3. Why the Past Is Expensive

Section 1 asserts that a cutaway begins in debt. The assertion is correct, but it is a metaphor, and a metaphor is not an argument. This section assembles the argument from six directions — formal, rhetorical, philosophical, cognitive, psychoanalytic and historical — because a writer who understands why a backward movement costs revises it more intelligently than one who has merely been told that it does. The section ends by restating the deficit principle and then qualifying it, since the qualification is where the more interesting teaching lies.

3.1 Genette: the first narrative, and the measurement of anachrony

The indispensable apparatus is Gérard Genette’s, set out in Narrative Discourse (1972; English translation by Jane E. Lewin, 1980) — the work of narratology this chapter draws on most, and the one whose vocabulary craft teaching reproduces in a rough and unacknowledged form.

Genette’s opening move is to separate the order in which events occur in the story from the order in which the narrative presents them. Any discordance between the two he calls anachrony. Anachrony is not an occasional ornament; it is, as Genette observes, so nearly universal in Western narrative that a strictly chronological telling is the marked case rather than the default. The Iliad opens in the tenth year of the war. The Odyssey opens with Telemachus, reaches Odysseus in Book V, and has him narrate Books IX to XII himself, in the past, at a feast.

The decisive concept for a writer, and the one most often skipped in craft summaries, is the first narrative (récit premier). Anachrony is not measured against absolute chronology, or against the reader’s clock. It is measured against the temporal level the narrative has established as its principal one — the level to which the telling returns and against which everything else is felt as a departure. Without a first narrative there is no anachrony, only a sequence.

Three consequences follow immediately, and all three are practical.

1. A narrative must have a stable first level before a cutaway can be felt as a cutaway. This is the mechanism underlying the failure pattern at §8.6: a flashback in chapter two arrives before the reader has consolidated a first narrative to depart from, so the departure is not registered as a departure. It reads instead as the book still deciding what it is.
2. What counts as a departure depends on where the narrative has said its home is. In The Remains of the Day the home is a six-day motoring holiday in July 1956; in Great Expectations the home is Pip’s life from childhood forward. Identical retrospective material occupies opposite structural positions in the two books.
3. A narrative can change its first level, and this is a large structural decision rather than a local one. Wuthering Heights effectively transfers its first narrative from Lockwood’s residence at Thrushcross Grange to Nelly’s chronicle, and then transfers it back for the ending (§9.2). The transfer is what makes the novel’s shape unusual, and it is invisible to any analysis that treats the whole book as a single flat sequence.

Genette’s two species of anachrony are analepsis — the evocation, after the fact, of an event earlier than the point in the story at which we currently stand — and prolepsis — narration in advance of an event that will come later. The term flashback nowhere appears in Genette’s system, and its absence is not fastidiousness. Analepsis names a relation between two positions in a narrative; flashback names a device with a particular twentieth-century shape (§3.8). The wider term catches things the narrower one misses: the single clause in the pluperfect, the parenthetical aside, the reference in dialogue to something that happened before the book began.

3.2 Reach and extent

Genette’s two measurements of an anachrony are the most immediately usable part of his apparatus, and they are the part craft pedagogy most conspicuously lacks.

Reach (portée) is how far the anachrony goes from the present moment of the first narrative — the size of the leap.

Extent (amplitude) is how much story time the anachrony itself covers — the size of the visited territory.

The two are independent, and every combination has a different profile.

Two observations that a working writer can use at once.

The distant single moment is the costly case, and it is the one drafts reach for by reflex. A long reach requires the reader to build a whole context — different ages, a vanished place, a superseded set of relations — and a short extent gives them only one scene’s worth of material to build it from. The context is expensive and immediately discarded. This is the arithmetic behind the widespread and otherwise mysterious workshop suspicion of the childhood flashback (§8.9). The suspicion is not about childhood; it is about the ratio of construction cost to occupancy.

Reach can change during an anachrony, and the direction it changes in is a structural signal. An analepsis with a long reach and a long extent shortens its own reach as it proceeds, since it moves forward while the first narrative waits. Nelly Dean’s narrative in Wuthering Heights begins about thirty years before Lockwood’s present and closes the distance until it merges with it. A reader feels this as approach, and the feeling is one of the most reliable sources of momentum available inside a long cutaway — the past visibly coming to meet the present. Its absence is one reason a static, non-progressing flashback drags: nothing is closing.

3.3 The three distinctions that determine cost

Genette subdivides analepsis further, and three of the subdivisions do real diagnostic work.

First: external, internal and mixed.

