A Reproducible Multi-Language Statistical Test of Phonetic Readings for the Phaistos Disc

The Phaistos Disc (ca. 1700 BCE) carries 241 sign impressions of 45 types arranged in 61 groups. Davis (2018) showed statistically (p = 1-3%) that the Disc and Linear A likely encode the same language but did not assign phonetic values. We complete that step by applying values from the published Daidalika Disc-Linear A signary (G. Owens, TEI Crete; Owens 2007) and testing the resulting reading against nine Bronze Age language hypotheses under a single Monte Carlo framework (100,000 iterations per language, seeded PRNG, uniform and frequency-weighted nulls). Eight hypotheses (Mycenaean Greek, Luwian, West Semitic, Proto-Finno-Ugric, Middle Egyptian, Hurrian, Etruscan, and Hittite via Rasmussen's independent sign values) show no evidence of relationship (uniform p = 0.101-1.000; FW p = 0.895-1.000). Only Minoan/Linear A survives under the uniform null (p = 0.032). The Minoan result does not survive Holm-Bonferroni correction for nine tests (p = 0.286), the FW null (p = 0.706), the alternative phonetic treatment of sign 02 (p = 0.056), or a leave-PD-inspired-out ablation that drops 10 of 34 LA reference entries flagged in the source data as curated against PD parallels (filtered p = 0.261). A uniform circularity control (p = 0.0047) confirms that the specific sign correspondences outperform random reshufflings of the same syllable inventory; a frequency-weighted reshuffling (p = 0.365) reveals that much of the overlap is attributable to frequency correlation. The eight negative verdicts are robust across all controls. The reading also exhibits productive affixation, a -ti suffix pattern (13/61 words), and a complementary qe prefix pattern paralleling known Linear A morphology. All analyses are implemented in an interactive open-source browser workbench enabling full independent replication.