Request Correction

A Computational Evaluation of Syllabic Hypotheses for Rongorongo: Evidence from N-gram Analysis

The study evaluates the hypothesis that the Rongorongo script of Easter Island functions as a syllabic substitution cipher where one symbol uniquely corresponds to one syllable. Using a genetic algorithm with a fitness function based on Rapa Nui n-gram statistics, we establish a performance baseline on controlled texts. Results show a strong correlation (0.75) between the algorithm’s fitness score and decipherment accuracy, identifying a "noise threshold" at 2,500,000 points. We further demonstrate that cross-corpus genre variance significantly impacts recovery rates, with accuracy dropping by more than half when mismatched linguistic statistics are applied. Application to the CEIPP transliteration yields scores well below the noise threshold (1.0M–1.3M), suggesting a lack of a simple syllabic signal. However, testing the rongopy transliteration produces scores in the "gray zone" (up to 2.7M) and reveals stable mappings for high-frequency glyphs 200 and 006. The consistency of these results across independent inscriptions suggests that while a pure syllabic model is insufficient, specific structural simplifications of the script may capture latent linguistic patterns.