the general case; the specific point here is technical. Hurston’s narration is not standard prose against which the speech is a deviation. It moves continuously between a high lyrical register and the characters’ idiom, frequently within a sentence, so that the boundary between narration and speech is porous in Bakhtin’s sense — the character zone (§3.4) rather than the quotation mark. A reader cannot locate a neutral position in the novel from which the dialect would read as marked, because the novel does not contain one.

That is the structural answer to the charge of exhibition, and it can be stated as a general principle: the asymmetry that makes phonetic dialect condescending is an asymmetry between narration and speech, not a property of the spelling. Close the gap between the two and the problem dissolves.

A note on translation and on texture. This is the clearest case in the chapter of a text whose effects do not survive paraphrase, let alone translation. The structural claims made here — the exclusion, the venue, the reversal — transfer to any language. The texture claims do not transfer at all.

9.5 Brontë, Wuthering Heights (1847) — dialogue at two removes

The novel’s central exchange is Catherine’s confession to Nelly Dean that she has accepted Edgar Linton, that it would degrade her to marry Heathcliff, and that she is Heathcliff nonetheless — whatever their souls are made of, his and hers are the same. Heathcliff is in the kitchen, unseen behind the settle. He leaves after the word degrade and before the rest. Nelly knows he is there and says nothing.

The passage is a small anthology of everything in this chapter.

Dialogue at two removes. Everything a reader has of Catherine’s speech has passed through Nelly’s memory, Nelly’s interests, and Lockwood’s transcription — and Lockwood is a fool. The chain is: Catherine speaks → Nelly hears and later reports → Lockwood records → the reader reads. Every link is an opportunity for distortion, and the novel exploits all of them.

This is worth teaching because it inverts the usual assumption about dialogue. §4.2 claims that speech is the novel’s least filtered material, the only element in which the character rather than the narrator selects the words. Wuthering Heights demonstrates that the claim is a convention, not a fact: a novel can foreground the reporting chain, and when it does, quoted speech becomes the least reliable material in the book rather than the most. Chapter 13 (The Viewpoint Shift) treats the machinery; Chapter 7 (Backstory and Flashback) treats the nested retrospection. What this chapter takes from it is a caution about test 7 — the transcript test assumes a transparent reporter, and where the reporter is not transparent, the test is measuring the reporter.

The under-taught point. Nelly is not a neutral conduit and is not a fully unreliable one either; she is an interested party with a servant’s precise knowledge of a household and a servant’s exclusion from its decisions. Her position as an unratified participant in the family’s life is what makes her a total witness to it — the structural fact §4.6 argues the five-domain model cannot express.

9.6 Saramago, Blindness (1995; trans. Giovanni Pontiero) — dialogue without the apparatus

Saramago abolishes quotation marks, paragraph breaks between speakers, and — largely — attribution. Speech runs on inside narrative paragraphs, marked by a comma and a capital letter where a new speaker begins. Characters have no names, only designations: the doctor’s wife, the first blind man, the girl with dark glasses.

The effects are precise and worth separating.

1. Speech loses its ontological privilege. In conventional presentation, quotation marks assert that these words were actually said, in a way the surrounding narration is not. Removing them dissolves the boundary: speech and narration become the same substance, and the reader stops experiencing dialogue as a distinct kind of evidence. In a novel about an epidemic that removes the ability to see who is speaking, this is not a typographic preference.

2. Attribution becomes the reader’s labour. Because speakers are rarely tagged, the reader must attribute turns by inference — from content, from register, from position in a sequence. This makes the transcript test (§7.1, test 7) into the reading experience itself: a reader of Blindness is continuously performing test 7, and the novel is only legible because Saramago’s voices are in fact differentiated. The device is a demonstration that the test measures something real.

3. Pace accelerates and orientation degrades. White space is the texture domain’s largest carrier (§4.5), and removing it produces long undifferentiated blocks that read fast and hold poorly. Readers report losing track of who is present — which is the represented experience.

4. The absence of names removes the vocative entirely. With no names, none of the relational work described at §8.7 is available. Saramago compensates by loading the relational domain onto the designations themselves: the doctor’s wife is defined by a relation, and the novel’s central irony is that the one person who can see is identified only by her marriage.

