28-09-2021, 05:39 PM
(This post was last modified: 28-09-2021, 06:49 PM by pfeaster.
Edit Reason: clarification
)
(28-09-2021, 02:38 PM)nablator wrote: You are not allowed to view links. Register or Login to view.Roughly 95% of vords having at least two characters preceding a benched gallows have a 'h' or 'q' among them.Thanks for sharing! As to that first exception, it looks as though [ol] followed by a benched gallows is pretty common, but almost always with a space inserted -- so I wonder whether it might be the spacing of [olcFholy] that's exceptional here rather than the glyph sequence as such. Of course, if it were [ol.cFholy] it wouldn't have fit your criteria, and so wouldn't violate the pattern.
Examples:
6r.13 qocThol
6v.2 chcKhy
6v.3 oochocKhy
6v.4 ShcKhy
6v.6 ShocThol chocTHhy
6v.15 chocKhy
First exception:
9v.5 olcFholy: no 'h', no 'q' before 'cFh'
I recently noticed one other curious pattern involving benched gallows preceded by [ch] or [Sh] (with apologies if this is already old news). If we calculate the probability that any given token of [ch] or [Sh] will be followed by [cKh] or [cTh], we get:
ch>cKh (243 tokens, 2.39%), ch>cTh (134 tokens, 1.32%), Sh>cKh (95 tokens, 2.23%), Sh>cTh (50 tokens, 1.17%)
But if we do the same for [n.ch/Sh] or [y.ch/Sh], we get:
n.ch>cKh (96 tokens, 7.46%), n.ch>cTh (46 tokens, 3.57%), n.Sh>cKh (38 tokens, 5.74%), n.Sh>cTh (17 tokens, 2.57%)
y.ch>cKh (77 tokens, 4.20%), y.ch>cTh (38 tokens, 2.07%), y.Sh>cKh (34 tokens, 4.10%), y.Sh>cTh (15 tokens, 1.81%)
In other words, when [n] or [y] precedes [ch] or [Sh], the likelihood of a benched gallows coming next in the sequence seems to increase, and to increase a good deal more for [n] than for [y], which implies that the issue isn't merely whether the [ch] or [Sh] is vord-initial. It looks like this might also be true for [cho] and [Sho], but the total glyph counts are so low for those combinations that it's harder to tell whether the differences are meaningful.
(PS. The counts I cite for y.ch>cKh and y.ch>cTh actually include one unspaced token apiece, as ych>cKh and ych>cTh; with spaces the token counts would be 76 and 37. But whether we include these or not doesn't much affect the percentages.)


![[+]](https://www.voynich.ninja/images/collapse_collapsed.png)