Tuesday, March 22, 2016

Discussion of Vladimir Diakoff's proposed sound changes

The following is a guest post by Vladimir Diakoff about his proposed Indo-European sound changes. This discussion is continued from this Language Hat post: THE INDO-EUROPEAN CONTROVERSY: AN INTERVIEW.

Pre-Laryngeal Plosive Loss Law

This new Indo-European phonetic law identifies the co-existence of forms with stops vs. forms with colored vowels. Since colored vowels (e, o, a) are known to have derived from PIE laryngeals and a combination of voiceless stops and laryngeals yielded voiceless aspirates in Indo-Iranian, it’s reasonable to postulate that it’s the PIE laryngeals that are responsible for the observed doublets. The split was apparently conditioned by the existence of forms in which a "T+H" sequence before V was a cluster (the ancestral state) vs. sequences in which T+H became a consonant phoneme or H+V became a (colored) vocalic phoneme. I hypothesize that in the pre-PIE there were no voiced aspirates but only voiced stops. In the environment immediately preceding a laryngeal, a voiced stop acquired aspiration through the phonemicization of the cluster voiced stop + laryngeal. There's some indication that palatovelars may also have gotten their palatal articulation from a contact with a laryngeal but this hypothesis is presently redundant as the “system” presented below can work without it. I don't know if there was a separate line of voiced laryngeals that may have voiced neighboring stops. I understand that we need to know in what paradigmatic contexts the phonemicization of a laryngeal occurred. I don't have an answer for this yet but the minimal requirement of conditioning (namely, cluster vs. phoneme), I think, is met.

Laryngeals were a group of phonemes, H1, H2, H3, distributed from front to back (in terms of their place of articulation. H1 keeps e as /e/, hence it must have been close to this vowel from the point of view of its place of articulation. H3 colors /e/ and makes it /o/, so it must have been labialized. H2 turns /e/ into a back vowel /a/, so it must have been an uvular or pharyngeal fricative. This is the simplest interpretation of the identity of laryngeals and it seems to fit the data best. H₁ is considered to have a palatal secondary articulation. This causes the following changes: H₁ > y in some contexts and H₁a > i in Sanskrit. This is an alternative explanation to the "schwa indogermanicum"/"syllabic laryngeal," as I reject the sound change H̥ > i in Sanskrit because H is not attested in any ‘father’ forms, all branches outside of Indo-Iranian (including Pal papaz ‘father’, Toch A pa:car, B pa:cer ‘father’) display an invariant /a/ and reconstruction *pH2-ter or *p-H2ter seems to be too overengineered for such a widely attested baby kin term root as *pa-.

Following many scholars, I assume the existence of old, pre-PIE *a and *o that were not products of laryngeal coloring. Hence, combination H1a, H1o, H2o, H3a are just as possible as H1e (= “new” e), H2e (= “new” a), H3e ((= “new” o).

Following A. G. E. Speirs and German Dziebel (see http://kinshipstudies.org/2014/07/23/indo-european-labiovelars-a-new-look/), I consider a labiovelar split best described for Greek (telos, polos, kyklos depending on the quality of the following vowel) of pre-PIE or PIE age. This assumption underlies a couple of etymologies below, but the vast majority of these etymologies don’t depend on whether one accepts this alternative chronology of the labiovelar split or not.

Pre-PIE pH1e > PIE *pe / *e

1. *pH1ekwo-/*kwH1ekwo- (Lat coquo, iecur).
Gk hepa:r (hepso ‘boil’), Lat iecur, Skrt yakrt, Slav *ikra ‘liver’ ~ IE *pekwo- ‘cook, bake’ (Lith kepu ‘bake’ > kepenos ‘liver’, Slav *pesti ‘bake’, *pesteni ‘liver’) > PIE *pH1ekw- ‘bake, cook’ > *pH1ekwr-/n- ‘liver’. The derivation of ‘liver’ from ‘bake, cook’ in Balto-Slavic is remarkable considering that it applies to both *pekw- and *kwep- root variants. In Greek, an overlooked pair hepa:r ~ hepso shows the same semantic link.

2. *pH1ek'w- (Lat pekus, equus)
IE *pek’us ‘livestock’ (including horses) (Skrt pasu, Lat pecus, Umbrian pequo, Lith pecus) ~ IE *H1ek’wo- ‘horse’ (Gk ‘ippos, Skrt asva, Lat equus, Toch yakwe, etc.) < *PIE *pH1ek'u-/*pH1ek'wo-.

3. *pH1egwh- (Lat bibo, e:brius)
IE *pH1egwH3- ‘drink’ > Hitt pa:si ‘swallows, drinks’, ekuzzi ‘drinks’, Pal ahu ‘drinks’, Luw u ‘drinks’, Toch yok ‘drink’ (< *ye:gwh), Arm empem 'drink', Lat e:brius 'drunk' (with a long grade because it's a "vrddhi" derivative), bibo: (< *pibo), Skrt pibati 'he drinks', payayate 'give someone a drink', OIr ibid 'he drinks', Gk pino: 'drink', posis 'drink', ne:pho 'be sober' (< *ne-egwho), Slav *pijo 'I drink', Lith puota 'drinking feast'.

Typically split into two sets (*H1egwh- and *peH3-/*piH3-), this group has traditionally posed a number of issues. The main one is -b- in Lat bibo:, Skrt pibati. It's currently assumed that they go to reduplicated *pipH3- and that H3 voiced the preceding consonant. But this interpretation is completely ad hoc and unfounded. The problem with *H1egwh-, on the other hand, is that it's rather sketchily attested outside of Anatolian and that Pal attests for a laryngeal -h- ostensibly corresponding to Hitt -k-. Finally, nobody has ever figured out why PIE would have two words meaning 'drink'.

Pre-Laryngeal Plosive Loss law (and Dziebel's hypothesis of a PIE labiovelar split) clarifies some (if not all) of these problems. -b- in Lat bibo:, Skrt pibati is the same -b- as in Lat e:brius going back to PIE *gwh (Gk ne:pho). *H1 reconstructed for *H1eghw- points to an earlier cluster *pH1-. Forms with long -o- (Lith puotas, Lat po:tas) go back to cluster *-gwH3- (see above for examples such as *mregwH2-, *mek'H2ter, *swetH2l-). Pal ahu shows a form after the plosive loss, while Hitt ekuzzi shows a phonemicized cluster *-gwH3-.

4. *pH1et- (Lat petere, iterare). IE *pet- ‘fly, rush, beseech, attack, demand’: Hitt pittar ‘wing’, Lat petere ‘rush, attack, beseech’ (hence petitio– > Eng. petition, repetere ‘do again’), Gk petomai ‘fly’, piptein ‘fall’, etc. ~ H1iter- ‘again’ (Lat iterare ‘repeat’, Hitt itar ‘road’, itrani ‘messenger’ < *pH1et-.

Pre-PIE *pH1a- > PIE *pa (InIr *i > 0) /*a

5. PIE *pH1ater. H1 does nor color ancestral /e/. And, now I can argue, it does not color ancestral /a/. So that's why we find /a/ in all the 'father' forms in IE languages (including the most divergent branches such as Pal papaz ‘father’, Toch A pa:car, B pa:cer ‘father’, Gk pappos ‘grandfather’, pater ‘father’) but Indo-Iranian. In Indo-Iranian we logically find /i/ in Skrt pita: ‘father’. Because H1 had a palatalizing effect, in Skrt sequence H1a gave /i/ and didn't turn p- into ph-.

6. PIE *pH1awo-/*pH1awH2o- 'grandfather, grandchild' (Hitt huhhas ‘grandfather’, Skrt putra ‘son’ [< *pH1autro-), Lat avus ‘grandfather’, avunculus ‘mother’s brother’, puclos ‘son, child’, ONorse afi ‘grandfather’, OIr aue ‘grandson’, amnair [< *aunater) ‘mother’s brother’, Welsh wyr ‘grandson’ (< *pwyr)

7. PIE *pH1awyo-: Gk pais ‘child, son’, Slav *uji, Lith avynas, OPrus awis ‘‘mother’s brother’.

Sets 5-7 coalesce at pre-PIE *pH1a- ‘father, grandfather, son, grandson’ enlarged with different affixes (Gk pappos ‘grandfather’ ~ pater ‘father’ ~ pais ‘son, child’, Toch A a:we ‘grandfather’ ~ Toch A a:p ‘father’, Toch B a:ppo ‘father’; Arm haw ‘grandfather’ ~ hayr ‘father’ (Gen. hawr); Old Irish aue ‘grandson’ ~ athir ‘father’; Old Norse afi ‘grandfather’ ~ fathir ‘father’). Hitt huhhas ‘grandfather’ actually attests for a laryngeal long postulated as responsible for the vocalic contrast between Gk pater and Skrt pita but never attested in forms with a literal meaning ‘father’.

Pre-PIE *pH2e- > PIE *pa / a

8. IE *peH2ur-/n- ‘fire’ (Hitt pahhur, Gen. pahwenas, Gk pyr, etc.) ~ *H2a-s-/t- ‘ashes, hearth’ (Hitt hassa ‘hearth’, Lat a:ra ‘fire-altar’, a:trium ‘room with a fireplace’, Czech vatra ‘fire’ < *pvatra, etc.) < *pH2es-/t-/*pH2eur-/-n-.

9. IE *H2ewis ‘bird’ (Lat. avis, Gk. aetos, Skr. vis, Av. vīš, Arm. hav, Lith. višta, Ir. aoi, Welshhwyad ‘duck’) ~ IE *pout- ‘bird’ (Slav *рutа ‘bird’, Lith putýtis ‘birdie’, pučiùtė ‘hen’, šìlо pùtinas ‘quail’, Skrt рṓtаs ‘animal cub’, Lith раũtаs ‘egg’ < PIE *pH2ow-t-. Note: Lith раũtаs ‘egg’ corresponds to Lat ovum, Gk oio:n, Slav *aje, Welsh wy ‘egg’, etc., which are often derived from the ‘bird’ root *H2owis.

10. IE *per- ‘strike, split, push’ (Arm harkanem, OIr orgaid ‘he kills’, Lith periu, Slav *pereti, Skrt sphurati) ~ IE *H2er- ‘to plow’ (Hitt hars- ‘work land for sowing’, Toch AB a:re, Gk aroo:, Lat aro:, OIr airim ‘plow’, etc.), *H2erH3tr- ‘plow’ (Gk arothron, Lith arklas, Slav *ralo, etc.) < *pH2er-.

Interestingly, Slavic thundergod Perun (< *per-) and his son Jarilo who plowed the soil may be derived from the same root *pH2er-.

11. IE *H2enH1- ‘breathe’ (Skrt aniti ‘he breathes’, Gk anemos ‘soul’, Lat animus ‘same’, Toch B a:nme ‘self, soul’, Arm holm (< *honm) 'wind' ~ IE *pneu- 'breathe' (Gk pneo: 'breathe, blow', pneuma 'breath', OEng fne:osan 'sneeze' (http://starling.rinet.ru/cgi-bin/etymology.cgi?single=1&basename=%2Fdata%2Fie%2Fpiet&text_number=1600&root=config) < *pH2enH1-/*pH2nH1-.

Pre-PIE *pH3e- > PIE *po / o

12. Another example of a possible PIE cluster *pH2- turning into phoneme *p or losing p if remaining a cluster is PIE *peH2-s ‘protect, feed’ vs. *H2owi- ‘sheep’ : Hitt pahs- ‘protect’, hawis ‘sheep’, Arm hoviw ‘shepherd’, Gk poimen ‘shepherd’, po:u ‘flock of sheep’ ~ o(F)is ‘sheep’, Lat pa:sco ‘put to graze’ ~ ovis ‘sheep’, Old Irish oi ‘sheep’. So we would reconstruct *poH2- > *pH2owi- (Hitt pahs- but *phawis > hawis).

