Language and Anthropology in the Americas

[Johanna Nichols]

----------------------------Original message----------------------------
AMR writes:
 
>  As for Nichols, her agument
>crucially depends on teh assumption that the many language families
>which most linguists do not regard as provably related are in
>fact UNrelated. For if they are related, then Greenberg is
>right and everybody goes home happy.
 
Not right.  My argument depends crucially on the assumption that the many
language families for which no probative evidence of relatedness has been
presented (despite the fact that such evidence has been sought by
comparativists) are SEPARATE STOCKS.  A stock (the term isn't crucial; this
is the one I use) is the oldest family-type grouping (clade in technical
phylogenetic terms) for which (a) genetic relatedness has been demonstrated
and (b) reconstruction of ancestral grammar, vocabulary, and sounds is
possible.  Some examples of stocks are:
 
        Semitic
        Chadic
        Basque
        Indo-European
        Uralic
        Austronesian
        Algic
        Yokutian (= Yokutsan + Miwok-Costanoan)
 
(A family deeper than the stock for which relatednes has been demonstrated
but reconstruction may well never be possible is Afroasiatic.  I would also
assign Indo-Uralic this kind of relatedness.)
 
I have estimated ages for large geographical groupings of languages,
notably those of the Americas, by determining the average greatest age for
stocks (ca. 6000 years) and the average number of initial branches of
stocks (about 1.5), and using those to calculate the number of 6000-year
'generations' (at 1.5 'offspring' per 'generation') required to give rise
to the present number of stocks.  In the case of the Americas I have also
estimated an immigration rate and computed the time required to give rise
to the present number of stocks by combined differentiation and
immigration.  Every time I have discussed this I have also calculated an
age based on only genetic differentiation and no immigration.  The ages I
get with these calculations are:
 
        * at least 20,000 years to form the American population of stocks
by differentiation and immigration; most recent calculation: ca. 40,000
 
        * over 50,000 years without immigration; most recent calculation:
at least 60,000
 
When discussing these figures I have made it clear that the second scenario
obtains if all the indigenous languages of the Americas are presumed to
descend from one ancestor, and the first scenario obtains if they descend
from several ancestors.  I have never maintained that each stock descends
from a separate immigrating ancestor; to my knowledge nobody has ever
advocated this; in fact it has been used by both Greenberg and myself as a
reduction ad absurdum showing the ludicrous consequences of assuming that
no deep genetic connections hold between the American stocks.
 
AMR gives a caricature of my position in the quote above.  My actual
position (in *Language* 66, 1990 and later work) is this:  If the languages
Greenberg groups together as "Amerind" descend from several ancestors, then
Greenberg is wrong.  If they descend (or are assumed to descend) from a
single ancestor, then Greenberg is wrong (because an immigration over
50,000 years ago is implausible, because just one immigration is
implausible, and because Greenberg explicitly claims an age of about 11,000
years for "Amerind").  (So anyone who assumes Greenberg is right about
"Amerind" must believe he is wrong about "Amerind".)
 
That is, I have always assumed deeper relatedness holds among the native
American language stocks.  And I believe my position is friendlier to
assumptions of deep genetic relatedness than either Greenberg's or AMR's
(or that of any long-range comparativist known to me).  My position is that
we can assume deeper relatedness exists and can work out rates of
diversification, immigration, etc. and therefore dates of colonization EVEN
BEFORE WE CAN IDENTIFY, MUCH LESS RECONSTRUCT, THE ANCESTORS and regardless
of the opinions of one's colleagues on the matter.
 
So no, I have never maintained that languages not now known or believed to
be related are all ultimately and completely UNrelated.  I don't think
anyone on earth believes this.  Of course no scientific linguist believes
it, but I don't think anyone else does either.  Even those who believe that
the world was created in its modern form in (what was it) 4004 BC by
deliberate design and without prior or subsequent evolution believe that
the modern languages descend from a single ancestor and that the
differentiation occurred after creation.  I mention this viewpoint not
because it is particularly relevant to scientific linguistics but by way of
showing how radical it is.  AMR, please stop attributing to your colleagues
a position that no scientific linguist takes and that even a radical
religious fundamentalist wouldn't take.
 
Johanna Nichols