--- In
cybalist@yahoogroups.com, "Patrick Ryan" <proto-
language@...> wrote:
> I had a similar problem with long and short vowels in my
approach to these problems.
>
> Without going into detail as to how I think they were
lengthened, I hypothesize a group of roots having the form *Ce:C-,
*Ca:C, and *Co:C, _without_ 'laryngeals' being involved.
You must mention the examples. I am very sure there are IE paradigms
based on root segments with long vowels, as *ste:w- -> *sté:w-/*stéw-
with a gradation originally completely parallel with é/zero. For
such roots there are also paradigms based on short /e/. For the few
items I would like to have underlyingly long /a:/, as 'nose'
and 'salt', I have no good evidence for a parallel independent /a/.
The same goes for the one item of /o:/ I imagine (but don't insist
upon), *gWo:w- 'cow'. Could you tell us what material you are
talking about?
> The problem I found with this is that where I expected Ca:C, I
frequently found *CaC, etc.
The weak grade of /a:/ will be /a/ by regular development. You must
mention the material if you want a reaction to your impression of it.
> After some thought, I applied what I think is a principle of
language development: economy of effort.
>
> Where there was no homonymous *CaC, *Ca:C could be shortened to
*CaC with no loss of root integrity; and frequently was so
shortened; when there was such a root, resistance to shortening the
vowel was much stronger.
I would like to see your expose in concrete terms. While I
understand your suggestion for a principle quite well, it does not
correspond to anything I have noticed. Therefore you need to
demonstrate it.
Jens