13. IE *H3estH1- ‘bone’ (Hitt hastai, Gk osteon, Lat os (Gen ossis), Skrt asthi, Arm oskr, Slav *kosti) ~ IE *pste(H)n- ‘breast’ (Gk ste:nion, Avest fstana, Skrt stana, Toch A passam) < *pH3estH-, or *kwH3estH- ‘breastbone’. In Slav *kosti k- has always been a puzzle, so the finding of p-cognates is in line with this data point. Alternation p-/k- may point to an ancient labiovelar *kw-.

14. IE *we(s)kwer- 'evening' (Gk hesperos, Lat vesper, Lith vakaras, Slav veceru, OIr fescor), Germ. *westi 'west' ~ IE *pos-t- 'before, late, later' (Lat post 'behind, after, later, Skrt pacca 'same', Lith pastaras 'late', Slav *posdh- 'late') < *pH3oskwe-/*pH3weskwe-.

15. IE *H3dont- ‘tooth’ (Sanskrit dantam, Arm atamn, Lat dentis, Gk odous, OEng tōþ, Gothic tunþus, Lithuanian dantis ‘tooth’, Slav *desna ‘gum’, etc.) ~ IE *ped-/*pod- ‘foot, under, below’ (Hitt patas, Lyc pede-, Lu pati-, Toch A peṃ, B paiyye, Sanskrit pāda, Arm otn, Gk podi, Latinpēs, Alb poshtë, Slav *peši, Lith pėda) < *pH3ed-/*pH3don-t-.

Semantically, this etymology implies that IE *H3dont- originally meant ‘lower tooth’ (in opposition to *gombho- ‘upper tooth’), hence the connection with ‘foot’. Judging by its simple morphology, the concept ‘foot’ looks primary. From it, the concept ‘under, below, bottom’ (comp. Slav *podu ‘bottom, foundation’, Lith рãdаs ‘sole of foot’, Latv раds ‘floor’, Skrt раdám ‘step, footprint, place’, Gk πέδον ‘soil’, Lat оррidum ‘place’, pessum ‘bottom’) is derived. The forms for ‘tooth’ are most complex morphologically and they carry an additional body-part affix –t- (also seen in Skrt yakrt ‘liver’). This suggests that they are tertially derived from the ‘under, below, bottom’ forms.

16. IE *H3emso- 'shoulder' (Skrt amsa, Gk o:mos, Lat umerus) ~ IE *pemsti- (OHG fust, OEng fyst [< *fumsti), Lith kumste, Slav *pensti 'hand, fist') < *kwH2ems-/*pH2ems-.

Pre-PIE *tH2e- > PIE *te / a

17. IE *H2ekwеH2- ‘water’ (Lat aqua, Goth ahwo: ‘water’) ~ IE *tek(w)- ‘run, flow’ (Lith teku ‘run, flow’, Slav *teku ‘same’, Skrt takti ‘rushes’, Toch B cake ‘river’, etc.) < *tH2ekw-.

18. IE *H2ek’s- ‘axis’ (Gk akso:n, Lat axis, Skrt aksas, OHG ahsa, Slav *osi, Lith asis) ~ IE *tek’s- (Gk tekto:n ‘carpenter’, Skrt taksati ‘he fashions, he constructs’, Lat texo: ‘weave, build’, Slav *tesati, Lith tasyti ‘hew’, OHG dehsala ‘axe’) < *tH2ek’s-. Note Gk akso:n vs. tekto:n suggesting that Proto-Greek had a normal sequence *k’s- and -t- in tekto:n is a late, local innovation.

19. IE *Hukw- ‘cooking pot’ (Skrt ukha ‘cooking pot’, Lat aulla ‘pot’ (< *auksla), Goth auhns‘oven’, Arm akut’ ‘hearth’, OHG ofan ‘oven’, OEng ofen ‘furnace’, Gk ipnos ‘oven’, Myc i-po-no‘cooking bowl’) ~ IE *tep- (Skrt tapati ‘warms up’, Lat tepeo: ‘be warm’, Slav *teplu ‘warm’, OIrish ten (< *tepn-) ‘fire’) < *tHekw-. Vocalism here still needs work under any scenario.

Pre-PIE *kH2e- > PIE *ke/k’e / a

20. IE *H2rtko-, *H2rkto- ‘bear’ (Hitt hartagga, Gk arktos, Skrt rksa, Lat ursus) ~ IE *k’rst-‘fur, animal hair, bristle’ (Lith siurkstus ‘crude, hard’, serys ‘bristles’, Slav *sirsti ‘fur, animal hair’, OHG hursti ‘cristas’) < *k'H2r-kst-.

Pre-PIE *kH3e- > PIE *ke/k’e / o

21. IE *k’er- ‘head, horn’. The belonging of Hitt harsar to the same group as Lat cerebrum, Gk kara, Skrt siras, etc. has been suggested multiple times. The phonetics has been an issue. My reconstruction *k’Her- makes the phonetics regular. It’s possible that the variant with a colored vowel is in fact IE *H3ers- ‘buttocks’ (Hitt arras, Gk orros ‘tail, rump, base of the spine’ < *orsos, Arm or 'rear-end'). Semantically, the connection between head and horn parallels the connection between rump and tail. And we've already discussed the tendency of Indo-Europeans to equate polar body parts (thigh ~ underarm, toe ~ finger, butt cheek and face cheek, etc.) So, the resulting protoform should be reconstructed as *k'H3er-s- or even *kH3er-s-.

22. IE *H3ekw- ‘eye’ (Gk ops ‘eye, face’, ophthalmos ‘ eye’ [< *opth-], Lat oculus, Slav *oko – the etymon is widely attested but many forms are contaminated, so I'll omit them here) ~ IE *skep-/*spek'- 'look, see' (Gk skeptomai ‘observe, look carefully, consider’, skopos ‘target, purpose, aim, spy’, Lat specio ‘I look, I see’, Skrt pasyati ‘look’, spasa ‘spy’, Avest spasyeiti ‘look’, OHG speho:n ‘regard, spy’) < *sk'H3ekwo-. Hitt sakuwa 'eyes' (dissimilated from *skekuwa) supports this link. PIE *spH3ekw- is also possible. In this case Skrt pasyati with a reflex of a palatovelar can be explained as caused by a cluster velar + y. In Lat specio the loss of -w- is expected before -y- (comp. socius < *sokwyos). In Greek, too, the labial component of a labiovelar is lost before -w-.

in light of a potential Uralic cognate/IE borrowing, namely pre-Finnic *koke 'perceive' pre-PIE reconstruction with two labiovelars (*s-kwekw-t-) seems to be quite possible, which suggests that s-mobile (already seen in Hitt sakuwa) was added to the root after the action of Pre-Laryngeal Plosive Loss Law.


23. PIE forms for 'wolf' (*wlkwo-) and 'dog' (*k'wen-) can now be brought together under the root *k'H3we-l-/-n-. PIE *k’H3we- ‘wolf, dog’ yielded two early affixal derivatives – *k’H3wen- ‘dog’ (Gk kuo:n, Lat canis, Slav *senka, Arm sun, etc.) and *k’H3wel-kwo- ‘wolf’. Skrt svaka ‘wolf’, vrka ‘wolf’ and Avest spaka ‘dog’, verka ‘wolf’ establish the possibility of the connection between the IE sets for the dog and the wolf. Skrt svaka/Avest spaka are morphologically archaic as they don’t have either the -n-, or the -l- (> InAr *r) suffix. Arm skund ‘young dog’ and Gk skulaks ‘same’ show that indeed affixes -n- and -l- were in alternation. OIr cuilen and Welsh colwyn ‘young dog’ (forms without the difficult s-mobile) also confirm that the l-suffix alternated with the n-suffix. (They are not product of dissimilation from *kunen, but independent l- formations from more basic *k’we-.)

So, basically, l- in lykos and lupus corresponds to the affixal -n- is kuo:n and canis. Gk lykos (< *lhykos) is the same as -ulaks in Gk skulaks. On a methodological note, if we reconstruct PIE *k'H3we- 'wolf, dog' we can easily compare it with Uralic *kujna 'wolf' and Eskimo-Aleut qenRa 'wolf'. Even the IE DOG form is presently compared with this "Nostratic" set but the fact that the IE term for 'wolf' can be shown to be related to the IE term for 'dog' makes this long-range equation quite intriguing. Afroasiatic has a very similar root *k(w)alp- ‘wolf, dog’ (Bomhard & Kerns, no. 319) that shows -l- where Uralic and Eskimo-Aleut have -n-. Bomhard and Dolgopolsky still believe that Afroasiatic is part of Nostratic. In Uralic, Eskimo-Aleut and Afroasiatic languages the same root means both ‘dog’ and ‘wolf’. The Indo-Europeans forms with the meaning ‘wolf’ (*wlkwko-) and ‘dog’ (*k’wo:n) are both of Proto-Indo-European age, so it would be unusual if they were unrelated. Taking this material at face value, the connection between PIE *wlkwko- and PIE *k’wo:n may mean that Indo-European preserved the traces of ancient Nostratic/Eurasiatic morphological variation, whereas Afroasiatic generalized -l- and Uralic and Eskimo-Aleut generalized -n-.

24. IE *k'leus- 'listen, hear' (Toch A klyos 'listen, hear', klots 'ear', B klyaus 'listen, hear', klautso 'ear', Arm lsem 'listen, hear', Gk kluo 'hear', kleo 'make famous', Lat cluo: 'to be called, be famous', OEng hlystan 'listen', hle:or 'cheek', OIr cluo 'fame', cluas 'ear', Lith klausyti 'listen' ~ IE *H3eus- 'ear' (Gk ous, Skrt usi, Goth auso, Lith ausis, Latin auris 'ear', audire 'listen', Slav *uxo, Arm unkn 'ear' < PIE *k'lH3eu-s, or *klH3eu-s. Semantics cannot be better (Tocharian, Celtic and Latin directly attest for one single root for 'hear, listen, ear', and notably both *k’leus- and *H2eus- root-shapes can take both meanings).

Pre-PIE *gH3e- > PIE *ge/g’e/o.

25. IE *genu- ‘knee’ (Hitt genu, Toch kanwem, B keni, Lat genu:, Gk gonu, Skrt ja:nu, Arn cunr) ~ IE *H3noghu- ‘leg, foot, nail’ (Skrt anghri ‘leg’, nakha ‘nail’, Gk onuks ‘nail’, Lat unguis ‘nail’, Lith naga ‘hoof’, nagas ‘nail’, Slav *noga ‘leg, foot’) < PIE *gH3enu-, *gH3nogh-.

Notably, Arm elungn ‘nail’ (< *enung-) shows the same morphology and dissimilation as OIr glun, Welsh glin, Bret glin, Gael gluin ‘knee’ (< *gnu:no < *gonu).

26. IE *gHnens-dho- ‘nest’ (Slav *gne:zdo, Lith lizdas [< *glinzdo < *gninzdo], Skrt nīḍás, nīḍám ‘resting place, abode’, Arm nist ‘sitting place’, Lat nīdus, OHG nest). Currently g- in Slavic remains unexplained.

Sets 25-26 may be related if the idea of a ‘resting place’ (from which the concept of ‘nest’ later emerged) evolved from the notion of ‘kneeling’. Balt -l- would then become supported by the same development as found in Arm elungen and Celt *gluno.

27. IE *genu-, *gonHdho- ‘cheek, chin, jaw’ (Toch A sanwem, Skrt hanu, Gk genus, gnathos ‘jaw’, Lat gena, Goth kinnu) ~ IE *H3ens- ‘mustache’ (Slav *(w)onsu ‘mustache’, OPruss, wanso ‘first beard’, OIr fés ‘beard’, find ‘hair’, Gk ἴονθος ‘youthful beard’ (*vi-vondho-)) < *gH3en-dh-. Note the shared morphology between Gk gnatos and ἴονθος, OIr find.

