17 Mar 2010

Missing honey

First, a summary of the problems with PIE (*)*melit 'honey' that ignore Occam's Razor:
  1. Numerous irregular deletions of word-final *-it.
  2. Morphologically unanalyzable in PIE.
  3. Wildly defective genitives (*mlitós → Hit milittas, Lat mellis, Gk μέλιτος, Ir melo, Arm melu).
  4. Chaotic Anatolian vocalism (Hit milit- vs. Luw mallit-).
  5. Shared Greco-Anatolian features (ie. genitive in *-t-os) as if by areal influence.
  6. Greek μελίτειον is mislabeled an Iranian term (as per Douglas and Adams).
  7. Unjustified derivatives based on just one or two languages (ie. **meluyo- 'bee', **melitih₂ 'bee', **mlit-ye- 'to gather honey', etc.).
  8. Total disregard for regular sound laws and accent throughout (eg. *mlit-ye- > βλίττω yet genitive *mlitós > μέλιτος).
  9. Finno-Ugric fossilizes early IIr *médʰu- but fails to show any trace of *mélit-.
  10. Word vanishes magically in all IE languages east of the Crimea, even in Tocharian.
Except for Ethan Osten (aka senori, thesenori or senoki), a now-blocked Bard College freshman with an attention-eating twitter addiction, no one can seriously be confused about probable loaning here. The root isn't sacrosanct. Even Jaan Puhvel in Hittite Etymological Dictionary, vol.6 as of 2004 notes the immense problems under the heading milit(t)-:
"The precise accordance between Hittite and (already Mycenaean) Greek [...], and cognates elsewhere [...], are at serious variance with the heteroclitic postulation *meli-t/mel-nes (e.g. IEW 723) based mainly on Lat. mel/mellis (which may rather take after its antonym fel 'gall')." (see link)
Now, I realized something else that was plaguing my subconscious. There are far too many missing words for 'honey' across Italy that could decisively resolve the issue. Does anyone else notice this convenience? Latin mel with the coincidently unexpected genitive is always cited. Yet what happened to Umbrian? Oscan? Venetic? South Picene? Knowing that translating Etruscan maθ as 'honey' is illegitimate[1], what is its real term if not none other than *mel? How did so many words for 'honey' vanish in the same geographical area? How can a typical offering to the gods be absent from all pertinent texts in numerous languages? It's rather frustrating to those that care.

While we can at least eliminate some potential non-IE sources of this term in the general search area like Akkadian dišpu and Sumerian lal, my mind is intuitively drawn to the poorly understood language of Hattic since linguists keep on finding comestible terms in -it (Hittite sepit-, a type of grain, and Greek alphit- 'barley') that have no decent Indo-European explanation and yet have a strange habit of lingering around Western Anatolia as if substratal.

I'm still looking for a good answer but so far my hunch is taking this path:
A Hattic term, *melit, had spawned both Anatolian *mélit 'honey'/*mlit-eu- 'sweet' on the one hand and Aegean *meli 'honey' on the other (with *malítʰau 'sweet' via Anatolian).

Anatolian languages loaned the Greek reflexes while Aegean gave birth to Etrusco-Rhaetic *mel 'honey'/*mlítʰu 'sweet' which yielded the Latin, Celtic and finally Germanic reflexes (via Celtic) once this branch first arrived in Italy.

Then for the pièce de resistance, another Aegean offshoot, Eteo-Cretan, inherited the word *mlítʰu 'sweet', attested by Greeks in the first half of the epithet Britomartis.
Interestingly, -it is said to form feminines in Hattic as in hanwasuit 'throne' so could a Hattic form *mel-it be analysed this way too?

But dare I say more on this holy day of drunkards. A toast to all: A tall glass o' mead for them stinkin' snakes... then another for them drunkin' shakes. Sláinte!


NOTES
[1] The term maθcva, found three times in the Liber Linteus with inanimate plural -cva, can therefore not be a mass noun like 'honey'. G & L Bonfante were no doubt aware of the grave problem when they abandoned the expected value of *'honeys' based on their very own grammatical sketch for the ad hoc ammendment 'full of inebriating drink'. Unproven, ad hoc, and biased by Indo-European *médʰu-.

