The evolution of P/F vs T/K raises important questions

[Jorge_Stolfi]

Therefore, in Stars, Herbal-A, and Herbal-B the first word of each parag (and possibly also the next few words) are very likely to be the name of the thing which the parag is about. As such, each of those words is very unlikely to be used again in the same parag.  It may be used a few times in other parags

But wouldn't this also be true for the K/T works? The point is that the P words behave differently than the K/T words in the same position.

What do you mean by "differently" and "same position"?

When you say that something is 'puffed' -- do you mean that a K or T is converted to a P, or that the P is just randomly appended to a line? At the very least, it seems like there are some unknown constraints on which words are 'Puffable'.

Indeed, I am still confused about how the "puffing" works in each of the three uses (1-3).  But I do believe that "puffing" mostly means replacing a gallows or gallows+e by p or f. Maybe it could also replace other letters, like d.

I don't think that type (2) puffing -- highlighting the head lines -- is deterministic, but it may not be purely random either.  There is that 5-10 word "puff-free" gap at the start of the head lines, whether the line-initial glyph has been puffed or not.  Consider our rule for capitalizing titles: only the "important" words are capitalized, while articles and common prepositions and conjunctions are left in lower case.

Assuming that a puff can possibly replace some of the ke/te as well as some of the k/t, the glyph frequencies after a puff should not resemble either those of k/t or those of ke/te.  But the frequencies of the glyphs before a puff (excluding the line-initial puffs) should resemble those of k/t, whch should resemble those of ke/te.  However, in Herbal-A, Herbal-B, and Stars, the words at the beginning of a paragraph are expected to be "special" -- which may mess up the glyph statistics.

I implemented all of your suggested changes below, and something interesting definitely is happening.  It doesn't seem to equalize P/F and Ke/Te, and increases the distance between normal P/F and K/T.

Thanks!  

Anyway, it is clear that the results depend a lot on how e, ee, and eee are counted and mapped.  In my current model, okedy and okeeedy count as one occurrence of ke each, but okeedy does not -- it counts as one simple k instead..

In addition, qok/qot words all become far more different. oK/oT words oddly seem mostly unaffected. Doesn't this seem like these are functioning like different classes of words? It seems odd that these types of substitutions would affect them unequally, right?

I have a strong feeling -- but not certainty yet -- that qo- means "and", and is used as "wa-" in Arabic, "ve-" in Hebrew, and "we-" in Ge'ez.  That is, an enumeration like English "A,B, and C" in Voynchese would be "A qoB qoC".  

Whereas  I have no good idea of what a prefix o- means.  One conjecture is that [aoy] denote pitch levels, which would explain why they seem to be so "fluid".  But they could also be letters, and then "okedy" and "kedy" would probably be unrelated words.

Have you seen my post about the sub-section structure of the Shennong Bencao?

I have not, can you share that?

The Shennong Bencao Jing (SBJ, The Divine Farmer's Classic Materia Medica, ~300 BCE) originally listed ~365 remedies and their uses.  It was traditionally divided into three sections: "high grade" -- harmless tonics that can be taken indefinitely; "medium grade" -- remedies that should be taken only for specific diseases; and "low grade" -- toxic substances that should be taken only as a last resort, in the smallest amounts possible, with a doctor's supervision.  Each section was divided into six sub-sections: Minerals, Herbs, Animals, Insects, Trees, and Grains.  

There is a significant difference in vocabulary between the main sections.  In the "High Grade" section the characters 久服 (in the context, "taking regularly") and 久食 ("eating regularly") usually appear, followed by general benefits like 轻身 ("making the body nimble") and 延年 ("lengthening one's years'). These benefits get scarcer in other sections, where they are replaced by diseases like 咳逆 ("coughing") Many other benefits and conditions have different frequencies in each section.

(However, in one digital file of the SBJ that I got, the organization is turned inside out: it has six sections based on nature, each divided into High, Medium, and Low grades.  Moreover, by 1400 the original manuscripts of the SBJ had been lost, and it survived only as integral quotations in later and much larger medical encyclopedias - which had their own organization.)

If the Stars and Herbals sections have similar organizations, that would impact the character statistics, with complicated effects on any data analysis based on them.

All the best, --stolfi