Liaisons in the Voynich text

[quimqu]

There are a few skaiin and *r kaiin, one *dkaiin though. Many *oo* (84) and *oa* (282).

Yes, "never" is too strong, my fault. I should have said "much rarer than expected". For example oo: there are 69 in the paragraphs, but with the amount of o in the text you would expect about 2700. Same story with the others, they are around 10 or 20 times rarer than they should be. That's what I call a forbidden pair. By the way, I only count the paragraphs, not the labels, so my numbers are a bit lower than voynichese.

How do you define with/without? By presence or absence of "glyph put between"?

No, by the ending of the previous word. If the previous word ends in n, r, d, s or m and the next one starts with a gallows, that's a clash (those pairs don't happen inside words). If it ends in o, y or l, no clash. Then I just count how many times the gallows word has the y in front in one case and in the other. Same logic for q and d.

But the insertion of a glyph obviously reduces the space between words: do you mean close even before the glyph was inserted?

I thought about this one. If the q had been added later into a gap, the o after it should be more or less where a normal o-word would be, and the q would be filling the gap. But that's not the case: the o of a qo-word is much further from the previous word than the o of a bare o-word, almost the whole width of the q. So the q was written in the flow, it pushes the word. The space I measure is the one in front of the q, and that one is a bit shorter than the space in front of a bare o. Anyway, I'd love to see what your spaces say, mine come from aligning glyphs on the scans and they are surely not perfect.

It's worth to check if the glyphs in question look notably different in ink or style, suggesting they might have been inserted in a 2nd pass.

Thanks for the link. I am not an expert in ink... but as I said to nablator the positions don't look like a second pass. Someone with better eyes could check a few qokaiin on the scans... About the why, I don't know. The rule itself is not complicated for the scribe (just look at the last glyph you wrote). What I find strange in a meaningless text is that it is not black or white, the rarer the pair, the more often he fixes it.

It's somewhat similar to the idea of @JoJo_Jost...

I see what you mean, a small piece that goes with the next word. The difference is that I don't read anything, I just count. The pattern is there whatever the glyphs mean (if they mean something).

In my work I considered "d" and "q" a "gallows glyph" so - ends "o,y = d/q before a/o" would conform.

Interesting. That would cover the q and d rows of the table. But the y and o rows go the other way: after n, r, d, s, m the gallows word takes a y or an o in front. So in your system, does the glyph in between always belong to the "other" class? If so we are maybe looking at the same thing from two sides. And yes, if you remove the spaces the thing disappears, the extra glyph is just one more glyph inside a longer word.