Inside the Voynich Network: graph analysis

[Jorge_Stolfi]

I have calculated normalized KPI for Voynich texts, natural language texts and generated texts.

I am intrigued  by how different Vietnamese comes out in all those plots.  It may be correct, but maybe it is something banal in the file format that is messing up the counting. 

Just to clarify, the file I gave you uses VIQR, a pre-Unicode encoding that uses many funky special characters to indicate the many diacritics of Vietnamese spelling; except that some accents were  remapped to avoid confusion with puctuation: "?"->"ß",   "("->"µ",    "."->"°".  Here is a sample:

sao ddu+'c ye^su no'i vo+'i o^ng giaß nhu+ ta muo^'n no' o+ß la°i cho

to+'i khi ta dde^'n thi` vie^°c gi` dde^'n ngu+o+i pha^`n ngu+o+i ha~y
cu+' theo ta va^°y co' lo+`i ddo^`n dda°i giu+~a anh em la` mo^n ddo^`
a^'y kho^ng phaßi che^'t nhu+ng ddu+'c ye^su dda~ kho^ng no'i vo+'i

Everything in the sample above other than space is to be treated as a letter.

Here is a Vietnamese sample again (maybe the same, not sure). It is a translation of the New Testament, probably from English.  This version is in the Latin-1 encoding, with upper case mapped to lower case,
all (actual) punctuation removed.  It is in 3 files because it the whole is  ~490'000 bytes and the forum limits each attachment to 200'000 byes.


  nwt1.txt (Size: 158.55 KB / Downloads: 8)

  nwt2.txt (Size: 158.21 KB / Downloads: 5)

  nwt3.txt (Size: 162.32 KB / Downloads: 7)

If you would rather have it in some other format (UTF-8, one word per line, whatever) just let me know.

All the best, --stolfi