Re: [tied] -osyo 4 (was: Nominative Loss. A strengthened theory?)

On Thu, 22 Apr 2004, Piotr Gasiorowski wrote:

>
> For the record, I said they were _fairly_ rare, not _extremely_ rare. To
> be
> sure, Ladefoged and Maddieson mention only Mixe and Yavapai as languages
> with lexically relevant overlong three-way length distinctions. They thus
> ignore (unjustifiably, IMO) the Low Saxon/Limburgish data I mentioned
> here
> yesterday, which may mean that they have overlooked other such cases as
> well. As for Estonian (and some of its Finnic-Saami cousins), overlength
> is
> more a matter of relative syllable quantity within a word than a contrast
> existing in the lexical representation. But then, what has been claimed
> here
> for PIE was merely _derived_ overlength, originally in complementary
> distribution with ordinary length plus a sonorant. Once it had become
> phonologised in the individual IE lineages (through the generalisation of
> sandhi variants), the contrast between overlength and ordinary length was
> either abandoned or converted into a qualitative contrast. That kind of
> thing is not bizarre at all.

Well, that's nice to know. I wonder if Sanskrit (pre-Proto-Indo-Iranian,
actually) compounds did not create a fourfold (or fivefold?) opposition
-a- : -a:- : -a:-a- (= -a-a:-?) : -a:-a:- the day they lost laryngeals?

I am not sure I do not myself have an opposition between

/pa/ <par> '(a) pair'
/pa:/ <parre> 'to pair'
/pa::/ <parrer> 'pair(s)' (prs.) or sg. 'person who pairs'
/pa:::/ <parrere> 'pairers' (pl. 'persons who pair')

Jens