With the sample RTF that I sent you I copy/pasted the word with the vowel into each display line of the high excerpt. On the 3rd line it chopped the vowel.
I did the same with the lower excerpt and it would hit and miss at chopping, the same as with the above.
The chop had happened on odd numbers for the first 11 of the display lines... being 3, 5, 7, 9 & 11. But then it did not happen until line 14, and again on line 22. I then looked at line one more carefully, and it had a very fractional chop at its top.
I don't see any visual reason, such as character combinations on the lines, that would cause it. Kerning position seems to be consistent. That it is usually odd numbers suggests to me that it is round-out metrics with the RichMemo display. That line one is only a shave, and it misses line 13, to hit 14, then does not hit again until 22, suggests that it is an accumulative value that eventually departs from serially odd figures.
I loaded the file into WordPad. It chopped the vowel on 3, 7, 9, 13 & 22. A slightly different series, but still showing an affinity to odd numerals.
I loaded it into PolyEdit. It never chopped the vowel on any line.
I loaded it into OpenOffice Writer. It also did not chop the vowel on any line.
Since PolyEdit and OpenOffice are more modern than WordPad, I suspect that it may be a fluke that was fixed for editors that came later.
Rick