From: dgkilday57
Message: 71305
Date: 2013-09-11
This is a reply to message #71297 [Gianfranco Forni].
You blame me for using 140+ sound changes to justify my 200+ etymologies.
I believe this amount of sound changes is by no means exceptional. Let's consider, for instance, Lydian, an Anatolian language that underwent major phonetic changes.
Lydian is still poorly understood: few Lydian lexemes and morphemes are understood with reasonable certainty.
However, in spite of this dearth of material to work on, Craig Melchert (in his "Anatolian Historical Phonology") uses 130+ sound changes to derive Lydian from PIE (namely, some 50 sound changes from PIE to Proto-Anatolian, and 80+ sound changes from Proto-Anatolian to Lydian).
Gianfranco Forni
--- In cybalist@yahoogroups.com, <cybalist@yahoogroups.com> wrote:Reply to D. G. Kilday
While my etymologies are based on Michelena and Trask's commonly accepted internal reconstruction of Pre-Basque, several of your criticisms are based on different, alternative reconstructions to Michelena's (namely, *mi instead of *ni for 'I', *ano instead of *a[h]o for 'mouth', *menti instead of *bendi for 'mountain', *arr instead of *ar 'male', *giza- instead of *gizon for 'man', *sut instead of *su for 'fire', *un- instead of *ur for 'water', *on instead of *bon for 'good'). As Octavià correctly pointed out in a parallel comment, such unorthodox reconstructions are unwarranted (unless you can point us to a publication of yours where such claims are substantiated by adequate proof, if any is available).
So if we exclude the above mis-reconstructions, we are left with 14 criticisms (namely zu 'you', aitzin 'before', aita 'father', argi 'bright', *anets 'breath', barre 'laughter', bederatzi 'nine', bero 'hot', biga 'two', etorri 'come', ezur 'bone', sudur 'nose', txori 'bird', ama 'mother').
For 3 of these (aitzin, aita, argi) the resemblance with Celtic is undeniable, so you simply dismiss them as loans. While I do agree that a few resemblances might be due to chance or loans, the cumulative amount of evidence provided by my article cannot be explained away as an unprecedented amount of loans for basic lexicon: see pages 306-307 of JIES #41 for a detailed statistical explanation and an empirical proof of this.
One etymology (txori) is jokingly justified as "luck", which I assume is intended as a lucky hit, i.e. chance resemblance. Well, any *individual* etymology might be explained away as such; it's the *set* of etymologies based on regular sound correspondences that cannot; so this criticism is very weak.
So, we're left with a mere 10 more-or-less grounded criticisms of my 209 etymologies (167 for basic lexicon + 42 additional ones in Annex 4): this amounts to less than 5%: hardly a substantial amount to justify the claim that you can "find no merit" in my theory.
Anyway, let me take a closer look at some of these residual criticisms.
For zu 'you', see page 281-282 of my reply to the critics in JIES #41, where I assume a conflation of singular *tu and plural *su-.
For barre 'laughter' and bero 'hot' you criticize *sw > b: this sound law is in fact supported by 6 etymologies (barre, beltz, ber-, bero, -bil and mehe), and an identical phonetic outcome can be found e.g. in Old Irish tuath-bil 'left turn', where the second element comes from *swelo- (according to Matasović's "Etymological Dictionary of Proto-Celtic" , page 363). As for 'dubious semantics' for barre, it is exactly the same semantics as in Proto-Celtic *swar-yo- 'laugh' (Matasović, page 361). On the other hand, your own semantics for bero 'hot' < 'black' looks pretty suspicious to me.
Nowhere does Matasovic' get *b- from *sw- IN AUSLAUT as you do. In the 200-page draft of his Celtic dictionary (formerly available at ieed.nl) he refers to the IE etymology of the *swelo- 'turn' words in the EIEC as "possible but unconvincing".
Finally, if your soundlaw is faulty, it hardly matters whether you have 1, 6, or 100 faulty etymologies erected upon it.
As for bederatzi 'nine', on pages 291-292 of JIES #41 I demoted it from "likely" to "tentative" etymology, exactly because it is still too speculative at this stage.
As for the alleged semantic permissiveness in my etymologies (a criticisms that recurs in your comment), please see Annex 5 in my article, which shows that 91% of them are based on an identical or very close semantic match.
Coming to your final sentence, "since you have obviously invested considerable time and personal reputation into this theory, I do not expect my criticism (or that of others) to have any effect": it is not a matter of personality or pride: it is just that, based on the factual analysis above (compounded with Octavià's reply), I would say your comments do not appear to be strong enough to dismiss my theory. Lat's see what comes up in further messages on this topic, anyway.
Gianfranco Forni
[DGK}: Now that I have a bootleg copy of your JIES paper, I can begin to criticize the etymologies in greater depth.
