Reconstruction talk:Proto-Slavic/melko

Definition from Wiktionary, the free dictionary
Jump to navigation Jump to search

What are the sources for asserting that mlijeko is a borrowing from Proto-Germanic? Outside of Wiktionary I have never seen this listed as a loanword. It is usually considered a cognate.

PIE *ǵ in *h₂melǵ couldn't have possibly yielded *k in Proto-Slavic, but it did in Germanic which was a source of dozens other words too. Native Satem reflex of that root can be seen in the verb *melzti "to milk" (> Serbo-Croatian mȕsti) < Proto-Balto-Slavic *melź-tej (> Lithuanian melžti). --Ivan Štambuk 12:27, 26 December 2010 (UTC)
There are many cases where there was no satem: gus' instead of zus', swekor instead of swesor, mleko instead of mlezo etc. And all of them are explained as borrowings, even though they're very basic vocabulary. Why isn't there a possibility that some internal reanalyzations/phonetical reassocations happened in pre-Slavic indoeuropean languages that did that? You often see -k/-g alternations (also -r/-l) in Indoeuropean roots; occasional lost of palatalization isn't something unique either. You see it in modern live languages a lot. But no, proto languages couldn't do that, they must conform to the table of sound changes 100% :) 09:07, 28 April 2013 (UTC)
*/k/ and */g/ and */ǵ/ are three different phonemes in Proto-Indo-European that don't simply "alternate" arbitrarily as you see fit. The reflex of PIE *h₂melǵ- was retained with in Slavic. What you're arguing is that the same PIE root simultaneously gave both a regular (*melzti) and an irregular (depalatalized *melko) reflex, which is absurd and silly. --Ivan Štambuk (talk) 18:33, 19 July 2013 (UTC)
On the other hand, it is strange that the medial -u- disappeared in Proto-Slavic. I would expect it to yield *melъko or *melъkъ. It doesn't seem like the Germanic languages lost the -u- until much later, after the Common Slavic period. —CodeCat 18:45, 19 July 2013 (UTC)
Well, the etymon used to be PGm *melkaz [1] before you replaced it with *meluks. That would be an adjective in Proto-Germanic, right? --Ivan Štambuk (talk) 19:00, 19 July 2013 (UTC)
That word could be a noun as well, morphologically. But that form doesn't reflect what the descendants actually show. -az is the nominative of masculine a-stems (o-stems in PIE), but the descendants are all feminine. What's more though is that the descendants (in particular Gothic) show remnants of an earlier consonant stem paradigm, which would have had an accusative in -ų (like u-stems) and a nominative in -s/z although it's hard to reconstruct. So that leads to the current form. Dropping of medial vowels is regular in Old Norse and Old English, but not in any of the other languages, which all preserve the medial -u-. So unless something happened in between, Proto-Slavic could not have borrowed this term directly from the Germanic people the Slavs were in contact with, but there must have been an intermediary in which the loss of -u- took place. Either that or the Slavs just understood the word wrong... —CodeCat 19:11, 19 July 2013 (UTC)
In other words, Old English meolc and Old Norse mjǫlk could reflect either *melkaz or *meluks. Orel also lists MLG melk and OHG melk in support of the former. It's much more economic to simply assume that Common Slavic *melko < Proto-Slavic *melka reflects Proto-Germanic *melkaz, rather than speculate about some unknown intermediary. --Ivan Štambuk (talk) 19:36, 19 July 2013 (UTC)
They can't reflect *melkaz because they are feminine and -az is strictly a masculine ending. A crucial piece of evidence is that Old Norse (and Icelandic) have u-mutation in the genitive singular, even though this is not the rule for nouns of this declension; the rule is in fact either no change or i-mutation. This indicates that the immediate pre-form of the Old Norse genitive mjǫlk (> Icelandic mjólk) must have had an -u- following the -e-, thus *meluk-, and it cannot have been *melk- as this would have become mjalk-. So there is no evidence whatsoever for a form *melkaz anywhere in Germanic, and even the evidence for *melk- (with a feminine declension) is rather scarse. Even if you ignore the Old Norse u-mutation, then the evidence for the form *melk- is still precisely in those languages that regularly syncopate medial vowels, so their evidence is not trustworthy. —CodeCat 19:51, 19 July 2013 (UTC)
None of these forms supportive of *melkaz have adjectival reflexes (i.e. not simply being used attributively)? --Ivan Štambuk (talk) 20:33, 19 July 2013 (UTC)
No, no adjectives, it's unambiguously a feminine noun. There is a class 3 strong verb *melkaną without the -u- apparently. But I don't really know how that works out. The -u- is really strange from an etymological point of view, because it's not in the PIE root, nor even in the related verb, yet all the Germanic languages show that it was there in the noun. —CodeCat 20:38, 19 July 2013 (UTC)
So, the adjective *melkaz would regularly be derived from *melkaną with no medial -u-, it was not retained in any of the daughters, but would easily account for the Proto-Slavic form *melka. I think that both theories are equally speculative and should be mentioned. --Ivan Štambuk (talk) 20:48, 19 July 2013 (UTC)
But what evidence is there for that adjective? To posit it just to account for the Slavic form is just silly. —CodeCat 20:54, 19 July 2013 (UTC)
It is explainable in terms of regular derivational morphology. Sure it has no direct descendants (according to you), but it's not impossible either. If Proto-Germanic speakers were to express the notion of "giving milk" or "milky", they would've come up with something like *melkaz. OTOH, derivation from *meluks presupposes two obscure sound changes, forcing you to speculate about an intermediary language. --Ivan Štambuk (talk) 21:11, 19 July 2013 (UTC)
That sounds more like a PIE word formation process than a PG one. Proto-Germanic derivation was affixal, ablaut was no longer productive. -az was certainly not a productive derivational suffix in Proto-Germanic. "Milky" would have been something like *melukugaz or *melukagaz. "giving milk" was *melukų gebandz. —CodeCat 21:34, 19 July 2013 (UTC)
I see. Then the possible cognates of PIE *h₂melǵos would have to be inspected. --Ivan Štambuk (talk) 21:43, 19 July 2013 (UTC)
  • I've updated the etymology per this discussion. Even if the derivation from *melka- is unlikely, we should mention it (with all its inherent problems) because it's commonly listed in etymological dictionaries. --Ivan Štambuk (talk) 14:19, 21 July 2013 (UTC)
  • What would the past participle of Proto-Germanic verb *melkaną. (that is reconstructible as far as I understood) look like? --Ivan Štambuk (talk) 14:19, 21 July 2013 (UTC)
    • It is a class 3 strong verb: *melkaną, present 3sg *milkidi, past 3sg *malk, past 3pl *mulkun, past participle *mulkanaz. Compare Dutch melken which preserves the strong conjugation more or less unaltered. —CodeCat 14:24, 21 July 2013 (UTC)