Gripping phonology (long)

Sai and I have nearly settled on a phonology for our gripping language.
This message is a rundown.

We've rejected several classes of motions we'd thought of earlier for the
phonemic inventory:

* large torsions = twistings of the hands with respect to each other:
awkward.  Small torsions are okay but occur so readily as articulatory side
effects of other things that we're not making them contrastive.
* rubs or slides perform by non-thumb fingers, and pressing points at
differing vertical distances with these fingers: awkward, in the way that
seems likely to induce RSI if done a lot.
* degree of palm contact: too hard to maintain, especially while walking.
And palm bumps, which on top of this are really visible.
* back and forth wrist motion: awkward.
* differential intensity: too hard to segment the continua, both
articulatory and perceptual, into distinct phones.
* thumb dispositions.  This is mostly for simplicity; putting the fingers in
any order aside from their natural nesting order vastly diminishes the
options available to the fingers out of their favoured position.

This leaves the following.

* who's dominant, i.e. who has their thumb on top.  Changing this is a big
and noticeable motion, so we don't want it to happen a lot, and it'd be
great if dominance was a state belonging to an entire conversation.  Since
various sorts of phones are asymmetrically available to the dominant and
subordinate hand, it'd be nice to relate hand dominance to an asymmetric
assignment of roles in the conversation, to exploit this.  Perhaps, for
instance, the main direction of information transfer is from dominant to
subordinate; maybe the dominant knows things to be secretly relayed about
the situation to the subordinate.

* global motions, involving the hand as a whole:
- rotating thumbward or pinkward (i.e. in the direction of the thumb resp.
pinkie finger, i.e. in the medial resp. lateral directions) at the wrist
- rotating thumbward or pinkward at the elbow

* finger motions, by far the largest group.

The distinctive features we can use are:
- articulator, i.e. which finger moves
- place of articulation, i.e. which position on the receiving hand is frobbed
  Of the potential places and articulators each pairs with relatively few of
the others, so these won't be independently completely phonemic; more on
this below.
- manner of articulation:
. whether this is a press (from a state of initial contact) or a hop
(involving an initial lift)
. whether the articulator finger uses the pad or the nail
. whether there's rubbing

The last two manners of articulation are only comfortable to use with the
thumb.  The press vs. hop distinction works only if the finger has a rest
position from which it doesn't have to move off the surface too much to do
either: otherwise we would get extra 'epenthetic' presses or rubs.  We're
taking this as a co-occurrence constraint: probably hops are only allowed if
there was a previous motion leaving the articulator close enough.  For the
nonthumb fingers, this is a nonconstraint; for the thumb there's something
to it.

We treat the thumb distinctly from the other four fingers, since it's
significantly more flexible and its motions are more asymmetric.  In
particular we haven't tried to pair up the dom thumb's motions with the sub
thumb's into phonemes yet.

We've picked out six places on the sub thumb that the dom thumb can press:
- the tipmost phalanx;
- the knuckle;
- the basemost phalanx;
- the knuckle (pressing this one makes the sub thumb extend in response);
- the metacarpal;
- below the metacarpal, nustling in near the wrist.
The last two of these allow nail presses.  There are also three rubs, along
either phalanx or the metacarpal, in either direction; the last of these can
be done with the nail.

On the dom thumb we haven't quite made up our minds but there are perhaps
six places again:
- points above the below-mentioned knuckle,
- the knuckle of the index finger, under the middle phalanx (pressing this
from the top again causes an extension);
- the basemost phalanx;
- the next knuckle, or the metacarpal (these are hard to distinguish);
- the fleshy bit between the index finger and the thumb;
- the underside of the thumb (pressing here lifts the two thumbs away from
the hands).
Are these presses from the top or the side?  Can we contrast the two?
There are two (three?) rubs, one along the basemost phalanx and a lower one.
 None of this is comfortable using the nails.

Next, the non-thumb finger articulators.  For ease I'll label these fingers
from 2, the index, to 5, the pinkie.  Potential points of articulation are
the knuckles, which I'll name the same as the fingers, and the gaps between
knuckles, which I'll name as the means of the knuckles on either side, as
integers plus 0.5.  This notation extends, e.g. 5.5 is the outside of the
hand  next to the pinkie.

Each finger can reasonably reach a range of positions, the range depending
on whether it's on the dom or sub hand, and also (we find) whether it's the
hand that prefers to be dom or sub when its owner clasps their own two hands
together: we've been calling these the on and off hands respectively.
Sometimes two fingers can reach the same position without too much
discomfort, but we deem telling the difference between these too hard.  So,
for each finger, we want only the easiest knuckle and the easiest gap it can
reach, and thus there are always just 4*2 = 8 articulator-place pairs
available.  (For simplicity we ignore the fact that we might be able to
enlarge the range by deviating from this with the extremal fingers.)

We find:
finger  f  on the     sub hand can reach positions  f + { 0,   0.5};
finger  f  on the on  dom hand can reach positions  f + {-0.5, 0  };
finger  f  on the off dom hand can reach positions  f + {-1,  -0.5}.

At this point the question arises of how to group these into phonemes
accounting for dominance variation, whether sender coded or receiver coded,
i.e. whether ensuring that the articulator resp. the place is constant in a
given phoneme.  (Dominance variation means we can't have it both ways.)
Using receiver coding would mean either that our phonemic range is narrowed
from 8 articulator-place pairs to the 6 in the intersection of all three
lists in the last paragraph, or that we strain some fingers to allow more;
neither is promising.  It would also mean that non-thumb presses on 1.5
would interfere with thumb presses there.  So we've gone for sender coding.

This leaves one question open, neither alternative quite satisfying.
Depending on the dominance and such, the two places that finger  f  may
press are either a knuckle  k  and the gap  k + 0.5 , or the knuckle  k  and
the gap  k - 0.5 .  So do we elevate as distinctive articulator and type of
place (knuckle vs. gap), and have the realisations sometimes switch order on
the hand?  Or do we go for two phonemes for each articulator distinguished
by order on the hand, so that each is sometimes a knuckle place, sometimes a
gap?  Dunno yet.

Summing up, there are roughly approximately 38 (dom) or 30 (sub) elementary
phones here.  Itemised they are
4 global motions;
18 (dom), 10 (sub) thumb presses breaking down as
  6 + 1?, 6 plain (the +1 conceptually collectively accounting for hops),
  2 + 1?, 0 with nail,
  3 * 2, 2 * 2 rubs counting direction,
  1 * 2, 0 * 2 nail rubs;
16 = 4 * 2 * 2 non-thumb presses, breaking down as articulator * (gap vs.
knuckle) * (press vs. hop).

I say "elementary" because we may want to recognise coarticulations of some
of these as single phonemes.  This is to be determined.

And once all of that is nailed down, we'll want a good concise ASCII
notation for all this...

Alex

Replies