kinship terms in Proto-Indo-European and Proto-Nakh

Darmstadt 1970).

Robert S.P. Beekes, Comparative Indo-European
Linguistics; an introduction
(Leiden 1995).

Johanna Nichols, “Chechen Phonology”, in: Phonologies of Asia and Africa, ed. Alan S Kaye (1996) 941-973.

Johanna Nichols
et Arbi Vagapov, Chechen English and English Chechen dictionary (2004).

Werner Beerle,
“A contribution to the Morphology of the simple verb in Chechen,” in: Studia
Caucasologia
I, ed. Fridrik Thordarson (Oslo 1998) 1-37.

Para Partchieva et Françoise Guérin, Parlons Tchétchène
– Ingouche
(1997).
  

 

 

Be Sociable, Share!

8 responses to “kinship terms in Proto-Indo-European and Proto-Nakh

  1. Yesterday, I had to delete a paragraph because the system kept giving me a “parser error” message. Now that I’ve read both posts through, I would like to bring that paragraph back to make sure my analysis is clear.

    We have a couple of good examples when Toch /ts/ corresponds to reflexes of /dh/ in other IE languages: Toch B tsaik next to Gk teikhos ‘wall’ (IE *dhoigh-), Toch AB tsäk next to Lat foveo ‘cherish’ (IE *dhegwh-). This suggests that Tocharian, just like Greek and Sanskrit, experienced a Grassmanian restriction on the co-occurrence of two voiced aspirates in adjacent syllables. Notably, in the case of tsäk and foveo, Sanskrit has dah- without aspiration just like duhita. Correspondingly, just like tsäk and tsaik presuppose Proto-Toch *dzäk and *dzaik, Toch A ckācer presupposes Proto-Toch *dzkācer > *tskācer > Toch A ckācer (before velar). Following Kloekhorst’s reconstruction of a hysterodynamic paradigm for the IE daughter term, we could postulate PIE *dṷegH2ter for Nominative (reflected in Toch A ckācer) and PIE *dugH2t- for hysterodynamic forms. It’s through the hysterodynamic forms characterized by zero-grade that the laryngeal succeeded in changing the front consonant from d- to dh- in many languages including Toch B.

  2. (contd.)

    “As far as I know both *dh and *d yield proto-Anatolian *d and subsequently change into proto-Lycian *t (Lycian kbi /cφi/ from Proto-Lycian *tṷi from PIE *dṷi is not aberrant), so it doesn’t compel us to assume that the PIE initial consonant was originally *d which was only secondarily aspirated by a distant laryngeal.”

    Good point. Anatolian seems to be indifferent in its treatment of pre-Anatolian plain voiced vs. voiced aspirates.

    “I can not see how we would get the long /ū/ in Germanic besides from a PIE *eu or *uH.”

    Honestly I don’t know. I lifted it from Kluge/Seebold, which has had the connection between brūdi and IE *mer- since 1800s. I guess some kind of “coloring” was going in this set under laryngeal influence. In Latin we see mariitus, in Lith marti, in OHG brūdi, so that PIE root *mer- yielded extended forms *mer-uH- and *mer-iH, with a possible third variant *mer-eH (frater, brothar, etc.) Divorcing OHG brūdi from Latin mariitus, Lith marti, Gk meirax, Crim Gothic marzus ‘wedding’ (?) will result in a non-parsimonious lexicological solution whereby OHG replaced an existing PIE word family with a new word *bhrūdh, for which, as far as I know, there are no IE parallels. As far as I understand, there’s no big problem with word-initial and word-medial mr- > br- in Germanic. You brought up the examples of OE bræġen and ON timbr, even without PGmc. *bremmanan. (But if mr > br is regular in Germanic, then it doesn’t matter if *bremmanan is onomatopoeic.) This makes the problem with the suffixal -ū- less of an obstacle to treat OHG brūdi as part of the PIE set *mer-. I don’t mean to say that the problem you identified so sharply is insignificant. I just think it’s the problem of IE-to-Germanic suffixation and not the problem of IE-to-Germanic root etymology.

    “Your suggestion that Proto-Latin *from- from *bhrom would turn into *form- is, to my mind, problematic, because it is unmotivated; A root form in *from- would have been perfectly acceptable to the speakers of Proto-Latin, as shown in frons, frontis, frons, frondis.”

