Comments:A Western isogloss. The isolated Karakh. form may be borrowed < Sak. käbä 'well-bred horse'; perhaps a better match in Turkic would be *Kojn 'sheep' (Лексика 431, ЭСТЯ 6, 24-26) ( > Mong. qoni(n) > Evk. konin etc., see TMN 3, 564, Щербак 1997, 139, Doerfer MT 37; despite KW 194 and Poppe 70, hardly a common Altaic *koni). In that case Turk. *Kojn < *Kobi-n, with a subsequent reconstruction of PA *gi̯abo(nV).
Comments:Kalm. kāčǝ- (WMong. kijači-, KW 222) 'to split (wood)' is probably borrowed from the Turk. derivative *Kɨjgač- (although the latter now means basically 'slant', the original meaning was 'cut aslant'). If the latter was the original meaning in PA, one could also compare Mong. *geji-, *geje- 'slanting, oblique' (otherwise see notes to *k`i̯úŋ(k)u).
Comments:KW 136, ОСНЯ 1, 230. In TM cf. also a variant *kilbe- / *kilt- 'glitter' (ТМС 1, 393, 394; hence Yak. kilbej-, kɨlbaj-; but Dolg. gilbej-, gɨlbaj- may be directly < Mong.; see Kaɫ. MEJ 103, Stachowski 86, 87). Since Mong. also has a variant *gilt- (MMong. (MA) qiltai-, WMong. giltagir, giltuji-, Kalm. giltǝgṛ, giltī-, KW ibid.), the TM forms may be borrowed from Mong. (see Doerfer MT 21-22, Rozycki 89). Note that Jpn. *kìrà- may also reflect PA *gari q. v.
Comments:In TM one would rather expect *gialeŋē, but the existing forms may be explained as secondarily monophthongized. Despite АПиПЯЯ 80, the Jpn. form rather belongs here (semantically it is exactly = Mong. ɣalaɣun, and in case *kàrí < *gāŕV we would rather expect *kàtí).
Comments:Cf. also some similar roots: *gokV,*ki̯omga, *k`éma (with possible mergers). Medial -m- in Turk. may be either due to these mergers, or a result of labial assimilation (*gEmürgen < *gEŋürgen). The Jpn. form must go back to a suffixed *gi̯aŋ(u)-gV.
Comments:The original meaning is well reconstructable as "to lag behind > become separated". The derivative *gi̯ā́la-kV 'separation, lagging' is reflected in PTM *giala-ku 'separated space, room' and the denominative *gal-gi- in Mongolian.