8 hours ago
tavie wrote:What I was cautioning about was connecting the lack of n.k with the lack of /nk/ within words, because n is almost always final.
Oh, ok, I see, an I think that you are right: for n and m the inside-word side tells us nothing, there are almost no n or m in the middle of a word, so "nk is forbidden" is just "n is final". The link has to stand on the glyphs that do appear in the middle of words. r, d and s do, and still almost never sit before a gallows inside a word (rk, dk, sk are 10%, 4% and 23% of what you'd expect). Those three are also the ones that take the y or o between words: y before a gallows after r 70%, after d 86%, after s 95%, against 28% after o, y, l. So for n and m I only have the between-words side; for r, d and s I have both, and they line up.
tavie wrote:It was Patrick Feaster I was thinking of
Thanks! Found it: Ruminations (2020), section 4, point 4. He says exactly that n_k is rare across a break because n and k are rarely next to each other anywhere, and that there is no good reason to treat within-word and across-break as separate things. So that part is his (and Smith & Ponzi's), I should have cited it. What I add on top is that the same word changes shape depending on the neighbour (taiin / ytaiin, okaiin / qokaiin, aiin / daiin), and which glyph comes in.
tavie wrote:I'd also suggest doing the analysis separately for different sections ... and also maybe consider the distorting effect of the position on the page.
Did both. Taking out the top line of each paragraph, the first word of the line (as the word before) and the last word of the line (as the word after), nothing moves: q 62% / 24%, d 89% / 34%, o 95% / 73%, y 80% / 31%. By section, on that same core text, the direction holds in every section with enough data; the strenght changes. q is weakest in Herbal (50% / 25%) and d strongest in Stars (73% / 6%).
tavie wrote:Despite the overwhelming y.q dominance, Stars and Balneo have small surpluses for y.l. For Stars, this is mainly ey.l while dy prefers dy.q.
I see the same thing from my side: in Stars, before an o-word, q comes 68% of the time after -dy and 61% after -ey. And the word before matters more than just its last glyph, I found that yesterday too (the glyph before the final y, and how the word starts, both change it). After daiin and aiin the q is low (20% and 10% in Stars), which is what the rule expects since n.o is a fine pair. kaiin is too rare there for me to say anything (8 cases).
I'll go through Emma's o / y / a posts, thanks for the pointer.