I've made some adjustments to my tree of features for describing the phonemes of English.
obs
+ -
whis nas
+ - + -
/ occ lab cons
/ + - + - + -
/ vel | m n pal \
glot + - | + - \
+ - g lab | dark lab \
h occ + - | + - + - low
+ - b d | r y w l + -
vel dent | pro high
+ - + - dent + - + -
k lab lab dark + - a dark i pro
+ - + - + - lab dark + - + -
p t f T S s + - + - o A e dark
v D Z z + -
u U
It's still just 14 features, though a couple of them have changed.
- cons = consonantal, tense
- dark = retracted tongue, half-round
- dent = dental
- glot = glottal
- high = high tongue, close
- lab = labial
- low = low tongue, open
- nas = nasal
- obs = obstruent
- occ = occlusive, stop, plosive
- pal = palatal
- pro = protracted tongue, front
- vel = velar
- whis = whispered, voiceless
The following list shows which feature values require additional feature values to be supplied before a phoneme can be identified. The tree above is simply a graphical representation of these dependencies, though it must be noted that something must require a value for [obs(truent)] in order to get the ball rolling.
- +cons > pal?
- -cons > low?
- +dent > lab?
- -dent > pal?
- -glot > occ?
- -high > pro?
- +low > pro?
- -low > high?
- +nas > lab?
- -nas > cons?
- +obs > mute?
- -obs > nas?
- +occ > vel?
- -occ > dent?
- +pal > dark?
- -pal > lab?
- -pro > dark?
- -vel > lab?
- +whis > glot?
- -whis > occ?