Curdled milk

PIE *twer
Gloss: ‘curdled milk, curds’ (item 125 in Bjørn 2017)

Attestations: Rus. tvaróg ‘quark’; Gr. τῡρός ‘curdled milk’; Ved. tuvara- ‘astringent’, Av. tūiri

Attestations only pertinent for the Graeco-Aryan and Slavic branchings, this produce term nonetheless appears to be reconstructable for late or late middle PIE. On the somewhat discontinuous semantics in Vedic, it may be noted that Middle Indic continuations retain the meaning ‘cheese’ (Turner 1966: 336). A case for an internal verbal derivational basis has not been satisfactorily posited (cf. Mallory & Adams 1997: 382f.). The phonetic structure is void of phonemes that represent the telltale developments of the pertinent dialects, and a wanderwort phenomenon cannot formally be excluded; the modern English word quark is similarly assumed to have traveled from Slavic through Middle High German.

External comparanda:
North Caucasian: *ˀV-twVr– ‘to become rolled up, turn sour, rot, putrefy’ > e.g. Archi tar-as ‘to roll up (of milk)’

Discussion: A North Caucasian origin of this term, as suggested by (S. Starostin 2009: 102) seems rather attractive as a technical term that emanated from the Caucasus, where the verbal derivational basis is transparent. The prefix here has to be discarded to fit the IE attestations, but this seems a minor problem; verbal roots normally interact more extensively with prefixes than nominals do (cf. Nikolayev & Starostin 1994: 82ff.). The possible linguistic fixpoints allow for a very concrete chronological hypothesis: The item likely departed from North Caucasian and entered either late PIE or, alternatively, differentiated IE languages still in close contact, and lastly reached the Germanic languages through Slavic.


PIE *ul ̥h2neh2
Gloss: ‘wool’ (item 127 in Bjørn 2017)

Attestations: (?)Hit. ḫulana-, (?)CLuw. ḫulana/i-; Lat. lāna; Welsh gwlan; Goth. wulla; Lith. vìlna; OCS vlŭna; Gr. λῆνος; Ved. úrṇa-, Av. varənā.

Notes: While the Anatolian connection recently has enjoyed special and thorough attention by Pinault (2016) who provides a relatively shallow PIE genesis for the term, the formal correspondences have also been questioned, e.g. by Kloekhorst on formal grounds (2008a: 357f.), but the overall resemblance and direct semantic match does make the comparison worth entertaining.

External comparanda:
NE Caucasian: *ƛ̱wähnɨ > e.g. Avar ƛ̱uh (S. Starostin 2009: 84)
Sumerian: ú-li-in / wux-li-in ‘colored twine/wool’ [(Whittaker 2004: 410-11)]
[Add. Semitic: Akk. hullanu ‘a blanket or wrap of linen or wool’ (CAD h: 229), suggested in a Facebook-thread]

Discussion: According to Mallory & Adams (1997: 648) wool is not a naturally occurring material, but was dependent on the domestication and breeding of certain strains of sheep that appear to have been kept only in the fourth millennium BCE. This chronology is of some influence on the spread of the IE languages whose earliest historical identification is Anatolian in the late third millennium BCE. The exact sequence of the PIE reconstruction is also of great consequence to the strength of this comparison, but may, if viable, even be informed by the NE Caucasian material (the connection is continued by Matasović 2012: 290f.); metathesis is required no matter what paradigm is preferred. Note, too, that words for ‘goat’ show an incredible resilience to reconstructional fixation and enter the same general semantic field as the present item. An inferred Hurrian noun *ḫul(a) (Kronasser 1967: 45) very closely resembles the Anatolian material, and several different scenarios are thus conceivable depending on the mutual compatibility of the IE forms; most probable seems the proposition that wool entered the IE languages as a loan on at least two different occasions, very likely from the Caucasian and Middle Eastern cultures, where it also appears early as borrowing into Sumerian with specialized meaning (Sahala 2009: 11f.); the apparent lack of Semitic comparanda is in this regard interesting. [Add. With the inclusion of the formally rather attractive Akkadian item, a Semitic ‘conductor’ may be found, although the semantic line connection Sumerian and Akkadian is less than clear.]