Chapter 9 (The Opening Pages) reads the same feature as a formal contract declared on page one. The addition here is that the device is semantic as well as formal: it alters what quoted speech means in the book, not merely how it looks.

A translation caution. Pontiero’s English retains Saramago’s punctuation practice, and the structural claims above hold across the languages. Claims about the specific rhythm of the run-on sentences do not.

9.7 James, The Portrait of a Lady (1881) — negotiation as the whole of the action

James’s dialogue is the most technically demanding in English fiction, and the reason is that almost every exchange is a negotiation in which the terms are never stated.

Take the exchange in which Madame Merle sets out her view of the self to Isabel: that no one is isolated, that the self overflows into everything belonging to it — the house, the clothes, the books, the company. Isabel resists: nothing that belongs to her is any measure of her; everything that belongs to her is on the contrary a limit and a barrier, and an arbitrary one.

What makes this dialogue rather than philosophy. Three things.

1. Both women want something. Madame Merle wants to know what Isabel is, and how far she can be shaped; Isabel wants to be understood as she understands herself. The abstraction is instrumental.
2. The reader’s information is asymmetric on rereading. Madame Merle is at this point already engaged in the project the novel will eventually disclose, and every abstract proposition she advances is, on a second reading, a piece of practical planning. This is the character-only implicature configuration described at §6.2, and it is the reason the novel improves so much on rereading.
3. Neither position is defeated. The novel’s plot arguably vindicates Madame Merle — Isabel is caught by the very material web she disclaims — and its moral sense is with Isabel throughout. Bakhtin’s polyphony (§3.4) applies in full: the argument is not adjudicated by the narration.

The technical signature. James’s characters conduct these negotiations through a set of moves that map onto §6.4 almost item for item: the question answered with a question; the too-precise reply; the significant repetition of the other’s word, held up for inspection; the unfinished sentence completed by the hearer. What distinguishes James is density — several mechanisms in a single turn — and the fact that his narration frequently reports the computation of implicature as an event: characters are shown working out what has just been meant, which is the theory of §6.2 dramatised.

Where it fails a test. The transcript test (§7.1, test 7) is a genuine problem in late James, where the principals share a register so completely that stripping attributions makes attribution hard. This is a real limitation, argued about by critics, and it is not obvious that it is a fault: the shared register is itself a representation of a milieu in which everyone has been trained to speak the same way.

9.8 Hemingway, “Hills Like White Elephants” (1927)

The canonical teaching case, and it earns the position. A man and a girl wait for a train at a Spanish station and conduct, in short turns and almost without attribution, a conversation about an operation that is never named. He assures her it is simple, that it is not really an operation at all, that they will be fine afterwards. She asks what happens afterwards. Eventually she asks him, seven times over in one sentence, to please stop talking.

Four features make it teachable. The narration withholds all interiority, so the reader computes implicatures from the same evidence the characters have. The subject is a hole in the text, so every turn is a swerve and the reader constructs the centre from the swerves. The repetitions are the pressure gauge — the man’s really and simple, her please — and the story’s climax is a flouting of manner rather than an event. And status inverts once: at the outset he is managing her, and after the sevenfold please he is not. The story can be analysed in a page, with every mechanism in §6.4 present except the ritual formula.

Its influence is also worth flagging as a hazard. The story is the origin of a great deal of the mid-century workshop orthodoxy that §16 questions, and students who take it as a model rather than as a case tend to produce dialogue in which nothing is named because the writer has not decided what it is.

9.9 A revision case: the doorway exchange

The chapter’s own worked example, using the household that runs through this series: a small shop, a father dead in March who kept its ledger for eleven years, a mother who changed the locks, and three adult siblings reconstructing what was withheld from them. Elena has driven across town to Dov’s front door; she does not yet know that the shop’s locks have been changed. Dov is in the doorway.

The canonical facts in play are established elsewhere in the series: the register drawer that sticks on the K, and the good pen the father used for the ledger only, are planted in Chapter 10 (The Prologue); the ledger is first seen, as an unremarked object, in Chapter 4 (The Descriptive Detail), and the changed handwriting after 9 March is discovered in Chapter 6 (The Chapter Break). This exchange plants the changed locks. The key in it is the new key to the shop door.