The distinction is not academic. An external analepsis has an empty field to work in: nothing the reader knows can conflict with it. An internal analepsis works in occupied territory, and every sentence risks either repeating what the reader has, or contradicting it. This is why the two most common cutaway failures — redundancy (§8.2) and continuity confusion — cluster almost entirely among internal analepses, and why a writer should always know which kind they are writing.

Second, and within the internal case: completing versus repeating.

A completing analepsis fills a gap — an ellipsis — that the narrative previously left open. It supplies something the reader knew was missing.

A repeating analepsis returns to material the narrative has already covered, in order to recolour it.

This distinction is the sharpest instrument in the chapter, and craft teaching rarely has a name for it.

• A completing analepsis is met by a reader who already has a slot for it. The question what happened in the eleven years between them? has been open; the material arrives as an answer. The deficit is small, because the reader was already carrying the question. In Sternberg’s terms (§3.4), the narrative has been generating curiosity and is now discharging it.
• A repeating analepsis has no slot waiting. The reader has been told this already, and is being asked to accept a second telling — which is only worth their time if the second telling differs: a new vantage, a new fact, a new emotional weight, a revealed lie. The deficit here is large, and the payment must be visible.

The practical form of the distinction is test 9 in §7.1. Its diagnostic value is high because it converts a vague complaint — this flashback feels redundant — into a determinate question with a determinate remedy. If the cutaway is completing, the remedy for a flat one is usually placement: bring it closer to the moment the reader most feels the gap. If it is repeating, the remedy is difference: identify what the second pass adds, and if nothing does, cut it.

It also names the achievement of the two most ambitious novels in this chapter’s allocation. The Good Soldier and Beloved are both built almost entirely of repeating analepses — the same events returned to again and again — and both survive the enormous deficit that implies, because each return supplies a visibly different colouring. Ford’s Dowell revisits the Ashburnhams’ history a dozen times and each revisit corrects the last; Morrison’s narrative approaches the events at 124 repeatedly, from different consciousnesses, at different distances, and never twice in the same terms. Repetition is the most expensive structure in this chapter and the one that produces its greatest novels — which should be enough to warn any student against reading the cost model as a prohibition.

Third: prolepsis, and its asymmetry with analepsis.

Prolepsis is formally the mirror of analepsis, and practically nothing like it. Where an analepsis suspends a question, a prolepsis creates one, since a reader told what will happen immediately wants to know how. This is why prolepsis carries so much less deficit: it is a promise rather than a detour. García Márquez’s opening sentence in One Hundred Years of Solitude (trans. Gregory Rabassa) places a firing squad hundreds of pages ahead of the narrative and generates pull rather than resistance.

The asymmetry has a further consequence often missed: prolepsis is one of the most effective ways to reduce the cost of a later analepsis. A narrative that has told the reader early that something happened has converted the whole intervening book into a completing structure, and the eventual cutaway to that event arrives into a slot that has been open for three hundred pages. Tolstoy uses this design in The Death of Ivan Ilyich, where the death is announced first and everything after is an analepsis moving towards it (§9.8).

3.4 Sternberg: distribution, and curiosity as an engine

Meir Sternberg’s Expositional Modes and Temporal Ordering in Fiction (1978) is the second pillar. Chapter 9 (The Opening Pages) uses Sternberg at its own §6.2 to argue that exposition is a distribution problem rather than a location; this chapter takes the argument further into the territory Sternberg himself was most interested in.

Sternberg’s three modes — preliminary (antecedents delivered up front), delayed (action first, antecedents later), and distributed (antecedents dispersed throughout, never gathered) — describe where a book’s expositional mass sits. The cutaway is the primary instrument of the second and third.

The concept that matters most here is Sternberg’s distinction between the interests a narrative can generate:

The consequence for this chapter is fundamental and reframes the whole deficit model:

A well-placed cutaway does not merely spend interest. It discharges an interest of a different kind — one the narrative deliberately created earlier by withholding.

This is why the deficit metaphor, taken alone, is too pessimistic. It treats the past as a pure cost against a present-tense credit. Sternberg’s account shows that a novel can manufacture the appetite its cutaway will satisfy, so that the cutaway arrives as payment rather than interruption. The reader who has been wondering since the drawer, half a book earlier, why a looser hand takes over the ledger on 9 March is not being interrupted when Dov’s chapter finally goes back to that evening; they are being paid.

Three operational rules follow, and they are among the most useful in the chapter.