Two etymologies of (P)IE lambs have external comparanda:

PIE *h1ln̥bh
Gloss: ‘lamb’ (item 36 in Bjørn 2017)

Attestations: Goth. lamb ‘sheep’; (?)Gr. ἒλαφος ‘red deer’

Notes: If rightly connected by Kroonen, this root has very limited and semantically difficult IE distribution, and for both branches necessarily different Ablaut grades (2013: 325f.). The traditional etymology for the Greek item departs from a possible Aeolic origin of the alternate form ἐλλός ‘fawn’ < *el-n-os, cf. Arm. eln ‘id.’, Lith. élnis ‘id.’, OCS elenĭ ‘id.’, Welsh elain ‘id.’ (e.g. Frisk 1960: 483ff.). Adams surely connects to these forms Toch. A yäl ‘gazelle’, B yal ‘id.’ and the rather elusive Hit. aliyan(a)- ‘(?)roebuck’ (2013:523 [note: contra Mallory & Adams, who connect it to Hit. ali- ‘soft’ (1997: 154)]), and further, more semantically tantalizing, Toch. B āl– ‘ram’ (2013:58). The labial obstruent in the main Greek lemma has then been compared to the identical ending of ἒρι-φος ‘youngling’ that is decisively derivational in Ved. vṛṣa-bhá– ‘bull’ from vr ́ṣan– ‘manly, powerful’ (Chantraine 1968: 333,372). This PIE suffix *-bhé– may be relevant for Germanic regardless of the heritage. Lehmann hesitantly connects all of the above from a root *el- ‘brown’ (1986: 226), which may as well be fitted with the initial laryngeal *h1-, but Germanic is formally difficult to include in this bulk. The Greek branch, along with its possible cognates, would surely help establish the root as of PIE origin, but the comparison is doubtful. [NOTE: See now also Vrieland 2017 in favor of the wider connection]

External comparanda:
NE Caucasian: *ɫVmbagV ‘sheep’ > Avar lémag, Akhvakh lãgi; Lezghian lap:ag

Discussion: What appears to be a new comparison nonetheless suggests itself straightforwardly with almost perfect alignment of three identical radicals; moreover, the Avar gloss is a) ‘sheep’ b) ‘1yr old sheep’ (NCED s.v.), mirroring the Germanic semantics. According to Nikolayev & Starostin the NE Caucasian form is likely to have been borrowed (ibid.). It is possible that this lexeme at an early stage complimented the proper PIE *h3ewi- ‘sheep’ (item 58), either as a loan from an unidentified source but in common with NE Caucasian, or as an internal derivation; the semantic fate in Germanic and possible elsewhere in IE surely indicates a kind of specialization. The lexeme was transferred into Finnish lammas ‘sheep’ from a Germanic language spoken in the Baltic region, either Gutnish or Gothic (Kroonen 2013: 326), both of which developed (or retained) the less specialized meaning. The connection may thus be compared to PIE *h2e(i)ĝ- ‘goat’ (item 40) as a lexical isogloss unique to NE Caucasian and western IE.


PIE *h2egw-no-
Gloss: ‘lamb’ (item 45 in Bjørn 2017)

Attestations: Lat. agnus; OIr. úan; OEng. ēanian ‘to yean’; OCS agnę; Gr. ἀμνός

Notes: Attestations reflect limited distribution, that nonetheless seems tightknit despite minor differences (Frisk 1969: 93f.). The item cannot be posited for the earlier stages of PIE, and appear to be a European regionalism.

External comparanda:
Semitic: *igl– ‘young animal’ (Gamkrelidze & Ivanov 1995: 769 [Add.: also Blažek 1996])

Discussion: Orel & Stolbova connect Semitic with cognates in Central Chadic, Cushitic, and Omotic to establish Afro-Asiatic *igal ‘cow, calf’ (1995: 247), and note should be made of the semantics that are consistently bovine except for the generalization that incurred very sporadically in Ge’ez to ‘young animal’ and possibly also Akkadian agalu ‘donkey’. Gamkrelidze & Ivanov choose to weigh the more general meaning (ibid.) that more readily resonates with the (P)IE reconstruction despite considerable evidence pointing to the conclusion of higher specialization (cf. also Diakonoff 1985: 129f.); the Semitic comparanda should consequently be rejected. Rather, the word may be connected with *h2e(i)ĝ- ‘goat’ (item 40), which requires both semantic drift and an explanation of the formally incompatible velars, that may be ascribed to its putative foreign nature.


