Re: Questions about Japanese historical phonology.

John Cowan said:
> Mark P. Line scripsit:
>
>> I tend to take with a grain of salt *any* claim of the type "language L
>> is
>> a member of family F" where there are no historical records as
>> supporting
>> evidence (what I call, but apparently nobody else calls, *prehistorical*
>> linguistics).
>
> So you accept the Romance-language hypothesis (which is amply documented)
> but are skeptical about the Indo-European one?
I'm skeptical about "Proto-Indo-European" having a referent, and about the
many proposed changes that are hypothesized to have happened in the
evolution of that language into "the Indo-European language family". If
somebody can show me a stable cladogram in which something that can be
recognized as our good, old IE family, I'll become less skeptical.

I'm not skeptical about Sinhala being more closely related to Welsh than
it is to Nahuatl.


>> I think that prehistorical linguistics should concentrate on finding
>> evidence for claims of the form "language A is more closely related to
>> language B than it is to language C"
>
> The difficulty is to create an appropriate scale of relatedness such
> that pairs of languages can be compared in this fashion.  Biological
> cladists tend to count characters, but one person's character is
> another's suite.
Yep. The key lies in the concept of uniformitarianism. I think
prehistorical linguistics should be using a Labovian variational model of
language change, not random casuistry based on classical philology.


>> rather than trying to invent
>> unfalsifiable, Rube-Goldberg-esque networks of changes that attempt to
>> describe apparent patterns in data from A, B and C. Discretizing the
>> speech varieties of the four-dimensional planetary population into
>> "families" seems both far-fetched and ill-supported by the available
>> data.
>
> The limitations of the Stammbaum have been well-understood for more than
> a century now.
Yes, which is why I would have expected a more cogent response to those
limitations before now.


>> In practice, though, these are
>> problems of scale that tend to go away as the empirical database grows
>> in
>> size.
>
> On the contrary, they tend to get worse as the database grows, because
> more and more obscurities come to light that were hidden by a more
> cursory search.  This is well-understood in comparative lexical work,
> but I think the principle is applicable more generally.
I was referring to biological cladistics, in which the results become more
stable as more characters are included to describe each taxon. I don't
know of any comparative lexical work that uses cladistic methodology, but
it wouldn't surprise me if somebody had thought to try it.


-- Mark

Replies