Fix phoneme chart display

defseg commented 5 years ago

abaz1241-1 (rounded radicals) kiri1256-1 (i^ u^ are super-high and should appear above i u) ersu1241-1 (tr etc. are a little messed up - and maybe we should have a consistent notation for retroflex affricates with rhotic release? cf. sang1321-1) toda1252-1 (string processing to recognize ɽr as a trill? we do that in Ersu too, seems like good enough notation since that's how they're realized in both) yele1255-1 (lw) bisu1244-2 (pl kl) acha1249-1 (kr) wutu1241-1 (sje, although I still don't know what this is supposed to be - maybe ç) weri1253-1 (no clue what ntz\ is supposed to be, use string processing) abkh1244-1 (uvulars are completely messed up) adze1240-2 (prenasalized glottal stop)

There are probably more - should write tests to ensure nothing ends up as unknown.

defseg commented 5 years ago

We already have tests! node tests/no_unfeaturalizable.js

Should maybe write internal docs.

Some of this will need to be handled in the featuralization pipeline. For example, ð̼ is unfeaturalizable in place because it's ended up as +distributed, but I think linguolabials are always apical.

defseg commented 5 years ago

zbua1234-2 - this probably needs to be fixed in featuralization too. Maybe just treat velarization on vowels identically to pharyngealization for now?