Several items meaning ‘goat’ have external comparanda, four are included here. See also Kroonen (2012: 245-247):

PIE *díg-
Gloss: ‘goat’ (item 15 in Bjørn 2017)

OHG ziga; Alb. dhi ‘fem. g.’; Gr. (Hes.) δίζα; Arm. tik ‘hide’; Ishkashimi (East Iranian) dec ‘goatskin bag’

Notes: The Greek form is problematic and requires either a glide to palatalize the velar, or, as has been suggested, the form in Hesychius, originally ascribed to Laconian, may rightfully be attributed to one of the lesser known IE Balkan languages, Thracian or Illyrian (Frisk 1960: 390ff.). All the data combined, this reconstruction still fails to paint the picture of a central PIE item, although proto-status certainly is possible.

External comparanda:
NE Caucasian: *tVqV > e.g. Ingush tɨqo
HU: Hurrian taɣə ‘man (male person)’
Kartvelian: *dqa > Georgian txa, Svan daq

The IE material does not seem to be particularly strong and lacks cognates in the decisive ancient branches. Proposed as a borrowing by Gamkrelidze & Ivanov (1995: 774) and Nichols (1997: 146), the exact nature of the reconstruction in Kartvelian is debated (cf. Fähnrich 2007: 125), but the Northeast Caucasian material does help establish the form in the region. The semantics of the Hurrian material questions its appurtenance, but a final rejection pends further illumination of the internal relationship. Ultimately, this item belongs in the very same category as the synonyms (items 21, 40, and 73, cf. § and be ascribed to a loan into the later strata of PIE.


PIE: *ghaid-o-
Gloss: ‘goat’ (item 21 in Bjørn 2017)

Lat. haetus ‘young g., kid’; Germ. *gait- > ON geit, Goth. gaits.

The connection of the Latin and Germanic forms seem beyond reproach, but remain isolated as a European regionalism. a-vocalism similarly makes a PIE origin of this item unlikely (Dolgopolsky 1987: 16).

External comparanda:
Afro-Asiatic: Semitic *gadi-̯ > Arab. jadyun, Heb. ge?ī
Afro-Asiatic: Berber aġăyd
NE Caucasian: Proto-Nakh *gāʒa, Lak gada ‘kid’

The Semitic and IE correspondence is difficult to ignore, but whereas Dolgopolsky considers it a direct loan from Proto-Semitic into PIE (1987: 14), Kroonen proposes a (likely extinct and unattested) third party origin for both, ultimately stemming from waves of early agriculturalists that first introduced the term to Semitic and later into European IE from an already present adstrate (2013: 163ff.).The dearth of proper PIE evidence affects both theories, but less detrimentally the latter. Nichols’ analysis of the NE Caucasian forms as old dialectal borrowings due to the internal inconsistencies (1997:129) seems to
corroborate the adstrate hypothesis; it is noteworthy, however, that Nikolayev & Starostin reconstructs a Proto-NC *gēʒ́wV that would remove the Caucasian item from comparison with PIE and Afro-Asiatic. It seems callous to posit the root for PIE proper and invites further inquiry into the ancient relations of European IE and its agricultural prehistory (cf. *h1ln̥bh– ‘lamb’, item 36, for a similar correspondence between Germanic and NE Caucasian).


PIE *h2e(i)ĝ-
Gloss: ‘goat’ (item 40 in Bjørn 2017)

Alternant 1: *h2eiĝ-
Attestations: Alb. dhi; Gr. αἲξ; Arm. aic; (?)Ved. eḍa– ‘kind of sheep’, Av. īzaena- ‘leathern’

The Vedic form requires analogy to fit the picture, but seems plausible (cf. Mayrhofer 1986: 264). Attestations only warrant reconstruction for late PIE.

