Comments:ОСНЯ 1, 306. The Turk. root may also be of a Persian origin (see TMN 3, 645-646). Mong. *küjil- < *küli-l- (with regular dissimilation). Kor. *kằrắm < *kòrắm (with vowel assimilation); cf. also Old Koguryo *kŭăl 'river' (see Miller 1979, 8).
Comments:EAS 47, 107, 141, KW 193, SKE 98, Лексика 575. Doerfer (TMN 3, 450) and Щербак (1997, 141) consider Mong. to be borrowed from Turkic, which cannot be excluded.
Comments:The Tung.-Jpn. match seems to be satisfactory, but the Turkic parallel is not quite certain: in PT one would rather expect *Kokuĺ with voiceless -k-, but available attestations seem to point rather to *-g-. {Cf. PE *kaɣǝ-(ru) 'peeled off walrus skin'.}
Comments:Doerfer MT 91 and Rozycki 130 consider Mong. to be borrowed < TM (Evk. kalim etc., see ТМС 1, 366-367); however, the direction of borrowing was probably reverse. This is one of several similar fish names, sometimes difficult to distinguish - see *kalu, *k`ile.
Comments:KW 182, Miller 1970, 129, АПиПЯЯ 293. Illich-Svitych (ОСНЯ 1,301, with Uralic parallels) compares Man. kola-, Evk. kūlū- 'to skin', but these are mongolisms (see ТМС 1,407); ТМ *xolda-ksa is a better match for the Mong. form.
Comments:АПиПЯЯ 288, ОСНЯ 1, 327 (with literature; Illich-Svitych also compares Mong. qoli- 'to mix', but does not list the Turkic and Japanese forms). Cf. also Mong. kölbe- 'to lie on one side' (KW 238), qolbir- `to slip sidewards'. The attribution of some forms is problematic: Turk. *Kul- may reflect a partial contamination with PA *gi̯ŭldo 'to stretch' (q. v.).; the Jpn. and Kor. forms, due to the merger of *r and *l, can actually also reflect PA *ki̯ṓr[i] q. v.; in Jpn. cf. also koro 'round log', korog- 'to roll, rotate'. Note that the MKor. form with -w-, -'- makes the affiliation of the Korean root questionable - unless it is a dissimilative development < *kurb-ɨr-, in which case it would be a morphological structure *k`úlo-bV- = PJ *kǝ́rǝ́mp-, Mong. kölbe-, qolbir-.
Comments:АПиПЯЯ 295. A Turk.-Kor. isogloss (cf. also Old Koguryo *kămul 'black', see Miller 1979, 8; perhaps also Manchu χumara- 'to sully', ТМС 1, 477). The comparison seems quite possible, although the scarcity of reflexes prevents a secure reconstruction of vocalism.