    The throwback I postulate must have occurred after ancestral morwH2- underwent the mromH2- > bromH2- transition in Latin. Initial brom- must have been difficult, so it changed into *bormH2- > *formH2 > form- after the loss of laryngeals. How do you explain f- in Lat frigo next to Gk ‘ριγο? We have to postulate *srigo > Lat brigo (comp. PIE *memso > Lat membrum) but then there was no assimilatory context to account for *brigo > frigo. My hypothesis can account for the unique Latin/Greek f-/s- isogloss by reconstructing *(s)mergH2- > *bregH2 > Lat frigo- and Gk *sbregH2 > *srigo-. Even without Slav *merg- suggesting PIE *(s)merg-, Lat f- in frigo is well explained by laryngeal throwback.

    “I do not see where you get the laryngeal from in your reconstruction because the Slavic laryngeal was probably part of the *-iH suffix and there is no need to assume it for other languages.”

    I assumed that Gk μύρμηξ contained a suffixal laryngeal *eH-k-s, which lengthened the vowel. So, again, like in the case of brūdi and marti, we have two versions of the suffix -iH- and -eH-. My hypothesis postulates the introduction of a laryngeal into a paradigm by means of a suffix. Once the suffix containing a laryngeal is attached to the root, root consonants become susceptible to “laryngeal/aspiration throwback” due to 1) phonetic proximity between H2 and aspiration and 2) mishearing of CVCh2- to sound like Ch2VC- provided that the front consonant can be aspirated.

    Finally, a note on methodology.

    “as Indo-Europeanists we first have to make the phonology work before we can point at semantic correspondances.”

    My approach doesn’t involve giving priority to phonetics over semantics. I’m trying to strike a balance. If we only follow phonetics, then we end up reconstructing unique protoforms to which only an arbitrary etymology can be given. The current reconstruction *bhreH2ter, for instance, forces Indo-Europeanists to keep furnishing etymologies (such as “fire-carrier” in Szemerenyi 1978) that are very naive and implausible semantically. We can keep our phonetic “house” tidy, but only at the expense of messing up the semantic “house.” One of the reasons why an anthropologist like myself got interested in IE kinship terms is precisely because when anthropologists or ethnoliguists (such as Needham or Friedrich) look at existing IE etymologies and corresponding lexical reconstructions of the PIE kinship system, they look plainly speaking odd and typologically opaque. Precisely because they are odd and opaque, other anthropologists (Pfeffer) or historians (Kullanda) begin to create fantastic interpretations of what PIE kinship system might have been knowing what ethnographic examples can teach us. But no anthropologist has attempted to put on a linguist’s hat and look at the state of primary IE material. This is what I try to do. Kinship terms provide a unique opportunity to map phonetic variation onto morphological and semantic ones.

    Another reason why I try to balance phonetic, morphological and semantic considerations is because phonetics only will take us down a circular argument alley. Only phonetically valid forms can form a cognate set, but then any diachronic phonetic law is dependent on the composition of the cognate set(s) it derives from. We end up depriving ourselves of the possibility of going beyond the simplest phonetic laws and create protoforms that are bound to reflect modern forms, rather than inform us how those modern forms came into being. Then enthusiasts of mass comparison lift those forms from an IE dictionary and compare them with a Dravidian form with similar sounds and of similar meaning. We end up having a form which is postulated to be 10-15,000 years old but which apparently hasn’t changed much from the form still used by present-day speakers. But the reason it hasn’t changed much is not because it’s inherently conservative phonetically and semantically but because we forced it to be such. IMO, existing Nostratic reconstructions are impossible because their primary IE input is underetymologized. I criticized Nostratic kinship term reconstructions in one of my papers.

  3. Dear Peter Alexander,

    Thank you very much for your response. I enjoy our exchanges immensely, but please don’t feel obligated to respond to my comments to the detriment of your more important objectives.

    “following Kloekhorst.”

    I’m glad you brought up Kloekhorst’s paper, which I found very interesting. Traditionally, the full grade was imagined to be *dheugH2ter but only imagined as it’s not attested anywhere. If Kloekhorst is right about Anatolian, then it provides a missing piece of the puzzle plus a surprising shape of the full grade. One immediate thought I have, however, is that, if Anatolian attests to the full grade *dhṷegH2ter, then it’s unlikely that the onset had a voiced aspirate in it. In those languages in which voiced aspirates are attested, aspiration tends to disappear between consonants or between a consonant and a semi-vowel. In Sanskrit, dvara ‘door’ has d- instead of dh-, while aspiration only emerges in hysterodynamic forms such as Lat foris ‘door’. This suggests that front consonant aspiration entered the paradigm rather late from oblique case forms that contained zero-grade. Provided that full grade preceded zero-grade, PIE must have had *dṷegH2ter ‘daughter’ and *dṷerH2- ‘door’.

