Hallo!

On Thu, 18 Dec 2003 23:20:33 -0500,
Rob Haden <[log in to unmask]> wrote:

> On Thu, 18 Dec 2003 22:08:14 +0100, =?iso-8859-1?Q?J=F6rg?= Rhiemeier
> <[log in to unmask]> wrote:
>
> >I think that Kartvelian is not as close to IE as is Uralic,
> >and some of the similarities between IE and Kartvelian
> >are due to contact.
>
> Certainly Kartvelian is not as closely related to IE as "Uralic" is.

Yes.

> However, as I said before, linguists currently believe that Kartvelian
> originated from northeastern Anatolia.  If IE originated from north(-east)
> of the Black Sea, it is doubtful that there were any meaningful contacts.
> There does seem to be evidence of contact with Hurro-Urartian, since both
> it and Kartvelian share and adjectival suffix in -Vli.  I'm not sure about
> other evidence, to be honest.  (Of course, a lambdaic adjectival suffix
> also exists in certain branches of IE, e.g. Latin.)  Also, Kartvelian seems
> to have had close contacts with the other Caucasian language groups, due to
> its large number of articulatory contrasts.  Some contact with Semitic
> could justify this as well.
>
> >No.  If PIE comes from Anatolia, one needs some *very* complex
> >migration patterns to explain the similarities between IE and Uralic.
>
> I completely agree!  I honestly can't see how anyone can claim that PIE
> came from Anatolia.

It doesn't seem that the cultures of Anatolian and central European
neolithic farmers match.  The former built small houses with square
plans; the latter built long houses (wherein long means about
20 metres and more) from the start.  Colin Renfrew picked Anatolia
because before the Black Sea Flood was discovered, it seemed the
only possibility.  No-one had areas on the board that are now
drowned beneath the Black Sea.

> >Yes.  And the similarities between IE and Uralic point at an IE origin
> >north of the Black Sea, too, no matter whether the languages are
> >similar due to common origin or contact.
>
> Right.  This makes sense, since IE would be more southerly than Uralic.
> However, I don't think that there was ever a Uralic Family per se.  Rather,
> I think "Proto-Uralic" was a looser association than PIE.

I have seen some bits of the debate on whether Proto-Uralic ever
existed or not.  I think the reconstruction is so lacking for two
reasons.  First, Uralic lacks ancient written records, which in the
case of IE bridge about half the time elapsed since the breakup
of the protolanguage.  One can easily see that Sanskrit and Latin
are related, for instance.  It is much more difficult with Hindi
and French, I mean without looking at their (fortunately known)
ancestors.  Second, there are fewer scholars working on Uralic
than on IE.  Most historical linguists choose their own family
as their main object of study.  And the speakers of IE languages
outnumber those of Uralic languages 100 to one.

But it is very well possible that (a) Proto-Uralic is even farther
back in the past than PIE and (b) many Uralic similarities are
due to contact.

> >And the neolithic farmers of central Europe don't come from Anatolia.
> >They are the descendants of refugees from the Black Sea Flood.
>
> Yes.  Language and history are often very intertwined!  Linguists don't
> notice that enough, in my opinion.  The movements of peoples is often
> highly significant in the development of languages.

Certainly.  When a language has spread from A to B, *someone* must
have actually made the trip - not necessarily in person, but perhaps
it were multiple generations involved.

> >I don't think that PIE had only one vowel phoneme.  The ablaut patterns
> >are as follows:
> >
> >strong grade   weak grade
> >
> >  á > e          a > @ > o/0
> >  í > ei         i > i
> >  ú > eu         u > u
>
> We don't *know* that for sure.  Linguistics, like any other science, is
> based on uncertainty.  All we're doing here is theorizing.

Of course.  Hence, add the asterisks that I have left out.

>       That said, the
> Ablaut patterns of PIE are usually easily distinguishable, and have been
> commonly accepted for decades.

Yes.

> However, it is readily apparent that PIE medial /i/ and /u/ derive solely
> from "zero-grade" Ablaut (e.g. leikW- "leave" vs. likWtós "left").  In the
> example, I think the root should properly be LYK[W].

I am not sure.  The pre-ablaut form might have been *laikW-, but,
as I have laid out above, it could just as well have been *likW-.
I tend to the latter version where another consonant follows the
semivowel (as in *leikW-), as there seems to have been a tendency
towards CVC roots.  In case such as *bhei- `bee', where the semivowel
is the last segment of the root, a diphthongal form *bai- is more
likely than *bi-.  And keep in mind that the vowels /i/ and /u/
most likely have been there besides /a/ for typological reasons.

