Re: [tied] About methodology...

----- Original Message -----

From: Glen Gordon

To: cybalist@egroups.com

Sent: Saturday, August 26, 2000 9:14 PM

Subject: [tied] About methodology...

ALL sciences can be called theories of logical deduction, not just mathematics. Deduction is the key to it all.

No, Glen, physics or geology do not study logical deduction but things like elementary particles or rocks. They of course make USE of logical deduction. They employ more-or-less idealised mathematical models whenever possible, but are still EMPIRICAL, not AXIOMATIC systems. Facts can falsify a physical theory but not a mathematical theorem (once it has been proved). I suggest we leave the scientific method alone for the time being. We've clarified our positions and can't get much farther than that.

I can't touch language, I can't measure language. These sciences are fully distinct in that respect from comparative linguistics.

Language has physical aspects and leaves physical traces. Linguistic behaviour is observable and linguistic performance can be measured in a number of ways. Human language has its universals and its internal logic. UNRECORDED languages are not tangible to us, but what else do we use as the basis for our inferences about them, if not tangible evidence? Even mummies sometimes speak. The Zagreb Mummy was wrapped in the longest Etruscan text that we have.

Would IE be as successful today if there wasn't such a booming interest in the Orient in the 19th century?

Remember, please, that many language families have been proposed and accepted in our scientific age, and that the addition of Anatolian to the IE family in the 20th century went down quite well, though it required a thorough revision of the Brugmannian model and forced IEsts to take the laryngeal theory seriously.

 

IE became REALLY successful when people like Grassmann and Verner, and finally the Neogrammarians, introduced the principle of rigorous regularity in demonstrating cognacy (Grassmann was a famous mathematician beside being a first-class linguist). The Oriental fascinations of early IE scholars actually prevented them from seeing that PIE was different from Sanskrit. It was the new interest taken in Armenian, Albanian, Baltic, Slavic, Celtic and Iranian studies in the second half of the 19th century that made IE reconstructions solid and credible. As for Jones and his celebrated "Discourse on the Hindus" (1786), he was indeed "intuiting" things without being able to prove them, like some people I know. He guessed, very correctly, that Sanskrit was related to Greek, Latin, "Gothic" and Celtic, but you'd be surprised to see the full list of languages he thought belonged together: "Ethiopian", Chinese, Japanese, Etruscan, Phoenician, and even "Peruvian" (lest you should suspect he was really anticipating Dene-Caucasian). He was a genius, in a way, a visionary and a challenger, but he lacked the right methodology.

Again, you have an arbitrarily set level of tolerance to probability that prevents you, personally, from accepting the relationship of Gilyak to Altaic. This is not an absolute view that all people may share.

There is no view, no matter how reasonable, which all people will share, such is Human Perversity. Creationists don't believe in evolution, for example. As a matter of fact, I have no pet hypothesis about the relationships of Nivkh, that's all. Why should I have one? I'm not a specialist in Nivkh and I certainly wouldn't venture an opinion based on a cursory examination of word lists or selected grammatical patterns. Greenberg's hubris has brought about enough confusion. I'd be doubly cautious if the selection were drawn from Nostratic or Eurasiatic studies.

Piotr: I see, Glen, you tacitly ASSUME that all languages are ultimately related in a non-trivial phylogenetic sense (i.e., that they form a neat family tree which is reconstructable at least in principle). However, except in highly controversial classifications, there aremany isolated languages and tiny groups, so Nivkh would be no exception. Is your tacit assumption warranted, then?

Glen: Yes, it is. You are now contradicting your statements. Does Occam's Razor mean anything at all? In the absence of absolute knowledge, it is simpler to assume that all human languages derive from a common ancestral language (instead of two, three, five hundred...), just as it is simpler to assume that all humans derive from a common ancestor. As such, it would be totally counter to Occam's Razor to assume that Gilyak is NOT related to any given language(s) and therefore "isolate". This is by far the unlikeliest scenario.

Contrariwise, my dear friend. And don't slash at me with Occam's Razor so recklessly; you may hurt yourself.

 

First, as you certainly realise, humans DON'T derive from a single ancestor. They could in theory derive, as in the Book of Genesis, from a single ancestral PAIR, but that's untenable from the point of view of population biology: even a specially protected species won't survive if its total reproducing population is too small. What you really wanted to say is, I presume, that all humans derive from a single interbreeding population (the size of which is difficult to estimate). This may well be true: most -- though, mind you, not all -- palaeoanthropologists reject the multiregional theory of human origins.

 

Secondly, I'm not postulating the existence of lots of independent family trees with roots in Sub-Saharan Africa, but rather a dynamic equilibrium, throughout human prehistory, between the complementary forces that can be labelled "split and divergence" versus "areal convergence", and whose combined affect is a "tangled bush" model John and I discussed some time ago. If there is any truth in it, there isn't really much to reconstruct beyond a certain temporal horizon, simply because the notions of "family tree" and "protolanguage" become meaningless, rendering the comparative method useless.

 

Thirdly, I don't posit any "entia praeter necessitatem", so do put that razor down. If it turns out that the extant human languages must be grouped into 500 families, each with the corresponding protolanguage, so be it -- facts must be respected; after all, you'll have to incorporate the very same families into your model as well. But what's parsimonious about creating macrofamilies, superfamilies, phyla and superphyla not warranted by reliable evidence?

 

I don't mean to assert that the last word has been said and that all attempts to demonstrate long-range groupings are futile. I'm sure there are things waiting to be discovered out there. People keep trying, sometimes with unexpected results like the recent reclassification of the languages of SE Asia. But I see much more hope for modest-size clusters like, say, Indo-Tyrrhenian than for hyperambitious projects like Nostratic or Dene-Whatever.

 

Piotr