28. IE *H3enbh- ‘hub, navel’ (Gk omphalos, Skrt nabhi ‘hub, navel, kin’, Avest naba-nazdista ‘next of kin’, Lat umbilicus, OHG nabalo, OPruss nabis) ~ IE *gen-/*gon- ‘beget’ (Skrt janati ‘give birth’, Gk genos, Lat gigno: ‘produce’, na:scor ‘to be born’, OEng cynn ‘race, family, kin’, cenna ‘produce’, OHG kind ‘child’, Arm cnanim ‘be born, bears’ < PIE *gH3en- ‘beget’, *gH3enbho- ‘kin, navel’. This analysis makes it possible to bring Hitt ha(n)s- (PAnatol *Ha(n)so- > hansatar, hassatar ‘family’) ‘give birth, beget’, Luw hamsa ‘grandchild’, into the same set with IE *gH3en-. Luw -m- may hide ancestral *-nbhs- > *mbs- (comp. Lat umbilicus, Gk omphalos) > *mms- > *-ms-. A “directive possessive” (Melchert, Oettinger) affix *-s(s)a- (see Puhvel’s Hittite Dictionary, “H”, p. 227) is an Anatolian innovation.

The semantically identical derivation of Hitt hassu- ‘king’ from *ha(n)s- and Germ. *kuningaz ‘king’ from *kun- ‘kin’ makes the connection all the more plausible. Hitt hassa-hanzassa ‘kith and kin’ or ‘child and grandchild’ (the former translation seems to be more compelling) is a full morphological parallel to Avest naba-nazdista.

29. IE *H3nomn ‘name’ (Hitt la:man [< *naman], Toch A nom, B nem, Skrt naman, Avest naman, Arm anun, Gk onoma, Lat no:men, OHG namo, Slav *enmen) ~ IE *gen- ‘know’ (Gk γνῶμα ‘sign’, Lat cogno:men ‘name’, Slav *znamen ‘sign, banner, badge’) < *g’H3nomen-. The shape of Hitt la:man (instead of expected hanaman**) goes back to *ghnomen.

Pre-PIE *bH1a- > PIE *bha/(y)a

30. IE *bheH2g’- ‘divide, distribute, allot’ (Gk phagein ‘eat, devour’, Skrt bhajati ‘divide, distribute’, bhaga ‘prosperity, happiness’, Avest baz- ‘bestow, divide’, Avest baga ‘share, happy lot’, OPers baga ‘god’, Slav *bogu ‘god’) ~ IE *yeH2g’- ‘worship’ (Skrt yajati ‘worship, sacrifice, make an offering’, Avest yaz- ‘same’, Gk hagios ‘holy’) < PIE *bH1eH2g'- or *bH1ag'-.

Pre-PIE *bH2e- > PIE *bhe/a.

31. IE *H2ener- 'man, husband' (Gk ane:r, Gen. anthropos, Arm haner 'husband's father', ayr [< *anir] 'man, husband') ~ IE *bhendh- 'relative, husband's father' (Gk pentheros 'husband's father', Skrt bandhu 'relative', Lith bendras 'companion' < *bH1en(dh)r)-/*bH1n(dh)r- < *bH1en-dh-. Note affixal consistency between Gen. anthropos and pentheros. Note precise semantic match between Arm haner and Gk pentheros.

32. IE *H2erH2mo-/*H2RH2mo- ‘arm’ (Skrt i:rma ‘arm’, Avest arema ‘forearm’, Arm armukn ‘elbow’, Lat armus ‘arm, forearm, shoulder blade’, Goth arms, OEng earm ‘arm’, OPruss irmo ‘arm’, Slav *ramo ‘shoulder’) ~ IE *bher-/*bherH2- ‘carry’ (Skrt bharati ‘carries’, bharitram ‘arm, shoulder’, bhari:man ‘carrying, keeping’, bharma ‘care’, Avest bereman ‘same’, Gk phero: ‘carry’, pherma ‘fetus’, Lat fero ‘carry’, offerumentum, Arm berem ‘carry’, Goth baira ‘carry’, Slav *birati ‘take’) < *bH2er-mo-, *bH2erH2-mo.

33. IE *bherg’- ‘white, shining, birch tree’ (Lith bersta ‘he turns white’, berzas ‘birch tree’, Skrt bu:rjas ‘Himalayan birch’, Goth bairhts ‘shining’, OHG birihha ‘birch tree’, Slav *bereza ‘birch tree’, Alb bardh, bardhe ‘white’, Lat farnus, fraxinus ‘ash tree’) ~ IE *H2erg’- ‘white, silver’ (Hitt harki ‘white, silver’, Toch A a:rki, B a:rkwi ‘white’, Gk arguros ‘silver’, arge:s ‘white, blinding’, Arm arcat’ ‘silver’, Skrt rajata ‘silver’, arjuna ‘white, light-colored, silver-colored’, Lat argentum ‘silver’, OIr argat ‘silver’) < *bH2erg'-.

A laryngeal is clearly responsible for the emergence of a voiced aspirate here. Notably Germanic, Baltic and Slavic that doesn't have a reflex of IE *H2erg'- is rich in reflexes of IE *bhergh'-/*bH2erg'-. Albanian bardh, bardhe 'white' is an exact morphological and semantic match for Hitt harki- 'white'. In Latin, argentum and fraxinus is an interesting doublet to analyze further.

Pre-PIE *dH1e- > PIE *dhe/e

34. IE *dhegho:m ‘earth’ (Hitt tēkan [Gen. tagnās], Toch A tkaṃ [Gen. tkanis], B keṃ ‘earth’, Gk khthṓn ‘earth’, khamaí ‘on the earth’, Skrt kṣā́ḥ (acc. kṣā́m, gen. jmáḥ) ‘earth’, Avest zā̊ (acc. ząm, gen. zǝmō) ‘earth’, Lat humus ‘earth’, Slav *zemĭ ‘earth’, *zmiji ‘snake’, Lith žemė ‘earth’, Alb dhe ‘earth’, dhemje ‘caterpillar’) ~ IE *H1egh(i)- ‘snake, chthonic creature’ (Gk ekhis 'viper', ekhinos 'hedgehog', Skrt ahi ‘snake’, Arm iz 'snake, viper', OHG egala 'leech', igil 'hedgehog', Slav *jez 'hedgehog', Lith ezys 'hedgehog') < *dH1egh-.

Pre-Laryngeal Plosive Loss Law is also relevant to the thorny problem of IE "earth" words. Hittite nominative and accusative singular te:kan points to *dʰ(e)ǵʰem. As Piotr Gasiorowski wrote about the PIE state, "the paradigm of ‘earth’ included nom.-acc. *dʰéǵʰōm, loc. *ǵʰdʰsém, and an oblique stem *ǵʰm- (in which the initial coronal was apparently dropped), e.g. in gen. *ǵʰmées. In some daughters the stem-shape of the locative, to which Schindler’s rules had applied, was generalized (cf. e.g. Gk χθών /khthǫ́ːn/, Skt acc. kṣā́am); in others the simple palatal of the oblique stem was apparently generalized (cf. e.g. Lat. humus); Anatolian and Tocharian generalized T(V)K- (cf. e.g. Hitt. dagān ‘on the ground’, Toch. A tkaṃ ‘earth’)." He also noted that we don’t know if the front consonant was an aspirate /dh/ or a voiced stop /d/. As I pointed out, metathesis and s-epenthesis proposed by Schindler and endorsed by Piotr are ad hoc solutions to this complex cognate set, and they should be avoided.

Pre-Laryngeal Plosive Loss Law explains the loss of a plosive as conditioned by the following laryngeal: nom-acc. *dH1egho:m, oblique *dH1ghe:m > Toch A tkam, B kem, Lat humus, Slav *zem-. Interestingly, the forms designating ‘snake, chthonic creature’ represent the original Nom-Acc. root shape *dH1egh-, which is otherwise lost from the subfamily of words meaning ‘earth’ outside of Hittite.

IE *H1egh- ‘snake, chthonic creature’ is usually taken to be a cognate of IE *H3egwh-/*H1ogwh- ‘snake’ (Toch B auk [< *aku-], Gk ophis). The latter suggests that PIE had an extended form *dH1eghu- and the cluster -ghu- later phonemicized into a labiovelar *gwh.

The extended form *dH1eghu- can also account for Gk ikhthus, Arm jukn, Lith zuvis ‘fish’.

The comparison between Gk khtho:n, khthamalos, khamai, ikhthus, on the one hand, and ekhis, ekhinos, ophis, on the other, suggests that -th- in khthamalos is likely unrelated to t- in Hitt te:kan but instead is a Greek-specific development similar to -t- in ptolis next to polis.

Pre-PIE *dH2e- > PIE *dhe/a

35. IE *dak’ru- ‘tear’ (Hitt ishahru-, Toch A ākär, pl. ākrunt, B akrūna (n. pl.), Skrt áśru- `tear', Avest asrū-azan- 'Tränen vergiessend', Gk dákru, Lat lacrima, OLat dacruma, Arm artasur (< *drak'ur-), Germ. *táxra-z, *tagrá-z, *tráxnu- (Goth tagr, OHG zahar, trahan), Lith ašarà).

This is a well-known cognate set but Pre-Laryngeal Plosive Loss Law suggests reconstruction *dH2ek'Hr-. Hittite ishahru attests for a laryngeal preceded by a later s-mobile (PAnat *hahru > Hitt *s-hahru (s-mobile) > *ishahru (vocalic prothesis). This is in line with current thinking. An alternative explanation would interpret the Hitt sequence sh- as a direct evidence for an unphonemicized cluster d + H2 (comp. Hitt siu ‘god’ with a palatalized *d- seen in Skrt dyavus). The absence of an aspirate in Greek suggests that in some ancient forms the vowel got colored, while d- didn't get aspiration. This may mean that the phenomenon of vowel coloring by a laryngeal emerged earlier than the phenomenon of the aspiration of a voiced stop by a laryngeal.

On the other hand, the form has a separate problem, namely the second -h- in Hitt ishahru corresponding to -k- in Gk dakru and other IE forms. Under Pre-Laryngeal Plosive Loss Law, the pre-PIE sequence VTH (vowel+stop+laryngeal) should manifest itself either as a long colored vowel or as VT. The vowel in Gk dakru is short, hence we have to reconstruct *dH2ek'Hr-.

Pre-PIE *dH3e- > PIE *dhe/(w)e

36. Hitt huwantes 'winds', Gk aFent- 'blowing', Lat ventus 'wind' ~ IE *dhwes- 'blow' (Skrt dhvaṁsati `to fall to pieces or dust, decay, be ruined, perish', ptc. dhvasta-; dhvasmán- m. `polluting, darkening', dhvasirá- `sprinkled, spattered', dhvasrá- `decaying, falling off', dhvasti- f. `ceasing, destruction', Lith dvē̃sti (dvẽsia, dvē̃sē) `den Geist aushauchen, (von Tiere) verenden, krepieren', Slav *duxnoti 'die', *dysati 'breathe', *duxu 'spirit' < *dH3we-s.

Pre-PIE *VCH > VC/ a:, o:, e:

37. IE *meH2ter ‘mother’ (Gk me:ter, Lat mater, etc.) < *makH2- (comp. OPruss moazo 'mother's sister', Lith masa 'husband's sister' < *mak'- < *makH2-)
38. IE *bhreH2ter 'brother' < *mregwH2ter (comp. Lith merga 'girl', Gk parthenos 'virgin, girl').
39. IE *seH2wl- 'sun' (Lat so:l, Gk he:lios, Slav *slunice, Lith saule, etc.) ~ IE *swet- 'light' (Hitt siwatt- 'day', Skrt cvetas 'white, shining', Lith sviesti 'shine', sviteti 'glimmer', OH hwiz 'white', Slav sweteti 'shine', *swetu 'light, day', *svetilu 'shiny, bright', Russ svetilo 'sun, moon') < *sewtH2l- < *swetH2l-.
40. IE *H1enH2ter 'husband's brother's wife' (Gk einater, Lat ianitrice:s, Skrt yatr, Lith jente, Slav *yentra, Arm niri) ~ *nepo:t 'grandson, nephew', *nepti- 'granddaughter, niece', Gk anepsios, anepsia 'cousin' < *H1enepH2ter.







