(pseudo-)generalizations and explanations

[Martin Haspelmath]

Many readers will have been shocked to read Matthew Dryer's statement
about the "value" of explanations and generalizations:

"I see the crosslinguistic generalizations as more valuable than their
explanations.
... having an untested and probably untestable hypothesis for why the
generalization exists
doesn't add a whole lot to the value of the generalization itself."

I think it's useful to be reminded how speculative most of our deeper
explanations are, but Matthew's statement sounds as if there was some
absolute value attached to kinds of scientific activities. But plainly,
different scientists pursue different goals. Many linguists are
primarily interested in deeper insights into the nature of language and
care about cross-linguistic generalizations only to the extent that they
provide evidence for or against certain hypotheses or theories. This
seems to be just as legitimate (and potentially valuable) as Matthew's
attitude, which apparently takes for granted the value of
cross-linguistic generalizations as a goal in themselves.

As for the danger of "detecting" (and even publishing)
pseudo-generalizations, I think we'll just have to live with it (nobody
is pefect). If we say that a generalization is significant at the 0.05
level, it means that there is a 95% chance that this is not a
pseudo-generalization. Instead of lamenting about the 5% failure rate,
we should be proud of the 95% success rate. Compared to the presumable
success rate of "micro-typology", which ventures to propose a universal
generalization on the basis of four or five languages (and often less),
95% is a staggering figure.

Martin

--
Martin Haspelmath (haspelmath at eva.mpg.de)
Max-Planck-Institut fuer evolutionaere Anthropologie, Deutscher Platz 6	
D-04103 Leipzig
Tel. (MPI) +49-341-3550 307, (priv.) +49-341-980 1616