From: giaforni
Message: 71346
Date: 2013-09-26
[Reply to DGK's post - Sep 26]
A first sight, your criticism looks devastating: it gives other group members the impression that my whole article is a house of cards, based on very shaky foundations: if you remove a few cards, the whole construction seems to collapse.
However, your simplistic line of reasoning has a major flaw.
Suppose we have an etymology for a word made of five PIE phonemes. Four phonemes out of five can be derived via regular sound laws that are supported by ample evidence; however, a single phoneme out of these five cannot yet be explained by a well-attested sound law. What should we do? You say we should throw away the whole etymology altogether, because the single phoneme we cannot account for makes the whole etymology invalid. I say we should consider the etymology valid (after all, we accounted for 80% of its phonemes) and we accept we have a doubtful (or missing) explanation for the remaining phoneme: this way, we don’t “throw out the baby with the bath water”.
As shown in appendix 2 of my article, on average my etymologies consist of 4.6 PIE phonemes each (incl. laryngeals). So throwing away the etymologies of “long” words just because a single phoneme in some of them is still not fully accounted for is decidedly too drastic. This is simply not the way historical linguistics made progress.
For instance, Verner’s law was established five decades after Grimm’s law for Germanic (adducing IE stress as a conditioning factor for the first Lautverschiebung); so for several decades many Germanic etymologies were based on seemingly arbitrary alternations; but this was no reason to discard them altogether.
Coming closer to our times, if you browse any recent etymological dictionary, you come across similar cases very often, where a single detail in an etymology is unclear, but the etymology is still proposed as valid; this is especially true of such languages as Armenian, Albanian, Lydian, or some divergent Iranian ones, which have undergone changes that are still not fully understood after decades of studies.
If you are trying to establish sound laws for a language for the first time, you are bound to find several examples you cannot (yet) explain. If you decided to discard any etymologies where you have even the slightest doubt, you will make no progress at all. I prefer to salvage etymologies where I can explain most sound changes, and leave a minor part of the explanation still open for further studies.
I hope other fellow linguists in this discussion group will side with my reasoning.
Gianfranco Forni
---In cybalist@yahoogroups.com, <cybalist@yahoogroups.com> wrote:
[GF:]
Dear group members,
let me try and summarize what's been going on with this topic so far. Since I posted my first message over two weeks ago, I received few replies. The most frequent ones are fairly weak criticisms, by D. G. Kilday, mostly based on:
- a rejection of key parts of Michelena's and Trask's commonly accepted internal reconstruction of Pre-Basque (it would be interesting to be pointed to some published material where such rejection is supported by some systematic evidence);
- an analysis of a very small percentage of my etymologies, which are either refuted on various grounds (incl. the critic's unorthodox reconstruction of Pre-Basque), or dismissed as "loans" when the similarity with other IE terms is too evident to be otherwise dismissed.
Comments by other fellow linguists would be very welcome.
[DGK:]
My detailed criticism of your individual etymologies does not depend on my theory of the prehistory of Basque, which is no more unorthodox than those of Menéndez Pidal or Tovar; at worst it is an "idiosyncratic" theory. I have used this theory only on occasion to suggest simple alternatives to your etymologies. My linguistic views are far more orthodox than Larry Trask's, who imagined not only words but whole languages as arising spontaneously ex nihilo.
Using the Addenda et Corrigenda which you recently posted to academia.edu, I find that many of your individual soundlaws are supported by only one etymology, and are thus by definition ad hoc. All etymologies involving ad-hoc soundlaws should be discarded on general principles. These are the etymologies numbered 1, 4, 14, 17, 18, 19, 23, 26, 27, 36, 39, 44, 48, 53, 55, 64, 67, 68, 74, 76, 81, 85, 92, 97, 104, 109, 111, 113, 119, 120, 122, 133, 146, 163, 206, 208, 228, 234, and 235.
Optional soundlaws are unacceptable unless you change your name to Seanfranco. Your anaptyctic vowel comes out whatever you want, so set #5 amounts to an optional soundlaw, and 38, 40, 90, 91, 118, 126, 144, 154, and 224 must be discarded. Your treatment of *-VtV- as yielding *-t- or *-d- is another, so 10, 29, and 30 must be discarded. Likewise your *-m- > *-u- under unspecified conditions eliminates 137 and 165.
When all the etymologies above are stricken, other soundlaws become supported by only one etymology, and this second group of ad-hoc etymologies must be discarded: 8, 9, 21, 24, 96, 107, 108, 112, 135, 143, 230, and 236. Since 71 and 149 involve an implicit optional soundlaw (retention or loss of *b-, of whatever source, before *o), they must both be discarded, stranding 15 as ad hoc, which must also go. Similarly 5, 166, and 203 involve optional behavior of intervocalic *-n- and must be discarded, stranding 202, which must go.
Striking all the additional etymologies above strands only 156 as an ad-hoc etymology to be rejected in the third run. However, the entire soundlaw set #8 is now supported by only 93 and 110. Many individual soundlaws are also supported by only two etymologies. These are #9a (only 207 and 219), #10a (31, 115), #12b (31, 32), #12c (22, 47), #13d (11, 106), #14c (150, 209), #15b (46, 98), #15k (54, 123), #15l (32, 78), #16e (94, 223), #16f (16, 106), #16h (69, 89), #17d (114, 158), #18a (84, 86), and #23b (73, 99). So far today I have been operating with general principles only. If any of the etymologies listed above in parentheses is questionable for any reason, not only it but its partner in parentheses must be discarded, for the partner will become an ad-hoc etymology. I will return to this point in the next installment of my criticism.