57 comments:

  1. OK, I'll try to open the discussion by asking for the conditioning factor(s) for the split.

    ReplyDelete
  2. Hi Eli, once again thank you for starting this. If I send you a complete (for now) list of etymologies that I'm proposing to support Pre-Laryngeal Plosive Loss Law, will you be able to publish it here on my behalf?

    In the meantime, to answer David's question, the split is conditioned by the existence of forms in which a "T+H" sequence before V is a cluster vs. sequences in which T+H is a consonant phoneme or H+V is a (colored) vocalic phoneme. I hypothesize that in the pre-PIE there were no voiced aspirates but only voiced stops. There's some indication that palatovelars may also have gotten their palatal articulation from a contact with a laryngeal. Laryngeals were a group of phonemes, H1, H2, H3 distributed from front to back (in terms of their place of articulation). (This is the simplest interpretation of the identity of laryngeals and it seems to fit the data best.) I don't know if there was a separate line of voiced laryngeals that may have voiced neighboring stops. I understand that we need to know in what paradigmatic contexts the phonemicization of a laryngeal occurred. I don't have an answer for this but the minimal requirement of conditioning (namely, cluster vs. phoneme), I think, is met.

    Over at Languagehat I offered a case in which the whole cluster sk- dropped before a reconstructed laryngeal (Gk skeptomai vs. opsomai < *skH3ep-t). Now in light of a potential Uralic cognate/IE borrowing pre-Finnic *koke 'perceive' pre-PIE reconstruction with two labiovelars (*s-kwekw-t-) seems to be preferable, which suggests that s-mobile was added to the root after the proposed Pre-Laryngeal Plosive Loss Law.

    However, this new etymology suggests that certain clusters may have been dropped as a whole in front of a laryngeal: IE *k'leus- 'listen, hear' (Toch A klyos 'listen, hear', klots 'ear', B klyaus 'listen, hear', klautso 'ear', Arm lsem 'listen, hear', Gk kluo 'hear', kleo 'make famous', Lat cluo: 'to be called, be famous', OEng hlystan 'listen', hle:or 'cheek', OIr cluo 'fame', cluas 'ear', Lith klausyti 'listen' ~ IE *H3eus- 'ear' (Gk ous, Skrt usi, Goth auso, Lith ausis, Latin auris 'ear', audire 'listen', Slav *uxo, Arm unkn 'ear' < PIE *k'lH3eu-s, or *klH3eu-s. Semantics cannot be better (Tocharian and Latin directly attest for one single root for 'hear, listen, ear').

    Another general example of the postulated law is Hitt huwantes 'winds', Gk aFent- 'blowing', Lat ventus 'wind' ~ IE *dhwes- 'blow' (Skrt dhvaṁsati `to fall to pieces or dust, decay, be ruined, perish', ptc. dhvasta-; dhvasmán- m. `polluting, darkening', dhvasirá- `sprinkled, spattered', dhvasrá- `decaying, falling off', dhvasti- f. `ceasing, destruction', Lith dvē̃sti (dvẽsia, dvē̃sē) `den Geist aushauchen, (von Tiere) verenden, krepieren', Slav *duxnoti 'die', *dysati 'breathe', *duxu 'spirit' < *dH3we-s.

    Among the well-known etymologies, Hitt ishahru 'tear', Gk dakruma, Toch akruna, Lith asru are likely candidates for the operation of the same law (*dH2ak'H2r- > PAnat *hahru > Hitt *s-hahru (s-mobile) > *ishahru (vocalic prothesis) but the absence of aspiration on d- in Gk dakruma is puzzling. One way to solve for it is to reconstruct PIE *dH1ak'Hr-/-n-, with an original -a- and a non-coloring, non-aspirating laryngeal H1.

    ReplyDelete
  3. Sorry, I had trouble posting with my LiveJournal account, so an older version of the last paragraph got published. Now the problem is solved. Here's an update:

    "Among the well-known etymologies, Hitt ishahru 'tear', Gk dakru, Toch akruna, Skrt assure, Lith asara are likely candidates for the operation of the same law (*dH2ak'H2r- > PAnat *hahru > Hitt *s-hahru (s-mobile) > *ishahru (vocalic prothesis) but the absence of aspiration on d- in Gk dakruma is puzzling. One way to solve for it is to reconstruct PIE *dH2ek'Hr-/-n- and to postulate that in some ancient forms the vowel got colored, while d- didn't get aspiration. This may mean that vowel coloring by a laryngeal preceded (emerged earlier than) the aspiration of a voiced stop by a laryngeal. On the other hand, the form has another problem, namely the second -h- in Hitt ishahru corresponding to -k- in Gk dakru and other IE forms. Under Pre-Laryngeal Plosive Loss Law VTH should give V: (long vowel), and the vowel in dakru is short. So this cognate set still needs more work.

    ReplyDelete
  4. Pre-Laryngeal Plosive Loss Law is also relevant to the problem of IE "earth" words. As we discussed at LanguageHat, Hittite nominative and accusative singular te:kan points to *dʰ(e)ǵʰem. As Piotr wrote, "the paradigm of ‘earth’ included nom.-acc. *dʰéǵʰōm, loc. *ǵʰdʰsém, and an oblique stem *ǵʰm- (in which the initial coronal was apparently dropped), e.g. in gen. *ǵʰmées. In some daughters the stem-shape of the locative, to which Schindler’s rules had applied, was generalized (cf. e.g. Gk χθών /khthǫ́ːn/, Skt acc. kṣā́am); in others the simple palatal of the oblique stem was apparently generalized (cf. e.g. Lat. humus); Anatolian and Tocharian generalized T(V)K- (cf. e.g. Hitt. dagān ‘on the ground’, Toch. A tkaṃ ‘earth’)." As I pointed out, metathesis and s-epenthesis are ad hoc solutions to this complex cognate set, and they should be avoided.

    Pre-Laryngeal Plosive Loss Law easily explains the loss of a plosive as conditioned by the following laryngeal: nom-acc. *dHegho:m, oblique *dHghe:m > Toch A tkam, B kem, Lat humus, Slav *zem-. It also allows one to consider in this context such forms as Gk ekhis 'viper', ekhinos 'hedgehog' (comp. Slav *zmiji 'snake', Alb dhemje 'caterpillar'), Arm iz 'snake, viper', OHG egala 'leech', igil 'hedgehog', Slav *jez 'hedgehog', Lith ezys 'hedgehog' as reflexes of the original Nom-Acc. root shape *dH1egh-.

    The comparison between Gk khtho:n, khthamalos, khamai, on the one hand, and ekhis, ekhinos, on the other, suggests that -th- in khthamalos is likely unrelated to t- in Hitt te:kan but instead is a Greek-specific development similar to -t- in ptolis next to polis.

    ReplyDelete
  5. @Vladimir: Yes, I can absolutely publish your list. That would be better, in fact; I just put my notes up as a placeholder.

    ReplyDelete
  6. This comment has been removed by the author.

    ReplyDelete
  7. Another interesting comparandum, now illustrating

    Pre-PIE *gwH3ye- > PIE *gwye-/PIE *ye-/yo.

    Pre-PIE *gwH3eiwo- > *gwyeH3wo-/*gwiH3wo- 'live, life' (Toch A śo-, B śāw- [< PT *śāw-] 'live'; B śātre 'grain', śaiyye 'sheep, goat', Skrt jī́vati `to live'; jínvati `to be active or lively; to urge on, impel'; gáya- m. `house, household, property'; jīrá- `quick, speedy, active'; jīvá- `living, alive', Gk bióō 'live', bíos 'life', ζωός 'alive', Lat vīvō 'live', Goth quius 'alive', Lith gīvénti, gīvuóti 'live, reside', Slav *ziti 'live', *zivo 'I live', OIr bith 'world', biu 'living', bethu `life') ~ PIE *yownko- 'young' (Skrt yúvan- 'young, youngster', yávīyān 'younger', yáviṣṭhas 'youngest', Lat iuvenis, iūnior, iuvencus 'calf', Goth juggs 'young', Lith jáunas 'young', Slav *junu 'young'). In the latter subset -no-, and -ko- are affixes. Full-grade vocalism is attested in Baltic (Lith jaunas) and in the Old Indic comparatives (yávīyān 'younger', yáviṣṭhas). Importantly, Greek does not have a direct reflex of PIE *yownko 'young' but Gk. ζωός 'alive', in which /dz/ is a regular reflex of PIE *y (and not of PIE *gw-), is always considered as part of the 'life' cognate set (as a dental doublet of bios). Gk ζωός is interpreted as derived from *gwyeH3wo-, which is fully compatible with *yownko-, under Pre-Laryngeal Plosive Loss Law.

    ReplyDelete
  8. 5. PIE *pH1ater. H1 does nor color ancestral /e/. And, now I can argue, it does not color ancestral /a/. So that's why we find /a/ in all the 'father' forms in IE languages (including the most divergent branches such as Pal papaz ‘father’, Toch A pa:car, B pa:cer ‘father’, Gk pappos ‘grandfather’, pater ‘father’) but Indo-Iranian. In Indo-Iranian we logically find /i/ in Skrt pita: ‘father’. Because H1 had a palatalizing effect, in Skrt sequence H1a gave /i/ and didn't turn p- into ph-.
    6. PIE *pH1awo-/*pH1awH2o- 'grandfather, grandchild' (Hitt huhhas ‘grandfather’, Skrt putra ‘son’ [< *pH1autro-), Lat avus ‘grandfather’, avunculus ‘mother’s brother’, puclos ‘son, child’, ONorse afi ‘grandfather’, OIr aue ‘grandson’, amnair [< *aunater) ‘mother’s brother’, Welsh wyr ‘grandson’ (< *pwyr)

    7. PIE *pH1awyo-: Gk pais ‘child, son’, Slav *uji, Lith avynas, OPrus awis ‘‘mother’s brother’.

    This analysis helps clarify another IE form without an etymology, namely Lat vi:tricus 'stepfather'. It derives from a zero-grade (comp. Welsh wyr above) form *pH1wiyo- + *ter + *ko-. The long -i:- is a perfect fit with the *-wiyos affix seen in Slav *uji, OPruss awis 'mother's brother'. The affixes -tr- + -ko- recur in Sorbian tryk 'father's brother'. Semantically, stepfathers are often equated with fathers and father's brothers as seen from Gk patruios 'stepfather'.

    As Hettrich noted (I quote from Dziebel), IE forms for mother's brother tend to exhibit the same root but different affixes in different branches. The affix -wiyo- seen in Balto-Slavic is not found elsewhere, and suffix -clo-/-tlo- seen in Italic and Celtic (Lat avunculus, Welsh eontr) is not found elsewhere. OHG o:heim, OEng team 'mother's brother' also have unique morphology. Hettrich interpreted this observation as an indication that the formation of mother's brother's terms in IE languages took place after the separation into individual branches on the basis of an underlying root meaning 'grandfather' (Lat avus). But now it's clear that Slav *uji is a morphological copycat of Lat vi:- in vi:tricus, which means that the root meaning 'grandfather' underlies not just forms for maternal kin but also forms for paternal kin as well.

    ReplyDelete
  9. The split was apparently conditioned by the existence of forms in which a "T+H" sequence before V was a cluster (the ancestral state) vs. sequences in which T+H became a consonant phoneme or H+V became a (colored) vocalic phoneme.

    What I'm waiting for is an explanation of what conditioned this latter split: when did T+H become a consonant phoneme, and when didn't it?

