That is a good question, Ron:
I think we can identify all (or at least the great overwhelming majority of)
*CV roots in PIE: *me, *no, *g(^)he, etc.
If we see -*i/*u or -*y/*w appended to one of these, then it is sure to
be -*y/-*w because PIE has no derivative suffixes in *V; though, of course,
there are some *CV's like -*Ha(:).
Thus *méi-, 'my', is properly *mey-.
If the bare root consists with *CV: (either 'underlying' or caused by
laryngal), *CV:(H), in the case of *CV:i/y-, -*y should be reconstructed
because PIE has no derivative suffixes in *V; the same is true of *CV(:)H-
except that when the consonantal nature of *H was still present, the
formative -yV was probably vocalized as -*i.
Any root that has the form *Céi/uC- must first be analyzed as having the
root *Cey-/*Cew- to which the root extension -*C has been added.
The short answer: *i and *u are never anything but avocalic reflexes of *y
and *w.
Patrick
----- Original Message -----
From: "etherman23" <etherman23@...>
To: <cybalist@yahoogroups.com>
Sent: Monday, February 11, 2008 9:34 PM
Subject: Re:Re: Re: Re: [tied] Re: The meaning of life: PIE. *gWiH3w-
--- In cybalist@yahoogroups.com, "Patrick Ryan" <proto-language@...>
wrote:
>
> Mr. Keene, tracer of lost vowels!
>
> What does it take for you to understand that PIE *i comes from
avocalic *y;
> and PIE *u comes from avocalic *w.
>
> Those are consonants.
How do you know that *w and *y are not non-syllabic variants of the
vowels *u and *i?