Fifty-two signs with a stutter

Claim (verbatim)

Barthel's catalogue gave rongorongo some six hundred numeric codes; Konstantin and Igor Pozdniakov ('Rapanui Writing and the Rapanui Language', Forum for Anthropology and Culture, 2007) argued the working inventory collapses to about 52 basic signs covering 99.7% of the corpus, with a frequency profile matching the syllable statistics of the Rapa Nui language. If that is right, the language should show through the wood in a second, independent way: Rapa Nui, like its Polynesian sisters, reduplicates pervasively — partial CV- and full CVCV-reduplication are core morphology in the reference grammar (Kieviet, A Grammar of Rapa Nui, Language Science Press, 2017) — so a syllable-riding script must carry an excess of immediate sign-doubling (AA) and alternating tetragrams (ABAB) that no unordered tally or ornament system produces. The claim: the corpus stutters where the language stutters, and the stutter concentrates in a small set of sign types — the texture of chant, not of decoration. Prediction: over the corpus reduced to the Pozdniakov basic inventory, AA and ABAB motifs will jointly occur at no less than twice their expectation under within-text order shuffling (p < 0.001), with a minority of sign types carrying the majority of doublings (primary clause: the twofold joint excess of AA and ABAB motifs against the permutation null; the verdict follows it). Exact computation: per-text permutation test over sign sequences, motif counts pooled across the corpus. Kill: the Barthel-coded corpus as digitized by the CEIPP, reduced per Pozdniakov and Pozdniakov (Forum for Anthropology and Culture 3, 2007), with the reduplication baseline taken from the open-access Kieviet grammar (Language Science Press, 2017).

Prediction clause (verbatim)

Prediction: over the corpus reduced to the Pozdniakov basic inventory, AA and ABAB motifs will jointly occur at no less than twice their expectation under within-text order shuffling (p < 0.001), with a minority of sign types carrying the majority of doublings (primary clause: the twofold joint excess of AA and ABAB motifs against the permutation null; the verdict follows it). Exact computation: per-text permutation test over sign sequences, motif counts pooled across the corpus.

Kill-dataset (verbatim)

Kill: the Barthel-coded corpus as digitized by the CEIPP, reduced per Pozdniakov and Pozdniakov (Forum for Anthropology and Culture 3, 2007), with the reduplication baseline taken from the open-access Kieviet grammar (Language Science Press, 2017).

Novelty / leakage triage

already answered in the literature

Pozdniakov and Pozdniakov 2007 already ran this comparison: they report doubled glyphs (AA) at only about 8% above chance in the corpus against about 50% above chance for doubled syllables in Rapanui, and ABAB glyph sequences at about twice chance against sevenfold for the language's reduplicated disyllables. The conjecture's joint twofold-excess prediction is directly answerable from — and partly contradicted by — these published numbers; the reduplication-signature test is not open.

• K. Pozdniakov and I. Pozdniakov, 'Rapanui Writing and the Rapanui Language: Preliminary Results of a Statistical Analysis', Forum for Anthropology and Culture 3 (2007), 3-36

Predictions

No prediction registered yet.