Distant linguistic relationships (was Re: Japanese from Tungus)

Hallo!

On Thu, 27 Jan 2005 18:57:15 +0000,
Ray Brown <ray.brown@...> wrote:

> On Wednesday, January 26, 2005, at 09:47 , Jörg Rhiemeier wrote:
>
> > Hallo!
> >
> > On Wed, 26 Jan 2005 19:03:11 +0000,
> > Ray Brown <ray.brown@...> wrote:
> >
> >> As far as I know, that Japanese and Korean are related is not proven.
> >
> > I agree to that.  I am also skeptical about their inclusion into
> > Altaic,
>
> So am I - very skeptical.
>
> > which I think has been done mainly for typological reasons.
>
> I think you are right.
Some linguists even reject Altaic altogether.  I think that Turkic,
Mongolic and Tungusic *are* related, but the time depth of that
relationship is much deeper than that of IE or Uralic, and there
has also been contact and borrowing between the three Altaic families.

> > (Greenberg does not include Japanese and Korean into Altaic, but he
> > nevertheless includes them into his Eurasiatic macrofamily.)
>
> Ooh - shades of Nostratic  ;)
Personally, I think that there is some evidence for a distant
relationship of Indo-European, Uralic-Yukaghir, Chukotko-Kamchatkan,
Eskimo-Aleut and Altaic (by which I mean just Turkic, Mongolic and
Tungusic), but the evidence is not overwhelming; and Greenberg's
two-volume work _Indo-European and its Closest Relatives_ is mainly
based on multilateral comparison, which actually proves nothing.

Of the language families mentioned, Uralic, Yukaghir, Chukotko-
Kamchatkan and Eskimo-Aleut seem to form a closer group.  This group
is called "Uralo-Siberian" by Michael Fortescue, and the evidence
he gives in his 1998 book _Language Relations across Bering Strait_
looks good to me, but I am no expert on the languages compared.

So, I propose the following structure of Eurasiatic (without
Korean, Japanese, Ainu and Gilyak), which I am trying to find
evidence for by system comparison (by which I don't mean something
Samuel Johnson would have called "milking the bull", but the
application of the comparative method to phonological and
grammatical systems):

A. Europic
  1. Indo-European
  2. Etruscan?
B. Uralo-Siberian
  1. Uralic
  2. Yukaghir
  3. Chukotko-Kamchatkan
  4. Eskimo-Aleut
C. Altaic
  1. Turkic
  2. Mongolic
  3. Tungusic

The time depths of the three groups are about 8000 years each,
that of Eurasiatic as a whole *at least* 10,000 years.

I am not sure whether Uralo-Siberian is closer to IE or to Altaic.
I somewhat tend towards the former, but the difference in time depth
is probably not significant.

The inclusion of Etruscan is uncertain as there is so frustratingly
little known of the language, though it looks in many ways similar
to IE without actually being IE.  (Bomhard and Kerns, in their
1994 book _The Nostratic Macrofamily_, treat it as an aberrant
IE language, but I don't agree with them.)

ObConlang: Albic would be A.3. in the above scheme, i.e., a branch
of Europic.

The Nostraticists consider Eurasiatic to be a subbranch of Nostratic;
the additional members of Nostratic would be Kartvelian, Sumerian,
Elamo-Dravidian (if those two are related at all) and Afro-Asiatic.
The inclusion of the latter two seems especially doubtful to me.

> > But as you wrote a few lines above:
> >
> >>       If one judges simply by similarities of structure, then a good
> >> case can be made out for a relationship between the Celtic and semitic
> >> languages; but few would take such a relationship seriously.
> >
> > Indeed.  In the case of Celtic and Semitic, we of course know that
> > the Celtic languages are Indo-European and acquired their "Semitic"
> > features secondarily, possibly from an unknown substratum.  (And the
> > Semitic languages are known to be Afro-Asiatic, which probably did
> > not display all of the "typically Semitic" features, either.)
>
> Precisely! Which is why I think positing relationships on grounds of
> typology alone is inherently unsound.
I whole-heartedly agree here.  A typological shared aberrance between
two (or more) languages may be a *hint* at a relationship, but in any
case, such hints require corroboration by the comparative method.

Greetings,

Jörg.