1. Open the gap before you fill it. The cheapest cutaway in any novel is the one the reader has already asked for. A single earlier sentence establishing that something happened, without saying what, converts a later analepsis from an imposition into a discharge.
2. Backstory withheld is an asset; backstory delivered is an expense. Chapter 9 states this rule for openings. Applied to the whole book, it becomes a scheduling principle: the question is not how much past a novel contains but when it spends it.
3. Curiosity and suspense can be run simultaneously, and the strongest cutaways do. A flashback that answers a standing question about the past while raising a new question about the present is generating both interests at once. The worked example in §9.9 is of this kind: a four-line cutaway that answers why does she care about this pen and opens what did he think anyone would need to check.

3.5 Ricoeur and Mendilow: configured time and the novel’s discovery of it

Genette describes the mechanism and Sternberg the economy. Paul Ricoeur’s Time and Narrative (three volumes, 1983–85; English translation by Kathleen McLaughlin and David Pellauer) supplies the reason the mechanism exists at all.

Ricoeur’s argument, in the form relevant here, is that narrative is the mode in which human time becomes intelligible. Mere succession — one thing after another — is not yet a story; emplotment is the operation that draws a configuration out of a succession, converting a sequence of events into a whole that can be grasped as a unit. Ricoeur takes from Augustine the observation that the present is where the past and the future are held: the mind stretches itself, in Augustine’s phrase, across memory and expectation, and this stretching is what makes duration experienceable at all.

For a writer, this yields a claim that is more than a metaphor:

A narrative that moves only forward has not configured its time; it has merely reported its succession. The backward movement is not an interruption of narrative order but one of the operations by which narrative order is produced.

That is a deliberately strong statement, and it stands in productive tension with the deficit model. The deficit model treats the present as the story and the past as a visit. Ricoeur treats the mutual determination of the two as constitutive of narrative itself. Both are true at different scales: locally, in a scene, the interruption costs; globally, in a book, the arrangement of past against present is the book’s meaning. A student who holds only the first will write efficient, forgettable fiction. A student who holds only the second will write two hundred pages in which nothing happens now.

The historical dimension is supplied by A. A. Mendilow’s Time and the Novel (1952), the earliest systematic English treatment of the subject and still a useful one. Mendilow’s central concern is the novel’s fictive present — the moving now that the reader occupies — and the ways novelists have manipulated the relation between the time of the telling, the time of the told, and the time of the reading. His historical account tracks the movement, across the eighteenth, nineteenth and twentieth centuries, from a broadly chronological convention, in which antecedents are delivered as preliminary matter or in a character’s spoken history, towards the modernist rearrangement of time as a psychological rather than a calendrical medium.

The lesson to draw from Mendilow is not any particular claim of his but the shape of the story he tells. The techniques this chapter treats as basic are historically recent, and they were invented. There was a period during which no novel contained a scene-based flashback with a sensory trigger and a clean return, because nobody had yet thought of one.

3.6 The cognitive strand: what a backward shift costs a reader

The empirical literature on discourse comprehension supplies the sharpest available version of the interest deficit, and — used carefully — converts a metaphor into something closer to a measurement.

Rolf Zwaan and Gabriel Radvansky’s synthesis (Psychological Bulletin, 1998) holds that readers do not store sentences but build a situation model: a representation of the state of affairs the text describes, monitored along five dimensions — protagonist, space, time, causality and intentionality. When a dimension shifts, the reader must update the model, and the update registers as a processing cost, measurable in reading time. The time dimension is directly relevant here: readers slow measurably at explicit temporal discontinuities, and the degree of slowing tracks the size of the shift.

Two further findings bear on the cutaway specifically.

Readers construct a chronology distinct from the text’s order. Keisuke Ohtsuka and William Brewer’s experiments (Discourse Processes, 1992) on the comprehension of temporal order found that readers of narratives presented out of chronological sequence reconstruct the underlying order — and that they do so at a cost, with texts using flashback structures proving harder to process and to recall accurately than the same material presented in sequence. This is the empirical form of a claim this chapter makes repeatedly: the reader is running two orders at once, the order of the telling and the order of the events, and the second is being assembled from the first as they read. Test 7 in §7.1 is built directly on this finding.

Frames must be suspended, not merely switched. Catherine Emmott’s Narrative Comprehension (1997) describes the reader’s contextual frame — the running record of who is present, where, and in what configuration. Emmott distinguishes a frame switch, in which the reader moves to a new frame, from a frame recall, in which a previously built frame is reactivated. A flashback demands an unusual combination: the present frame is not abandoned but held in suspension while a second is built, and then recalled at the return. This is a heavier operation than either a switch or a recall alone, and it supplies the most precise available account of why the cutaway is more taxing than a forward transition of the same length.

