Rob Haden wrote:
> I was wondering if anyone could tell me what the most common conditions 
> are
> for compensatory lengthening to arise.  Thanks!
I'd say: almost always due to loss of some segment. Frequent in -VC1C2V- 
esp. where C1 is a non-stop, so -V[fricative, nasal, liquid, semivowel, h, 
?]CV- could > V:CV

There's also non-phonemic automatic lengthing of stressed V, or of V in 
monosyllables, or automatic length before e.g. voiced segments (like Engl.) 
but I'm not sure those would be considered compensatory.

There's also coalescence of two V in hiatus (i.e. V0V-- esp. like vowels) so 
that e.g. -aa- > a:, -ao- maybe > O: --but again, I'm not sure this would be 
called compensatory either.

My favorite language, Leti of Indonesia, has long V the are derived 
(historically at least) as follows:

*sú?at 'measure' >
**sú?ata (echo V to preserve final C) >
(probably then ?-loss > **sú(w)ata >
**súwta (?*sú?ta) (delete unstressed V) >
however, it still has derived/suffixed forms /suát-/ so there is some 
question as to what's the base or underlying form....