Scott Alexander, curated
← Back to curation

Your Book Review: How Language Began

Quality
80
Excellent
Claude Shift
58
Moderate
RWI
2
of 10

Summary

Contest-finalist review of Daniel Everett's How Language Began, structured as a war between two gods: Chomsky (whose dominance is measured in the comic unit of 'Chomskys' -- 500k citations -- and whose contrarian doctrines are grammar-supremacy, recursion, sudden Merge-mutation origin, and the E-language/I-language dodge) and Everett, the anti-Chomsky. Tells Everett's story (born-again missionary -> accidental linguist among the Piraha -> loses his faith twice, in God and in Chomsky), the immediacy-of-experience principle and the Pirahas' refusal of recursion/creation-myths/hearsay (the Jesus anecdotes), and the vicious recursion-war backlash. Then the book's thesis: language is an invented cultural technology, not innate -- symbols via Peirce's index->icon->symbol progression, gradual emergence from Homo erectus a million-plus years ago, no language-specific brain area. Closes with the sharp modern payoff: LLMs' statistical-learning success vindicates the cultural/communication view over Chomsky, though Chomsky might still describe what's inside them.

Why this score

Quality 80 · Excellent. Excellent. A wide-ranging, information-dense, genuinely gripping review that does real work -- linguistics history, the Piraha ethnography, the recursion feud, and a timely LLM payoff -- with vivid writing and the memorable 'Chomskys' device. The strongest guest review of the run; clears Excellent on scope and synthesis.

Claude’s paradigm shift 58 · Moderate. Notable. The content is Everett's/the field's; the reviewer's fresh contribution is the synthesis and especially the LLMs-vindicate-statistical-learning framing.

Real-world impact 2 · Minor. Minor. The Chomsky-vs-statistical-learning debate is consequential for AI/linguistics, but this review's own reach is discourse-level. Within-niche.