plural, classifiers & sampling

[Elena Maslova]

At 09:27 11/11/98 +0100, MORAVCSIK Edit wrote:

> The point that David Gil made yesterday seems to me to be very
important.
> Taking off from Johanna Nichols' earlier remark, David pointed
out that
> when testing implicational universals, it is not enough that
the
> sample be genetically and areally balanced but one must make
sure that
> the features serving as implicans and implicatum are not
themselves
> areally (or genetically) clustered.

In fact, the need to balance samples areally and genetically
represents a compromise solution intended to overcome our
inability to *afford* random (or total) samples. If someone tests
a universal against a really random sample (which turns to be
genetically or areally unbalanced from someone's else point of
view), the results are anyhow more reliable than those obtained
on the basis of a balanced non-random sample. If it is the total
population, one can claim something about the *existing*
languages for sure (and the results of this level of reliability
are rare, if existent).

Now, if one has investigated the *total population* of type A
(HAVE CLICKS in Edith's example) languages, and all these
languages belong to type B (HAVE V-O AGREEMENT in Edith's
example), one can safely claim that the implication A > B holds
for all existing languages, independently of whether or not the
total population is genetically and/or areally unbalanced (as is
inevitably the case for areally restricted features). The
question is, of course, whether this conclusion is theoretically
intersting, but it is, beyond doubt, reliable.

To say the truth, I have been wondering for a long time why
geography is getting such a major role as a linguistic
explanatory factor - i.e., if some geographical patterning is
observed, linguistics appears to be not required anymore.
Consider David's example:

> Imagine the following.  Here's the map of the world.  Along
comes the
> number-marking genie, who gets to fling one big blob of
number-marking at the
> world.  She hits Africa, Europe, and much of Asia.  Next comes
the classifier
> genie, who gets to fling one medium sized blob of classifiers
at the world --
> she hits East Asia.  Is it that surprising (ie statistically
significant)
> that the two blobs happened not to overlap?  Not at all.  If
what we're
> counting is blobs, rather than points on the globe (or
individual languages),
> then what looks like a typological generalization turns out to
be the result
> of an accident.

Now, with genies it is always the case that their discoveries are
either accepted (if the society is ready to accept them and
*needs* this discovery) or not noticed. The society that has
already accepted the discovery of number-marking and a society
which does not have such a tool may well react differently to the
discovery of classifiers (by their own or a neighbor genie) - due
to essentially *linguistic* factors. Then, it is not an accident
(from the linguistic point of view) that the second society (in
another geographical area) did accept the new discovery and
integrated it in their languages, while the first one did not.
Accordingly, it is not an accident, from the linguistic point of
view that "the two blobs" do not overlap - and it may well have a
revealing and even *correct* linguistic explanation.

Lena

Elena Maslova
University of Bielefeld
mailto:lena at lh.bicos.de
http://doma.bicos.de/elevine/