Alternant 2: *h2eĝ-
Lith. ožýs; OCS azno ‘goat-skin’; Alb. edh; Ved. aja-, Av. aza-

The homophonous verbal root *h2eĝ- ‘to lead’ (item 43) has quite naturally been
suggested as the derivational base, but the likeness to (1) complicates this connection (cf. also Mallory & Adams 1997: 229). This form also brands cognates in Balto-Slavic, but still fails to secure the decisive old branches for ancient strata.

External comparanda:
North Caucasian: *Hējʒ́u (cf. *ʡējʒ́wē of NCED s.v. ‘goat, she-goat’)

The variant forms within (P)IE do suggest a foreign source, which, indeed, may
be found in North Caucasian, as suggested by S. Starostin (2009: 80 fn.8). This is certainly also suggested by the phonological compositions that are close to being superimposable, especially on reconstruction (1) with the diphtongue. The second reconstruction may under this paradigm be explained as either stemming from folk-etymological analogy with the homophonous verbal root *h2eĝ– ‘to lead’, or as the natural yet inconsequent treatment of a foreign sequence in (P)IE (cf. Matasović 2012: 290 fn.16). Further phonological confusion is encountered if PIE *h2egw-no- ‘lamb’ (item 45) is considered a derivative to the present form.


PIE *kaĝo-
Gloss: ‘goat’ (item 73 in Bjørn 2017)

Old Low German hōken; OCS koza; Alb. kedh, kec ‘kid’

Very limited distribution questions ancient PIE moorings for this item. Connections with PIE *h2e(i)ĝ- (item 40) are formally impossible (cf. Mallory & Adams 1997: 229). The reconstructed a-vocalism is noteworthy.

External comparanda:
NE Caucasian: *qoVcV > e.g. Lezgian ʁec

The Slavic forms may, according to Derksen, be borrowed from a Turkic language
relatively late (2008: 242), ostensibly within the first millennium CE. S. Starostin proposes this Northeast Caucasian connection (2009: 81), but the formal correspondence is not very attractive.


PIE *kŵon-
Gloss: ‘dog’ (item 72 in Bjørn 2017)

Hit. LUkuwan-/kun- ‘dog-man’, HLuw. zú-wa/i-n(i)-; Toch. A & B ku; Lat. canis; OIr. cū; Goth. hunds; OPrus. sunis; Rus. súka ‘bitch’; Gr. κύων; Arm. šown; Ved. śva ́-, Av. spā-.

An early extension of the root accounts for the Germanic stock, which is comparable to Latvian sùntena ‘big dog’ and Armenian skund ‘small dog’ (Mallory & Adams 1997: 168), but this formation need not be old. It may be noted that there is a different treatment of the sequence *k̂w- in the centum languages than in ‘horse’ (item 32).

External comparanda:
Uralic: *kVn– > e.g. Selkup kanan, Udmurt kyjon, North Saami gaidne ‘wolf’
Yukaghir: *keδe ‘wolf, dog’

The semantic gap from ‘wolf’ to ‘dog’ is hardly insurmountable, cf. that the shift also occurs in certain IE branches (Mallory & Adams 1997: 168), and the somewhat scanty attestations from across the Uralic pallet (Samoyedic, Volgaic, and Saami) may thus be considered likely comparanda. Chronologically the Uralic proto-language is required, making a transfer from dialectal IE implausible. The term may have migrated as far as China, where, according to Gamkrelidze & Ivanov, it was adopted around the middle of the second millennium BCE, cf. Old Chinese k’iwen (1995: 507), but this, obviously, has to be seen as a secondary transfer, possibly from Proto-Tocharian. This protracted journey suggests that an earlier loan into Uralic from PIE is probable, although the relative scarcity of Uralic material makes an exact reconstruction difficult, which further makes the comparison with exact IE strata difficult; that it cannot be compared to a satəm continuant, e.g. Indo-Iranian, is, however, very clear. This comparison requires that Selkup k- is the reflex of PIE *k̂- (cf. the discussion of ‘catfish’, item 78), and a case for an Indo-Uralic item could easily be assembled.


Two items with the meaning ‘pig’ have proposed external relations.

