A Phonemic Feature Tree for English (1)

Here's something I've been working on off and on for quite a while now.

  1. dent = dental
  2. glot = glottal
  3. high = high tongue body, close
  4. lab = labial
  5. lat = lateral
  6. low = low tongue body, open
  7. mute = voiceless
  8. nas = nasal
  9. obs = obstruent
  10. occ = occlusive, stop, plosive
  11. pal = palatal, front
  12. rnd = round
  13. tns = tense, consonantal
  14. vel = velar

These 14 features with the inherent connections listed below describe all and only the phonemes of the General American dialect of English.

I'm basically using binary features (with the "+" and the "-"), but I've added a little something to that. Where you see a "?" below, that's an unvalued feature. It's looking for either a "+" or a "-", so something else has to be pulled from the creativity of the speaker before you have something usable at the next level of combination. The wedges indicate that the valued feature (on the left) always implies the unvalued feature (on the right).

  • +obs > mute?
  • +mute > glot?
  • -glot > occ?
  • +occ > vel?
  • -vel > lab?
  • -occ > dent?
  • +dent > lab?
  • -dent > pal?
  • -mute > occ?
  • -obs > nas?
  • +nas > lab?
  • -nas > tns?
  • +tns > lat?
  • -lat > rnd?
  • -rnd > pal?
  • -tns > high?
  • +high > pal?
  • -high > low?
  • +low > rnd?
  • -low > pal?

Starting with the unvalued feature [occ?] (presumably selected by the word or syllable in which the sound in question occurs) and following the paths dictated by the connections listed above, you can draw a tree that looks something like this. (Sorry I don't have time to draw this up as a nice clean image.)

              obs
             +   -
          mute    nas
         +   -    +  -
      glot  occ    \  \
     +   - +   -    \  \
     h  / vel  dent  \  \
       / +  -   +  -  \  \
    occ  g lab lab pal lab\
   +   -   + - + - + - + - \
 vel  dent b d v D Z z m n  tns
+  -   +  -                +   -
k lab lab pal            lat   high
  + - + - + -           +  -   +   -
  p t f T S s           l rnd  pal low
                         +  -  + - +  -
                         w pal i u |  pal
                           + -    rnd + -
                           y r    + - e U
                                  o pal
                                    + -
                                    a A

These features are by no means universal. You'd certainly see similar features in other languages, but only a language with this exact phonemic inventory would have these exact features. That's the point, actually. This kind of feature representation eliminates the need for a phonemic inventory.

I've made a few adjustments to these features since I first posted. I'm trying to make it so that if a given feature doesn't have a "+" or "-" value, you assume "-" for pronunciation. That would make the absolute default speech sound above be /U/. Sounds about right, huh?

There's more to it than just this tree of course. I still need to cover the phonotactics, stress, and syllabification. I'm pretty sure those are just more details to add into this mode of analysis though. I'll get to it soon. The next time I do laundry maybe?

Labels: , ,

0 Comments:

Post a Comment