sh_ and ch_ compose the same words

[-JKP-]

I have a partly written blog about this based on notes I jotted down in 2008. I am never going to have time to finish it or to create the graphs. I have 41 other unfinished blogs I need to get done, so instead I'll just post a very short textual summary of what I observed.


Frequency and Similarities:

• The ch patterns are about twice as frequent as the sh patterns (it's slightly more than twice, but twice is close enough to observe that the percentage balance remains about the same even when you look at the text that follows these glyphs).
  
  
• When you look at the statistics for common patterns, like chy, chey, chol, chor, chod, cheey, etc., the percentage frequency and order is very similar to shey, shy, shol, sheey, shor, etc.
  
  
• When you look at the statistics for less common patterns, like chos, chaiin choT, choy, etc., the percentage frequency and order is somewhat similar to shos, shaiin, shoT, shoy, etc.
  

So, except for ch being more frequent, the most common sequences do look like they might be interchangeable if one looks only at the patterns that follow ch or sh.

Beginning position is somewhat similar in overall numbers, but a bit higher for sh in terms of percentages:

• About 53% of ch sequences  are at the beginning of tokens (close to 6,000 instances).
  
• About 70% sh sequences are at the beginnings of tokens (a little more than 3,000 instances).
  

This is probably not a big enough difference to be significant, but I mention it for the record.



But... when you look at what precedes ch or sh, here is something to consider, which is why I stated upthread that ch and sh behave differently...


Preceding and following sequences differ in their general patterns. Here are some examples:

• k precedes shy less than half as often as one would expect if one uses the same general frequency parameters as for chy.
  
  
• t precedes shy only 1/4 as often as one would expect if using the same general frequency parameters for chy.
  

Similarly, when ch or sh are preceded by d, dsh is almost as numerous as dch (83% rather than approx. 48% if it were similar to "following" patterns)

When chol and shol follow k, a difference is noted in the other direction. If the percentages were similar to sequences following sh, one would expect about 38 instances of kshol, but there are only about 9 (depending on transcript).


Summary

• sh sequences are slightly more likely to be at the beginning of tokens than ch sequences in terms of percentages.
  
• What follows ch or sh is more-or-less similar in both composition and frequency if one takes a broad view of the patterns.
  
• What precedes ch or sh diverges from the percentages one sees for sequences that follow ch or sh and they can diverge in either direction.
  

These observations do not negate the possibility that ch and sh are interchangeable (it depends on how one parses Voynichese), but they are worth keeping in mind.


P.S., I also looked into whether ee was analogous to ch or sh, but it appears to have its own patterns and is not as similar to either one as ch and sh are to each other.