Transition Grammar and Cognitive Domain Profiles
in the Voynich Manuscript:
Evidence for a Functional Recording System from
Six-Role Cluster Analysis
Roble Mumin
∗1
1
Independent AI Researcher and AI Consultant
Manuscript Draft, March 2026 — v0.98
Abstract
Building on the functional cluster classifica-
tion established in Mumin (2026a), we ex-
tend a six-role taxonomy—derived from the
original seven-role framework through an em-
pirical model selection process—to character-
ize the transition grammar and domain struc-
ture of the Voynich Manuscript. The six
roles (R1 Initiator-like, R2 Closure-like, R3
Link-like, R4 Content, R5 Temporal-like, R6
Reference-like) are defined entirely by transi-
tion statistics and positional behavior among
the highest-frequency clusters (25.8% of cor-
pus tokens; 47 classified clusters)
1
; the sin-
gle strongest structural signal is the R2→R1
transition (z= +9.71vs. 1,000 shuffled
baselines), which anchors the packet gram-
mar. At paragraph level, 61.3% of role
transitions conform to a finite-state packet
grammar (prospectively specified threshold:
60%), a result consistent with—but not con-
clusive evidence for—a packet-level gram-
mar. Entropy decomposition supports the
dual-channel prediction: role-sequence en-
tropyH(structural) = 2.38bits is lower than
within-role cluster entropyH(variant) = 2.77
bits. Section-specific role profiles differ sig-
nificantly across all eight manuscript sec-
tions (Kruskal-Wallis,k= 7tests, allp <
0.0001;…打开文档引用此文档
Roble Mumin “Voynich Grammar and Domain Profiles”.
2026-03.
访问地址：文档页面
current summary qualified; independent peer review not verified