[sh] and Paragraph First Line

[MarcoP]

Hi Emma,
I ran a script I put together in 2021, when I first learned of Patrick's research. These are plots show frequencies according to vertical and horizontal paragraph position. Figures are expressed as % of words containing a feature. I considered all paragraph text, ignoring uncertain spaces.


   
Word-initial 'sh' (comparable with blue in figure 3 of Patrick's paper)

---

   
'sh' in any word position

---

   
'chdy' in any word position (~95% of occurrences are word final):

I would say that the plot above confirms that chdy is more frequent in the first line of paragraphs (though of course numbers are about one order of magnitude smaller than those for 'sh'). About 10% of the occurrences are of the pchdy type, and these of course concentrate in the first line. But also excluding 'pchdy', a preference for the first line is still there:

   
plot for pchdy

---

   
Xchdy excluding pchdy;

---

   
word initial 'chdy' (basically, the word 'chdy')

---

   
sum of the previous two (i.e. all occurrences of chdy, excluding pchdy)

It's interesting that the isolated peak at the middle of the last row appears to be due to the word 'chdy' specifically. Contrary to all other 'chdy' patterns, the stand-alone word chdy does not seem to show a preference for the first line.

As always, it's possible I made errors, but at least the plot for initial 'sh' appears to be consistent with Patrick's plot.