----- Original Message -----
From: "H.M. Hubey" <hubeyh@...>
To: <Nostratica@yahoogroups.com>
Sent: Friday, January 31, 2003 10:21 PM
Subject: [Nostratica] One, first, head, finger


> That heuristic (called regular sound change before Kessler) seems to be the only one, not that it is not a sound heuristic.

What ought we to call it "after Kessler"? Is Kessler's research some kind of watershed in historical linguistics? :)

> Would it not behoove us to think about "principles" that can be applied consistently accross languages instead of making them along the way as we see fit?

Not as we see fit but as observation and analysis compel us. In reconstructing language relationships we apply the principles of comparative method as rigorously as possible. If the method yields no consistent reconstruction of a section of the protolanguage and a systematic derivation of the attested languages from it, the relationship remains undemonstrated. No other method known to me even remotely approaches the rigour and reliability of this approach. It would be brilliant to have something even better, but what would that be? Just speculating about hypothetical principles won't get us far. Loose talk about datamining and clustering isn't even the beginning of a method. Develop one, try to apply it and show that it works and yields believable results. Other people have already tried, but with little success.

> How can something as complex as language family relationships be based on a single heuristic?

Not on a single heuristic, but on careful comparative analysis in which that heuristic (or working assumption) plays an important role: it allows one to filter off a lot of noise, so that the useful signal can be isolated. Again, if you have a better idea, it's up to you to tell us what it is and to show that it might work.

Piotr