Re: Introduction and sandhi scripting question.

Edgard Bikelis wrote:
 The problem really is sandhi, about
> which there are two alternatives: or I write a long list of original
> status and outcome, for instance:
>
> aa, a:
> ae, a:
> ...
> bt, pt
>
>     Or I try generalizing, like I wrote on the grammar. But I tried, and
> that is the source of the pain I mentioned ; ). It needs much more
> talent to write a small and efficient code...
I know nada/zero/zilch about writing computer code-- but these things are
very (well, relatively) easy to do with standard generative phonology rules
with Greek-letter variables--The form of these particular changes would be
_transformational rules_ since they're changing two segments >> one.

(using a,b etc. for alpha, beta...)

Like-V length:  SD: 1[+V aF]2[+V aF] ==> SC: 1[+V aF +long]  2>0

Unlike-V coalescence:
SD: 1[+V aF]2[+V bF] ==> SC: 1[+V aF +long] 2>0

(where aF--aF means 'same features', aF--bF 'different features'-- I'm not
totally sure it's a permitted abbreviatory technique, but it works for me
;-) )

Your C-voicing rule could be: [+C +voi] --> [+C -voi]/__[+C -voi] (that's
probably way too general, but you get the idea)

(Development of my Gwr is full of such rules ;-(( )

In computer coding, is there no way to delineate "major classes", e.g. all
high vowels, like/unlike V, all vl/vd stops etc.-- other than just making
lists?  I suppose one would use something like:

all voiced stops (bdg etc) represented by "G", then have general rules that
operate on "G"...????

Don't know if this will be of use-- but are you familiar with "IPA Zounds"?
http://zounds.artefact.org.nz/

(BTW-- I agree 100%, your pdf's are very nicely done. I'm envious.