    “Tocharian B tkācer points to initial *dh because *d would either become *ts or disappear. As the second branch off of Indo-European I think we must take heed of this evidence.”

    Yes, but Toch A has ckācer. Usually, front c- is explained as emerging by assimilation with the /c/ of the suffix. But Kloekhorst is not convinced and suggests palatalization caused by -e- in the newly-reconstructed form *dhṷegH2ter. This indicates that he treats Toch A as derived from the primary PIE nominative form (full grade), while Toch B from the secondary PIE oblique form (zero grade). What remains problematic is that this interpretation posits a distant assimilation of dh to e. I will venture another interpretation, namely that Toch A derives from PIE *dṷegH2ter via *tskacer > *ckacer. This interpretation is phonetically plausible (dental affricate /ts/ shifting to alveolar affricate /c/ before a velar) and makes both traditional and Kloekhorst’s hypotheses about the origin of c- in ckācer redundant.

  4. P.A. Kerkhof

    Dear German Dziebel,

    My apologies for my silence and once again thank you for commenting; I need to put a lot of time in my thesis right now, so I hope you won’t be offended if I took some time to think things over, before I reacted on your interesting views.

    There are a lot of things I like in your reasoning and you bring some interesting problems to the fore. I want to react on some of them.

    In the case of the initial consonant in the word for “daughter” I’d stick with the original reconstruction of *dh. Tocharian B tkācer points to initial *dh because *d would either become *ts or disappear. As the second branch off of Indo-European I think we must take heed of this evidence.

    The Lycian form of “daughter” you quote is especially interesting, because the PIE form, that must lie at the base of the Anatolian form, probably had a full grade in the root; following Kloekhorst, I’d posit: Lycian kbatra, /cφatra/ from Proto-Lycian *tṷetra (a-umlaut gives the /a/ in the Lycian root) from PIE * dhuegh2tr-eh2. The rest of the Indo-European languages turned this dhuegh2tr secondarily into a hysterodynamic word. The fact that perhaps the suffix *-h2ter had a nominative gradation *-h2tr in Proto-Indo-Hittite makes the suffix even more enigmatic. As far as I know both *dh and *d yield proto-Anatolian *d and subsequently change into proto-Lycian *t (Lycian kbi /cφi/ from Proto-Lycian *tṷi from PIE *dṷi is not aberrant), so it doesn’t compel us to assume that the PIE initial consonant was originally *d which was only secondarily aspirated by a distant laryngeal.

    Connecting Germanic brūðiz to Lithuanian martí from *mortiH is perhaps a nice semantic match, but as Indo-Europeanists we first have to make the phonology work before we can point at semantic correspondances. I can not see how we would get the long /ū/ in Germanic besides from a PIE *eu or *uH. This long /ū/ or its predecessor which isn’t present in the purported cognates makes the genetic connection highly dubious and, to my mind, makes it more likely that we are dealing with a substratum word *bhrūdh or bhruH which got a PIE suffix -ti.

    Latin formica indeed has a messy set of cognates with a part of the cognates building on *ṷ(o)rm-o/i, probably under influence of PIE *ṷrmis. The fact that a considerable amount of Indo-European languages join in this lexical blending makes it plausible that *ṷ(o)rm-o/i dates back to PIE. The phonetic proximity of /w/ and /f/ could have brought the initial /f/ into Latin via an intermediary language. A scenario which involves distance dissimilation of m-m to f-m is also plausible. Your suggestion that Proto-Latin *from- from *bhrom would turn into *form- is, to my mind, problematic, because it is unmotivated; A root form in *from- would have been perfectly acceptable to the speakers of Proto-Latin, as shown in frons, frontis, frons, frondis. Furthermore, I do not see where you get the laryngeal from in your reconstruction because the Slavic laryngeal was probably part of the *-iH suffix and there is no need to assume it for other languages.

    As you can probably see, I still have some serious problems with your “laryngeal throwback” or “aspiration throwback” theory because the empirical basis for assuming aspiration caused by laryngeals in languages other than the Indo-Iranian languages is rather small. Therefore, Ockham’s razor advises us to stick with the hypothesis with the least new assumptions.

