The model in this chapter, with its two channels, twelve mechanisms and ten tests, describes a particular and datable ideal of dialogue: compressed, indirect and heavy with subtext. That ideal took shape in the reception of Chekhov, in Hemingway’s stories and in the writing workshops that codified both, and it is a technical achievement worth learning, because it engages the reader’s inference and produces character without assertion. It is not a description of dialogue as such. Austen’s gap was social, a matter of what could be said in a given drawing room, not psychological. Dostoevsky’s characters say what they think, at length, and the gap lies between their self-accounts and the truth. Compton-Burnett’s speak a syntax no one has spoken, and Wilde’s say what no one says aloud. Each of these would fail several of the tests, and the failures are why they are read.
The tests therefore identify what an exchange is doing, not whether it is good. A writer who turns their findings into verdicts will cut the passages where the conventions are under strain, which is often where the book is most alive. What survives the change of fashion is not the preference for obliquity but the social grammar underneath it, which held for Austen and holds for McCarthy: speakers want things from each other, protect each other’s face until they stop, mark their refusals, and design what they say for everyone in the room. Dialogue may be as stylised as a book requires, provided it still does what talk does.
See also: the Reference coda, which develops the historical argument; The Descriptive Detail (Chapter 4), the next chapter, on the detail that means more than it names.