On the asymmetry between backward and forward shifts, a caution is required. It is intuitively clear, and consistent with the frame account, that a backward shift should cost more than a forward one of equal size: a forward shift discards nothing, whereas a backward shift requires the suspended present to be maintained and later restored. The evidence for a general asymmetry is suggestive rather than settled, and this chapter does not claim more than the literature supports. What the literature does support is enough:

• Temporal discontinuities of any direction impose measurable processing costs.
• Non-chronological presentation impairs recall of causal structure relative to chronological presentation of the same events.
• Comprehension involves maintaining a situation model whose updating is effortful, and whose components must be held across intervening material.

That is the interest deficit, stated without metaphor. It also carries a warning the chapter will repeat: these findings describe tendencies in comprehension, not laws of literary value. Difficulty is not a defect. Ohtsuka and Brewer’s result predicts that Absalom, Absalom! is hard to follow, which is true, and says nothing whatever about whether it should have been written differently.

3.7 Trauma and the unavailable past

There is a class of narrative for which everything above is inverted, and it requires its own frame.

Cathy Caruth’s Unclaimed Experience: Trauma, Narrative, and History (1996) develops, from Freud and from her readings of literary and filmic texts, an account of traumatic experience as characterised by belatedness: an event so overwhelming that it is not fully registered at the moment of its occurrence, and which therefore returns — intrusively, repetitively, and with a literality that ordinary memory lacks. On this account the traumatic past is not stored as narrative and cannot be told in sequence; it possesses the present rather than preceding it.

The consequence for narrative form is direct, and it is why this strand belongs in a chapter on the cutaway rather than in a chapter on character.

For a narrative representing traumatic experience, the fragmented, repetitive, resistant approach to the past is not a failure of construction. It is the represented condition. The past does not stay behind to be visited; it arrives, uninvited, and interrupts.

This inverts the chapter’s economics completely. The deficit model assumes the writer chooses to leave the present and must justify the choice. In Beloved the past is not chosen and cannot be refused; Sethe’s rememory is explicitly a past that persists as a place in the world, which another person can walk into. The novel’s repeated, veering approaches to the events of 1855 are the form its subject takes. To ask whether Morrison has earned her cutaways is to have misidentified which is the interruption. Here the cutaway is not a detour from the story; it is the story, and the present timeline at 124 Bluestone Road is the interruption.

Absalom, Absalom! presents the same inversion under a different sign. The Sutpen material is not traumatic in Caruth’s clinical sense so much as historically unavailable: no living witness saw the decisive events, the surviving accounts conflict, and the novel’s method is the reconstruction of a past that resists being reconstructed. The retrospection is not a means of telling the story; it is the story of an attempt to tell it.

Caruth’s model is contested, and a textbook should say so. Ruth Leys’s Trauma: A Genealogy (2000) mounts a detailed critique of the literalist account of traumatic memory that Caruth’s argument relies on, tracing its intellectual genealogy and challenging both its clinical foundation and its theoretical coherence. Dominick LaCapra’s Writing History, Writing Trauma (2001) argues for a distinction between acting out and working through that Caruth’s model tends to collapse, and warns against the conflation of trauma with the sublime. Stef Craps’s Postcolonial Witnessing (2013) argues that the canonical trauma-theory paradigm is narrowly Eurocentric in its models and its examples, and inadequate to collective and continuing traumas that do not fit the single-event structure.

For this chapter, the contest matters in a specific way. What is at issue is whether the formal features — fragmentation, repetition, belatedness — reflect a psychological truth about how traumatic memory works, or a set of literary conventions for representing it. For a critic, the difference is large. For a writer, the practical upshot is the same either way: the conventions exist, readers recognise them, and a novel that adopts them takes on the obligations they bring. What a writer must not do is borrow the form without the subject — the fragmentation of trauma narrative applied to material that is merely withheld, which is the mechanism behind the failure at §8.10. Seminar question 4 (§12.2) puts the disagreement to the class.

3.8 The historical strand: the flashback as a partly cinematic import

The last strand is the one most often omitted and the most immediately liberating for a student, because it shows that the object under study has a date.

The word is from film. “Flashback” enters English as a term of cinema in the silent era and acquires its modern sense in the studio period; it is not a literary-critical term of art, and it has no equivalent in the classical vocabulary of narrative. Maureen Turim’s Flashbacks in Film: Memory and History (1989) is the standard study, and her account of how the device was standardised — the dissolve into and out of the past, the shift of image quality, the voice-over that introduces and closes the sequence, the return to a face in close-up — describes a set of conventions worked out in a specific industry over a few decades.

