Here's how AutoCAD stores the text content of your "after" Mtext object, which is in the (1 . "Text content") entry in entity data, or the TextString VLA property:
"TITLE\\P \\PGENERAL\\P \\P\\pxi-3,l3,t3;A.\tLINE1\n\\PB.\tLINE 2 \n\\PC.\tLINE 3\\P\\pi0,l0,tz;\\PSPECIFIC\\P\\P\\pi-3,l3,t3;1.\tLINE 1\\P2.\tLINE 2\\P3.\tLINE 3\\P4.\tLINE 4"
The \\P's are hard returns, and the \n's are new-line characters [why they're not also \\P's, I couldn't say, but it's probably related to being within bulleted content]. The \t's are Tabs. The \\pxi-3,l3,t3; must be about the beginning of letter-designated bullets, the \\pi0,l0,tz; about the end of those, and the \\pi-3,l3,t3; about the beginning of number-designated bullets [note the absence of the x that's in the letter-based one]. I don't know exactly what the components of those bullet-related codes mean. The A. B. C. and 1. 2. 3. 4. bullets themselves look like just plain text characters, but they're clearly under the influence of those bullet-triggering codes, because if you edit out a bulleted line, the subsequent lines get different bullet letters/ numbers.
It may be possible to get a routine to inject those kinds of codes into the string content of Mtext that doesn't have them. But having a routine find where to put them would be a real challenge. There would need to be some type of constants to look for. I assume there would not always be three and only three letter-designated ones, so it couldn't just count through a certain number of hard returns. And I assume the "TITLE" would not be that word. Would the headings "GENERAL" and "SPECIFIC" always be there, always exactly like that? Would there always be blank lines between the letter-designated ones but not between the number-designated ones?
Kent Cooper, AIA