Comments: PAT *q̇ǝrǝq̇ǝ; PAK *kʷǝ́rq:ʷǝ / *gʷǝ́rq:ʷǝ (in compound with *pśa 'neck'). The Kab. word has a specific meaning "part of neck between the first cervical vertebra and the skull".
Despite some phonetic irregularities (expectable in expressive stems of this type) it is hard to separate the PAT and PAK forms.
Comments: Cf. also Abkh. Bz. -c-ħa-, Ub. 1st p. sǝ́-q̇Iǝ-n. In PAT and PAK present is a compound with 'tooth' as a first component: PAT *cǝ-ħa-, PAK *c:a-q:a-. PAT *ħ reflects rather PWC *q̇Í (with palatalization due to Ablaut?); the normal reflex q̇ is present in the reduplicated form PAT *q̇aq̇A (Abkh. á-q̇aq̇a-ra, Abaz. q̇aq̇-ra) = Ub. q̇Iaq̇Iǝ- 'to chew'.
Comments: PAT *ħʷa-: labialisation is probably secondary (as in Ad.), under influence of PAT *(a)ʕʷV- 'be heard, reach one's ears' (see *q́:IʷV).
PAK *ʡa- (preserved only in Kab.: in Ad. the root also has acquired secondary labialisation under influence of *ʡʷa-/*ʡʷǝ- 'be heard, reach one's ears'; there are also some contexts where labialisation is present in Kab., too - see Шагиров 2, 159-160). The noun *ʡʷǝ 'mouth' should be probably kept apart from this root. On the component žǝ- in Kab. see *(a)č:ʷV.
Ub. 1 p. sǝ-q̇á-n; the root is also used as a noun q̇a 'word, speech'.
Comments: PAT *qaʕa (cf. also Bzyb. á-χā); PAK *ʡa-c̣ʷǝ́. The PAT and Ub. forms are reduplicated (note that PAT *qaʕa reflects an early dissimilation *q̇́Iaq̇́IV- > *q́Iaq̇́IV-).
See Balkarov 1967, 115 (PAK - Ub.). The PAK form is a compound with *c̣ʷǝ 'good' q.v.; the first part, however, (despite Dumézil 1963 and Shagirov 2, 158) has certainly nothing to do with *ʡa 'hand' (as well as q̇a- in Ub. q̇aq̇ǝ́).
Comments: PAK *ʡa; Ub. def. á-q̇aṗa. For the etymology of -ṗa see *ṗA; the root q̇a- is discoverable in many compounds (q̇a-(n)cʷá 'finger' etc.).
External correspondences suggest that this root must have been labialised, but labialisation was for some reason lost in PWC. Indeed, we find the labialised form *q̇́Iʷa (~*q:́Iʷa) within the compound *q̇́Iʷa-(p)ƛ̣a (~*q:́Iʷ-) 'arms, embrace; to embrace' q.v.
AT parallels for this root suggested by several authors are not persuasive (see Шагиров 2, 154 - 155 with literature).
Comments: PAT *ħʷa (used only with preverbs). PAT *ħʷ can have several PWC sources, thus the reconstruction of a glottalized affricate is only one of the possibilities.
Meaning:1 to tan, rumple, soften (leather) 2 to thresh
Abaza:ħ:ʷa-ra 1
Adyghe:ʡʷa-n 2
Kabardian:ʡʷa-n 2,1
Comments: PAK *ʡʷa (the meaning 'to tan, soften' is preserved in Kab. ʡʷa-tǝ-n and Ad. ša-ʡʷā-ca, Kab. fa-ʡʷā-ʒa 'implement for softening leather' - see Kuipers 1975, 73).
Comments: PAT *qʷǝ (cf. also Bz. a-χʷǝ́; in Abaz. the meaning 'a hair' is expressed by a derivate form qʷǝ-c). PAK *q:ʷǝ-jǝ́ (where -jǝ is a privative suffix) - this comparison seems irreproachable to us, whereas the comparison of AA and Ub. forms with PAK *xʷǝ-č̣́a- 'to wash (head)', proposed by Shagirov (2, 77) is worse semantically and impossible phonetically.
The old derivate *q̇Iʷǝ-cǝ ( > Abaz. qʷǝc) with some irregular processes (loss or metathesis of labialisation) is also reflected in PAT *qǝcʷǝ > Abkh. a-χć̌ǝ́ (Bzyb. a-χć̌ǝ́) 'braid, plait', Abaz. qǝčʷ "bird's comb", PAK *ʎa-q:ǝca (*ʎa 'foot') 'shaggy fetlocks'.
Comments: PAK *q:ʷǝ-śħá (the second component is *śħa 'head'); in Kab. the word is preserved only as an ethnonym ('Balkarian' or 'Ossetian') or in toponyms (see Шагиров 1, 236). Despite Shagirov (ibid.) and Gan 1909, 68 the first component q:ʷǝ- should be kept apart from PAK *q:ʷa 'gorge, ravine, valley' - the latter has a separate Ub. parallel, see *q:ʷa.
Comments: PAT *qʷa-; PAK *q:ʷa-(n)č́á (a compound with *č́V 'bend', see Shagirov 1, 230); Ub. a-q̇Iʷǝ́-n. The root can be also reduplicated: PAT *qʷaqʷa 'hooked, curved' (Abkh. á-χʷaχʷa, Abaz. qʷaqʷa); Ub. q̇Iʷǝ́q̇Iʷǝ- (sǝ-q̇Iʷǝ́q̇Iʷǝ-n) 'to squat, huddle up'.
The comparison of PAT *qʷa- and PAK *q:ʷa- belongs to Abdokov (1973,55), and is certainly better than Klimov's (1967, 306) comparison of PAT *qʷa- with PAK *wǝ-xʷa- 'to bend' (absolutely impossible phonetically). Abdokov adds also Ub. q̇ʷarṭa- (sǝ-q̇ʷarṭá-n) 'to bend, be(come) crooked', but for phonetic reasons it seems better to choose Ub. q̇Iʷǝ- 'to bend' (thus following Shagirov 1,230). [Shagirov compares also Abkh. a-r-q̇ʷ-ra 'to bend (trans.), Abaz. čǝ-na-r-q̇ʷ-ra 'to bend', but this is a misunderstanding: the root here is PAT *q̇ʷǝ 'to lower, go down', and the meaning 'bend' is metaphorical (applied to fortune, circumstances and the like).]
Besides PWC *q̇IʷV- the WC languages also reflect a phonetically close variant: *q̇́IʷV-, present in a compound *q̇́IʷV-rṭV / *q̇́IʷV-nṭV: Ub. q̇ʷarṭá- (see above), PAK *ʡʷanṭa- / *ʡʷarṭa- (Ad. ʡʷanṭa-, Shaps. ʡʷarṭa- 'to bend, crook', Kab. ʡʷanṭa- 'to twist').
Comments: PAT *c̣ʷǝ-ʕʷanǝ (for the first part see *c̣ʷǝ 'pointed stick'); PAK *ʡʷǝ́na; Ub. def. a-nǝ́q̇́. The correspondence PAT *ʕʷ : PAK *ʡʷ : Ub. q̇́ is regular and points to PWC *q̇́ʷ.