Hallo conlangers!

On Thursday 14 February 2013 05:15:03 Herman Miller wrote:

> On 2/12/2013 1:56 PM, Alex Fink wrote:
> > Diphthongs being retained while simple vowels are lost is certainly
> > plausible enough.  Also, the change you're looking for might be
> > stress-conditioned.  Something like that immediate post-stress vowels
> > are lost; or the same restricted to word-finally; or vowels in
> > certain weak positions in feet are lost.
> 
> Yes, stress might explain some of the alternations. I'll have to think
> about that. Verbs have an imperfective suffix, -sé or -ŏs. When preceded
> by a voice suffix it's just -s.

This looks quite a bit like Indo-European ablaut, and may indeed
be the result of accent alternations.  (The origin of PIE ablaut
is not perfectly understood, but the main conditioning factor
seems to have been accent.)
 
> >> So far so good. But the inanimate ablative suffix is -öl after
> >> consonants and -l after vowels, and it's -ol after consonants and
> >> -l after vowels for abstract nouns. This starts to get tricky.
> >> 
> >> *siaḍu-ile>  śaṛ-öl *zakiğ-ile>  zaķi-l ?
> >> 
> >> *vieze-elu>  véz-ol *guleviğ-elu>  gulvi-l ?
> > 
> > I, for one, would not be satisfied with -öl and -ol having been
> > _further apart_ in the proto-language, as with this -ile and -elu; I
> > would want to draw them together.
> 
> It does seem that at some point they should have been more similar, but
> that might have been earlier than Proto-Jardic.

Yes.  The suffixes are too similar to resemble each other
accidentally, and certainly have a common origin and thus should
be more similar in Proto-Jardic.
 
> > That is, if you do want to keep your modern table of inflections, it
> > probably dictates what the most likely vowel coalescences from
> > proto-Jardic to Jarda should have been!

Yep.

> >     E.g., here, in the ablative,
> > one wants to explain modern _ö_<  *Vin+Vabl, _o_<  *Vabs+Vabl, where
> > Vin and Vabs were the former theme vowels of the inanimate and
> > abstracts, and Vabl was the initial vowel of the ablative clitic.
> > Then one could set up the proto-ablative as *VlV, and account for the
> > patterns of vowel loss using some kind of metrical explanation.  E.g.
> > acute is stress, possibly secondary, below, and we suppose that a
> > diphthong attracted it but otherwise it defaulted to the end syllable
> > in words of a suitably canonical stem shape. animate: *...C+VlV́>
> > C-lü inanimate: *...CV́+VlV>  C-öl abstract: *...CV́+VlV>  C-ol
> 
> I'm thinking it might simplify some things if length is distinctive in
> Proto-Jardic.

Why not?  There are many languages where vowel length is
distinctive, and even families where it is distinctive in the
common ancestor but not in the daughters (e.g., Romance).

>       Say that inanimate nouns end in -i or -ī, where short -i
> was lost (*siaḍi-olu > śaṛ-öl) but long -ī remained as -i (*zakī-olu >
> zaķi-l). Abstract nouns might have ended in -o, or some other short
> vowel that didn't mutate the -o in -olu.

That is my opinion, too.
 
> > This supposes there was a simple *V that could have become -ü,
> > though!
> 
> Yes, and I guess typically that would be *u. (On the one hand, that
> would explain why the /l/ in -lü didn't get palatalized. But then modern
> /u/ needs to come from somewhere else.) So maybe the animate nouns had a
> stress on the ending, -olú instead of -ólu.

This is possible, though I wonder how that came about.  Perhaps
a lost enclitic pronoun or whatever that dislocated stress to the
right?
 
> > The latter explanation actually stands a chance of being good, here,
> > I think.  That is, perhaps the dative and instrumental were actual
> > cases in proto-Jardic, while the ablative was made of a clitic
> > following an oblique case (maybe the same one as the antecedent of
> > the modern absolutive or genitive?).  Then we'd just have to suppose
> > the oblique case ended in V for both inanimate and abstract, while in
> > other cases the inanimate used a theme vowel and the abstract
> > didn't.
> 
> A clitic might work. Cases might develop from postpositions, but I don't
> think Jarda is the sort of language that would have postpositions (it's
> pretty consistently head-initial).

That a language is head-initial now doesn't mean it always was.
It may have been head-final earlier.  Proto-Celtic probably was
head-final (at least, Celtiberian was, and so was PIE).  My own
Proto-Hesperic is also head-final, and while Old Albic is pretty
consistently head-initial, the old head-final word order shows
in compounds, and in the secondary case endings which evolved
from Proto-Hesperic postpositions.
 
> > I may as well append a few things I meant to say regarding your
> > earlier Proto-Jardic threads but never got around to:
> > 
> > - In a language where there are lots of nonsystematic ablauting
> > procedures, appearing in only a few word-families each, one expects
> > the ablaut to be _old_.

Indeed.  The less systematic a pattern is, the older it probably is.

> >     If I were you, I would not be trying to
> > explain all the Jarda ablaut in Proto-Jarda!  I'd be trying to
> > explain some of it, especially the more contemporarily systematic
> > parts, but then leave the residue as ablaut in the proto-language, to
> > have arisen at an even earlier stage. (Differential treatment in
> > loans of course helps too, but that is likelier to give all-out
> > doublets than cases with systematic derivation-looking function.)
> 
> I haven't actually determined how old Proto-Jardic is: it could be old
> enough to account for the ablaut. But I'm thinking more like
> Proto-Germanic or Proto-Romance rather than Proto-Indo-European. So
> there's still a lot of space for earlier ablaut processes.

That is, I think, a good idea.  The Proto-Jardic forms you have
given look about as close to their Jarda outcomes as Latin forms
to Romance ones.  Having Jardic as a branch of a larger family
opens up a lot of possibilities (one of the reasons why I decided
a few years ago that Albic would be just a branch of a larger
family, Hesperic).
 
> > - IMO the idea of having a whole proto-series of retroflexes to
> > explain your modern _ṛ_ is unlikely.  It is very rare for whole
> > place-of-articulation series to implode that catastrophically (manner
> > series do it a bit more).  Spontaneous retroflexion of a rhotic, on
> > the other hand, *is* likely; it helps exaggerate the so-called
> > flatness which is one of the perceptual cues of these sounds (a
> > lowering of whatever formant it is).  I suspect you just had an
> > alveolar *r of some sort in the proto-language, which has gone to 
> > modern Jarda _ṛ_.  Similarly:
> 
> Modern Jarda has an ordinary trilled /r/ as well, and I'm assuming that
> was a trilled alveolar /r/ in Proto-Jardic. But looking at initial
> consonant clusters in modern Jarda, it does seem likely that /ṛ/ was
> some kind of approximant to begin with. You never see /ṛ/ as the first
> consonant in an initial cluster (e.g. *ṛw or *ṛj), which would be
> expected if it had been some kind of stop. It could have been a lateral,
> but Jarda has enough laterals as it is. Or a tap, like /ɾ/ vs. /r/ in
> Spanish.

So you have an alveolar and a retroflex rhotic.  This *could* be
the residue of a merger of an alveolar and a retroflex series (or
whatever may have done in the other retroflexes), but that is not
necessary.  Not knowing the phonology you have in mind, I cannot
tell which scenario makes the most sense, though.

--
... brought to you by the Weeping Elf
http://www.joerg-rhiemeier.de/Conlang/index.html
"Bêsel asa Éam, a Éam atha cvanthal a cvanth atha Éamal." - SiM 1:1