8 comments:

  1. There also occurs in Palaic a certain malitannaš to which Carruba remarks in the lexicon portion of his Das Palaische: Texte, Grammatik, Lexikon (1970:63):

    ma-li-ta-an-na-aš "honighaltig" o.dgl.; Ableitung aus einem *malit, vgl. heth. melit, luw. malli(t)-, wobei das Suffix noch unklar ist: Dat-Lok. Pl. 2A Vs.14, 19; B2 12 (ma-l[i- usw.); D7 (ma]-li-ta-an-[na-aš).

    Kammenhuber, Pal.19; Laroche, RHA XIII (1955) 75.

    I mean, it's not like it adds much new interpretation to the dataset; like he says 'wobei das Suffix noch unklar ist', but given the weird association of Palaic with the Hattic Ritual texts (as far as we're able to tell, anyway), that might have something to help your argument there if you want some sociolinguistic reason (e.g. some possible cult connection).

    Oh and I just remembered as I wrote that last part (brainstorm!), we should not forget that one of μέλι's earliest attestations in Greek comes from an offering context in Knossos (KN Gg 702):

    .1 pa-si-te-o-i me-ri *209(VAS) 1
    .2 da-pu2-ri-to-jo , po-ti-ni-ja me-ri *209(VAS) 1

    I wouldn't draw any serious conclusions on this evidence alone, but hey, it's nice food for thought.

    ReplyDelete
  2. Is it really necessary to post Ethan's twitter and talk about his supposed 'twitter addiction'? If you don't agree with someone, fine, go ahead and post his name, and link of his google profile.

    But none of the people reading this blog have any reason to even read his twitter.

    I tried reading the discussion you had with him. And while I agree, that there is no reason to assume from the Irish example that this comes from *melit- because mil is hardly a phonetically complex enough word to constitute proof that it is an old word, rather than a loan. There's nothing in the word mil that proves there was a *t at the end (yet, the vocalism does show that it had an *i, but Greek has the same thing).

    Greek works though.

    *melit gen. *melit-os

    word final *t is lost in greek (for example the 3sg of the secondary endings *-t).

    The fact that in the gentive this *t shows up, I think, is a rather strong point of favor of the word being loaned into greek as *melit-. While the word is still clearly of foreign origin it is no surprised that we have no ablaut in the root. And even if we did have

    *mélit gen. *mlitós it is hardly out of the range of possibility to assume that a paradigm meli gen. blitos motivated the Greeks to analogically fix it.

    I don't know. All I can tell from this is, that Greek probably did have *melit as it's original form, while the Latin and Celtic form are a lot more difficult.

    Proposing that this word is Purely PIE seems out of the question. The word is clearly foreign, and we can, at least, find no proof that it existed in all branches.

    But if I understood Ethan's argument correctly, he wanted to show that *melit- had regular correspondences in several languages, rather than 'random' loanword behavior.

    And while I think his case in Irish is rather weak, I think it's strong for Greek.

    And I don't think it implies that the word is Indo-European.

    ReplyDelete
  3. Phoenix: "There's nothing in the word mil that proves there was a *t at the end (yet, the vocalism does show that it had an *i, but Greek has the same thing)."

    Covered already under #3.

    OIr melo acts as a **u-stem** genitive and absolutely nothing like the original genitive (*)*mlitós, while milis 'sweet' being a separate word shouldn't contaminate our assessment of 'honey'.


    "Greek works though."

    Nope. Greek μέλιτος requires assuming an accent shift and also a restoration of the initial zerograded syllable in D&A's original genitive *mlitós. Again, problems and assumptions.

    Ethan Osten persisted that these cognates were regular based on nothing but a pompous self-assuredness. His factless bickering isn't welcome on Paleoglot. Neither is relativism. Real opinions are justified by real facts.

    ReplyDelete
  4. Mattitiahu,

    Yes, I've seen malitannaš and strangely it looks a lot like a theoretical Aegean derivative **malítʰau-na '(that) of the sweet'. The suffix -na is used in Etruscan to mean 'that related to, pertaining to, consisting of' and is also found in Minoan where, in my view, it acts as a genitive case marker.

    ReplyDelete
  5. Nope. Greek μέλιτος requires assuming an accent shift and also a restoration of the initial zerograded syllable in D&A's original genitive *mlitós. Again, problems and assumptions.