    ReplyDelete
  10. Here are a few points that may answer your question. 1). PIE had a mobile accent that shifted within a paradigm; 2) cluster T+H is the ancestral condition, while TH as a single phoneme is a derived one; 3) judging by the 'earth' example (which everybody agrees had a plosive loss but only my Pre-Laryngeal Plosive Loss Law explains why the plosive got lost), different root shapes correlate with different accent placements (*dH1e'ghom vs. *(dH1)ge'm; 4) one of the accent placements (either on the syllable immediately following the cluster, or on any other syllable down the string) favored the retention of the the ancestral cluster condition, which resulted in plosive loss and vowel coloring; 4) leveling took place across paradigms that may have obfuscated the original accent-based conditioning.

    ReplyDelete
  11. While we are pondering what pre-PIE paradigmatic conditioning caused the split described by the Pre-Laryngeal Plosive Loss Law, here's another potential etymology that involves two "famous" PIE laryngeal roots: PIE *steH2- 'stand up', present *stisteH2ti- (Gk histe:mi 'I stand', statos 'standing', Lat sisto:, Skrt tisthati 'stand', sthitas 'standing', Toch B ste [< *stH2o-] 'is', OIr tau [< *sta:io:-] 'I am') ~ PIE *H1esti- 'be, is' (Hitt ēszi, Skrt ásti, Gk ἐστί 'is', eimi 'am', Slav *estu 'is', *jesmi 'am', Goth ist) < pre-PIE *stH1estH1a(ti)-. There's some sort of dissimilation going on in this reduplicative present that involves dentals (s, t) and laryngeals (H1...H1). It's not 100% clear to me why Skrt has -th- in tisthati and shtitas. If th is from t + H2, then why is it followed by -i- (Gk -a-), which come from syllabic H (in the conventional reconstruction) or H1a in my reconstruction (see above). The situation is apparently the same as in Skrt duhita: ~ Gk thugate:r, and I explain -h- as a throwback from *dh, while the effect of H1 is palatalization in Proto-Indo-Iranian *-jh- (but not the aspiration). By analogy, Skrt tisthati could be from *sthesteti, with a throwback.

    ReplyDelete
    Replies
    1. It's not 100% clear to me why Skrt has -th- in tisthati and shtitas. If th is from t + H2, then why is it followed by -i- (Gk -a-), which come from syllabic H (in the conventional reconstruction) or H1a in my reconstruction (see above).

      Oh, that's easy. There was no such thing as a syllabic laryngeal, just like there was no such thing as a syllabic *s. Instead, the i is an epenthetic vowel inserted into the syllable-initial consonant cluster *sth₂t-.

      Delete
    2. I remember you argued that in Skrt pita the lack of aspiration on p (not phita*) can be explained as the result of i-epenthesis that presumably occurred prior to the development of voiceless aspirates out of clusters laryngeal+voiceless stop in Indic. But in Skrt sthita both -i- and -th- are in attendance. Why?

      Delete
    3. This comment has been removed by the author.

      Delete
    4. Shouldn't we expect both? *sth₂°t- > *stʰ°t- > stʰit-.

      We don't find that in pita- because the laryngeal wasn't there in the nominative and was on the other side of the epenthetic vowel in the oblique cases: the nominative was *ptēr > *p°tār > pitā́, the genitive was *p°h₂trós > *p°t°́rs > pitúr.

      Delete
  12. Another good one to add to the following:

    "Pre-PIE *VCH > VC/ a:, o:, e:

    37. IE *meH2ter ‘mother’ (Gk me:ter, Lat mater, etc.) < *makH2- (comp. OPruss moazo 'mother's sister', Lith masa 'husband's sister' < *mak'- < *makH2-)
    38. IE *bhreH2ter 'brother' < *mregwH2ter (comp. Lith merga 'girl', Gk parthenos 'virgin, girl').
    39. IE *seH2wl- 'sun' (Lat so:l, Gk he:lios, Slav *slunice, Lith saule, etc.) ~ IE *swet- 'light' (Hitt siwatt- 'day', Skrt cvetas 'white, shining', Lith sviesti 'shine', sviteti 'glimmer', OH hwiz 'white', Slav sweteti 'shine', *swetu 'light, day', *svetilu 'shiny, bright', Russ svetilo 'sun, moon') < *sewtH2l- < *swetH2l-.
    40. IE *H1enH2ter 'husband's brother's wife' (Gk einater, Lat ianitrice:s, Skrt yatr, Lith jente, Slav *yentra, Arm niri) ~ *nepo:t 'grandson, nephew', *nepti- 'granddaughter, niece', Gk anepsios, anepsia 'cousin' < *H1enepH2ter."

    PIE *(s)teH4- (Hitt ta:yezzi 'steals', Toch B ene-stai 'in secret', Skrt sta:yu: 'thief', Gk te:taomai 'deprive, rob', Slav *tajo 'hides', *taj 'secret', *tati 'thief', OIr ta:id 'thief') ~ PIE *(s)teg- (Skrt sthagati 'cover', sthaga 'cunning, fraudulent, dishonest', Marathi thak 'thief', Gk stego: 'cover', tegos 'roof', Lat tego: 'cover') < *(s)tegH2-. This reconstruction explains the origin of Skrt -th- in sthagati as shifted forward from *stegH2- or even dissimilated from *stH2egH2-. Semantically, the overall is perfect. There's no need to reconstruct *H4.

    ReplyDelete
    Replies
    1. 39 requires so many metatheses and random insertions or losses of *t that I don't even want to count them: there's just no point. :-| Honestly, I'm not sure you even notice how many metatheses you propose all over the place.

      Also, congratulations on discovering a whole new sound law, namely PIE *sw- > Proto-Germanic *hw-! Now you only need to explain why there are words that begin with *sw- in Proto-Germanic.

      We've discussed 40 already. I note you haven't responded to my criticisms.

      There's no need to reconstruct *H4.

      Uh, of course not. There are good reasons why it's not mainstream.

      Delete
  13. Replies
    1. I edited it a few times when it was my notes, and most recently to replace my notes with Vladimir's guest post that he sent to me (that was sometime after March 27). I haven't edited it since then.

      Delete
  14. @David

    "Also, congratulations on discovering a whole new sound law, namely PIE *sw- > Proto-Germanic *hw-! Now you only need to explain why there are words that begin with *sw- in Proto-Germanic."

    This is a good catch. I will take it off for now. I want to keep the list as single-mindedly focused on my Pre-Laryngeal Plosive Loss Law for now (without drawing too much on Dziebel's pre-PIE labiovelar split and IE *s ~ stem *s identity proposals).

    "We've discussed 40 already. I note you haven't responded to my criticisms."

    I don't think we did. Could you repeat your objections?

    "Is this post becoming longer?"

    I'm adding new things into the comments section, nit into the post body.

    ReplyDelete
  15. IE *wiHro- ‘male, man’ (Toch wir, Skrt vi:ra, Avest vi:ra, Lat vir, OIr fear, Goth wair, Lith vyras) ~ IE *vepr- ‘boar’ (Slav *vepri, Latv vepris ‘castrated boar’) < *wepH1r-. For semantics, see PIE *wers- > Skrt vrsan ‘manly’, Gk arse:n ‘male’, Lat verre:s ‘boar’.

    ReplyDelete
  16. "in light of a potential Uralic cognate/IE borrowing, namely pre-Finnic *koke 'perceive' pre-PIE reconstruction with two labiovelars (*s-kwekw-t-) seems to be quite possible,"

    Perhaps possible, but not at all necessary to assume: the loan etymology that has been suggested presumes that *h₃ was word-initially substituted as *k in Uralic.

    Also, the semantics are somewhat weak: the abstract 'to perceive' is only a late meaning in literary Finnish, and most reflexes point to an original meaning 'to examine, to check (e.g. traps)'. It's quite possible that the words have nothing to do with each other at all.

    ReplyDelete
  17. Looking a bit closer, one main problem with the hypothesis that I did not see addressed seems to be that some clusters of stop + laryngeal are usually reconstructed for PIE, e.g. *kh₂eyd- 'to hit', *preuth₂- 'to splash'.

    (Though all word-initial cases that I can dig up seem to involve a voiceless stop + *h₂. Perhaps it was only other cases where an earlier stop was lost.)

    ReplyDelete
  18. @protouralic

    "Looking a bit closer, one main problem with the hypothesis that I did not see addressed seems to be that some clusters of stop + laryngeal are usually reconstructed for PIE, e.g. *kh₂eyd- 'to hit', *preuth₂- 'to splash'."

    Yes. I've noticed this potential issue. For now, I settled on reconstructing IE *kh2eyd- as *keH2yd- with the secondary shift of h2 to the front in some branches (after the end of action of the Pre-Laryngeal Plosive Loss Law) reflected in Skrt khidati 'tear, press'. The law is supposed to affect all stops as illustrated by H1ek'wo- vs. *pek'wo- above (< *pH1ek'wo-), *pH3ed-, etc.

    "Also, the semantics are somewhat weak: the abstract 'to perceive' is only a late meaning in literary Finnish, and most reflexes point to an original meaning 'to examine, to check (e.g. traps)'. It's quite possible that the words have nothing to do with each other at all."

    Any connection with Uralic is highly tentative and may fall apart at a closer examination.
    Thanks for calling out a problem here.

    ReplyDelete
  19. Also, as I'm reading around a bit, not everybody connects Lat caedo with Skrt khidati. LIV (Lexikon der indogermanischen Verben), for instance, keeps the two forms separate.

    ReplyDelete
  20. @protouralic

    I wonder if clusters TH1, TH2, TH3 (where T is any voiceless stop) are indeed reconstructible for PIE, then why voiceless aspirates are not? I would assume that, once a laryngeal came in contact with a voiceless stop, it would phonemicize as a voiceless aspirate. At least this is how people explain the origin of voiceless aspirates in Indic. In Indic, the condition for the emergence of voiceless aspirates is the shift of a laryngeal into a position immediately following a voiceless stop. I wonder if the examples of TH2 that you have in mind resulted from a similar shift in post-PIE times.

    ReplyDelete
  21. Looking at Byrd's list of roots in which he reconstructs a cluster TH1, TH2 or TH3 for PIE (Byrd, The Indo-European Syllable, 134), two of them deserve a comment. PIE *petH2- 'fly' is reconstructed above as *pH1et-, with plosive loss attested in such forms as Lat iterate and Hitt itar 'road'. Another interesting case is furnished by Byrd's root *pletH2- 'spread out'. Among its reflexes is such a body part term as Skrt prtha 'flat hand' which attests for an earlier cluster tH2. But Pre-Laryngeal Plosive Loss Law allows one to connect *pletH2- with another PIE root *H3el- 'elbow' of which Skrt has a reflex aratnis 'elbow' without a voiceless aspirate. The *H3el- root is complex morphologically and semantically (comp. Toch A a:lem 'palms of the hands', Arm olin 'spine, shoulder', Gk o:lene 'forearm') but notably Balto-Slavic has identical morphological formations such as *plekt- 'shoulder' and *o:lekt- 'elbow' (Lith uolektis, Slav *olkuti), which further supports the cognation between *plektH2- and *H3el-. The resulting pre-PIE root would be reconstructed as *pH3l-/*peH3l- with extensions *pH3lek-, *pH3let-, *pH3len-. The direct comparison between Skrt prtha and aratnis shows that one of the extended roots must have contained two laryngeal clusters - one resulted in a loss of a preceding plosive (and then the loss of the laryngeal with vowel coloring), the other one got phonemicized (in Indic) into a voiceless aspirate.

    ReplyDelete
  22. Re: IE *kh2eyd- as likely *keH2yd- with the secondary shift of h2 to the front in as reflected in Skrt khidati 'tear, press'.

