The 'Chinese' Theory: For and Against

[Jorge_Stolfi]

The only thing your analysis shows is that a broader claim like “sh is generally higher everywhere” is refuted. But as I said, that’s not what I claimed in the first place.  However, you’re not refuting my actual finding regarding the matched form pairs under investigation.

If q-attraction was indeed stronger for sh-words than for sh-words, independently of the suffix, that would be a strong argument against "unencripted natural language" -- not just East Asian, but any European language as well.  

(It could perhaps be a Semitic or Bantu language, where grammatical inflections affect prefixes rather than suffixes, and thus the prefix of a word may affect the next-word frequencies.  But even there  I bet that we would not see an absolute dominance.)

But since we agree that the q-attraction difference between sh and ch depends on the whole words, not just on their initial letters, this result no longer contradicts the "natural language" theory.  Instead it is even one more bit of evidence for it.  Again, which words can follow any given word is normally determined by grammar and semantics, and these attributes depend on the whole word, not on the letters that it happens to use.

Is this phenomenon compatible with the Chinese one?

There are many statistical "oddities" of the SPS text that would have to be explained eventually.  

Even if not absolute, the larger q-attraction of shXXX compared to chXXX for many common XXX suffixes is strange.   

Someone recently complained that the SPS has too few repeated words and phrases compared to the SPS (which is ironic, since one of the arguments against "natural language" used to be "the VMS has too many repeated words and phrases)

But even more intriguing than your q-atraction differential is why the absolute counts of shXXX and chXXX are so similar, for that set of suffixes XXX that I listed (and maybe others, I haven't checked).  At the moment I cannot think how that could happen under the "unencrypted natural language" thesis.   

If the numbers were just proportional, it could be due to random loss of the plume from the sh; but for the counts to be equal that loss had to have happened with exactly 50% probability.    I thought of some weak cipher that swaps ch- and sh- on evey other word; but that would have broken the 气 = chedy crib that seem to work qute reliably.  Or maybe the author swapped ch- and sh- when he switched from Language A to Language B, and there is just about the same amount of parag text in each language...

But I am violating my own advice #2 here. I should have checked whether the same occurs in East Asian texts, at least in the pinyin reading of the SBJ.  That is, whether there are initial consonants A and B, and a set ZZZ of suffixes,  such that the token counts for Azzz and Bzzz are almost equal, for any suffix zzz in that set.  

According to my advice, I may well be surprised. Like I was ~25 years ago, when I set out to prove that Voynichese was not natural language because You are not allowed to view links. Register or Login to view. -- and then, almost as an afterthought, I did that plot for Vietnamese...

That said, for many other oddities (like the number of repetitions) the problem is not the lack of explanation, but that there are many possible explanations.  The mapping of the SBJ to the SPS was affected by many processes and variables, like pronunciation variation, sandhi, phoneme confusion by the Author, multiple ways to encode the tones, selective transcription errors, uncertain word spaces, etc.   Sometimes it seems that either the Author or the Scribe could not decide whether a syllable was (say) okedy or okeedy, so he wrote down both.  It would be futile to guess a combination of those factors that explained each oddity.

So my plan now is to first produce enough proof of SPS=SBJ to convince everyone.  The most sensible path after that seems to be to figure out the language, and how its phonemes were meant to be encoded in the Voynichese script.  Then we would be in a better position to explain those statistical oddities.

All the best,--stolfi