> >The o-grade appeared where vowel deletion would have resulted
> >in an inadmissible consonant cluster; later it was paradigmatized
> >as a separate grade from the zero grade.
>
> PIE was fairly lenient on consonant clusters.

Which suggests that quite a number of vowels were lost.

>       However, a powerful constant
> in PIE phonotactics seems to be that clusters of more than two consonants
> was illegal.  The only exceptions to the rule seem to be sequences of 3
> consonants, where one of them is a sonant (and thus, isn't really a
> consonant, so there technically may not be any violations to the rule).
>
> Speaking of consonant clusters, the fact that there are so many initial
> clusters in PIE makes it almost a certainty that it *did* have penultimate
> (non-initial, in any case) accent at some point.

Yes.  Non-initial, in any case.  And the more archaic PIE accent
patterns indeed point at a penultimate accent.

> More about Ablaut... it seems to me that there are probably several
> different kinds of Ablaut going on in PIE just before it ceased to be a
> coherent language or Sprachbund.  If you want, we can try to classify the
> different types.  One that seems obvious to me is the root-internal
> Ablaut.  Another is the "thematic" Ablaut, which might actually be two
> kinds, one for nouns and one for verbs.

Yes, nominal thematic vowels have nothing to do with verbal thematic
vowels, except that the same phoneme was involved.  I think the verbal
thematic vowel was a 3rd person object marker (the same morpheme
as the 3rd person stative ending *-e < *-a) that later became
a transitivity marker and yet later a semantically empty stem-forming
element.  The thematic nouns might have been adjectives originally,
or just nouns that happened to end in *-a.  There were noun stems ending
in *-i and *-u, so there probably also were noun stems ending in *-a.
That's all.

> Certainly PIE exhibits some morphological processes that are similar to
> those of Afrasian languages.  In the Semitic languages, the oldest verbal
> distinction is the perfect-imperfect (or past-nonpast) one.  Besides the
> fact that one conjugation uses prefixes for marking person and the other
> uses suffixes, each also uses a different form of the consonantal root.
> Compare Arabic ?aktubu "I wrote" vs. katabtu "I write/am writing".

As Steg already pointed out, the other way 'round.  And the Afro-Asiatic
stative endings closely resemble the IE ones:

        AA       IE

1st    *-ku     *-h2a
2nd    *-t(k)V: *-th2a
3rd    *-a      *-a

Here, AA *k apparently corresponds to IE *h2.  (The vowel V in the
AA 2nd person is *a for masculine and *i for feminine.)

Miguel Carrasquer had the brilliant idea of tracing the stative
endings back to the AA prefix conjugation via a copula *-ku that was
suffixed to stative verbs:

1st  *?a-ku > *-ku
2nd  *ta-ku > *-tku + gender markers *a/*i > *-tka:/*-tki:
3rd  *ya-ku > *-a

>       I think
> it is possible that the differences in root forms is originally due to both
> stress-accent placement and (later) paradigmatic levelling.  Furthermore, I
> think a similar process occurred in (pre-) PIE.  What do you think?

Possible.

> >I also think that it is (mainly) due to accent.  Accented vowels
> >became strong grade, unaccented vowels weak grade.
>
> While I agree, there seem to be some exceptions in "commonly accepted
> reconstructed PIE".  For example, what about root nouns (a more ancient
> class) with genitives in -ós (e.g. Greek kuôn "dog," gen. kunós)?

I don't know.  If you ask me, the ending-stressed cases originally
had bisyllabic (and possibly not monomorphemic) case endings from
which later final vowels were lost, e.g. genitive *-asa.
This caused the accent to sit on the (first syllable of the) ending.
But why the genitive ending has an *o rather than the expected *e,
I don't know.

> >True.  And not always the assumption of a simpler system is the
> >simpler explanation.
>
> True.  Occam's razor always wins.

Right.

> >Yes.  The evidence is still very weak.
>
> I'm inclined to agree with you.  Linguistics is, by the very nature of its
> subject matter, an imprecise science.  Effectively, each of us speaks a
> different language.  However, with many other people, one's personal
> language differs only slightly.  But with others, the differences are great
> that we cannot understand one another.  So to try to describe language
> change in terms of something resembling mathematical axioms is in vain, for
> any linguistics theory will always be "messy".

Very true.

> Has anyone considered the possibility of, or tried to implement, a
> sophisticated AI program for looking at different languages and trying to
> find connections between them.  I think it would be fascinating to try to
> do that.

It would be.

Greetings,

Jörg.