Comment: The Arch. form can belong both here and to the PL *ʔaχär- 'sleep, lie' q.v. The original meaning of PL *ʔaqɨ- must have been 'to fall, to become spilled, scattered', cf. Ag. l-aqa-s (Bursh. aqi-s) 'to scatter, spill, get scattered' (cf. also in Tab. without the d- preverb aq- 'to fall'), Arch. c-a=χa- 'to fall; to throw' (with an expressive preverb) and possibly also Lezg. al-aq- 'to stick, adhere', Ag. ʕ-aq- (Bursh. ʕ-aqi-) 'to mix'.
Cf. also Tab. al-a=q-, Düb. h-a=q-, Ag. ʕat:-aq- 'to meet, encounter' (*"to fall upon") - which may have obtained its meaning due to the influence of PL *ʔiqɨ- 'to find' (q.v.), or may, in fact, provide a link between these two roots (stems): see PEC *=iq_wV.
Comment: Cf. Arch. dur. darq̇:Iur. Unclear is the loss of pharyngealisation in Rut. (cf. also l-a=q:e- 'to unlock'). Besides the forms mentioned above, we must note a strange development in lit. Tabasaran and the Southern dialect: qI-a=q̇I- (with irregular -q̇I-, opposed to regular -q:I- in Northern Tab.). This could be a separate root, connected with Arch. erq̇Iʷin- 'to conceal, put away' - however, the Archi form itself may be related to d-aq̇:Ia-s (in most forms -q̇Iʷ- here occurs in a syllable-final position - like inf. erq̇Iʷmus etc. - and it may in fact represent the morphonemic *q̇:Iʷ).
Comment: The root is used as an adverb (Lezg. aq̇ʷan), postposition (Arch. eq̇en) or suffix (Tab. -q̇an, cf. also Lezg. -q̇ʷan with the same meaning and function). Historically this is obviously a gerundial form (cf. the same development in Andi).
Meaning:1 to take off, peel (skin etc.), to skin 2 to kill
Lezghian:al-až- 1
Rutul:q-a=ǯe- 1
Archi:a=č:a- 2
Comment: For *-r- cf. Rut. dur. qa-raǯe-r. The verb belongs to the strong class. It is possible to derive from this root (with Ablaut) the PL noun *ʔärč: "wineskin" ( < *'peeled skin'), reflected in Tab. erǯ (Düb. irǯi) 'wineskin; bellows', Ag. irž 'bellows' (Bursh. erǯ 'wineskin').
Comment: In Tab. and Ag. the stem is used mostly with preverbs; without a preverb cf. Ag. Bursh. arki- 'to get into, find oneself (somewhere)'. With other preverbs cf. Ag. Bursh. qat:-arki- 'to lag behind; to lose, be defeated', Rich. k:et:-ik- id., ald-ark- 'to turn round', Tab. Düb. q-a=k- 'to return'.
Comment: Cf. also Bud. term. ʕake-ǯi, Arch. dur. =erku-r, Ag. Bursh. alt:-arki- 'to fall; to lie down, go to bed' (also ut:-arki-,ʁ-arki-,ʕ-arki-). Without a preverb cf. Ag. Bursh. arki- 'to fall down'; all other languages use only the root combined with preverbs. The verb belongs to the strong conjugation class. It should be distinguished from PL *ʔarke- 'to go, move (away)' q.v. (although sometimes contaminations are happening).
Most forms reflect the PL vowel *-a- in the first syllable; Archi, however. reflects *-e-. Cf., perhaps, also Tab. -e=k- in k:-e=k- 'to hide, to conceal' (*'to put down, to let lie') and with some other preverbs.
Comment: Cf. also Ag. Tp. (Shaumyan) ilgʷan- 'to stay' (pointing to *-n-conjugation in PL); perhaps also Rut. g-ä=ga- 'to lose, be defeated' (fronting of the vowel probably conditioned by the preverb g( ́) < *ƛ̣:-; another possibility would be to compare the Rut. form with Tab. e=g- 'to become tired'). Medial -l- in Ag. is a modification of *-r- in the -n-conjugation. The root demonstrates the PL *a/i Ablaut.
Comment: Cf. also Arch. dur. =arḳu-r. In Tsakh. the root forms the aorist and imperative part of the suppletive paradigm of the verb "to go". It is also used with another preverb (ʔ-) and the causative auxiliary verb -ās in the meaning 'to drive, urge' (aḳan-ās). Reasons for -n-conjugation in Tsakh. are not clear. Since there are no Ag. or Tab. forms, it is impossible to establish whether there was a medial *-r- in PL (external data, in particular, Dargwa, suggest such a possibility).
Meaning:1 to fall 2 to go down, roll down 3 to get into
Lezghian:aw-ax- 2
Tabasaran:a=x- 1
Agul:arx- 3
Tsakhur:q:aʔ-a=xa- 1
Budukh:araxa- / a=xu-
Comment: The verb is very productive (used with many preverbs, especially in Tab. and Ag. - cf. Ag. ʁ-arx- 'to lie down, to sleep' etc.), but sometimes tends to contaminate with the roots *ʔeʎe- 'to put' and *ʔeʎ:ʷɨ- 'to lie, to put' which are to be strictly distinguished.