On 2012-03-29 06:32, Eric Christopherson wrote:

> On Mar 26, 2012, at 10:40 AM, Alex Fink wrote:
> > On Mon, 26 Mar 2012 17:24:21 +0200,
> > BPJ<[log in to unmask]> wrote:
> >
> > > On 2012-03-26 02:46, Eric Christopherson wrote:
> > > \> Oh, yeah. Does anyone know how the*onset* of
> > > such a syllable was affected by the closedness of
> > > it?
> > >
> > > Yes, that's what I described: if the syllable was
> > > closed
> > >
> > > [...] Apparently the onset, and probably also the
> > > nucleus, became relatively shorter when there was
> > > a coda.
> >
> > Right, presumably the *how* of the matter was some
> > sort of isochrony rule: (non-initial) syllables
> > were tendentially all kept at the same length (with
> > perhaps some exception for proto-long onsets). So
> > if there was a coda consonant, that entailed less
> > room for the onset (or the V).
> OK, so the stops that would end up as fricatives were
> originally short stops; and I'm guessing they
> contrasted with the forerunners of the stops that now
> are either short or long. Is that assumption correct?
> If so, it reminds me of Sami's system of three
> distinctive lengths. I've been trying to wrap my head
> around consonant gradation in Sami today, but I
> haven't been able to discern if it's directly
> correlated to the Finnish gradations or not.

AFAIU Finnish and Sami consonant gradation are
basically the same at least in origin, though both
have been obscured by later changes.

There were only two underlying quantities, but each of
them was realized shorter/weaker in the onset of a
closed syllable than in the onset of an open syllable,
so there were three surface grades, only two of which
were realized as a quantity distinction; the weakest
grade was relized as a qualitative distinction:
voiced fricatives vs. short and long voiceless stops
or long sonorant vs. sonorant followed by long or short stop.

Estonian later actually acquired three phonemic
quantities for both vowels and consonants when former
allophonic length distinctions due to word/syllable
metrics were phonemicized due to apocope. This was a
separate but similar development (for which I've been
unable to find an online account); both involve the
phonemicization of former allophonic isochrony
phenomena.  Estonian synchronic morphology becomes
rather convoluted when trying to describe both
kinds of gradation as a single phenomenon.