    Greetz,

    Peter Alexander

  5. Hi,

    Recently, in genetics blogging circles, there’s been a debate regarding a possible “Daghestani” component in Indo-Aryans. If you’re interested, check out http://dienekes.blogspot.com/2010/12/solution-to-problem-of-indo-aryan.html and http://dienekes.blogspot.com/2010/12/y-chromosomes-and-mtdna-from.html

    Also, for the possible mr-/br- or, better, mr-/bhr-…H2- connection, I forgot to mention Lat formica next to Arm mrǰiwn, Gk μύρμηξ ‘ant’, etc. All from mormH2- (or morwH2- in light of Slav *morwi-) and the Latin form from *mromH2- > *bromH2- > *from- and back to the original CVC structure in formica. This is a difficult cognate set, with some irregularities due to assimilation/dissimilation, but otherwise the presence of f- in Latin formica seems to be of the same secondary nature as I postulate for *bhreH2ter (Latin frater).

    “This scenario may help to explain why the IE words for “father” and “mother” are missing from the Anatolian lexicon; one may assume that they entered PIE at a date when the Anatolian branch had already branched off.”

    I like the idea of *pHter being an innovation in IE. It’s reflexes are also notably missing in Balto-Slavic, which agrees with Anatolian in having forms derived from *at- or *ta- instead. Same for Goth fadar, which is used only once as a vocative form, the referential one being atta. But I don’t think *pHter is a borrowing.

  6. Thank you for you extended response. I apologize for possibly posting twice but I’ve problems posting and can’t see if it got posted or disappeared. If you have to delete one of my comments, please delete the first one, as the last one has a typo corrected.

    “original PIE words beginning with *mr- are very rare, so the empirical base for positing a pan-PIE sound change is very limited.”

    I agree. However, if we take the potential mr > bhr change in *bhreH2ter in consideration we now may turn to words that show initial bhr/mr and not just initial br/mr. For instance, an isolated (afaik) Latin form frigo ‘I freeze’ may be connected to Slav *merz- ‘to freeze’ (from *mregH2-). This seems to be a better etymology for the Latin form than a singular case of s > f change as may be suggested by Gk ‘rigo- (form may be not accurately rendered).

    “Old English bræġen from Pgmc. *bragnam from *mroǵhnos. This would, as far as I know, be the only instance of this initial change in Germanic.”

    Don’t forget Germ *bruudi ‘bride’ and IE *mer- (with or without *bhreH2ter), which seem to be an accepted cognate set (see Kluge 2002 or later).

    I agree with you that we need more examples of aspiration derived from H2 outside of Sanskrit. And I do know the textbook example of πλατύς from *plth2us with -t- instead of -th-. Without trying to make a strong claim, I’d like to suggest that the hypothesis of a “laryngeal throwback” (= aspiration throwback) or a suprasegmental redistribution of vocalic and consonantal reflexes of laryngeals, as tentatively attested in *bhreH2ter, makes it possible that πλατύς comes from *plth2us through *phlatus (comp. thugater), with the onset aspiration not surviving between p and l. (You wouldn’t question the reality of Verner’s Law because clusters -st- didn’t change in Germanic dialects to -zt-.) So, we obtain:

    *dugH2ter > thugater
    *merH2ter > frater
    *plH2tos > p(h)latus

    Note that in Gk thugater the front consonant is aspirated, while in Skrt duhita it’s not. Grassmann’s Law tends to affect the first aspirate, hence it looks like th- in thugater is a secondary throwback, while d- in duhita looks natural/regular. Of course, Goth dauhtar and Osc fuutir point to PIE *dh- but this only pushes the throwback to the PIE level. In Greek, as in Sanskrit, H2 gives both consonantal and vocalic reflexes, the only difference being that in duhita they are proximal (h and i), while in thugater they are distal (th and a).

    So, by analogy with PIE *mreH2ter ‘brother’, I would reconstruct PIE *dugH2ter/*duegHter ‘daughter’. This makes no difference for Balto-Slavic and Celtic and will make Lyc kbatra a more natural outcome (comp. Lyc kbi and Lat dui ‘two’ with an unaspirated onset).

    There’s no doubt that one needs to bolster new etymologies with more and more phonetic and morphological matches. But we got to look in the right place. The etymology of *bhreH2ter that I suggest is very satisfactory on typological grounds (unlike the zillions of earlier etymologies), on morphological grounds (you don’t get better matches than Latv marsa and Lat fratria) and intriguing on phonological grounds. I don’t have your depth of understanding of IE phonology:), but, for an anthropologist, I’m rather well-trained and practiced in IE historical linguistics (having taken classes from the likes of Garrett, Baldi, Kazazis and spending years on reading up:==) to suggest a hypothesis worthy of future exploration.

    “I don’t really see why a formation *w+ašo would be different from a formation p+h2ter, making /h2ter/ not a suffix but instead an independ lexeme.”

