Archive for November, 2008

A gripping language, and a question about suprasegmental analysis

Sunday, November 23rd, 2008
[xposted from CONLANG-L]

On Fri, Nov 21, 2008 at 11:48 AM, Sai Emrys wrote:
> recently we started figuring out how we might be able to make a conlang entirely mediated by touch (of the sort where we could talk to each other discretely, masked by normal behavior like holding hands).

So, we discussed this again more recently.

To specify the domain better, the language we're trying to make should be:
* able to communicate simple and maybe meta* things (doesn't need to be capable of Shakespeare or neuroscience)
* communicable entirely by the speakers' hands being grasped together or the like (as is socially normal for couples in most situations - though I'd like to expand this to other forms of casual touch also)

By "meta", I mean that the grip-language may occur in parallel to an ongoing, and separately sensible, acoustic language - and would act as some sort of meta-commentary to it in real time.


First, one thing came up that's a philosophical? question of analysis.

One phonetic feature of the domain is that the primary two grips (opposite hands gripping, thumbs same direction, palms together, fingers interlaced) are symmetrically asymmetric - A's thumb is either outside or inside B's.

Switching between these two grips (let's call them A or B dominant based on whose thumb is on the outside) is a relatively elaborate cascade or disengage-re├źngage process, thus seems like something that would not be done frequently.

Alex's analogy for this was to vowel harmony & suprasegmental features more generally, which I think is apt.

The question is, does one analyze the words [k2r2m] vs [korom] as: a) being phonemically /k2r2m/ vs /korom/, with an non-semantic rule that vowels are supposed to be frontness-harmonic, or b) being phonemically both /k$r$m/ where $ signifies a mid rounded vowel, frontness unspecified, and frontness is a separate bit property of the whole word

Another example from ASL is hand dominance. E.g. HELP is dominant hand /A/ resting on base /B/; dominance is a non-phonological property in ASL (except in explicitly visual-spatial context). One could however analyze this as actually being two distinct signs, left A on right B vs right A on left B, with some handwaving about some signers preferring one over another form, but being allophonic.

However, suppose that I were to create ASL', in which using reverse dominance to one's true dominance carries ironic pragma. How then would one analyze it - as being a feature of each phone, of each phoneme, of each "word" (granted that 'word' is a bit ambiguous in ASL), or of a sentence / utterance overall? At some level it is specified, and at the levels below that it is not.

My preference is to analyze this sort of thing as being a bit "belonging to" the level at which it changes meaning - so if e.g. [k2r2m] vs [korom] is cat vs dog, then that's to the word itself; if it's ironic vs normal then it's to the utterance overall (unless it's just that word that's emphasizedly ironic, in which case the word again); and if yet it's indicative of deferential vs superior politeness marking, then certainly to the entire utterance or even the discourse.

I'd be interested to read y'alls' thoughts on this.


Second, we made a preliminary pass at enumerating the phonological inventory. This is divided into a few semi-parallelized channels:

* grip: A-dominant, B-dominant; possibly other variants also, not fully enumerated

* thumb disposition: default, dominant thumb under sub thumb (sub-dominant?), and dominant pointer over sub thumb (double dominant?)
- I do double dominant by leaving dom thumb as is, and just moving dom pointer over the tip of sub thumb, in a somewhat side-by-side position)

* disposition transitions: short-short, short-stroke, or stroke-* (I found stroke-stroke and stroke-short to be too hard to reliably do differently)
- short = minimal contact w/ other finger except as needed to transition
- stroke = stroke up or down other finger during that segment of the transition

* motions:
- 1..5th knuckle press (coded by recipient's knuckle, thumb = 1st)
- 1..4th gap press (1st gap = thumb web)
- 1..4th short gap press (gap press is made w/ finger extended, short gap press w/ finger pad pulled back to be against the fleshier bit)
- ? 1..5th finger squeeze (coded by squeezer's lower-ordinal squeezing finger, e.g. dom 1st squeeze = squeeze sub thumb w/ thumb & pointer)
- ? finger separation (only possible from double-dominant grip)
- ? some subset of the combinations thereof

* elbow-dominance (walking hand-in-hand, dominant elbow is in front)

* torsion (?neutral, dominant out, and dominant in - e.g. dominant out has the whole dominant thumb base outside the sub thumb)
- ? possibly these can be characterized as motions instead of states


Some possible issues with the domain:
* for me (though not for Alex), fourth and fifth finger action is not entirely seperable (so there will be noise between the two)
* we have different grip dominance preference (interlace your fingers together - which way do you prefer? I like my right thumb dominant, he likes left), so one of us is always a bit awkward with a grip
* Alex dislikes the double dominant position for being too squeezy, for making thumb usefulness worse, and magnifying grip asymmetry
* thumb disposition and grip both significantly affect the motions one can do, and the perception of them; one issue e.g. is whether to code recipient xor presser finger as phonological

Suggestions?

Anyone done similar?

Any languages for deaf-blind worth stealing from (e.g. that aren't just some originally-for-sighted sign language done using recipient hands to feel the signer's)?