>>would we say Latin was well-nigh a 3-vowel language?
> I don't think we disagree about any real facts, so this is all about when
> to use the adverb "well-nigh".
I'm happy with "well-nigh" for Sanskrit, where the selection of vocalic or
consonantal "allophone" is purely phonetically conditioned, and exceptions
ar rare and without apparent pattern.
In Latin, however, the situation is somewhat different. In a given
sequence, the choice of how to realise each segment phonetically is not
purely phonetic. It depends on wider knowledge, such as semantics, lexica,
and morphology.
Consider for example:
-ERVI-
Is this to be taken as -E # RVI-, which triggers the realisation -erui-
or as -ER # VI- which triggers the realisation -ervi-?
(Examples are eruit and fervit. We need to recognise the simplex ruit)
Or consider
-NSVE-
Is this to ge taken as -NS # VE, giving -nsve
or as -N # SVE, giving -nsue
(Examples are dicens tacensve, and consuesco. We need to recognise the
bound morpheme -ve)
Or consider
-LVI-
Normally this is to be taken as -lui-, but not in the word volvit, which
retains the consonantal form, in order to disambiguate it from voluit.
So I don't think we can put Latin in the same class as Sanskrit.
Peter