    That's hardly a problem though. In consonant stems -os is the only ending that Greek knows. And ablaut in the root of consonant stems is unheard of in any form of Greek.

    No ablaut in a Greek noun can not be used as evidence that it isn't Indo-European. If that were the case χθών 'earth' wouldn't be Indo-European either, as Hitt. clearly has nom. tekan gen. taknas (which is probably /tknas/).

    I'm sure this wasn't the implication of what you were trying to say, as you must be aware that root-ablaut in Greek is non-existent in Greek. (I bet there are a few examples that do have it, but I can not think of any right now).

    All in all though, I'm not sure whether the reconstructed genitive is correct.

    It's the typical Beekes method of reconstructing paradigms. 'Some words of this root in some languages have zero grade so it must be from the Genitive with ablaut.' And while I agree that he must have a good case at times, it's a bit dangerous.

    But nevertheless, the word *mélit remains difficult. And it's true that only Greek and Hittite show the proper form with an *-it suffix. All the other ones went on to 'innovate' the root.

    I agree that it's highly implausible that they replaced the 'suffix' *-it with whatever they liked, while *-it wasn't a suffix in these languages in the first place.

    It's much more likely this word came in these languages as *mel and was than expended wiht new suffixes to be able to inflect the word.

    Just one thing I'm not completely clear on now. Hittite and Greek do show words with ablaut. βλιττω is difficult to explain any differently.

    I have a feeling that you're going to explain this form with Etruscan Syncope, but could you illustrate how you imagine this happened?

    surely it's 'some form of melit'+ie/o-

    But what is this 'some form of melit' and how does it work semantically? Would that be derived from the word you mentioned in your earlier post *mliθ 'sweet' But how does the meaning 'to cut out the comb of bees, take the honey' develop from 'sweet'+ a rather semantically neutral verbal derivation *-ie/o?

    ReplyDelete
  6. "No ablaut in a Greek noun can not be used as evidence that it isn't Indo-European."

    Yes, absolutely not. We both agree that the individual points I raise when isolated aren't persuasive, but when all the forced assumptions are taken as a whole properly, (*)*mélit becomes dubious in a way that *dʰgʰōm can never be.


    "Hittite and Greek do show words with ablaut. βλιττω is difficult to explain any differently."

    It's farfetched to build a verb **mlit-ye- based only on one word in Greek so the more sane options are that 1) the word is innovative in Greek alone, or that 2) its borrowed from another language, such as from Anatolian where ablauted *mlit-eu- 'sweet' (> Hittite miliddu-/maliddu) is already attested.

    ReplyDelete
  7. It's farfetched to build a verb **mlit-ye- based only on one word in Greek so the more sane options are that 1) the word is innovative in Greek alone, or that 2) its borrowed from another language, such as from Anatolian where ablauted *mlit-eu- 'sweet' (> Hittite miliddu-/maliddu) is already attested.

    Sure it'd be insane to reconstruct that form for anything earlier than Proto-Greek.

    But I do wonder how Proto-Greek came up with this form. I doubt ablaut was still a productive system at the time that one could speak of 'Greek'. And if it was it still would not be usual for words with a *-ie/o suffix, as there are plenty words with this suffix that have a full-grade.

    And the other problem remains. I do not see how the meaning developed from 'honey' or 'sweet'. Sure there's a semantic relation between the two, but the meaning is very specialised.

    It would be lovely if the explanation was somewhere in between. If Hittite would have a meaning for a word with this root in the zero-grade which could be plausibly changed to get the meaning that we find in the Greek word.

    ReplyDelete
  8. The form *mlit-ye- is distinctly Indo-European but as you note (and I didn't realize that until you mentioned it), the zerograded initial syllable and shifted accent is atypical for denominal presents in *-ye-. Great catch!

    So my sights are on other Indo-European languages for the source of Greek βλίττω, namely Anatolian ones where a source for μέλι can already be found without appeal to the assumptive PIE roots. Note that Hittite already shows zerograded *mlit- in maliddu 'sweet'.

    Additionally, *mlit-ye- may actually be represented in Hittite-Luwian (Puhvel, Hittite Etymological Dictionary - Words beginning with M, p.157: Hier. *maliti(ya)- 'sweeten with honey').

    This just leaves the issue of semantic shift.

    ReplyDelete