Comments:Цинциус 1984, 125-126, Дыбо 12. A Western isogloss. In PT *k- would be expected; initial voicing is probably due to the influence of another Altaic root, *kĕ̀ŋi 'empty' (PT *geŋiŕ 'nasal cavity'), cf. its reflexes in Mong. (keŋkeji- 'be wide and empty'; Щербак 1997, 127 actually regards the latter as borrowed from Turkic - which is quite dubious; see TMN 3, 612-613).