<snipping the ad hominem here and in what follows; let me only say that I'm genuinely interested in Nostratic studies and if I see my role on this list as the devil's advocate among partisans of Nostratic, it is not out of contempt for anything or anybody, but in the interest of maintaining a healthy balance of opinion>.... </snip>> For argument, I mentioned that IE roots are canonically CVC- (Kurylowicz would turn in his grave to think that *g^enH- was being put forward by you as an IE root) with only a few exceptions for *shorter* function words, and that *g^enH- must be interpreted as CVC+H-. Accordingly, it cannot be the root --- even if all attested forms pointed unequivocally to *g^enH- as their basis.> Now, a question to Piotr: do you believe that some IE roots have the form CVCC? Do you dispute the idea that all (with the exception noted above) IE roots have the form CVC?Yes, I dispute it. By the way, it's Benveniste's idea of "root", not Kurylowicz's. For me, the root is primarily a synchronic concept -- the unanalysable central morpheme of a word, not a "primitive form" in the diachronic or glottogenetic sense. Canonical roots are analytic artifacts. Of course you can insists that any morpheme of the form CVRC _must_ be analysed into CVR+C or whatever your preconceived idea of "canonical" forms. I prefer not to insist on such divisions if there is no independent evidence of the shorter forms. For example, I treat *leuk- or *derk^- as roots, since there is no reason (other than a Procrustean love of invariable shapes) to segment them into *leu-k- or *der-k^-. There are lots of languages permitting C(C)V(R)C morphemes, so why shouldn't PIE have been one of them?> I offered Old Indian jánati. This word does *not* require *g^enH- as its basis; it can be derived very nicely from *g^en-; and therefore proves my contention that howevermany words might be derived from *g^enH-, there is every reason to believe that some were derived from *g^en-, which validates the general CVC-rule for IE roots.It is a circular speculation, not a general rule. It is not the "standard theory" of IE word-formation either. The reconstruction *g^enh1- has the advantage of sufficiently explaining all the PIE derivatives of the root (including, crucially, its characteristic nil grade). The environments in which *g^en- seems to be a possible analysis are prevocalic (here, of course, the loss of *h1 leaves no traces). Preconsonantally, *h1 is systematically reflected in the "diagnostic" branches.> Piotr misrepresents what he did. He mentioned a number of Slavic forms that may or may not be related to zrak, having to do with 'shine'.> In Pokorny, these words are derived from 3. *g^her-. Pokorny also has 2. *g^her-, 'scratch, scrape', just before it. Obviously, zrak could just as easily be derived from 2. as from 3. *g^her-, which, in effect, is what I am suggesting.First, let me assure you that the forms I gave _are_ related to <zrák>. The Czech word is not an isolated relict that could be interpreted "just as easily" any odd way, but a member of a large word-family. Its cognates occur so densely in Slavic, and all the historical stages of semantic shifts that have affected them are so well documented, that there can be no mistake.> Piotr asserts what he should attempt to prove when he says: "I also explained the origin of related Slavic forms and their meaning." What Piotr did was to cite some words that *may* be related but are not necessarily related.His explanation is just speculation!> Now I do not claim to know much about Slavic linguistics. But I have a Russian dictionary that lists Russian zarosl6, 'underbrush, thicket', that looks to me as if it might be a reflex of 2. *g^her-. But I will not insist on it. I simply do not know.That you don't know much about Slavic is evident. <zarosl'> = {za+rost+l'}. The root here is {rost-} < Proto-Slavic *orst- 'grow', and {za-} is a prefix of prepositional origin.> I cited an example, Greek gennáo:, 'I beget (said of a father)'. Regardless of what Piotr's Polish dictionary may say, my dictionary "Liddell-Hart", says "BEGET".Why should a Polish dictionary be inferior to Liddel & Hart? But those interested in what the latter have to say may check this link for themselves:As you will see, the verb may refer to either parent, and when it is used of the father, the quotations given show that "to beget" in this context means simply "to father" and does not focus on the physical process that leads to becoming a father.> Piotr claims that this root is a denominal verb: "Actually, it is a denominal verb, from <genna:> 'birth, origin', hence its rather general meaning."