The classic definition of Proto-Nostratic is "the most recent common
ancestor of Indo-European, Uralic, Altaic, Dravidian, Kartvelian and
Afroasiatic". Subdivisions may vary somewhat from author to author,
and several tiny groups or linguistic isolates (most notably
Japanese, Korean, Nivkh, Elamite and Yukaghir) are also made to fit
into this broad family relationship.

Some Nostraticists (especially Bomhard & Kerns) consider the
membership of Sumerian as possible, while emphasising the loose
relation of Afroasiatic to the rest of the superfamily (reflecting a
particularly ancient split).

Greenberg regards traditional Nostratic as comprising languages that
are ultimately related but do not constitute a valid family. His
alternative proposal is the "Eurasiatic" macrofamily that overlaps
Nostratic by including the core grouping of Indo-European, Uralic-
Yukaghir and Altaic (with Korean and Japanese). However, Greenberg
excludes Dravidian, Kartvelian and Afroasiatic, and includes Ainu,
Nivkh, Chukotian and Eskimo-Aleut instead.

Starostin proposes a superfamily he calls "Eurasian" (very different
from Greenberg's Eurasiatic), divided into Hamito-Semitic (=
Afroasiatic), Nostratic and "Sino-Caucasian". Starostin's "Nostratic
proper" (without Afroasiatic) consists of Dravidian and, as its
sister branch, a core of more closely related families: Indo-
European, Uralic, Altaic and Kartvelian. He is also sympathetic to
the inclusion of Eskimo-Aleut and "Chukchee-Kamchatkan" in Nostratic.

Piotr

--- In nostratic@..., ieforum@... wrote:
> The question is the very idea of Nostratic as a common Proto-
language
> for a number of families. But what are those families sharing
> Nostratic features?
>
> There are obvious similarities in phonetics and morphology between
IE
> and Semitic (particularly after the Glottalic theory was
introduced).
> There are also cognates in the vocabulary between IE and Semitic,
IE
> & Caucasian, IE & Altaic etc.
>
> But still the number of members of the Nostratic macrofamily does
not
> remain stable. First what we should do, to my mind, is to define
> which families are definite relatives. Any opinions?