PIE *súH
Gloss: ‘pig’ (item 119 in Bjørn 2017)

Toch. B suwo; Lat. sūs; Welsh hwch; ON sýr ‘sow’; OPrus. swintian; OCS svinū; Alb. thi; Gr. ὗς, σῦς (var.); Ved. sūkará-, Av. hū-.

Only lacking in Anatolian (possibly hiding behind Sumerograms) and Armenian, the word has old status in PIE and has been proposed connected to the verbal root *seuH– ‘to bear, bring forth’. Although Mallory & Adams criticize that it appears not to be restricted to the sow (1997: 425), semantic widening may well have rendered the term more general already before the breakup of PIE. More problematic is the fate of the laryngeal that is missing, formally inadmissibly, in some derivational forms, and a more onomatopoetic origin may alternatively be posited (e.g. Mańczak 2000: 232f.), a notion which Beekes denies (2010: 1425).

External comparanda:
Kartvelian: *(h)ešw– > Georgian ešw– ‘tusk’, Mingrelian o-sk-u ‘pigsty’
Sumerian: šáḫ(a)

The semantics of Kartvelian are marked by great dispersal, but Klimov seems right to connect ‘tusk, fang’ and ‘pigsty’ under a common denominator ‘(wild) boar’ (1998: 48), according to Fähnrich also attested in Old Georgian (2007: 152), which suggests an early transfer to account for the divergence that nonetheless cannot directly be honored since the item is lacking in Svan and thus, conservatively, may only be secured for Georgian-Zan; a loss in the Svan language is, of course, conceivable, and must be posited for the relation to be of PIE date as Gamkrelidze & Ivanov suggest (1995: 774). Kaiser & Shevoroshkin prefers common heritage from Nostratic with the common meaning ‘to give birth’ (1986: 369), but this scenario requires that Kartvelian independently innovated the same meaning, which is hardly preferable to a loan from early middle PIE where the meaning already had been attained. A borrowing in the opposite direction, i.e. from (an antecedent of) Kartvelian is theoretically possible, and may be corroborated by the archaeological data (Anthony 2007: 285f.). The Sumerian comparandum is semantically unproblematic, but likely requires the reconstructed PIE laryngeal questioned by some. The item may well be the same in all three families (cf. Sahala 2009: 10) and have travelled from a Middle Eastern origin at an early date, although the exact origin, possibly related to the spread of agriculture, remains elusive.


PIE *ĝhor-io-
Gloss: ‘pig’ (item 18 in Bjørn 2017)

Alb. derr ‘pig’; Gr. χοῖρος ‘piglet’.

Notes: Slight irregularities in Albanian does not challenge the ultimate adherence to the Greek form (Demiraj 1997: 131f.), but the uniqueness of the two Balkan forms cause Beekes to suggest an unspecified substrate origin for the term (2010: 1640f.). Possibly derived from the noun *ĝher- ‘bristle’ (Mallory & Adams 1997: 425), this word is obviously an innovation and cannot be projected back onto PIE proper.

External comparanda:
Kartvelian: Georgian-Zan *γor– > Georgian γor-, Mingrelian, Laz γeǯ– (Fähnrich 2007: 491f.)
Afro-Asiatic: Chadic *γr

Discussion: Tentatively posited by Gamkrelidze & Ivanov as a mutually dialectal loan relation (1995: 774 fn.15), this connection suffers from sporadic attestations in both IE and Kartvelian. An ostensible connection must consequently be assigned to a late stratum, likely dialectal, which is geographically difficult. Alternatively, Kaiser & Shevoroshkin suggest ancient relations to Afro-Asiatic, notably without Semitic cognates (1986: 378 fn.8), but the lack of PIE continuation makes a tentative connection to agricultural substrate preferable.

Cow / Bull

Two words with very plausible external relations denote the bovine species:

PIE *gwṓu
Gloss: ‘cow’ (item 30 in Bjørn 2017)

HLuw. wa/i-wa/i-(i); Toch. A ko, Toch. B keu; Umb. bum (acc.); OIr. bō; ON kýr; Latv. gùovs; OCS go-mĭno ‘threshing floor’; Alb. ka; Myc. qo-u-, Gr. βοῦς; Arm. kov; Ved. gáu-, Av. gāuš.

