Fluent chant, dead reading

Claim (verbatim)

The Peruvian slave raids of 1862–63 carried off well over a thousand Rapanui, the literate chanters among them, and the epidemics that followed took the island to 111 people by 1877; whatever rongorongo literacy was, it died in that decade. So when Bishop Tepano Jaussen sat the Rapanui labourer Metoro Tauʻa Ure over the tablets of his Tahiti collection around 1873 — Aruku Kurenga (B) and Mamari (C) among them — and Metoro produced fluent chant, the question is what was performing: a reading, or recitational competence draped over pictures. Barthel (1958), who printed the chants aligned to the glyph lines, mined them for genuine readings; Jacques Guy and others have argued they are improvised picture-naming. The claim takes the sceptical side and gives it a measurable signature: picture-naming is stable at the sign (a nameable shape gets a same-ish name each time) and unstable at the string (the same sign sequence fails to elicit the same wording twice), which is the inverse of what reading produces — and so the Jaussen material, useless for decipherment, becomes a measurement of the loss itself. Prediction: aligning Barthel's edition of Metoro's chants to the glyph lines, sign pairs that recur three or more times within the chanted tablets will receive lexically matching glosses at all their occurrences in fewer than half of cases, while pictorially transparent single signs keep a stable modal gloss across sessions (primary clause: the sub-50% repeated-bigram gloss consistency; the verdict follows it). Exact computation: a consistency tabulation over the aligned chants, bigrams typed by Barthel base numbers, glosses matched on content words. Kill: the Jaussen list (Jaussen, L'île de Pâques: historique et écriture, 1893) and Barthel's aligned edition of Metoro's chants in Grundlagen zur Entzifferung der Osterinselschrift (1958), with the session history as documented in Fischer's Rongorongo: The Easter Island Script (1997).

Prediction clause (verbatim)

Prediction: aligning Barthel's edition of Metoro's chants to the glyph lines, sign pairs that recur three or more times within the chanted tablets will receive lexically matching glosses at all their occurrences in fewer than half of cases, while pictorially transparent single signs keep a stable modal gloss across sessions (primary clause: the sub-50% repeated-bigram gloss consistency; the verdict follows it). Exact computation: a consistency tabulation over the aligned chants, bigrams typed by Barthel base numbers, glosses matched on content words.

Kill-dataset (verbatim)

Kill: the Jaussen list (Jaussen, L'île de Pâques: historique et écriture, 1893) and Barthel's aligned edition of Metoro's chants in Grundlagen zur Entzifferung der Osterinselschrift (1958), with the session history as documented in Fischer's Rongorongo: The Easter Island Script (1997).

Novelty / leakage triage

anticipated in the literature — this exact test has never been run

Guy's 1999 study argued exactly this direction: Metoro improvised picture-namings rather than reading — he chanted the Mamari calendar backwards and missed the full-moon pictogram — and the Jaussen material cannot ground decipherment; Fischer documents the sessions. The proposed measurement is new: no one has tabulated sign-level versus string-level gloss consistency (the sub-50% repeated-bigram criterion) over Barthel's aligned edition of the chants.

• J.B.M. Guy, 'Peut-on se fonder sur le temoignage de Metoro pour dechiffrer les rongo-rongo?', Journal de la Societe des Oceanistes 108 (1999), 125-132
• S.R. Fischer, Rongorongo: The Easter Island Script (Oxford, 1997)

Predictions

No prediction registered yet.