    There's another intriguing case that I would adduce here. PIE *deH2ywer 'husband's brother' (Skrt devar, Gk dae:r, Lat le:vir, OHG zeihhur, Lith dieveris, Latv dieveris, Slav *de:veri etc.) can be compared to Slav *de:va 'girl, daughter, virgin', *de:ti 'children', Latv dels 'son' Lat fi:lia-/-us 'son, daughter', Gk the:lus 'female',Skrt dhe:nus 'cow' (all from PIE *dheH(y)- 'suckle, nurse'). The root should be reconstructed as *deH2(y)- 'suckle, nurse' with a secondary shift of the laryngeal back into a position immediately following a voiced stop > dH2 > dh- in Latin, Skrt and Greek, and forward into a position immediately preceding suffixal -w- in OHG zeihhur (with Verschaerfung). if the original form were dH2eywer**, we would have expected Gk aer** 'husband's brother' (by Pre-Laryngeal Plosive Loss Law).

    ReplyDelete
  23. Another etymology illustrating Pre-Laryngeal Plosive Loss Law: PIE *per-tH2- 'the young of an animal" (Skrt prthuka- 'young animal', Arm orth 'calf', Gk portis, portaks, ports 'young cow, heifer, calf', OEng fearr 'bull' ~ PIE *H3er- 'rise, grow' (Slav *orsti 'grow', *orstoku 'sprout, shoot', Gk ornu:mi 'raise, stir up, set out', ormenos 'shoot, sprout', Arm yarmen 'rise, awake', armn 'root' (< *ormen)) < pre-PIE *pH3er-tH2-, *pH3er-n, *pH3er-men. Slav *orsti is morphologically divergent next to Gk ornu:mi and Arm yarmen, but is identical to Skrt prthuka- and Arm orth. The p-forms above, if taken in isolation, lack a verbal or adjectival source, which the *H3er- set readily provides them with.

    ReplyDelete
  24. For the treatment of -CH- medially we currently have:

    1. IE *meH2ter ‘mother’ (Gk me:ter, Lat mater, etc.) < *makH2- (comp. OPruss moazo 'mother's sister', Lith masa 'husband's sister' < *mak'- < *makH2-)
    2. IE *bhreH2ter 'brother' < *mregwH2ter (comp. Lith merga 'girl', Gk parthenos 'virgin, girl').
    IE *H1enH2ter 'husband's brother's wife' (Gk einater, Lat ianitrice:s, Skrt yatr, Lith jente, Slav *yentra, Arm niri) ~ *nepo:t 'grandson, nephew', *nepti- 'granddaughter, niece', Gk anepsios, anepsia 'cousin' < *H1enepH2ter."
    3. PIE *(s)teH4- (Hitt ta:yezzi 'steals', Toch B ene-stai 'in secret', Skrt sta:yu: 'thief', Gk te:taomai 'deprive, rob', Slav *tajo 'hides', *taj 'secret', *tati 'thief', OIr ta:id 'thief') ~ PIE *(s)teg- (Skrt sthagati 'cover', sthaga 'cunning, fraudulent, dishonest', Marathi thak 'thief', Gk stego: 'cover', tegos 'roof', Lat tego: 'cover') < *(s)tegH2-. This reconstruction explains the origin of Skrt -th- in sthagati as shifted forward from *stegH2- or even dissimilated from *stH2egH2-.
    IE 4. *wiHro- ‘male, man’ (Toch wir, Skrt vi:ra, Avest vi:ra, Lat vir, OIr fear, Goth wair, Lith vyras) ~ IE *vepr- ‘boar’ (Slav *vepri, Latv vepris ‘castrated boar’) < *wepH1r-. For semantics, see PIE *wers- > Skrt vrsan ‘manly’, Gk arse:n ‘male’, Lat verre:s ‘boar’.

    Another case can be adduced:

    5. IE *med- 'measure' (Gk medomai 'be mindful of', Arm mit 'thought, reason', Lat editor 'mediate', OIr midithir 'judges', Goth mitan 'measure', OHG muezzin 'measure, compare, evaluate') ~ IE me:tis/*meH1tis 'measure' (Skrt ma:ti 'measure', Gk me:tis 'plan', Lat me:tor 'measure', Alb mat 'measure', Goth me:l 'time', OEng mae:l 'measure, mark, appointed time) < PIE *medH1-. H1 merges with -d- when phonemicizaton occurs but "knocks out" -d- and lengthens the vowel in the case of cluster treatment. The quality of the laryngeal is unmistakably H1 (and not H2, for example) because in the case of *med- -d- is not aspirated and in the case of *me:tis the lengthened vowel stays uncolored.

    ReplyDelete
  25. Next to pre-PIE *pH3le-k-t- (PIE *plek-t- 'shoulder' and *o:le-k-t- 'elbow') one can cite another pair of roots with a very similar development. PIE *pleH1- 'fill', *pleH1-no-, *pleH1-to-, *pleH1-ro-, etc. (Lat ple:nus 'full', ple:rus 'most of', ple:b:es 'common people', Skrt prnati 'fill', pu:rna- 'full', Gk ple:to- 'become full', ple:re:s 'full', ple:thus 'crowd', Arm lnu 'fill', lir ‘fullness,’ OIr lin 'number' (< *ple:n-), loth 'stallion', Lith pilnas 'full', Slav *plunu 'full', *plodu 'fruit', *pledmen 'tribe') ~ PIE *H1leu-dh- 'grow, free, people' (Lat li:ber 'free', li:be:ri 'children', Skrt rodhati 'grows', Gk eleutheros 'free', Goth liudan 'grow', OHG liut 'person', liuti 'people', Lith liaudis 'common people', Slav *liudu, *ljudije 'people') < pre-PIE *pH1el-, *pH1le- 'grow'.

    Something intriguing is going on in the second segment of forms such as Gk eleutheros, Lat ple:be:s and li:be:ri. -u- in eleutheros may have metathesized from an u-stem form *h1ledhu- (comp. Gk ple:th-u-s) or it's a separate suffix and the form should be reconstructed as *pH1lH1e-wo- next to well-attested *pH1leH1ro- and *pH1leno- root-shapes. Pre-Laryngeal Plosive Loss Law predicts that the group *-eH1- reconstructed from forms such as Lat ple:nus and Gk ple:re:s presupposes earlier *-eCH1-. The only suffixal consonant that matches this expectation is the dental in Slav *pledmen (with a short -e-), which, however, corresponds to -b- in ple:be:s and -th- in Gk eleutheros and ple:thus and -dh- in Skrt rodhati. The Gk and Skrt forms point to *dh, or *dH2 and not *d/*dH1. Or the original form was *pH1ledH1u- and the laryngeals dissimilated from H1…H1 into H1…H2 in forms such as eleutheros, ple:thus and rodhati. If this was the case, then forms such as ple:nus, ple:re:s should be reconstructed as *pH1ledh1no-, *pH1ledH1ro-.

    A point should have been earlier that the Armenian and Celtic p-less forms don't represent parallel, secondary developments in those two branches, but rather the generalization of a pre-PIE alternation.

    ReplyDelete
    Replies
    1. Interestingly, Lat li:ber 'free', li:beri 'children' shows an unusual development of PIE diphthong -eu-. The normal development seen in Lat lu:x 'light' (Gk leukos 'light, white', Skrt roca- 'lightning') is contrasted with an irregular development -*eu- > oi followed by regular -ei- > -i:-. Meyer-Brugge, Indo-European Linguistics, 94, attributes eu > -i:- to an environment between -l- and a labial, but comparison with ple:be:s shows that li:beri may have, instead, evolved from *leH1dh-, with H1 yielding a glide that later merged with -e- into an expected Latin -i:-.

      Delete
  26. Pre-PIE *pH2el-, *pH2le-, *pH2leu-, *pH2lei- 'white, shining' > PIE *pH2elu- 'grey, white' (Skrt palitás 'old, grey-haired', Gk πολιός 'grey', Lat раllеō 'I'm pale', pallidus 'pale', OIr líath 'grey' (*pleitos), Arm alik' 'white', Slav *роlvu 'fallow, pale', Lith раl̃vаs 'fallow', OHG falo 'pale, light-haired', OSax falu 'fallow' (*falwaz)) ~ PIE *leuko- 'shine, bright' (Toch A, B luk 'light', Lat lu:x, lu:men 'light', lu:na 'moon' (*luksna), Arm loys 'light', luc'anem 'to light', Celt *loukos 'white' (Welsh llug 'shimmer'), Gk leukos 'bright, white, pale, weak', OHG lioht 'light', Lith laukas 'field, open space', Slav *luci 'beam of light').

    The -ko- affixation of the 'light' forms is likely the same as the -ko- affix found in one of the IE terms for 'mouse', *pH2eluko- (or *pelHxus in Mallory/Adams, 334), derived from the first root *pH2el- (Irish luch, Welsh llyg 'shrew', Old Breton loc, Lith pele, Wakhi purk). Again, Celtic forms show generalized p-less outcomes (Welsh llug and llyg) of PIE *pH2elu-k- regardless of which semantic subset they belong to.

    The identity of the front laryngeal and vowel is debatable. I put -H2e- because of Lat palleo but -H2o- (Gk polios) is possible, too. -H1a- is another possibility, although, in this case, we would expect pilitas** in Skrt.

    ReplyDelete
  27. A rare case of the application of Pre-Laryngeal Plosive Loss Law to sonorants. (comp. also *k'lH2eus- 'hear, listen, ear' above):

    pre-PIE *mH1er- (*mH1er-gh-, *mH1er-kh-, *mH1er-gw-, *mH1er-gwh-) 'flicker, dim' < PIE *mergh, *merkh- (Skrt markas 'eclipse', Lith murgai 'visions, reveries', mirgė́ti 'flicker', márgas 'spotted', merkti 'blink', Latv mir̂gu, mir̂dzêt 'flicker', Goth maurgins 'morning', ONorse myrkr 'dark', mjorkvi 'dense fog', Slav *mriknoti 'grow dim', *sumerki 'twilight') ~ PIE *H1re-gw-, *H1re-gwh- 'darkness' (Toch A arkant, B erkent 'black', Skrt rajas 'vapor, mist, clouds, gloom, darkness', rajani: 'night', Gk orphnos 'dark', orphne: 'darkness', erebos 'underworld', Goth riqwis 'darkness', ONorse rokkr 'dark', Arm erek 'evening', erek 'yesterday').

    ReplyDelete
  28. Now, a potential case of a dual loss of a plosive in an environment immediately preceding a laryngeal in the beginning and middle segments of a word.

    Pre-PIE *pH1elegwH2-, *pH1legwH2- 'float, be light': PIE *pleu- 'float, swim, wash' (Toch AB plus 'soar', Toch B plus 'float', Skrt plavate 'swim', Gl ple(F)o 'swim', Arm luanam 'wash', Lat pluit 'it's raining', OIr lu:id 'moves', OHG flouwen 'wash', Lith plauti 'wash, rinse', Slav *plаviti 'swim') ~ PIE *H1legwh- 'light' (Toch B lancutse 'light', Skrt raghu-, laghu- 'fleet, fast', Gk elaphros 'light, fast', elakhus 'small', Lat levis 'light', OIr laigiu 'lighter, poorer', Goth leihts 'light', OHG lungar 'rapid', Lith lengvas 'light, easy, slight', Slav *li:guku 'light, easy', Alb lehte 'light, soft, slight, nimble'.

    The semantic connection is perfect. Importantly, IE terms for 'lung' (the lightest internal organ that floats) trace back to both *pleu- and *H1legwh- derived forms: Gk pleumo:n, Lat pulmo: 'lung', Skrt kloman 'right lung' (< *ploman) as well as OHG lunga 'lung', Arm lanjk' 'breast', ORuss legkoe 'lung'.

    It's possible, as EIEC, 353, argues, that two different but similarly sounding roots (*H1lengh- 'fast' and *H1legwh- 'light') are lumped up above. If this is the case, then PIE *pleu- should be compared with *H1legwh- only. However, the fact the IE terms for 'lung' above have the -n- suggests that the two roots are the same and that a secondary nasal infixation occurred in a few descendant forms.

