From: Patrick Ryan
Message: 43426
Date: 2006-02-14
There are two major arguments advanced by "mainstream" linguists _against_ the possibility of valid comparisons among reconstructed language(-familie)s when the comparisons are theoretically placed at a time-depth greater than 10kbp (+/- 1-5ky; this varies so widely and capriciously that the propositions for which it defines the time-frames should per se be suspect as artificial and the result of dogma vs. scientific determination):
1) at this advanced time-depth(s!), transformational phonetic developments have erased the evidence of _any_ discernible real connections;
2) at this advanced time-depth(s!), substitutional vocabulary developments have so depleted vocabulary inventories that no real cognates can exist.
Both of these propositions are false, in my opinion, but they will not be discussed at this juncture. I will take them both up if interest is shown by the lists in continuing the discussion.
In spite of the theoretical (virtually) absolute limits set by the assumption of these processes, alleged cognates, composed of sets of morphemes related by regular sound changes and semantically reasonably relatable or identical meanings have been found by Nostraticists like M�ller, Illich-Svitych, Bomhardt (and others) principally but also by longer-range comparisonists like Starostin, Sr. and Jr. (and Bengtson). Additional _possible_ cognates have been identified by presumptive classificationists like Greenberg and Ruhlen on a necessarily less rigorous identification of purely apparent similarities.
Ringe's principal 'contribution' to this debate was to suggest that
1) at this advanced time-depth, any 'only apparent' cognates of meaning and morpheme shape, which do 'appear' to exist in spite of the two major processes so dearly beloved by 'mainstream' linguists, are mere non-significant 'look-alikes', and are the result of purely random processes (COINCIDENCE of both form and meaning!) divorced from any causal connections.
Ringe attempted to "prove" this proposition immediately above by the introduction of supposed statistical analysis, first, in his own formulations, which failed, and then in conjunction with 'statisticians'.
He discusses this in his own words at
http://www.tlg.uci.edu/~opoudjis/Work/ringe.html
and his own words, in my opinion, should prevent any scientifically objective linguist, from taking his statistical theory seriously, and any conclusions which flow from it.
Ringe says:
"Only a tiny fraction are, maybe only 10%, but that's still rather a lot. I didn't find such a hard core in Illich-Svitych, I've been through it and I don't find it there. That's really what I'm talking about. It is possible to get some kind of measure of what the range of chance is and I'm working on that, and I've published a number of things along those lines, and I'm still trying to improve the mathematics, and improve the application of the mathematics to data, to come up with a set of guidelines that will work across the board, so we can sit down and apply the guidelines and come up with a good estimate at the other end."
From this, it should be apparent to any objective reader that Ringe really has no statistical model but rather a series of purportedly 'statistical' models that he continues to tinker with because they DO NOT conform to reality BY HIS OWN admission.
Where he has presented these 'statistical models', he has, in spite of his own candid admissions of failure here, asserted their validity; and linguists, eager for any reason to discount comparisons at longer time-depths, have eagerly embraced them.
Knowing his own failures, Ringe then goes on to state:
"Actually Ann (Taylor) is giving the presentation, we're working jointly with a computer scientist whose mathematics we don't understand, it's a little too complex. I've got enough understanding to field questions if they're not technical. I hope nobody's going to ask how the algorithm works because the last explanation I got lasted 15 minutes, and I went along with it pretty well for 15 minutes and then she said "and then you just do it all with dynamic programming" and I can't even visualize that!"
Should this explanation of the latest version (of which I am aware) of 'Ringe's' statistics give any objective linguist any confidence in the results obtained by it?
I think it should be remembered that I did not introduce "random look-alikes" as an explanation for anything though I certainly will admit the extremely unlikely statistical possibility.
There is something to legitimately criticize in almost every one of the paragraphs attributed to Ringe in the interview referenced but I believe, sincerely, he has been damned in his own words. I think the probability is near 100% that he knows and understands NOTHING about statistics
If any linguist wishes to explain away purported cognates at a time-depths greater than roughly 15ybp, he would do well not to invoke Ringe! and 'his' statistics, which are, of course, not really HIS by creation or even understanding.