Comment: A Shakhdagh isogloss. Formally the word resembles a gerund (*saʔa-j), thus the root may have been originally verbal. In that case *s- may be analyzed as a preverb (rather frequent in Shakhdagh).
Comment: Reasons for voicing in Tab. (cf. also Düb. ʒakur) are unclear; external evidence clearly speaks in favour of PL voiceless *s- (cf. also Ag. Fit. sakʷa-na).
Comment: Oblique base *sarsa- (or *sorsa- with Ablaut which probably passed into the direct base in Rut.), cf. Lezg. sara- ( < *sarsa- with dissimilation).
Comment: The root is preserved only in Tsakh.; it obviously was modified semantically under the influence of PL *c:imc: 'ant'. However, external data show that it is an independent root.
Comment: Obl. vowel base unclear (in many languages only the original plural form 'teeth' is preserved and passed into singular, thus in Tab., Ag., Rut. and. Ud.). For PL *-l: cf. Arch. sot, pl. sot:-or (where -t:- < *-l:-). Belongs to the 4th class in all class-distinguishing languages.
See Лексика 1971, 111; Гигинейшвили 1977, 70; Талибов 1980, 295.
Comment: Cf. also Lezg. Khl. sɨw 'summer pasture', Ag. Bursh. su (NB: s:u in the final version of the MSU recordings is obviously a misspelling). Obl. stem *sɨwä-, cf. Lezg. suwa- (Khl. swa-), Tab. siwi-, Ag. suwa-, Rut. sɨwa-. The Ud. form (with suffixed plural marker -ruχ) goes back to *wɨsa - which, in view of the external evidence, seems to be the most archaic variant.
See Лексика 1971, 183; Гигинейшвили 1977, 133; Талибов 1980, 314.
Comment: Cf. also Lezg. Khl. sur, Ag. Bursh., Fit. sür, Tsakh. prop., Gelm. sɨwij. The most archaic form is preserved in Tsakh. Gelm., all other dialects and languages have contracted reflexes. 3d class in Rut. and Tsakh. [N.B.: the latest MSU recording of the Ag. Bursh. form is s:ür, which is obviously faulty: the word is a member of the minimal pair sür 'wild onion' : s:ür 'throat'].
Comment: Evidence for strong *s: : Ag. Bursh. s:a-d; Arch. (in compound numerals) s:e-j-; Ud. sa. It is probable, that Rut. sije-nä 'all' is derived from the same root (as well as Rut. Ikhr. sa-sa-na 'every').
Comment: The root is used only in compounds (with *ʡiš:ʷ 'night' > Lezg. jif/sen-fiz < *sen-jifi-z/ or *jiq: 'day' in Arch. /with exceptional development -q:- > -ʁ- in cluster/; with *ʔeʎ:a 'evening, late' in Tsakh., cf. also Gelm. san-ixa, Tsakh. prop. sanɨxa 'yesterday'.
Comment: Obl. base *s(:)antɨ- (cf. Tsakh. santɨ-) - or *s(:)ontɨ- with Ablaut (if Lezg. -u- reflects the vocalism of the obl. base, which is rather probable). 3d class in Tsakh., but 4th class in Rut.
Despite Vinogradova-Klimov 1979, 157 the root is hardly borrowed from Arm. sand (the Arm. word itself has no IE etymology and is very probably borrowed from a Caucasian source).
Comment: Cf. also Tab. Düb. sejri / sēri, Khür. (Magometov) s:eri-di, Ag. Bursh. sará-ji. Ag. sa-jaʁa < *s:ar-jiq:a (a compound with *jiq: 'day'); other forms are originally locative (but with a non-trivial development *-d > -ṭ in Tab. dialects).
Comment: Cf. also Ag. Fit. sim. The Kryz. form has a more specific meaning: "a sheep or goat having once born a young one". The component -ʁal in Lezg. is not clear etymologically.In Tab. the form contains a diminutive -ag; the vowel in the 1st syllable is a result of reduction and assimilation to -m- already after suffixation (sum-ag < *sim-ag). There is not enough evidence to choose between *s:- or *s- in initial position, or to reconstruct the vocalic obl. base (there are only the Ag. forms: Burk. semu-, Fit. sima-).
Comment: Cf. Lezg. pl. sep:-erar, Lezg. Khl. seb (sep:-erar) "abuse", Tab. Kand. sib 'gall', Düb. s:iw 'gall bladder', Ag. Bursh. sew (probably a misspelling instead of *s:ew); Rut. säb 'gall secreted while vomiting'. 3d class in Rut. and Arch.; obl. base *s:äm:ä- (cf. Ag. sebi-, Rut. sä́bä-).