Creyeditor wrote: ↑16 Apr 2024 18:10
/ɛ ɔ/ are lax mid vowels and /e o/ are tense mid vowels.
Thank you. Sorry, I hadn't noticed yet that this is based on PHOIBLE's own system of features. Their whole approach as a whole is not quite what I'm used to.
About your last point, when you said "/ɲ/ and /h/ also frequently cooccur with them", does "them" refer to voiced stops, or to tense mid vowels?
In regard to point 3, "Lax mid vowels frequently cooccur with voiced fricatives. Maybe an effect of Bantu?":
PHOIBLE counts about 969 languages with any of /ɛ,ɔ/, of which 662 also have at least one voiced fricative (assuming /β,v,ð,z,ʒ,ʐ,ʝ,ɣ,ʁ,ʕ,ɦ/). So that's indeed a majority cooccurrence by about two thirds.
Not sure if and how much it has to do with Bantu. While it's true that Atlantic-Congo languages make up a majority of the result (with 254 languages), this family is in itself probably the largest in the dataset by far (with 423 entries out of 2186), and there are languages from families across the board that show a similar pattern at roughly the same proportion (with Indo-European at 80 out of 149, and Sino-Tibetan at 43 out of 100).
I can't claim any expertise in all this, but my guess is that it's more of a cross-linguistic pattern, with Bantu only being a potential culprit in the Atlantic-Congo family itself, due to how large it is.
As for point 4, "Voiced stops frequently cooccur with tense mid vowels. No real idea for an explanation here.":
The result seems to show a similar pattern. Languages with any permutation and number of /e,o/ and /b,d,ɖ,ɟ,ɡ,ɢ/ count 1198, versus 379 that have any of /e,o/ but no voiced stops, and 265 that have any voiced stop but neither of /e,o/. Languages come from a wide range of families, with Atlantic-Congo having a pretty large representation at 369 , but Indo-European with 103 and Sino-Tibetan at 77 come in second and third again.
All in all, it's probably also just a cross-linguistic pattern. Maybe there's something more interesting to be dug up by calculating how strong the pattern is proportionally within each family, but I can't afford the time to do that math right away.
Why the pattern would occur at all though, is entirely without my wheelhouse.