Version A — the on-the-nose draft

Dov stood in the doorway. “I know you want the shop key,” he said. “I’ve been dreading this conversation.”

“I’ve had a key to this shop since I was twelve,” Elena said. “It’s humiliating to be offered one by you.”

“It won’t fit any more. Ma changed the locks in March, right after the funeral. I’ve felt caught in the middle ever since — I don’t want to betray her, but I don’t want to lie to you either.”

“Why would she do that? Doesn’t she trust me?”

“She said you’d ask that. She knew you’d take it personally, and I do think she’s right that you take things personally. But she asked me not to explain, and I feel I have to respect that, even though I know it’s hurting you.” He stepped back from the door. “I’m sorry. I really am.”

Diagnosis

Note that Version A contains all the necessary information, and is not badly written at the sentence level. Its failure is entirely one of channel: it runs on one, and it therefore has nothing for a reader to do. Note also that Version A is more emotionally explicit than Version B and communicates less emotion — which is the chapter’s central claim in a single comparison.

Version B

“You’ll want the shop key,” Dov said, not moving from the doorway.

Elena held out her hand. “I’ve had a key since I was twelve.”

“Not this key.” He turned it over once between two fingers before he set it in her palm, face down, like he was folding a hand of cards. “Ma changed the locks in March.”

“Did she.” Elena closed her fist around it. “Did she say why.”

“She said you’d ask that.” Dov stepped back into the hall. “She didn’t say I had to answer.”

Audit

What the revision did, procedurally

1. Converted every statement of feeling into an act. I’ve been dreading this conversation becomes not moving from the doorway. It’s humiliating to be offered one becomes an outstretched hand. The relational domain is carried by physical position throughout (§4.3), which costs no dialogue at all.
2. Applied the third-person deflection (§6.4, mechanism 8). Dov’s I feel I have to respect that becomes She didn’t say I had to answer — the same refusal, with principalship shifted to an absent party. He is now animating someone else’s instruction, which is both a shield and a disclosure: a man who needs the instruction to be someone else’s.
3. Applied the significant repetition (§6.4, mechanism 11). Did she. Did she say why. Elena returns Dov’s own verb, twice, flat. The missing question marks are not a typographic affectation; they mark utterances that are not requests for information, because she has already drawn her conclusion.
4. Made the information cost something (§7.1, test 4). The locks fact arrives inside a correction — Not this key — so that Elena is contradicted before she is informed, and the contradiction is what makes the information land.
5. Added the contradicting action beat (§6.3). Turning the key over like he was folding a hand of cards supplies the reader with counter-evidence to Dov’s compliance: a man who is conceding does not handle the concession that carefully.
6. Deleted the apology. Version A’s I’m sorry. I really am closes the exchange. Version B’s last line opens one, because it announces that a reason exists and is being withheld — the distinction between planting and withholding that Chapter 9 (The Opening Pages) draws for the stakes contract.
7. Differentiated the registers. Dov’s turns are declarative and complete; Elena’s are two and three words long and structurally interrogative without being questions. Two variables, per §7.2 — sentence length and relation to questions — and no verbal tics.

What Version B has not solved

Three things, and they should be said plainly.

It is conspicuously written. The card simile and the missing question marks are visible technique, and a reader alert to the contemporary literary register will recognise the moves. Version B passes test 7 against Version A and does less well against the shelf it sits on — the same qualification Chapter 9 (The Opening Pages) makes of its own revision case.

It leans hard on one mechanism. Three of the five turns run on deflection or repetition. A novel in which every exchange is built this way would be exhausting, and Dov in particular is at risk of becoming a man who only ever declines — which is the over-delivery described at §6.4, mechanism 1.

It does not serve the theme domain except by implication. The claim in the audit table — that the exchange stages what family members owe each other — is available to a reader who has the rest of the book. In isolation it is thin, and a writer should be honest about which domains an exchange serves in situ and which it serves only in the analysis.

And one thing it does that Version A could not. Version B leaves a gap that a later exchange can close (§6.5). Dov’s reason is now a withheld object with a location in the book. Version A spent it.