    I can only see in -ter/H2ter a modifier. It can’t be a root because it’s found across so many different kinship terms (brother, father, mother, daughter, son, husband’s brother’s wife, etc.) that share no semantic pattern. Nakh ašo is only ‘sibling’ just like *me-/*meH2- or *p-/*pH2- are.

    My contention is that IE kinship terms haven’t been properly reconstructed, hence comparison with other language families – as part of a putative inherited fund or as a borrowing – can be premature.

  7. P.A. Kerkhof

    Dear German Dziebel,

    Thank you for your comment. Concerning the relation between PIE mer- and PIE *breh2ter, I see some problems. First I’d like to point at the fact that original PIE words beginning with *mr- are very rare, so the empirical base for positing a pan-PIE soundchange is very limited. If one leaves Greek βροτός, which is a backformation to ἄμβροτος from *(a-)mrotos from PIE *n-mr-to-s, out of consideration we only have βραχύς from PIE *mrǵhús and βρεχμός from *mreǵhmos left. It is possible that the latter has a cognate in Germanic and thus participated in this */mr-/ to */br-/ change; Old English bræġen from Pgmc. *bragnam from *mroǵhnos. This would, as far as I know, be the only instance of this initial change in Germanic, since Pgmc. *bremmanan can’t be connected to μορμύρω because of the obvious onomatopaeic nature of the word. Therefore I’d rather look for a different etymology for this word. Ofcourse we know for certain that the assimmilation /mr/ to /br/ took place wordmedially; PIE *temro to ON timbr. But the easiest explanation for initial Germanic /br-/ to correspond with Greek /φρ-/ is to rely on our reconstruction of PIE /bhr-/.

    This brings us to the aspiration that was, in your opinion, caused by a laryngeal. The only languages in which h2 certainly caused aspiration are the Indo-Iranian languages. The change can be posited for greek to explain PIE *ṷoidth2e to Grk οἶσθα but then why doesnt it happen in the case of πλατύς from *plth2us. The most sensible solution would, to my mind, be to restrict aspiration by h2 to Indo-Iranian and regard the perfect ending of Greek as a Greek idiosyncracy. Besides, /h2/ only aspirates a consonant in Indo-Iranian when it immediately follows the consonant. With PIE *bhreh2ter this isn’t the case. We can regard the e-grade as secondary but that places the laryngeal after the /r/ and not after the /b/. Then we´re left with the fact that the other IE languages don’t show traces of aspiration by laryngeals. In short, the explanation which needs the fewest new assumptions is to stick with our reconstruction of PIE /bhr-/ and therefore, to my mind, this explanation should be preferred.

    I don’t really see why a formation *w+ašo would be different from a formation p+h2ter, making /h2ter/ not a suffix but instead an independ lexeme. I know that my rant didnt constitute a viable theory but thats exactly what I said :). Finally, I agree with you on the point that the neogrammarians view on kinship systems was pretty limited. Fortunately, the typology of kinship systems has been greatly extended since then by work from anthropologists like yourself :).

    PS
    If didnt understand your remarks correctly, I apologize.

  8. Regarding *bhreh2ter “brother,” it’s likely that it’s related to IE *mer- ‘affine’ (Latin mariitus ‘husband’, Germ *bruudi ‘bride’, Latv marsa ‘brother’s wife’ – here is the clear morphosemantic link between the two cognate sets as marsa is identical to Lat fratria ‘brother’s wife’, with suffixal -r- regularly dropping in Baltic), so that the original form was *mreH2ter, which naturally turned into *breH2ter (Greek and some other dialects suggest that mr->mbr->br may have been common in PIE) and then the second laryngeal caused b- to acquire aspiration, hence *bhreH2ter. The latter development is supported by *dhugH2ter ‘daughter’ yielding duhita in Sanskrit, with /h/ usually coming from PIE *gh, and thugater in Greek/dauhtar in Gothic, pointing to PIE *dh-.

    The problem with IE kinship terms is that they are semantically complex and hence cognate forms were usually divorced into separate cognate sets by the Neo-Grammarians, who couldn’t imagine that ‘brother’ and ‘affine’ could be related notions.

    It’s unlikely that p- m- were class prefixes, as besides meH2ter and pH2ter there are no other examples. -ter or -Hter, on the other hand, seem to be some kind of kinship suffix. It’s position at the end of a term, rather than at the onset, is directly the opposite from North Caucasians examples you cited.

Leave a Reply

Your email address will not be published. Required fields are marked *