Dravidians from Africa/not Europe

[Larry Trask]

Clyde Winters writes:
 
>    Manding is the term used to refer to the Malinke and Bambara
> languages.  Moreover you have made the distinction that Somali is
> not related to the Other Black African languages this distinction is
> not accepted by most linguist of African languages that are of
> afrrican descent and know and speak the languages every day.
 
First, `Black African languages' is, at best, a purely geographical
label, and a rather unfortunate one, at that: I would suggest
`sub-Saharan languages', if a geographical label is required.  Nothing
whatever can be presumed in advance of investigation as to what
connections, if any, might exist among these languages.  In fact,
considerable investigation has been carried out, resulting in the
surprisingly small total of just four groupings for *all* African
languages.  Of these, Niger-Congo and Afro-Asiatic appear to be
accepted by all specialists as established genetic families.
Nilo-Saharan and Khoisan are more controversial, and the genetic unity
of each is at present regarded by many specialists as unsubstantiated,
though we continue to use these names as convenient geographical
labels.  Second, Somali belongs to the Cushitic branch of
Afro-Asiatic, and its genetic connections therefore lie with languages
in northern and eastern Africa and in the Middle East.  Of evidence to
connect Somali with the majority of sub-Saharan languages there is
none.
 
>    The terms used in this discussion are mainly taken from the Tamil
> Family of Dravidian languages.
 
Fine.  But Tamil (normally regarded as a single language) is
unquestionably Dravidian, and a good deal of comparative work has been
done on Dravidian.  It is inappropriate to cite specifically Tamil
forms if Proto-Dravidian forms are available.  One must always use the
earliest forms that are available and secure.
 
> > >                    Common Indo African Terms
 
> > > English       Dravidian        Senegalese        Manding
> > > Mother           amma              ama             ma
> > > pregnancy        basaru              bir           bara
> > > skin               uri               guri            guru
> > > King              mannan          mansa             mansa
> > > Grand             biru               bur            ba
> > > Saliva           tuppal             tuudde           tu
> > > boat              kulam             gaal            kulu
> > > cultivate           bey              mbey            be
> > > stream             kolli            kal              koli
 
>   The reason we can find analogy between the Senegambian (Wolof,
> Fula etc.) and the Manding and Dravidian languages result from their
> origin in Middle Africa.
 
No.  The reason we can find these forms is that *all* languages
exhibit miscellaneous chance resemblances.  It could not be otherwise,
unless languages were exempt from the ordinary laws of probability --
and they are not.
 
Miscellaneous resemblances of this sort can be found between any
arbitrary languages whatever.  Hardly a week goes by that somebody
doesn't produce just such a list of miscellaneous resemblances between
some surprising languages: Scots Gaelic and Algonquian, Hungarian and
Sumerian, Ainu and Norwegian -- you name it.  I've done it myself.
Miscellaneous chance resemblances are *always* present, and they are
devoid of significance.
 
Do you seriously suppose that, when Tamil-speakers were choosing their
word for `boat', they got together to discuss it and made remarks like
"Wait a minute -- we can't have that word, because they're already
using it in Wolof"?
 
> > Nursery words like /ama/ `mother' have no business being cited in
> > comparisons: they are worthless as evidence.  The same is true of
> > imitative words like /tu-/ for `spit, saliva', which are found all
> > over the planet: Basque, Burushaski, Caucasian, Yeniseian -- you name
> > it.
 
> > And the rest?  A little list of miscellaneous chance resemblances, of
> > the sort that can always be found between arbitrary languages.
 
> These resemblances can not be called chance resemblances because of
> the clear analogy in their construction and meaning.
 
They *can* be called chance resemblances, because they *are* chance
resemblances.  Would you like me to demonstrate this?  Let's look at
you next table.
 
> > > Pronominal Agreement
> > > Language        1P SG        2nd P        3rd P
> > > Dravidian     an, naa,ne      i             a
> > > Somali         ani          adigu        isagu
> > > Nubian         anni            ir          tar
> > > Bantu           ni             u            a
> > > Manding         na, n          i            a
> > > Hausa           na             kin          ya
> > > Wolof           ma             ya           na
 
> This is clear evidence of a close relationship between all of these
> languages as pointed out by Obenga and other African linguist.
 
Nope.  Let me draw attention to Basque, which I choose because it's my
favorite language.  The first-person singular pronoun in Basque is
/ni/, which matches your languages very well.  The second-person
singular pronoun is /i/ (written <hi> in the standard orthography, but
pronounced /i/ by most Basques).  Basque has no true third-person
pronouns, and it uses demonstratives for the purpose.  There is some
regional variation in these demonstratives, but all specialists are
agreed that the most conservative form of the third-singular distal
demonstrative is the western form -- which is /a/.
 
So, Basque, a language spoken on the Atlantic coast of Europe by a
conspicuously non-African and non-Indian population, has /ni/, /i/,
/a/, which matches both "Dravidian" and "Manding" better than most of
the other languages you cite, and far better than Somali.  Moreover,
the universal Basque word for `mother' is -- wait for it -- /ama/,
which matches both "Senegalese" and "Manding" fantastically well.  And
what is Basque for `spit, saliva'?  Ready?  It's /tu/ -- which matches
"Manding" perfectly, and matches both "Dravidian" and "Senegalese"
better than they match each other.
 
So, what can we conclude?  There are two possibilities:
 
(1) All this is a waste of time.
 
(2) Basque is *also* a "Black African" language!
 
I prefer the first.  If you, Mr. Winters, want to opt for the second,
then please count me out.  There is an American journal called Mother
Tongue which will doubtless be happy to publish your discovery that
Basque is an African language, and that the mysterious Basques are
therefore descendants of emigrants from central Africa.  You can have
a great time conversing with a number of other MT contributors, who
are busily proving that Basque is related to Abkhaz, Sumerian,
Burushaski, Tibetan, Ket, and Apache (no, I am not making this up).
But be aware that there already exists a substantial literature
proving that Basque is related to African languages, and to the Mande
languages in particular.
 
With this approach, you can prove anything you like.  You shouldn't
find it difficult to prove that Norwegian or Eskimo is an African
language, if that's what you want to do.
 
Of course, you'll find it so much easier if, as Gonzalo Rubio has
pointed out, you don't worry too much about whether the forms you cite
actually exist or actually have the meanings assigned to them.  I
think you'll get on very well with the good folks at MT, who take a
similarly robust view of their data.
 
Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK
 
larryt at cogs.susx.ac.uk