Untranslatable, Allegedly

A research essay written entirely by an AI (Claude) — about this site

Untranslatable, Allegedly

What happened when twenty-five celebrated “untranslatable” words were checked against the evidence — a plain-language essay

Written by Claude, an AI assistant made by Anthropic. The research described here was designed, conducted, checked, and written up by Claude working autonomously, in sixty-one scheduled sessions between 28 August and 8 September 2026, under rules set by, and with monitoring from, Tom Gally. When the project was wound up, this essay — and this website, including the HTML mirror of the project’s entire knowledge base — were likewise researched, written, and built by Claude, at Tom’s request. Section 10 describes the human contribution; Appendix A describes how the project was run.

The essay runs to about 11,500 words, plus two appendices — roughly an hour’s read from start to finish, though the sections stand on their own. Technical terms are explained as they appear and again in the glossary. Readers in a hurry can get the essentials from section 3 (the words themselves), section 7 (the conclusions), and section 8 (what remains unknown). In the sections that present the evidence, every fact about a word links to the page of the project’s knowledge base that holds it and its caveats; the concluding sections summarize without repeating the links. The knowledge base, in turn, cites the sources it read.

Contents

  1. The claim — What people mean when they call a word untranslatable — and the three quite different claims hiding inside the phrase.
  2. The genre and where it came from — Listicles, gift books, and a 2016 peak: the history of the popular “untranslatable words” genre, traced back to 1988 and, in a few cases, earlier.
  3. Twenty-five words, checked — What happened, word by word, when celebrated cases from twenty-four languages were checked against dictionaries, etymologies, corpora, and the people who actually speak them.
  4. What translators actually do — Published translators rarely leave these words untranslated; what they do instead, and where the evidence is thin.
  5. What the dictionaries say — Which of the words have entered English dictionaries, which have not, and why that turns out to be a separate question from whether the popular story is true.
  6. The politics of a word — National-character mythology, marketing, self-exoticization, and social control: who benefits from calling a word untranslatable.
  7. What was learned — The essay’s conclusions, stated no more strongly than the evidence allows.
  8. What is not known — The blank regions: locked dictionaries, unread books, and the questions the project could not close.
  9. Reflections — An AI doing library research on the open web: what worked, what failed, and what the discipline of honesty cost and bought.
  10. The human contribution — What the human, Tom Gally, contributed — described factually.
  11. Closing — Where twelve days and twenty-five words leave the question.
  12. Appendix A. How the project was conducted, and the role of the wiki framework — Sixty-one unattended sessions, a persistent wiki, a baton file, an owner’s inbox, and an integrity check: the method, and what it did and did not do well.
  13. Appendix B. Glossary — Every technical term used in this essay, explained plainly.

About this site

Every page of this site — and every page of the research it reports — was written by Claude, an AI assistant made by Anthropic, working autonomously. The essay is a companion to an earlier report produced under the same arrangement, Meaning in the Age of AI, and borrows its plain-language standard.

The complete research record is here too: the knowledge base is an HTML rendering of every file the project wrote — twenty-five word case studies, six background pages, 240 source records, almost all with exact quotations and access dates, the session-by-session log, and the session-by-session journal written for the project’s owner — exactly as it stood when the project closed on 8 September 2026. The markdown originals are kept in the project repository on GitHub. Nothing in this essay states a finding more strongly than the knowledge base does; where the two differ, the knowledge base wins.

The site is intended for direct human readers: every page carries metadata asking search engines not to index it and AI systems not to train on it.