Albanian shows some irregularity in the exact type of velar (Mallory & Adams 1997: 134ff.), but does not question the PIE reconstruction. The phonetic complements in Hittite (GUD-us, GUD-un) do allow a continuation of this root, but cannot be known for certain (Gamkrelidze & Ivanov 1995: 483), but the Hieroglyphic Luwian attestation appears solidify evidence for the Anatolian branch (Melchert 2003: 195).

External comparanda:
Semitic: *ġi ‘to bellow’ (Schott 1936: 66f.)
Egyptian: gw ‘bull’ (Mallory & Adams 1997: 135)
Kartvelian: *pur (attestations see Klimov 1998: 206)
NW Caucasian: *č’:amə > e.g. Kabardian gwaw ‘bull’ (Nichols 1997: 143)
NE Caucasian: *ʒ ̣̌VW > e.g. Chechen-Ingush govr ‘horse’ (ibid.)
Sumerian: gu4 ~ gud ‘bull’

It is likely that the PIE root originally was generic only for the species (and not also for the gender), denoting both ‘cow’ and ‘bull’, i.e. ‘cattle’ (cf. Gamkrelidze & Ivanov 1995: 482), which is practically demanded by the external comparanda. According to Mallory & Adams the domestication of the cow began during the Neolithic revolution in the Middle East in the 7th millennium BC and had spread through Europe by the inception of the fourth (1997: 137). Considering the rather extensive list of proposed comparanda, the possibility of onomatopoetic homophony may justifiably be argued, but the ostensible motivation is elusive and perhaps best represented by the Semitic verbal connection proposed by Schott, but it is weak compared to all other comparanda that clearly relate a very narrow semantic value to each other; the only possible scenario that could comprise this root as viable would be as the primordial verbal root from which the term was originally derived. Against this notion, the Egyptian evidence stands comparatively stronger denoting the animal itself and seems strengthened by further Afro-Asiatic cognates in West Chadic *warar ‘vicious bull’ (Orel & Stolbova 1995: 527). These forms may gain further credibility through the similarly positive situation in Sumerian (cf. Sahala 2009: 7) that represents the earliest attestation and is perpetuated as Sumerograms in Hittite where it curiously eclipses a word that might ultimately be related. Further suggested borrowings include Old Chinese (Mallory & Adams 1997: 135), but especially noteworthy is the proposed connection to Altaic which includes a demonstrable semantic generalization that allows a shift to other quadrupeds, cf. as a link
Manchurian geo ‘mare; cow; female (of quadrupeds)’ and consequently Classical Mongol gegün ‘mare’ (Gamkrelidze & Ivanov 1995: 491f.), particularly relevant for the inclusion of the NE Caucasian comparandum suggested by Nichols (1997: 143). The NW Caucasian attestation, obviously semantically incompatible with its western neighbors, is formally interesting from a PIE point of view as it retains an initial labiovelar. Labiality is also the proposed recipient rendition of *gw– in Kartvelian [ADD: Nichols doesn’t consider the forms related, but for the sake of argument I’ll leave it here as my own suggestion], which necessitates either an early borrowing directly from PIE, a centum continuant, or a non-IE language with similar retention of the feature, e.g. NW Caucasian, but the proposed concordance poorly matches ‘to extinguish’ (item 110) where PIE *gw– yields Kartvelian *kw-, and is therefore quite improbable. All things considered the item may well be foreign in PIE, but it is of considerable age; it seems to significantly predate the spread of other agricultural terms, e.g. *(s)teuros ‘bull’ (item 118), but may, too, be connected with the same general wave of technological innovation, as widespread borrowings certainly suggest. The somewhat difficult Anatolian material could be secondarily introduced, but there is no imperative to favor such an interpretation.


PIE *(s)teuros
Gloss: ‘bull’ (item 118 in Bjørn 2017)

PIE (1) *steuros

Attestations: OHG stior; Av. staora– ‘large (domestic) animal (i.e. horse, ass, cow, camel)’

Notes: The suggested Avestan correlate to OHG stior may also derive from *steh2uro– ‘big’ (Kroonen 2013: 478), which would render the Germanic forms isolated.