    It's also possible that -gwh- in *H1legwh- and -w- in *pleu- are two different ancient affixes added to the more basic root *pH1el-/*pH1le-. But the example of Lat levis (next to Lat pluit) suggests that -gwh- directly corresponds to -w- through the action of Pre-Laryngeal Plosive Loss Law whereby -*gwH2- or *-gwH3- yielded -w- in some cases (when the protoform was a cluster) and *-gwh- in others (when the protoform for phonemicize into an aspirated labiovelar). This interpretation may cast a light on the dual reflex of *gwh in Latin: sometimes as -v- (nivis Gen. 'snow') and sometimes as -gu- (nix Nom. 'snow', ninguit 'it snows).

    ReplyDelete
  29. Pre-Laryngeal Plosive Loss Law offers a glimpse into the evolution of the PIE numeral system.

    First, it suggests an etymological connection between PIE forms for ‘five’ and ‘nine’: PIE *penkwe- ‘five’ (Toch A päñ, B piś, Skrt pañca, Gk πέντε, Lat quīnque, Arm hing, Goth fimf, Lith penkì, Slav *pętĭ, Alb pesë) ~ PIE *H1newn- ‘nine’ (Toch A nu, Skrt nava, Arm inn [< *enun], Gk enne(F)a, Lat novem, Goth niun, Lith devynì, Slav *devętĭ, Alb nëntë) < *pre-PIE *pH1enkH1we-n-/*pH1nekH1we-n-. The plosive loss apparently happened prior to the development of PIE labiovelars *kw from a pre-PIE combination of k+w as evidenced by the widely-attested retention of -w- in the ‘nine’ forms. (If Plosive Loss occurred while kw was already a single plosive phoneme, we would have seen ‘nine’ forms without -w- across the IE dialects.) There’s an intriguing alternation between kw and p in IE forms for ‘five’. In forms where we would expect to find /p/ we often see /kw/ (Lat qui:nque and OIr coic, Welsh pump) ‘five’, while in forms where we would expect /kw/ we find /p/ (Goth fimf ‘five’). This is usually explained as a result of later processes of contamination, and Pre-Laryngeal Plosive Loss Law does not seem to add anything new here.

    From the semantic standpoint, the fact that the PIE *H3ekto: ‘eight’ is considered to have meant at some point in history ‘four’ as well (as seen in Avest ašti- ‘width of four fingers’) indicates that the connection between ‘five’ and ‘nine’ is also semantically possible. This may be explained as a vestige of a base-4 numeral system (comp.: Gamkrelidze & Ivanov 747).

    Second, Pre-Laryngeal Plosive Loss Law can explain such aberrant forms as Arm vec’ and OPruss uschts ‘six’ (undeniably related to PIE *swek’s ‘six’: Toch A sak, B skas, Skrt sas, Gk heks, Lat sex, OIr se, Goth saihs, Lith šeši, Slav *šesti, Alb gjashte) in which the lack of s- makes them a bridge between PIE terms for ‘six’ and ‘eight’ (*H3ek’to: > Toch A okat, B okt, Arm ut’, Skrt astau, Gk okto:, Lat okto, OIr ocht, Goth ahtau, Lith aštuoni, Slav *osmi). In pre-PIE times there were forms *sH3ek’-t, *sH3ek’-s- with an undifferentiated meaning ‘six, eight’ (and possibly ‘four’). Note the “thorny” alternation between Gk heks and okto: paralleling Gk akso:n and tekto:n (< *tH2ek-t, *tH2ek-s above).

    ReplyDelete
  30. Next to the pair of IE *bherg’- ‘white, shining, birch tree’ ~ IE *H2erg’- ‘white, silver’, we also find IE *bheuH- 'be' (Skrt bhávati 'is, there is, happens, prospers, becomes', Avest bavaiti `becomes, originates; happens, will be', bhūmán- 'fullness, wealth, bulk, mass, wealth', pra-bhúḥ 'mighty, salient', Arm bois 'sprout, herb, plant', busanim 'burst forth, spring forth', Gr. φύω 'beget', φύομαι `become, grow', φυτόν 'growth, plant, kid, child, ulcer', φυή 'growth; nature, character', φῦμα 'plant, growth, ulcer', φύσις `nature', φῦλον `stem, gender, sex, kind of', Lat fuī 'I have been', futūrus 'future, about to be', OIr buith 'be', Goth bauan 'stay, dwell, inhabit', Lith bū́ti 'be', būklas (*būtla-) 'nest, den, hideout, lair of wild animals', Slav *byti'`become, be' ~ IE *H2eug'- 'grow, increase' (Toch A oksiš `grows', A okšu, В aukšu `old', Skrt ojman 'strength', Gk ἀ(F)έξω `grow, increase', ἀέξομαι `grows', αὐξάνω `grow, increase', Lat augeo: 'increase', Goth aukan 'increase', Lith augti 'increase, grow', áukštas 'high') > *Pre-PIE bH2eug'H1-. The semantic connection is straightforward. The meaning 'grow, swell' underlies PIE *bheuH- 'be' (Pokorny 146-50). There are also some clear morphological parallels such as Skrt bhūmán- 'fullness, wealth, bulk, mass, wealth' next to ojman 'strength' (< *bH2eug'H1men). There's asymmetry in the way the two stops followed by laryngeals get lost in both forms: in PIE *bheuH- the second stop is lost, while the first stop is retained (with the laryngeal phonemicized as *bh), while in *H2eug'- it's the other way around.

    ReplyDelete
  31. Further support for Pre-Laryngeal Loss Law comes from an Indo-Europeanist Kroonen as quoted by Kloekhorst (http://www.uni-marburg.de/fb10/iksl/sprachwissenschaft/dwnld/kloekhorst.pdf): " Moreover, in his fthc. Etymological Dictionary of Proto-Germanic, my colleague Guus Kroonen (p.c.) will propose to connect the word for ‘night’ with the root *dhngw- as attested in the Germanic words for ‘dark’ (OHG tunkal, OSax. dunkar) and in Hitt. dankui- ‘dark’. His idea is that the root originally was *dhnegw-, in which the cluster *dhnV- was simplified to *nV- in PIE already. We could therefore imagine that when in e.g. an original hysterodynamic paradigm *dhnégw-t, *dhngw-ét-m, *dhngw-t-és the nom.sg. form regularly developed into *négw-t, the whole paradigm was reshaped. We may have to assume that on the one hand, the stem *négw-t was used as a basis for a static paradigm (*négw-t, *négw-t-m, *négw-t-s), or that the root *negw- was generalized throughout the paradigm, yielding *négw-t,*nogw-ét-m, *nogw-t-és (with regular change of unaccented *e > *o as described in section 8). The former paradigm could have yielded the Hitt. gen.sg. form nekuz, whereas the latter could have been the basis for*nokwt- as attested in the other IE languages." Kroonen does not explain why "the cluster *dhnV- was simplified to *nV- in PIE already" - an explanation offered by my PLPLL - but he definitely found another etymology that supports it.

    ReplyDelete
  32. After a long hiatus, a couple more:

    Pre-PIE *kH1rebH2-, *kH1rebH2-w- 'cover, hide, pile up' > PIE *H1rebh- (Gk erepho: 'cover with a roof', orоphе: 'roof', OHG rippi, ribbi, ONorse rif 'rib', Slav *rebro 'rib' < *reb-tro) ~ PIE *kreH2w- (Gk krupto: 'hide, conceal', krupha, krubda 'secretly', Lith krauju 'pile up', kru:va 'heap', Slav *kryti 'cover, hide', *krovu 'roof'). Clusters stop + laryngeal are operating in both anlaut and inlaut. In inlaut, the loss of a stop (*b) results in the eventual coloring of the vowel by the following H2. When the laryngeal is lost, a diphthong -au- is formed from the new colored vowel /a/ and the originally suffixal -w-. The fusion of a stop + laryngeal sequence results in the formation of a voiced aspirate *bh. *-bh- seen in Gk krupto: may not be the same -*bh- seen in Gk erepho: unless krupto: is metathesized from *krebh-w-to-.

    Pre-PIE *kH3reg'H1-i- 'bend, stretch' > PIE *H1reg'- 'extend, straight, right' (Toch AB rak 'strethc out, cover', Skrt rjyati 'stretches out', Avest razayeiti 'adjusts, arranges', Gk orego: 'stretch', Lat rego: 'move in a straight line', rectus 'straight', OIr rigid 'stretches', OHG recchan 'stretch out', Lith rezti 'stretch') ~ Balto-Slav *kreiwos (Lith kreivas 'bent, crooked', Latv kreiss 'left', kreilis 'left-handed person', Slav *krivu 'bent, crooked'). Similarly to the one above, this set shows clusters stop + laryngeal in two places in a word. The loss of *g' in Balto-Slavic results in the formation of a diphthong -ei' but there's no vowel coloring because H1 that followed -g'- is a non-coloring laryngeal.

    ReplyDelete
  33. From the domain of body parts here's an amazing case of how Pre-Laryngeal Plosive Loss Law can tie together two seemingly incompatible isoglosses with exactly the same meaning creating therefore a more continuous distribution:

    Pre-PIE *nH2egwH2r- 'kidney' (and some other paired organs) > PIE *H1negwhro- (Gk nephros 'kidneys', Lat (Praenestine) nefro:ne:s 'tesicles, OHG nioro, ONorse ny:ra 'kidney' ) ~ PIE *H2eHer- 'kidney, heart' (Hitt hahari 'some paired organ', Toch B arance 'heart', Lat re:ne:s 'kidneys', OIr a:ru 'kidney, gland'. The morphology of Lat re:ne:s (either from *nwre:ne:s < *newre:ne:s or from *h2eH2re:ne:s) is identical to the morphology of Lat nefro:ne:s and Toch B arance.

    ReplyDelete
  34. Pre-PIE *dH2egwH2-mr-, *dH2egwH2-r- 'to burn, be hot' > PIE *dhegwh- (Skrt dáhati 'burns', Avest dažaiti, Gk τέφρᾱ 'ashes', Lat favilla 'hot ashes, smoldering coals', febris 'fever', MIr daig 'fire', Bret devi 'burn', Goth dags, OHG tac, ONorse dagr 'day', Lith degu 'burn', dãgas 'heat, summer heat', Slav *zego (< *gego < *dego) 'burn', Alb djek 'I burn') ~ PIE *H2eH2- 'burn, be hot', *H2eH2mr- 'day' (Toch A omäl, В emalle 'hot', Skrt ahar, Gen. ahnas 'day', Gk Att ἡμέρᾱ, Gen. -ατος, ἀμ̄έρα, Hom. ἦμαρ ‎'day', Arm awr < *a:mo:r 'day'). h- in Gk ἡμέρᾱ is analogical after ἑσπέρα 'evening'. Lat febris could be from *femris.

    ReplyDelete
  35. Pre-PIE *kwH2opH2-i- 'pay, abound in' > PIE *kwoi-, *kwoineH2- 'pay, price' (Skrt cayate 'pays, punishes', Avest ka:y- 'pay, repent', kae:na: 'vengeance, hatred', Arm kʿēn 'hate', Gk tino: 'pay, repent', poine: 'compensation for crime, blood price', tisis 'payment, punishment, vengeance', MIr cin 'guilt, debt', Lith káinа, kainà 'price', Slav *kajo 'repent', *ce:na 'price') ~ PIE *H2op- 'wealth' (Hitt hap- 'be in abundance', happar- 'price, cost, trade', happarai- 'sell, transfer, give back', happinant- 'wealthy', Skrt apnas 'possessions, wealth', Gk (Hom.) aphneios 'wealthy', Lat opes 'possessions, wealth', officium (< *opi-ficium) 'debt, duty', OHG uoben 'fulfill'. The loss of a medial stop followed by a laryngeal in a PIE stem *kwH2opH2i- resulted in the formation of a diphthong -oi- seen in Gk poine:. This laryngeal explains -ph- in Hom. aphneios.