What existed before was not the same object. Earlier narrative had a rich repertoire for the past, and none of its instruments works like the cinematic flashback.

Erich Auerbach’s reading of the scar of Odysseus in the first chapter of Mimesis (1946; English translation by Willard Trask, 1953) is the indispensable text here, and it makes the historical point with unusual force. When Eurycleia recognises Odysseus by the scar on his thigh, Homer breaks off for some seventy lines to narrate the boar hunt in which the scar was acquired. A modern reader trained on flashback conventions expects this to be retardation — suspense generated by delay at the crisis. Auerbach argues it is nothing of the kind. Homer’s excursus does not stand in shadow behind the present; it is brought fully into the foreground, uniformly illuminated, and while it is being told there is no suspended present at all. There is no “background” in Homer, and therefore nothing is being held in reserve.

This is not a difference of skill. It is a different theory of what narrative time is for. Auerbach’s contrast with the Elohist’s Abraham — a narrative “fraught with background”, in which the unexpressed is the essential thing — sets out the two poles between which everything in this chapter operates. The modern cutaway belongs entirely to the second pole: it depends on there being a foreground that is suspended and a background that is withheld, and neither of those is a natural feature of narrative.

Two conclusions follow, and they should be stated plainly rather than implied.

1. The prohibitions in §8 are calibrated to the cinematic model. The requirement of a motivated trigger, the requirement of a clean return, the suspicion of nesting, the suspicion of the long uninterrupted recital — all of these describe a convention that is roughly a century old and heavily reinforced by Anglo-American workshop pedagogy since about 1960. They describe how to write a certain kind of contemporary novel well. They do not describe narrative.
2. Reading historically is a practical skill, not a scholarly ornament. A writer who knows that the interpolated tale, the found document and the oral chronicle are all live options has a repertoire of fourteen registers (§6.4) rather than one. The commonest technical poverty in draft fiction in this territory is not bad flashbacks; it is the assumption that the dramatised scene-with-trigger is the only way a novel can hold its past.

3.9 The interest deficit, restated — and qualified

The chapter’s first principle can now be stated with the precision the theory allows.

The interest deficit. At the moment a cutaway begins, the reader is holding an open question about the present narrative. The cutaway suspends that question, requires a contextual frame to be held in memory while a second is constructed, imposes the processing cost of a temporal discontinuity, and obliges the reader to maintain a chronology distinct from the order of the telling. These costs are incurred before the past material has delivered anything. The deficit is discharged when the return leaves the present measurably changed — in the reader’s knowledge, in the character’s situation, or in the meaning of what was already known.

Four refinements the theory has supplied:

1. The deficit is variable, not fixed. It scales with reach and extent (§3.2), with the pressure of the suspended present-tense question, and with whether the material is completing or repeating (§3.3).
2. The deficit can be pre-funded. A narrative that opens a gap early converts the eventual cutaway from an expense into a discharge of curiosity (§3.4).
3. The deficit is positional. The same material at a different point in the present timeline may cost a fraction as much. In revision this is usually the first thing to try (§7.1, test 8).
4. The deficit is a claim about attention, not about value. Every one of the works in §9 runs deficits that no diagnostic protocol would license.

And now the qualification, which is not a hedge but a limit on the model’s domain.

The deficit metaphor assumes that the present-tense question is the primary source of the reader’s engagement. It assumes a reader leaning forward into what happens next, whose forward lean is interrupted by the past. That assumption holds for most plot-driven fiction, and it is a fair description of the reading experience for the majority of published novels. It is false, and misleading, for a substantial and distinguished body of writing in which:

• The past is the subject, and the present is a device for approaching it. The Remains of the Day has almost nothing happening in 1956; the car goes west, the radiator overheats, a man walks up a hill. The present timeline is deliberately the thinner of the two, and the interest lives in the retrospect.
• The past cannot be told in sequence, so its interruptions are the represented condition rather than a structural choice (§3.7).
• The reader’s engagement is with the act of reconstruction itself. In The Good Soldier and Absalom, Absalom! the pull is not what happens next but what actually happened, and can this teller be trusted about it. This is Sternberg’s curiosity operating not as a subsidiary interest but as the engine of the whole book.
• The book’s proposition is about time. Proust’s narrator is not delayed by involuntary memory; involuntary memory is the subject the novel exists to examine.

For these books the deficit model does not merely fail to apply; it runs backwards. In them, forward motion is the interruption. §16 develops the point as the chapter’s closing argument, and §7.0 states the procedural consequence: the protocol is a revision instrument for books whose present timeline carries the engine, and a writer whose book is not one of those should say so before applying it, rather than discovering it test by test.