This graph is based on Takahashi's transcription. All ch- and Sh- words with at least 50 occurrences are represented, together with their Sh-/ch- counterpart (which sometimes has less than 50 occurrences). I think that these numbers confirm the "replaceability" pointed out by nablator, and that they suggest that the two members of each couple are related.
E.g. chedy (501 occurrences) appears to be related with Shedy (427 occurrences). The two words tend to occur in the same pages. For instance You are not allowed to view links. Register or Login to view., we find reduplicating and quasi-reduplicating sequences like shedy.shedy.chedy and chedy.shedy.
I believe this is somehow similar to what happens with ok- vs qok-. Like qok- adds something to ok-, Sh- adds something to ch-, but in this case the addition is not a whole character like q, but a single stroke, the superscript-curl above the bench. If the script is phonetic, the curl could be equivalent to a whole character, or a few characters, and be an abbreviation; or it could be some kind of lesser modification, like an accent or an umlaut.
E.g. chedy (501 occurrences) appears to be related with Shedy (427 occurrences). The two words tend to occur in the same pages. For instance You are not allowed to view links. Register or Login to view., we find reduplicating and quasi-reduplicating sequences like shedy.shedy.chedy and chedy.shedy.
I believe this is somehow similar to what happens with ok- vs qok-. Like qok- adds something to ok-, Sh- adds something to ch-, but in this case the addition is not a whole character like q, but a single stroke, the superscript-curl above the bench. If the script is phonetic, the curl could be equivalent to a whole character, or a few characters, and be an abbreviation; or it could be some kind of lesser modification, like an accent or an umlaut.
![[+]](https://www.voynich.ninja/images/collapse_collapsed.png)