    ReplyDelete
  36. Pre-PIE *dH1eg'h-/*dH1og'h- 'out of, athwart' > PIE *dH3g'h-mo- (Skrt jihma 'athwart, oblique', Gk dokhmos 'slanting, oblique') ~ PIE *H1eg'h-s (Gk eks 'from, out of', Lat ex 'out of', OIr ess 'out', Lith iš, ìž 'from, out of', Latv iz 'out, from', Slav *jiz 'out of', Alb ith 'behind'). The formal alternation here is identical to the one found in such a well-established but unexplained set *dn̥ǵʰwéh₂s 'tongue' (Skrt jihva, Lat lingua, Goth tuggo:, OPruss insuwis, Slav *jenzyku). Skrt -i- in jihma is the same -i- as in pita 'father' above: this shows that in Sanskrit the combination of H1 + any vowel yields -i-. (Currently, it's assumed that all laryngeals are vocalized in Sanskrit as /i/.)

    ReplyDelete
  37. Pre-PIE *mH1egH2- 'I, me' > PIE *(H1)egH2-om 'I' (Hitt u:k, Skrt aham, Gk ego:, Lat ego, Goth ik, OHG ih, Slav *azu, OLith eš, Lith aš) ~ PIE *meg'h-yo 'me' (Hitt ammuk 'me' (Acc., Dat.), Toch B nuk 'I' (< *muk, under influence from oblique n-forms), Skrt mahya 'me' (Dat.), Lat mihi: 'me' (Dat.), Goth mik 'me' (Acc.)).

    ReplyDelete
  38. Pre-PIE *pebH1eti 'drink' > Hitt pa:si 'swallows' (< *pabH1si), Gk pi:no: 'I drink' (< *pobH1i-), OPruss *poieiti 'drink', Slav *pijo 'drink', *pojiti (< *pobH1i-). The medial stop is preserved in such forms as Skrt pibati, Lat bibo:, OIr ibid (< *pibid).

    There are two more forms that may be related to the one above:

    1. Pre-PIE *pobH1i-wo-/-men/-no- > Skrt payi- 'swell, be full' (as attested in payate 'swell', Perf. pi:pa:ya, Caus.-Fact. Pres. pinvati, payas 'milk'), Avest pae:man 'mother's milk', Gk pi:(F)o:n 'fat', piar 'fat', Lith pyti 'give milk', pienas 'milk'.
    2. Pre-PIE sp(H2)e/obH1i-ro- 'be sated, prosper' > Hitt ispa:i 'get filled, be sated', Skrt spha:yate 'grows fat, increases', sphira 'fat', Lat spe:s 'hope', pro-sper 'lucky', OHG spuon 'thrive, prosper', OEng spo:wan 'thrive', Lith speti 'predict, foretell, be on time', Slav *speti 'thrive, prosper', sporu 'rich'.

    ReplyDelete
  39. No need to reconstruct *peH3- as forms such as Lat po:to 'I drink' can be derived from *pobH1to-.

    ReplyDelete
  40. Pre-PIE *dH2wer-to, *dH2wer-gh- 'turn, rotate': PIE *dhwer- 'door, gate' (Skrt dvā́ras 'door', Gk θύρᾱ 'door', θαιρός (< *dhveri̯os) 'door hinge', Lat forēs 'French door', foris 'door', Goth daúr 'gate', OHG turi 'doors', Lith dùrys 'door', Slav *dveri 'door', dvoru 'court, yard', Alb derë 'door') ~ PIE *H2wer-to, *H2wer-g- (Hitt hurki 'whee;, Toch A warkant 'wheel, circle', warto 'garden, grove', Skrt vartati 'rotates', vrtis 'fence', Lat verto: 'I turn over', Lith vartiti 'turn', vartai 'doors', OPruss warto 'door', Slav *vriteti 'turn', vorota 'gate', ORuss vorotu 'neck', Alb vathe 'fence, yard').

    The semantics cannot be better, and it's a very important cognate set for the study of the evolution of IE wheeled transport vocabulary. The loss of a plosive before H2 is very transparent here: Hittite attests for h- (hurki, just like harki < *bherg'h- above). Latin forēs, foris shows that aspiration merged with -w- after the loss of d-.

    ReplyDelete
  41. Pre-PIE *g'H2wedH1-r/n 'wild animal' > *g'hwer- (Gk θήρ, Lat ferus 'wild', Lith žvėrìs, OPruss swirins 'wild animal', Slav *zve:ri 'wild animal') ~ PIE *H2wedr-, Gen. H2wednos (Hitt huetar, heutnas 'wild animal', pl. huita:r 'wild animals', ONorse vitnir 'animal, wolf').

    It's a double loss of a plosive here, the second one explaining the vowel length in Gk θήρ and Slav *zve:ri. Consistently with the antiquity of PLPLL, it predates the formation of Western IE labiovelars (Gk θήρ) from a cluster "velar + w".

    ReplyDelete
  42. Pre-PIE *nH3odH2s- 'nose, smell' > PIE *noH2s- 'nose' (Lat na:ris, Lith nosis, Slav *nosu) ~ PIE *H3od- 'smell' (Gk ὄζω, Lith uodziu, uosti, Lat odo:s).

    ReplyDelete
  43. Next to such previously described (see above) pre-laryngeal consonant cluster losses such as PIE *k'lH3eu-s 'hear, ear' and *spH3ek'wo-/*skH3ek'wo- 'inspect, eye', there's Pre-PIE *snH1eigwH2o- 'snow, ice' > PIE *sneigwho- 'snow' (Gk νείφει 'it snows', νίφα 'snow', νιφάδες 'snowflakes', Lat nīvit, ninguit 'it snows', niх, Gen. nivis 'snow', Goth snaiws 'snow', OHG snîwit, ONorse snjor 'snow', snigid 'it snows, rains', OPruss snaygis, Lith sniẽgas 'snow', snìgti 'it snows', snaĩgala 'snowflake', Latv snìegs 'snow', Slav *sne:gu 'snow') ~ PIE *H1eig- 'ice' (Hitt eka 'cold, frost, ice', egai, igai 'freeze', ekuna, ikuna 'cold', ekunes 'become cold', ekunima 'coldness', MIr aig 'ice' (< *yega), Welsh ia 'ice', ONorse jaki 'ice-floe', jokull 'icicle, glacier'). The semantics is perfect. The morphology is shared (Lith snaĩgala 'snowflake' next to ONorse jokull 'icicle, glacier'). The ice set preserved the original cluster -gw- later phonemicized into labiovelar -gw- as seen in Gk νίφα. Aspiration seen in Gk νίφα (< *sneijgwheH2-) must have come from a laryngeal H2.

    ReplyDelete
  44. It may not quite belong here, but footnote 32 in this paper explains the Slavic *struj- words for "uncle" as a borrowing from Iranian (and the Lithuanian one as a borrowing from East Slavic) for a reason explained in the text.

    ReplyDelete
  45. Thank you David. I wasn't aware of this paper and honestly did not know about Eugen Hill in the first place. I haven't been posting much on this topic - not much interest plus I've run out of forms suggestive of the law. :) But I haven't changed my mind. What I did notice from many of the pairs is that the forms with elided consonant tend to be morphological and semantic derivatives of the forms with the consonant retained (e.g., peH2- 'protect' > *(p)H2owis 'sheep' (the protected one)). This may give us a clue as to why the consonant drops - the question you raised initially.

    ReplyDelete
  46. The moment I said I ran out of examples and here it comes: Pre-PIE *dH3wer- door, open and close like a door' > PIE *dhwer- (Skrt dvā́ras 'doors', Gk θύρᾱ 'door', θαιρός (< *dhweRi̯os) 'door hinge', Lat forēs 'two-leaf door', foris 'door', Goth daúr 'gates', OHG turi 'doors', Lith dùrys 'door', Slav *dviri 'door', *dvoru 'courtyard', Alb derë 'door') ~ PIE *wer- 'open' (Skrt apavr̥ṇṓti 'opens, locks', Gk ἀ(F)είρω 'I tie'; Hes. ἄορον ̇ θυρωρόν, ἀ(F)ορτή 'bag', Lat aperio 'unlock', operio 'lock', Lith veriù, vérti 'pull a thread, open, close, lock, unlock', OPruss etwerreis 'open!', Goth warjan 'obstruct', Slav *vre:ti 'close, lock, pass through', *vere:ja 'door leaf, door frame'. I tentatively reconstruct H3, but H2 is possible, too. With H2, there will be an onset cluster *dH2w-. What makes this etymology string is the fact that the otherwise isolated semantically and morphologically Gk θαιρός (< *dhweRi̯os) matches exactly on those aspects Slav *vere:ja < *vereyo-.

    ReplyDelete
  47. Pre-PIE *dH1wLH1-dh-, -gh-, -n- 'steadfast, long-lasting' > PIE *dLH1gho- 'long' (Hitt daluki 'long', Skrt dīrghá- `long', Gk dolikhos 'long', en-delekhḗs `continuous', Goth tulgus 'firm, steady', OSax tulgu 'very', OEng tylgest 'best', Lat longus 'long', Lith ìlga- `long', Slav *dlugu- 'long-lasting, long', *dilinu 'long', Alb glate 'long') ~ PIE *H2ulH1-eH1- 'be strong' (Lat valeo: 'be strong, healthy', Olr. fallnaithir 'to rule', Welsh gwaladr 'ruler', Olr. flaith 'rule', Welsh gulat 'rule', Olr. fal 'rule', OPr. weldisnan 'inheritance' [acc.sg.], Lith. veldeti 'innherit, Slav vladeti 'to rule', Goth waldan 'to rule', OHG walten 'to dominate', Toch A wal, gen.sg. lant, Toch B walo, gen. lante 'king' ).

    This root *dH1wLH1- was very productive and has generated a) several morphologically distinct offshoots shared across branches (comp. -t- in Alb glate 'long' and Welsh gulat 'rule', or a voiced aspirate -dh-/-gh- in Slav *vladeti and Gk dolikhos or -n- in Slav *dilinu, Lat longus but OIr fallnaithir); b) onset cluster reductions (Lith ìlga-, Lat longus, Toch A lant).

    Both PIE sets have been difficult to reconstruct. H2- in *H2ulH1-eH1- is suggested by -a- in Olr. fallnaithir but Welsh W. gwaladr 'ruler' points to an onset without a laryngeal. There's a secure syllabic -L- in the 'long' set and -L- could have given -a- in Old Irish. H1- is consistent with all the other 'strong' forms (-a- in lat valeo: can stand for any laryngeal). An unaspirated d- in the 'long' set guarantees that the onset did not contain H2, as it would have aspirated the d-. But it's consistent with H1-. -w- pervasively attested in the 'strong' set can be "worked into" the 'long' set without too much difficulty. The variation around the syllabic L seen in Slav *dlugu (not **dligu) vs. Lith ìlga- and Skrt dīrghá- (not **du:rgha) may point to the same -w- because Hitt daluki agrees strongly with the Slavic form, so we can reconstruct an intermediary protoform *dwL- on the basis of Slavic and Hittite. Gk dolikhos can easily be *dFolikhos < *dFelekhos, just like Goth tulgus can be from *twulgus < *twLgo-.

    Semantically, the two concepts - 'long' and 'strong' - are very compatible. Goth tulgus 'firm' proves this. 'steadfast and long-lasting' because 'full of vitality' and 'in good health' can be reconstructed for Pre-PIE Pre-PIE *dH1wLH1-. The semantic development into 'rule', 'own' and 'power' is also understandable as all of those notions presuppose an ability to last and outlast.

    ReplyDelete
  48. Addendum: no. 10 above.

    Arm harkanem 'split, fell', OIr orgaid ‘he kills’ is compared to Hitt harkzi 'is destroyed' (EIEC, 158).

    ReplyDelete