Re: a "natural language" ?

Mark J. Reed wrote:
> I keep running across that term "sandhi".  Does that just mean the rules
> that determine which allophone is chosen in a given phonetic
> environment, or is something else going on?
>
It can be purely phonetic, but also morphophonemic (changes to the base
form). In most cases, it probably involves assimilation of one sort or
another.

Consider:
Skt. root {ruc} 'shine': present (with guna ="add -a- to root") >
rocati
r-a-uc+a+ti
r.guna.uc+theme V+3pers.

root {bhu} 'be' :
present bhavati = bh.guna.u+a+ti in this case the a-u > av because a vowel
follows.

Treatment of aspirates/voicing in clusters:
budh- 'enlighten' + -ta 'participle' > buddha

There's a prefix /niC-/ 'un-' (I forget what the underlying C is) that shows
up as nir-, nis.- and others depending on the following C or V.

I think Kash r-metathesis is an example too: nimbur 'remember' + -to
'future' > nimbutro; also in several irregular (fossilized) formations like
sotrelo < week /sor 'seven' + lero 'day'/.

I'm not sure how tone sandhi works; probably something like:
High followed by Low > Falling, or maybe mid-low, or high-mid. I'll have to
work on this for Gwr, probably re-inventing the wheel in the process :-))))

Replies