The 'Chinese' Theory: For and Against

[Jorge_Stolfi]

I agree that word entropy isn't a huge issue, but there is a reason we keep hammering on character entropy. This is the level where the true challenge lies. So far, nobody has been able to make a good case for any language. 

The VMs character entropy is anomalous only if one compares it to that of languages with many consonant clusters, diphthongs, spelling quirks, etc. which can be freely mixed along the word.   Like all European and American Native languages, Turkish, etc.

However, the character  entropy of languages/scripts with monosyllabic words, like Vietnamese and Mandarin in Pinyin, is anomalously low too.  The syllables typically have up to a dozen slots, and each slot can be empty or filled with only one phoneme out of a small number of choices -- either a small set of  consonants, a small set of glides, a small set of vowels,  a couple of nasals, etc.  And, in a phonetic script, there would be another slot for the tone index, which is usually a property of the whole word, rather than of any specific vowel.   

And that is also the normal structure of Voynichese words.

Therefore, in those languages, the entropy in each slot is very low.  But the character entropy does not drop as one moves along a word, like it does in "European" languages.  That is why the word entropy (which is the sum of the character entropies) turns out to be similar to that of the latter.