1. Basque _aho_, _aha-_ 'mouth'; PIE recte *h3éh1os, *h3eh1es-, archaic gen. sg. *h3h1sós 'id.' (cf. de Vaan, EDL 436; Pokorny, IEW 784-5). Tavi was correct in pointing out the inadequacy of my earlier criticism, but his protoform *Cabo hardly agrees with your theory. The Northern Basque combining form is attested in _ahamen_ and _ahazuri_. In order to salvage your theory with this comparandum, we must posit PIE *h3e- > Bq. a- (against your *h3(e/o)- > o- in Table 1), intervocalic *h1 > Bq. -h- (against your arbitrary adding and dropping of aitches on the grounds that Trask considered them "non-phonemic"), and not only loss of word-final postvocalic *-s, but loss of postvoc. *-s- when a consonant immediately follows.
2. Bq. _aitzin_ 'frontside, before', etc. Since Celtic origin is highly plausible, this has no value for your theory.
3. Bq. _aita_ 'father'. Ditto.
4. Bq. _anai(a)_ 'brother (of a male)', protoform *annai-; alleged PIE deverbal noun from *h2ed-g^n.h1-je/o- 'to be born near' with simplex *g^n.h1jetor 'he is born' > Celtic *ganjetor > Old Irish _-gainethar_. You do not provide the explicit form of the alleged noun, which is supposed to have retained the *-j- of the present stem. I can find no such noun in Celtic, which does reflect *g^n.h1-tó- in Gaulish _Cintugna:tos_ 'Firstborn' and *g^neh1-tó- in Middle Welsh _gnawt_ 'relative' (IEW 373-4). There is no good reason why IE-speakers would drag a present-stem /j/-suffix into a deverbal noun when appropriate nouns were already available. If your Vascoid IE reflected PIE *n.h{x} as *na: you would have *h2ed-g^n.h1-tós becoming *(h2)adg^na:tos and by your own rules (including *-dgn- > *-nn- designed for this particular word, plus those required to salvage _aho_) *annato.
5. Bq. _ar_ 'male'; PIE *h2/4né:r 'man, hero'. If _ar_ indeed reflected *anar as you claim, the northern form would be *ahar, and in fact under #166 you refer northern _zahar_ 'old' (southern _zar_) to *zanar. But then you say *-ar in *zanar may be identical to _ar_ 'man, male', whose protoform you insist is *anar, so the REAL protoform of _zahar_ must be *zananar. And around in circles we go. At least I understand your strategy for getting published: Make your papers so hopelessly convoluted that reviewers fall asleep before finding the internal inconsistencies!
Not that you would care, but this comparandum also involves a semantic difficulty, since Bq. _ar_ is usually employed in compounds like _oilar_ 'male chicken, rooster', and similar compounds with *h2/4né:r would be ludicrous.
6. Bq. _argi_ 'bright; light'. Almost certainly of Celtic origin, and useless for your theory.
7. Bq. _hats_ 'breath'; PIE recte cum accentu *h2n.h1-tó- 'breath', multo melius *h2énh1-tu-. If the protoform were *anets, northern Basque would have *ahets not _hats_. By your own rules accented *-tós would not become *-tsu, but unaccented *-tus would. However, my improvement on the PIE side does not fix your problem on the Basque side.
8. Bq. _-ba_ 'kinship suffix'; PIE *-kWeh2- recte 'FEMALE kinship suffix' in Old Cornish _modere-b_ 'aunt', etc., *-kWo- 'MALE kinship suffix' in Sanskrit _ma:tr.-ka-_ 'mother's brother'. Your own rules do not allow *-o- and *-a- to fall together, yet we have Bq. _osaba_ 'uncle' and several other males. Moreover Bq. _izeba_ 'aunt' has a northern variant _izeka_ and a southern variant _izeko_. The latter has probably just substituted the very common _-ko_, but _izeka_ has a different suffix _-ka_ which is closer phonetically to IE *-kWa: than _-ba_ is. At any rate since Bq. _-ba_ is gender-neutral I prefer to derive it from West Mediterranean *-wa observed in Etruscan _ruva_ 'brother' and _nacn(u)va_, a word sometimes found in apposition with _ati_ 'mother'.
9. Bq. _barre_ 'laughter'; alleged PIE deverbal noun from *swer-je/o-, better zero-grade *swr.h{x}-je/o- yielding Celtic *swarjo: 'I laugh'. This suffers from the same problem as #4. You have provided no explicit protoform for the noun and no justification for incorporating a present-stem /j/-suffix into it. Nor have you explained how Bq. -e could result from such a deverbal noun.
Your alternative to *swer(h{x})- is the very flimsy *sward- 'to laugh', erected by Pokorny with a question mark (IEW 1040) on some Brittonic forms and Greek _sardánios_ 'bitter (of laughter)'. The Celtic words are assigned to *swarjo: by Matasovic', and Beekes (EDG 1308) does not even mention P.'s proposal for the Greek word. But even if you could justify this root, and add an additional rule *-rd- > -rr- to your pile, your protoform would be a thematic *swa(:)rdo- with no explanation of Bq. -e.
10. Bq. _bederatzi_ 'nine', allegedly from *bede-eta-oitsu 'one and eight' with irregular *-t- > *-d- > -r- and *-u > -i, containing irregular *-tsu from oxytone *-tó: in the reflex of PIE 'eight'. I believe you have (wisely) already withdrawn this one.
Of your first ten comparanda, only the first has even a slight sliver of merit, and that only when one of your soundlaws is revised and two more are added. Not an auspicious start!
Douglas G. Kilday