Louis Menand’s review of two new writing books — Florence Hazrat’s On the Mark and Sarah L. Kaufman’s Verb Your Enthusiasm — is mostly about a quieter claim: changes in the technology of writing always track changes in who gets to read. Punctuation exists to make writing easier to read; easier reading means more readers; the audience widens and the old monopoly on literacy breaks.

His best evidence is textual. The famous commas in the opening of Pride and Prejudice do no grammatical work at all — no non-restrictive clause, no series, no natural pause. And Austen probably didn’t write them.

  • Austen’s surviving manuscripts show little comma or paragraph use; she used dashes. The published Persuasion replaces five dashes and three ampersands and lowercases four nouns.
  • The grammar of her later novels is believed to have been cleaned up by William Gifford, an editor at her publisher John Murray. The polished, quasi-epigrammatic sentences we call Austenian “may not be sentences Austen actually wrote.”
  • The period was invented by Aristophanes of Byzantium, head of the Library of Alexandria, in the second or third century BCE.
  • The earliest known question mark is in a fifth-century document in Syriac, a language close to Aramaic. The exclamation mark dates from the mid-fourteenth century.
  • The printing press, around 1440, “let loose a flurry of signs” — semicolon, quotation marks, dash, ellipsis. No new marks since 1600, unless you count emojis.
  • Hazrat’s candidate for the greatest world-changing mark is the space, invented in the eighth century by Irish monks to separate words.
  • Menand’s verdict on Hazrat’s ambition (“the history of punctuation is the history of humanity”) is that the genre invites overreach — but the narrow version holds: writing technology correlates with the distribution of social power. “Literacy is no longer the monopoly of a privileged class. Today, even the gatekeepers are gone, or have turned into paywalls.”

The AI section is why the piece is worth keeping. We have long treated writing as substitute speech — imagining a person behind the words, saying “That’s Jane Austen.” Once programs generate Austen-like sentences at volume, that anthropomorphizing stops making sense, and words may come to be handled the way we handle numbers: as marks to decode.

The closing distinction is the sharp one. The usual worry is that we mistake a machine’s poem for a person’s. The more dispiriting prospect is that we dismiss a person’s poem as AI slop.