PIE (2) *tauros

Attestations: (?)Lat. taurus; Gaul. tarvos; ON þjórr; OPrus. tauris ‘bison’; OCS turŭ ‘aurochs, bull’; Alb. ter; Gr. ταῦρος; Khot. ttura– ‘mountain goat’

Notes: Celtic shows metathesis, but fits otherwise unproblematically with the European cognates. Germanic has *-e- instead of *-a-, which has been explained as analogy to *steur-, but see the discussion immediately below for a different solution. [Add. Latin is expected to show metathesis of the sequence *auRV > aRvV, suggesting that it might be a relatelively late borrowing (Weiss 2009: 157).] The Khotanese item, however, stands out semantically, but fits with the Avestan continuation of PIE (1) *steur- if part of a Proto-Iranian shift as suggested by Bailey (1979: 132).

External comparanda:
Semitic: *tawr > Akk. šūru, Arab. tawr, Heb. šôr, etc. (Bomhard 1981: 416)
Semitic (alt.): Arab. ṭaraḳa ‘to impregnate (of camels)’ (Schott 1936: 78)
NE Caucasian: *stw– (Nichols 1997: 143)

Discussion: The very obvious correlation between the Semitic and IE forms certainly suggests them as related, obvious even to philologists (Levin 1995, cf. § 3.2.6). The nature of this relation, however, is disputed. Schott’s attempt at a connection with Arabic ṭaraḳa ‘to impregnate (of camels)’ to substantiate internal semitic derivation (1936: 78) is improbable, potentially even more so in light of Militarev & Kogan’s Afro-Asiatic reconstruction (2005:307ff.) that nonetheless requires some semantic leniency, covering both ‘elephant’ (Central Chadic) and ‘rhinoceros’ (Northern Omotic), but it is noteworthy that Orel & Stolbova do not include to form (1995), and Lipińsky seems to share this uncertainty of origin (1997: 561) [Add. This could shed new light on the Avestan comparandum, although it becomes chronologically strained.]. The two disparate PIE words, although clearly related, seem to have entered the language at two different stages; *steuros likely prior to *tauros, where the latter is phonetically less complex and also brands a-vocalism. Dolgopolsky initially considered the IE complex descendants of a single form borrowed directly from proto-Semitic into PIE (1987: 14), but later suggested only partial common heritage for Semitic and PIE that was subsequently confounded by a loan (1993: 244). Weak Indo-Iranian evidence, however, as well as a complete lack of Anatolian and Tocharian cognates, question the antiquity of the noun in PIE while the Semitic provenance itself may be inhibited by the questionable nature of the Afro-Asiatic evidence for the root. Alternatively, Kroonen suggests that the e-vocalism in Germanic *steur- and *teur– is a consequence of raising from an otherwise unidentified donor language (cf. Haarmann 1994: 269f. and the treatment of *ghaid-, item 21) that similarly shows up in Etruscan thevru– (2013: 540), and may additionally account for the Germanic forms with initial s- from a foreign spirant not found at the recipient stage of the language (2013: 478). Nichols’s suggestion of a NE Caucasian*stw– (1997: 143) is tantalizing as it seemingly reflects the sporadic s-forms in IE, but the root is not included by Nikolayev & Starostin (NCED) and has not been validated.


PIE *h1ék̂w-os
Gloss: ‘horse’ (item 32 in Bjørn 2017)

HLuw. azu(wa), Lyc. esbe-; Toch. A yuk, Toch. B yakwe; OLat. equos; Gaul. epo-; ON jór, Goth. aíhwa-; Lith. ašvíenis ‘stallion’; Myc. i-qo, Gr. ἳππος; Arm. ēš; Ved. áśva-, Av. aspa