> If Piotr thinks *g^enna:- is a PIE form, I beg to differ.PIE? I mean Greek <genna:>, from which <gennao:> is derived:> [Piotr] asserted that Sumerian tu(d) meant 'beget'. I told him that it probably meant simply 'bear' (female aspect of the process), and cited reasons to believe that.> He returned by impugning my ability to interpret the Sumerian data; and I assert for the record that Piotr has no idea what training I have had in Sumerology. He just wanted to denigrate my interpretation.> I showed him that Thomsen, an authority whom HE cites, seems to agree by defining tud in her grammar as "Thomsen defines tud as 'to bear, to fashion' on pg. 320 of her grammar. Sorry for you but no 'to beget'."> Piotr passes that by in silence because, I believe, he has only heard the name Thomsen, and never read her book. If he has, he is very forgetful.It seems what we really disagree about is the precise meaning of "beget". You concentrate on the physiological aspect ("emitting semen into a vagina") and I on the abstract relation ("becoming the father of"). Naturally, the latter can't happen without the former in the physical world, but that doesn't mean that the "physiological" meaning is older or more fundamental than the "relational" one, or that the two meanings are not lexically separable. What I said in the first place was that the meaning of Sumerian tu(d) was roughly the same as of PIE *g^enh1-. It so happens that meanings like 'bear' or 'produce' are covered by the latter. Curiously, at your site you refer visitors to John Halloran's "Sumerian Lexicon". Whatever its merits, it gives "to beget" as one of the senses of <tu(d)>.> Anyone who questions whether words derived from a verb 'plait' can serve as a basis for words meaning 'basket' really should be doing something other than comparative work.> And yet, Piotr questioned whether OHG kratto, 'basket', could be derived from IE *ger-, 'plait'. He asks for evidence from other branches: I gave him Old Indian gárta-H, 'wagon-seat', which we know were made of withe. He ignores it.> I doubt that Northern Europe needed to borrow a word for wickerwork from the invading Romans since burning people in baskets was one of their favorite pastimes.> Latin cra:tis, 'wickerwork', comes from another IE root, *ker-t-. One has nothing to do with the other. Since we have OHG hurd-, why would they have needed to borrow anyhthing.With the same logic, <egg> cannot be an Old Norse loanword, since the Anglo-Saxons had eggs prior to Scandinavian settlement in England, and <eagle> cannot have been borowed from Old French, because there have always been eagles in Britain. I simply asked for solid evidence. You provided two forms without as much as attempting to analyse them. You think in terms of roots -- that is, you match consonantal skeletons and ask no further questions if they seem to fit. I'm more interested in details like ablaut grades and derivational processes. Do you claim that Old Indic garta- is related to OHG kratto? OK, but _just how_ are they related? what sort of derivatives are they? You seem to be saying "Germanic kr- matches Indic gar- and the rest be damned". Well, it doesn't happen to be my favourite approach. I accept that two words are related if I understand the nature of the relation.> Piotr also asserts that the sequence s-k-r is iconic.> If words beginning with skr- in a number of languages have to do with 'scratching, scoring' etc. it is because they are s-mobile forms of roots with, without the s-, meant 'scratching, scoring', etc.I wouldn't say that iconic words cannot have ordinary etymologies. My point is that, being iconic, they follow their own patterns of development (note, for example, the almost general dispalatalisation of OE scr- in "scratchy" verbs) and are of weak evidential value in linguistic comparison, let alone _distant_ comparison. The etyma you refer to are notoriously irregular, involving numerous variants: *(s)ker-, *skrebH-, *skrob-, *skorb-, *gerbH-, *gHrebH-, and the like. The common denominator is the phonetic makeup rather than a shared underlying root (though *ker-, for one, _is_ a bona fide PIE morpheme).> So, that is Piotr's argument. Evidently, poor mortals looking at Pokorny, simply cannot get the right idea. It takes special training.Sorry, but it's true. You can't learn PIE from Pokorny. You need some training and a lot of practice.> So, let me ask since I can defer to the magister: what formal training in IE linguistics have you had, Piotr?I have had some at academic level. I got my PhD for work in historical linguistics. I teach historical phonology and the history of English at Poznan University in Poland.Piotr