Comments: PAT *a- 'definite article', *a- in *a-rǝjǝ 'this', *a-nǝjǝ 'that' (Abkh., Abaz. ari,ani). In PAK the root was used independently (*a 'that', 'this'); in Ub. it is also used as a definite article (as in PAT) and within complex pronoun stems (a-j-da 'that' etc.).
Comments: In Abkh. the meaning of interrogative pronouns is normally expressed by verbal suffixes -da ('who'), -j ('what'). Since d- and j- are class markers (animate/unanimate respectively), the only morpheme to which we can ascribe the interrogative meaning is -a (probably reduced in *-ja > -j). The same morpheme can be seen in -a-ba- 'where' and independent pronouns d-a-rban 'who', j-a-rban 'what' (the -rban suffix is not very clear etymologically - perhaps it contains the root *-ba- 'to see' q.v.?). In Abaz. the situation is analoguous - only in the function of independent pronouns we have d-zač̣ʷǝ-j-a 'who' and j-zač̣ʷǝ-j-a (or ač̣ʷǝja) 'what'. These are derived (by suffixing the same hypothetic interrogative morpheme *-a) from d-zač̣ʷ,j-zač̣ʷ 'each' (ač̣ʷ probably had also existed) where -zač̣ʷ = Abkh. -zač̣ǝ 'single' (from *zA 'one').
Proto-West-Caucasian:*-a- (/-0-, -na-)
Meaning:1 inanim. sing. subj./poss. class marker 2 3d p. subj. trans.
Comments: PAT *(a)ća (cf. also Bzyb. áća). The initial vowel is probably a trace of the original pronominal prefix (aj- in Abaz. is a reciprocal prefix). Cf. also the synonymous PAT *taća (Abkh. a-táca, Bzyb. a-táća, Abaz. taca; for the first component *ta- cf. perhaps Abkh. a-tá-ḳʷaž̌, Abaz. ta-žʷ 'old woman'; or maybe *taća is an independent kinship term, corresponding to the isolated (within EC) Av. durc 'son-in-law' ? [other similar forms - God., Botl., Tind., Cham. durc, Gunz., Bezht. dursa - are all more or less recent loanwords from Avar]).
Despite Abdokov (1974, 62) the PAT root can not be related to PAK *nǝsa 'daughter-in-law'; see the criticism in Shagirov 1, 288.
Meaning:1 word, speech 2 to say,speak 3 to take oath,swear
Abkhaz:áž̌a 1
Abaza:ažʷa 1
Kabardian:žǝ-ʡa- 2
Ubykh:čǝ-q̇á- 3
Comments: PAT *azʷa (cf. Bzyb. áža); Ub. sǝ-č(ǝ)q̇á-n (used also as a noun: čǝq̇a 'oath'). The PAT root is also used in a compound *cʷa-zʷa- 'to speak' (see under *c:ʷa-). In PAK *č́:- would be expected, but only the Kab. form is known (Kab. ž can go back to PAK *č́:, *č:, *ǯ́). Still, the exact parallelism between Kab. žǝʡa- and Ub. čǝq̇a- (where the second part goes back to *q̇́Ia- 'say' q.v.) makes this etymology quite plausible.
The comparison of Kab. žǝ-ʡa- and Abkh. áž̌a was first proposed by Mészaros (1934, 308), who also compared Ub. maśa 'word' which is impossible for phonetic reasons. See Shagirov 1, 198.
Comments: In Ub. two roots are usually confused: (a)gʷǝ́ 'small' and gʷǝ- 'to grind' (q.v.); see Vogt 129 about this distinction. The root 'small' seems to have no correlates in WC languages, but has possible parallels in EC.
Comments: PAT *matǝ (with secondary deglottalisation). Klimov (1967, 302) suggests that the root may be cognate with PAK *-nṭa in *Lanṭá 'flexible', but this is highly dubious. Ub. anṭá < *am(ǝ)ṭa with assimilation.
Comments: PAT *χ́anǝta (in Abkh. there also exists a variant á-χ́amta). PAK *want:áʁʷǝ (also borrowed in Abkh. a-wadaʕw).
A very difficult case. PAT *χ́- can not in any way correspond to PAK *w-, and Ub. ǵ- to either, so we must assume that in all subgroups we have compounds with different initial components (but the nature of these components is not clear). The situation is aggravated by irregularity of correspondence between PAT *-t- , PAK *-t:- and Ub. -ṭ- and by the existence in Abaz. of a synonymous adjective ħat-la 'heavy' (does it have any connection with *χ́anǝta ?).
Comments: PAT *abV (cf. also Abkh. plur. ába-ča); PAK *at:a,*t:ǝ (cf. also Ad. /in compounds/ āt(a) ). Ub. def. form á-tʷ.
The correspondences are regular, thus it seems reasonable to accept Deeters's etymology (Deeters 1963,39), pace Shagirov (1, 56-57). Cf. also Mészáros 1934,208 (Abkh.-Ub.).