Only the Slavic and Albanian branches miss this lexeme, and the horse can safely be ascribed to the earliest strata of PIE on internal evidence alone, despite Dolgopolsky’s attempt to discredit the inherent nature of the Anatolian forms (1993: 240). The root gives valuable insights to the intricacies of the velar series with the co-occurrence of the palatal *– with the labial element *-w-: While satəm languages retain the independence of the labial element, basically [+pal] + [+lab], the centum languages fuse the phonemes into one, [+vel, +lab], and subsequently follow the rules governing the inherited labiovelars. The Greek forms remain, although clearly related, problematic, probably reflecting some kind of dialectal borrowing (Beekes 2008: 597f.), although Sihler suggests that at least the geminate may be a regular outcome of the inherited palatal-labial cluster (1995: 159f.). Internal derivation has been proposed from a root ‘swift’ attractive due to a common Graeco-Aryan collocation, ‘swift horses’, Gr. ὠκέες ἳπποι, Ved. áśvāḥ āśávaḥ, but the connection is formally difficult (Mallory & Adams 1997: 273), if not impossible (Beekes 2008: 597f.).

External comparanda:
Kartvelian: e.g. Georgian ačua (children’s language)
North Caucasian: *ɦɨ[n]čwĭ
NW Caucasian: *cʷ̣̌́ ǝ > e.g. Abkhaz, Ubykh ačy (pl.)
NE Caucasian: Avar, Lak ču, Akhvakh ičwa, Andi iča, etc.
HU: Hur. eššǝ

The domestication of the horse represents a central pillar in Anthony‘s archaeological location of the PIE speakers on the Pontic Steppes (2007: 193-224), and intense and selective exploitation of the horse was widespread from around 4.500 BCE in the region from southern Ukraine through to Kazakhstan (Mallory & Adams 1997: 273ff.). It is thus noteworthy that Uralic did not borrow the PIE word. A common Proto-North Caucasian reconstruction has been posited, *ɦɨ[n]čwĭ, without convincing internal derivation, although the authors do not explicitly make the connection with PIE (NCED s.v. ‘horse’). Such a reconstruction, however, gives depth of time comparable to PIE, and invalidates Colarusso’s intricate and very idiosyncratic analysis of NW Caucasian and PIE with the aim of uniting the phyla in Proto-Pontic (2003: 41ff.). The Hurrian form is proposed as a loan from the satəm Mitanni-Aryan superstrate that brandished a particularly specialized equestrian vocabulary, especially owing to the assibilated geminate (Gamkrelidze & Ivanov 1995: 809). A similar origin has been suggested for Semitic, cf. Akkadian sisū and Ugarit ssw (despite confounding factors, cf. Militarev & Kogan 2005: 261ff.), Sumerian sí-sí (Sahala 2009: 10), some NE Caucasian languages (Dolgopolsky 1987: 19) as well as Abkhaz (Mallory & Adams 1997: 274), but these latter can just as well be explained from the bulk of Caucasian attestations treated below. According to Nikolayev & Starostin the sporadic Kartvelian forms can be ascribed to intra-Caucasian loan relations (NCED s.v. ‘horse’), and are thus considered secondary and beyond the scope of the present paper. The evidence presented by Dolgopolsky to substantiate the entry into Caucasian languages as an early Proto-Indo-Iranian loan rests solely with the palatal treatment of the internal velar in the Daghestanian dialects (1987: 19), but the argument is mute because palatals are reconstructed for PIE and there is no evidence to suggest that a proposed Proto-North Caucasian recipient language would not have treated them as such (§ Indeed, if horses were a trademark of ancient PIE culture (predating the split of Anatolian) with concomitant mercantile prowess, and the PIE speakers interacted intensely with the North Caucasian linguistic area at an early date, it is very likely that the ‘horse’ represented a sufficiently valuable item to be transferred from PIE and subsequently thrive in the Caucasus (cf. Matasović 2012: 291). Tantalizing both in the initial laryngeal PNC *ɦ– to mirror PIE *h1– and the internal sequence of palatalized velar plus labial glide, the only aberrant element in the North Caucasian stock is the nasal found in the Lezgian branch of NE Caucasian (reconstructed *ʡɨnšʷ [~ħ-], see NCED s.v. ‘horse’), but the circumstances are internally unclear. Alternative solutions, albeit less attractive, are chronological adjustments either (1) further back into prehistory and approaching Nostratic, or (2) later as a dialectal PIE loan, as suggested by Dolgopolsky (1987: 19).