待翻譯:A Glyph Is Not a Letter, a Token Is Not a Word, a Space Is Not a Space: What the Units of Voynichese Are Not
AI 服務暫時不可用,以下為來源摘要,待恢復後補全翻譯:arXiv:2608.17096v1 Announce Type: new Abstract: The Voynich manuscript (Beinecke MS 408) is usually analysed on three unstated assumptions: that its glyphs are letters, that the strings between blanks are words, and that every blank is a word space. We test all three against the Zandbergen-Landini transliteration with matched prose, cipher, and pseudo-text controls and quire-level resampling. None holds, and the failures share a shape: the order in Voynichese sits at the edges of tokens and at graded boundaries between them, not in the succession of tokens themselves. Glyph regularity is too strong for one-to-one substitution of any tested plaintext (conditional entropy 2.7 bits against about 3.5 for Latin, Italian, and English) and resolves instead onto a quire-stable scale of recurrent multi-symbol units. Tokens form a plausible vocabulary, yet the identity of one token predicts the next by under 1% of token entropy, below every matched control (2-10%), while the glyphs at token edges share 0.2 bits of mutual information, more than in any prose control. Blanks fall into two regimes: the separators transcribers marked uncertain behave like word-internal junctures, are physically narrower on the page (AUC 0.905 from independent image coordinates, with the same sign in a small blind ink audit), and are crossed by learned units even when every space is erased before learning. This profile is also what discriminates. A published Voynich-imitating cipher and a self-citation text generator both reproduce the low entropy, the unit scale, the weak token order, and the null result of a calibrated substitution attack; neither reproduces the edge-glyph coupling or the open, hapax-rich vocabulary (70% singleton types against 41% and 59-60%). Any account of the manuscript must therefore earn, rather than assume, the step from glyphs, tokens, and separators to letters, words, and word spaces, and these are the measurements on which to do so.
AI 服務暫時不可用,以下為來源正文,待恢復後補全翻譯。
--> [Submitted on 17 Aug 2026] Title:A Glyph Is Not a Letter, a Token Is Not a Word, a Space Is Not a Space: What the Units of Voynichese Are Not View a PDF of the paper titled A Glyph Is Not a Letter, a Token Is Not a Word, a Space Is Not a Space: What the Units of Voynichese Are Not, by Liudmila Rozanova and 1 other authors View PDF HTML (experimental) Abstract:The Voynich manuscript (Beinecke MS 408) is usually analysed on three unstated assumptions: that its glyphs are letters, that the strings between blanks are words, and that every blank is a word space. We test all three against the Zandbergen-Landini transliteration with matched prose, cipher, and pseudo-text controls and quire-level resampling. None holds, and the failures share a shape: the order in Voynichese sits at the edges of tokens and at graded boundaries between them, not in the succession of tokens themselves. Glyph regularity is too strong for one-to-one substitution of any tested plaintext (conditional entropy 2.7 bits against about 3.5 for Latin, Italian, and English) and resolves instead onto a quire-stable scale of recurrent multi-symbol units. Tokens form a plausible vocabulary, yet the identity of one token predicts the next by under 1% of token entropy, below every matched control (2-10%), while the glyphs at token edges share 0.2 bits of mutual information, more than in any prose control. Blanks fall into two regimes: the separators transcribers marked uncertain behave like word-internal junctures, are physically narrower on the page (AUC 0.905 from independent image coordinates, with the same sign in a small blind ink audit), and are crossed by learned units even when every space is erased before learning. This profile is also what discriminates. A published Voynich-imitating cipher and a self-citation text generator both reproduce the low entropy, the unit scale, the weak token order, and the null result of a calibrated substitution attack; neither reproduces the edge-glyph coupling or the open, hapax-rich vocabulary (70% singleton types against 41% and 59-60%). Any account of the manuscript must therefore earn, rather than assume, the step from glyphs, tokens, and separators to letters, words, and word spaces, and these are the measurements on which to do so. Comments: 33 pages, 7 figures, 3 appendices. Analysis code and data are included as ancillary files and mirrored at this https URL Subjects: Computation and Language (cs.CL) Cite as: arXiv:2608.17096 [cs.CL] (or arXiv:2608.17096v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2608.17096 arXiv-issued DOI via DataCite (pending registration) Submission history From: Alexander Temerev [view email] [v1] Mon, 17 Aug 2026 20:05:12 UTC (6,897 KB) Full-text links: Access Paper: View a PDF of the paper titled A Glyph Is Not a Letter, a Token Is Not a Word, a Space Is Not a Space: What the Units of Voynichese Are Not, by Liudmila Rozanova and 1 other authors View PDF HTML (experimental) TeX Source view license Ancillary-file links: Ancillary files (details): README.md analysis/direct_pixel/analyze_direct_morphometry.py analysis/direct_pixel/measure_direct_pixels.py analysis/reproduce_cipher_calibration.py analysis/reproduce_dialect_linestart.py analysis/reproduce_direct_pixel.py analysis/reproduce_edge_order.py analysis/reproduce_headlines.py analysis/reproduce_naibbe_control.py analysis/reproduce_scale_transition.py analysis/reproduce_selfcitation_control.py analysis/reproduce_space_sensitivity.py analysis/reproduce_unit_inventory.py analysis/reproduce_unit_scale.py data/controls/naibbe/LICENSE_greshko.txt data/controls/naibbe/PROVENANCE.md data/controls/naibbe/greshko_naibbe_tables.csv data/controls/naibbe/greshko_nathist_output_ciphertext.txt data/controls/naibbe/greshko_nathist_pre_encryption_respaced_plaintext.txt data/controls/naibbe/naibbe_caesar_ciphertext_lines.txt data/controls/naibbe/naibbe_tables.json data/controls/selfcitation/selfcite_seed1.txt data/controls/selfcitation/selfcite_seed2.txt data/controls/selfcitation/selfcite_seed3.txt data/direct_pixel/estimator_robustness.csv data/direct_pixel/results_unblinded.csv data/direct_pixel/threshold_sensitivity.csv data/v101/voyn_101.txt data/voynich-units/ZL3b.txt data/voynich-units/herbal/culpeper_en.txt data/voynich-units/herbal/pliny_the_elder.naturalis_historia.part.4.books_16-20.tess data/voynich-units/herbal/pliny_the_elder.naturalis_historia.part.5.books_21-25.tess data/voynich-units/herbal/pliny_the_elder.naturalis_historia.part.6.books_26-30.tess data/voynich-units/italian/26166.txt data/voynich-units/italian/27608.txt data/voynich-units/latin/caesar.de_bello_gallico.part.1.tess data/voynich-units/latin/caesar.de_bello_gallico.part.2.tess data/voynich-units/latin/caesar.de_bello_gallico.part.3.tess data/voynich-units/latin/celsus.de_medicina.part.1.tess data/voynich-units/latin/celsus.de_medicina.part.2.tess data/voynich-units/latin/celsus.de_medicina.part.3.tess data/voynich-units/latin/celsus.de_medicina.part.4.tess data/voynich-units/latin/celsus.de_medicina.part.5.tess data/voynich-units/morphometry_voynichese/voynichese_boxes/f100r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f100v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f102r1.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f102r2.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f102v1.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f102v2.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f103r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f103v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f104r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f104v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f105r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f105v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f106r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f106v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f107r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f107v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f108r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f108v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f10r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f10v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f111r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f111v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f112r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f112v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f113r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f113v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f114r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f114v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f115r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f115v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f116r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f11r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f11v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f13r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f13v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f14r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f14v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f15r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f15v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f16r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f16v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f17r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f17v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f18r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f18v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f19r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f19v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f1r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f1v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f20r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f20v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f21r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f21v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f22r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f22v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f23r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f23v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f24r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f24v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f25r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f25v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f26r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f26v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f27r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f27v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f28r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f28v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f29r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f29v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f2r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f2v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f30r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f30v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f31r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f31v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f32r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f32v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f33r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f33v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f34r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f34v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f35r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f35v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f36r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f36v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f37r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f37v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f38r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f38v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f39r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f39v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f3r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f3v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f40r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f40v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f41r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f41v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f42r.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f42v.js data/voynich-units/morphometry_voynichese/voynichese_boxes/f43r.js d [truncated for AI cost control]