> Is it possible that PIE had a system like this:
>
> h2 - a - a:
> h1 - e - e:
> y - i - i:
> h3 - o - o:
> w - u - u:
PIE certain had the patterns y - i and w - u.
The long vowels all seem to be later developments (even if within PIE), for
example, from -iH or -uH.
The laryngeals are seldom distinguished - often only in Greek. Syllabic
h1, h2 and h3 all appear as i in Sanskrit, and a in most other languages.
It is mostly when they are next to e that we can distinguish them. The
prothetic vowels in Greek often preserve a three-way distinction, but even
here there is a lot of "vowel-colouring" by analogy and so on, that can
distort the evidence.
So we can't say PIE had the patterns h2 - a, h1 - e etc.
It is better to think of h2e > a and h1e > e and h3e > o and so
on.
Peter