Comment: 3d class in Kryz. The obl. bases (Tab. hüli-, Ag. ħüli-) are not enough for a precise PL reconstruction.
Since the meanings 'sea' and 'moisture, liquid' sometimes interchange (cf., for example, the reflexes of PEC *ɦwmǟɦwā), it seems tempting to compare also Arch. hoIlo-t:u- 'liquid' (and perhaps as an old dialectal variant, with another frequent semantic change 'liquid -> green' - Arch. oIlow-t:u- 'green'). There are no other probable hypotheses about the origin of these Archi words.
Comment: Cf. also Ag. Bursh. ħarčal, Burk. ħarčaj, Tp. ħürčal. An expressive root, known only in Tab. and Ag., and with not quite regular correspondences (with secondary devoicing in Ag.).
Comment: Tab. wič̣ri < *hʷiṭiri, as seen from the more archaic Düb. form hüṭür / hüṭer 'leech'. The root is attested only in Tab., but has reliable external parallels.
Comment: In Tsakh. the stem is used only for the 3d/4th classes (opposed to ǯu-/ǯe-, see PL *-iž); in Lezg. and Tab. it is used for the 3d person (while the stem *-iž is used for the 1st and 2d persons). The original usage is probably the Tsakhur one.
The PL paradigm can be reconstructed as *ʔičʷ (dir. stem), *ča- (obl. stem, cf. Tab. gen. ča-n, Ag. Bursh. ič̌,če-).
Comment: This archaic pronominal stem is preserved only in the Archi genitive wit (where w- is a class prefix, like in w-is (1st p. sing.), ulu ( < w-olo, 1st p. plur.), w-iš (2d p. plur.)). Note that the Ag. (Fit.) genitive wit has nothing to do with the Archi form: it contains the PL attributive morpheme *-t:V > -t, common to all pronouns (zi-t,xi-t,či-t,kü-t).
Comment: In Arch. cf. erg. ej-mi. Tsakh. je-d historically = Arch. ej-t:u-r (with an adjectival suffix, typical for Archi kinship terms). A nursery word, but having probable external parallels.
Comment: Cf. also Tab. Kand. juḳ, Düb. jaḳa, Ag. Fit. jiḳʷ. 4th class in all class-distinguishing languages. Obl. base *jaḳʷɨ- (cf. Lezg. jaḳu-, Tab. jeḳu-, Ag. Fit. jiḳu, Tsakh. jaḳu-).
Comment: Cf. also Tab. Düb. jac, Ag. Bursh. jacʷ, Fit. wic, Burk. wec. The original medial nasal *-m- is preserved (as -n- before a sibilant) only in Arch. All other languages have lost it, but Agul has preserved the compensatory labialisation (Proto-Ag. *jawc; in Rich. bec regularly < *wec, cf. the forms in other dialects). All correspondences are regular (including the fronting *a > e in Rich. and Burk. before a hissing affricate, and the development *ja- > *ji- in a closed syllable in Tab.).
The obl. base can be reconstructed as *jamca- (cf. Rut. jasa-, Tsakh. jaca-, Arch. erg. ansa); 3d class in all class-distinguishing languages.
Comment: Cf. also Lezg. Khl. jatur 'thigh, hip'. 4th class in Shakhdagh. The medial resonant is not reflected anywhere (in the combination *-mt- it could have been preserved as -nt- only in Arch.), but may be postulated on external evidence.
Comment: The Ag. form is from the Burshag dialect. Cf. also Tab. Düb. jaq:-li, Ag. Fit. jaq:al-t 'high'. In Lezg. there occurred a contamination with *[hI]alχä- 'long' q.v. (that is why we have jarʁi instead of *jarq:i; cf. also Lezg. Khl. t:aq:aj 'high', which contains a preverb t:-).
The root is actually verbal, which is shown by verbal prefixation (q:-) in Kryz.; it is also very probable that the Tab. verb di-ji=ʁ-uz (Düb. dī=q:-)'to stand' reflects the same root.