10. Dialogue Across Forms

The model was developed for the novel. It transfers with adjustments that need stating, because students routinely import novel-dialogue habits into forms that punish them.

10.1 The short story

The story has proportionally less room and a different relation to its own exposition, and four consequences follow.

The important asymmetry is the first. A novel can teach the reader that a character deflects and then, three hundred pages later, cash the training. A story must produce the same effect by convention — the reader arrives already knowing that in a literary short story, a conversation about hills is not about hills. This is a real dependency on genre expectation, and it is why the form’s dialogue reads as mannered when transplanted.

10.2 The novella and the long novel

The two long forms fail in opposite directions.

The novella’s characteristic dialogue problem is that it is written by novelists: an exchange that takes four pages to arrive at its turn has spent a fiftieth of the book on approach. The novella wants exchanges that begin at the second or third turn — the discipline of §2.2 applied ruthlessly.

The long novel’s problem is the inverse and less often named. Over eight hundred pages, a subtext-heavy dialogue practice becomes monotonous, and the great long novels do not sustain one. War and Peace, Middlemarch and The Brothers Karamazov all alternate between exchanges of extreme obliquity and exchanges in which characters simply argue at length. The alternation is what makes the oblique passages readable. A book’s dialogue practice must vary across its length for the same reason a page’s density must vary across a page — emphasis is relational, a point Chapter 9 (The Opening Pages) makes about the density ceiling and which applies here without modification.

10.3 Series and sequence

Series dialogue faces a specific hazard: the relational domain is the series’ principal accumulating asset, and it is the one most easily squandered. What makes a fourth-volume exchange between two recurring characters land is three volumes of history, and the temptation is to spend that history in every book by having the characters refer to it.

Three principles.

1. Re-establish the voices completely and the history not at all. Readers who finished the previous volume eighteen months ago retain the sound of a character and almost none of the plot.
2. Recap in the relational channel, not the informational one. Not after what happened in Prague, but two people behaving towards each other in a way that only a shared history explains.
3. Ration the gap-closing move across the whole sequence, not per volume. A series in which each book ends with a character finally saying what they mean has trained the reader that they always will.

10.4 The screen comparison, and a caution about borrowing

Contemporary craft advice on dialogue is imported heavily from screenwriting, and most of it imports badly, for one structural reason: the screenplay has no narrator and the novel does.

Two specific mis-imports need naming. “Enter late, leave early” transfers cleanly and is among the best advice in the craft literature; it is the discipline of §2.2. “Let the subtext play” does not transfer, because on screen an actor supplies what the line withholds, and on the page nothing supplies it unless the prose does. A novelist who writes a bare exchange expecting the reader to feel what an actor would have shown has written a screenplay without a cast.

The third and least noticed difference is that screen dialogue is heard and prose dialogue is read, which is why disfluencies work in one and not the other (§3.6): an um is inaudible in performance and conspicuous in print.

10.5 Translated fiction

The series’ standing distinction applies with unusual force here. Structural claims about dialogue survive translation; texture claims do not, and register claims survive only partially.

What transfers:

• The sequence of turns, and therefore every adjacency-pair and preference-organisation effect (§3.2).
• Who evades, who deflects, who is interrupted, who is silent — the whole of §4.6 and most of §6.4.
• Status moves realised through acts: correction, interruption, refusal, the granting or withholding of an exit.

What does not:

• Idiolect and the transcript test. A translator produces the dialogue in their own language’s registers, and voice differentiation is remade rather than transferred. Garnett’s Dostoevsky, Pevear and Volokhonsky’s Dostoevsky and the intervening versions differ substantially in how alike the brothers Karamazov sound.
• Dialect, wholly. There is no equivalence between a regional English and a regional Russian, and translators’ solutions — invented dialects, class-marked English registers, standardisation — are all lossy, and are all interpretations.
• Address systems. Languages with a T/V distinction carry a relational value in every second-person pronoun that English cannot render without an added clause. The move from vous to tu is a relational event of the kind §4.3 describes, and in English it must be replaced by something else or lost.

The practical instruction for a student: when analysing translated dialogue, run tests 1–6 and 10 with confidence, and run tests 7, 8 and 9 only with the translator’s name attached to the finding.

10.6 Concordance: concept to novel

A study aid: each concept with the work that displays it best, and the section that treats it.