On that note, as this thread does seem to be slowing down, I might as well post interesting stuff that I've found more recently IMD anyways now.
Probably the most interesting bit I have found recently is that specifically everyday speech, there is a lenition without neither voicing nor frication of unstressed intervocalic /p/, /tʃ/, and /k/ in unstressed intervocalic positions that seems to actually be far more consistent than I had previously assumed. This combines with that in the same register /b/, /dʒ/, and /ɡ/ are also voiceless lenis in this same position, likewise far more consistently than I had previously assumed. Also, /p/, /tʃ/, and /k/ do not induce preglottalization here either.
The result is essentially a pure vowel length contrast between /b/, /dʒ/, and /ɡ/ and /p/, /tʃ/, and /k/ here in everyday speech, to the point that in a narrow analysis that treated vowel length, except in morpheme-final syllables without obstruents in their coda, as phonemic would likewise treat these two sets as merging in morpheme-medial unstressed intervocalic positions.
I would still probably not use such an analysis, though, as high registers much more clearly distinguish the two sets in that position, both through the voicing of the lenis set and through less lenition of the fortis set.
(I should note that this does vary heavily depending on the exact dialect in use, e.g. my ex seems to have far less of this than people from back in the suburb where I grew up, where I heard a lot of this when I listened for it.)
On another note, one thing I noticed is that final /n/ tends to be elided more than I had thought before my dialect; it had seemed to be primarily an intervocalic thing, but rather from paying attention to things closely since then, it seems to be sporadically to very frequently, depending on register, elided finally regardless of what follows it or even utterance-finally. My parents, for instance, do seem to not infrequently drop it finally, albeit not as much as I catch myself doing. It does seem to be somewhat lexicalized, as this seems to happen much more with grammar words and with effectively-grammatical constructions than with content words, as far as I have noticed so far. (I for instance drop final /n/ extremely frequently for can, in, than, kind, and like compared to normal content words.)
Dibotahamdn duthma jallni agaynni ra hgitn lakrhmi.
Amuhawr jalla vowa vta hlakrhi hdm duthmi xaja.
Irdro. Irdro. Irdro. Irdro. Irdro. Irdro. Irdro.