A key to understand the VMS

[Torsten]

Many thanks for the effort!

I don't think that the conclusion can be drawn just yet.

The figure is defined by edit distance, and the edit distance of course depends fully on orthography.

For this Chinese text, upper case and lower case have been kept seperate, and as a result the very frequent
word 'wo3' ends up in two spots at almost opposite points in the figure.

The intonation can be represented in different ways. Here, it is done with diacritics over the vowel, which is of course a valid option, but I wonder (from a closer look at the figure) if it was consistent.

Also, this graph considers composite words (two syllables) while the structure in Mandarin occurs at the level
of syllables.

The problem with these types of experiments is exactly the large number of permutations that one could try.

Yes, the graph is not perfect. The same is true for the graph of the VMS. For instance we have the limits of the transcription alphabet used. Because of this EVA alphabet words like [shain] and [sain] are handled as similar but not words like [dain] and [chain]. We also know for sure that the transcription contains errors. But all this detail problems doesn't change the big picture. For languages the graphs show smaller networks and in the case of Chinese the networks for short words are connected to a larger network. The graph for the VMS is obviously different since he shows shows only one network. This means that all the words follow the logic of the VMS. The VMS contains only 229 word types which differ in more then two letters from any other word type  (see You are not allowed to view links. Register or Login to view.). But even for this word types it is possible to split them into two or more common VMS words.
If we would assume that the VMS text contains language and that a word of the VMS stands for a word in this language this would imply that the text of the VMS didn't contain any loan words. A language that didn't use names. In my eyes this is a contradiction.

There was indeed a problem with two letter words. I have removed the rule not to compare words with two or less letters:

   


  VoynichPinyin.gephi.zip (Size: 106.45 KB / Downloads: 42)