Distinct from the suppletive verb =χʷi- [inf., imv.] / =χʷa-l- [imperf.] / χu-l-lˈe [perf.] 'to die (pl. subj.)' [Kibrik et al. 1977a 2: 74; Kibrik et al. 1977a 3: 243; Kibrik & Kodzasov 1988: 170; Kibrik et al. 1977b: 334, 386; Chumakina et al. 2007].

    Distinct from h=atʼ- 'to die (pl. subj.)' [Kibrik et al. 1999: 885].

    Another (apparently less frequent) verb for 'to die' is kʸečmiš-x- [Kibrik et al. 1999: 880], which consists of the borrowed Azerbaijani adjective kečmiš 'last, past' plus the Tsakhur verb ɨx- 'to become'.

    Distinct from h=atʼ- 'to die (pl. subj.)' [Kibrik & Kodzasov 1988: 170; Dirr 1913: 165, 241].

    Distinct from class=ɨrqʼ- / class=qʼɨrqʼ- 'to kill (pl. obj.) / to die (pl. subj.)' [Dirr 1912: 144, 166].

    Distinct from class=ɨrqʼ- / class=qʼɨrqʼ- 'to die (pl. subj.)' [Kibrik & Kodzasov 1988: 170].

    The same in the Khudig subdialect: kʼ-i- 'to die' [Shaumyan 1941: 183].

    The same in the Usug subdialect: kʼ-e- 'to die' [Shaumyan 1941: 183].

    The same in the Khanag subdialect: l=ikʼ-ˈ [imperf.] / qːa=kʼ-ˈ [perf.] / yikʼ [imv.] / kʼ- [inf.] 'to die (sg. subj.) / to kill (sg. obj.)', as opposed to yiχ-ˈ 'to die (pl. subj.) / to kill (pl. obj.)' [Uslar 1979: 754, 800, 1008; Dirr 1905: 180, 190, 245]. The same in the Khyuryuk subdialect: l=ikʼ-ˈ [imperf.] / qːa=kʼ-ˈ [perf.] / yikʼ [imv.] / kʼ- [inf.] {кIув} 'to die (sg. subj.) / to kill (sg. obj.)', as opposed to yiχ-ˈ {йихув} 'to die (pl. subj.) / to kill (pl. obj.)' [Genko 2005: 80, 112].

    The same in the Khiv and Khoredzh subdialects: yikʼ-ˈ {йикIуб} 'to die (sg. subj.) / to kill (sg. obj.)', as opposed to yiχ-ˈ {йихуб} 'to die (pl. subj.) / to kill (pl. obj.) / to beat up (pl. obj.)' [Genko 2005: 80]. The same in Literary Tabasaran: yikʼ-ˈ [imperf.] / ʁa=kʼ-ˈ [perf.] {йибкIуб} 'to die (sg. subj.) / to kill (sg. obj.)', as opposed to yiχ-ˈ {йихуб} 'to die (pl. subj.) / to kill (pl. obj.)' [Khanmagomedov & Shalbuzov 2001: 184].

    The same in Literary Lezgi: r=eqʼˈi- [imperf.] / qʼˈe- [perf.] / qʼi- ~ qʼi-n-iqʼ [masdar] / yiqʼ [imv.] {кьин, кьиникь} 'to die / to kill' [Talibov & Gadzhiev 1966: 205; Gadzhiev 1950: 883; Haspelmath 1993: 289, 503, 518, 522; Gyulmagomedov 2004, 1: 442]. The masdar qʼi-n-iqʼ is reduplicated.

    NCED: 661. Distribution: The Lezgian data on the verbs for 'to die' and 'to kill' can be summarized as follows (the slash sign "/" denotes lability):

'TO DIE/TO KILL'	Proto-CA-Udi	Archi	Kryts	Budukh	Tsakhur	Rutul	Aghul	Tabasaran	Lezgi
*ʔiƛʼe [NCED: 661]	'die/kill' imperf.	'die' sg.	'die/kill'	'die'	'die', 'kill' sg. (different prefixes)	'die/kill' sg. (redupl. = pl.)	'die/kill'	'die/kill' sg.	'die/kill'
*ʔilχʷe [NCED: 635]		'die' pl.						'die/kill' pl.
*ʔatʼʷɨ- [NCED: 271]				'kill'	'die', 'kill' pl. (different prefixes)
*ʔarčːa- [NCED: 265]		'kill'
*VpʼV	*pʼu- 'die/kill' perf.

    The root *ʔiƛʼe can safely be reconstructed as the Proto-Lezgian labile verb for 'to die / to kill', at least with singular subject ('to die') / object ('to kill') and at least as the imperfective stem.

    The only languages that suppletively discriminate between the imperfective and perfective stems are Caucasian Albanian and Udi. Formally, such a situation could reflect a Proto-Lezgian feature, but CA-Udi *pʼu- (with the ablaut variant *upʼ-) is isolated within Lezgian and possesses rather scant external comparanda (the Khinalug verb iːb-i 'to kill' and the imperative stem iːb-i 'to die'), so it is preferable to regard Caucasian Albanian-Udi *pʼu- as a secondary complication of the verbal paradigm in question.

    Archi and some Nuclear Lezgian lects demonstrate the lexical opposition between verbs with singular and plural subject/object. Formally, this could be a late innovation of areal origin, but there are actually no reasons not to project such a opposition onto the Proto-Lezgian level. If so, the correspondence sg. *ʔiƛʼe / pl. *ʔilχʷe between Archi and Tabasaran should be reflecting the Proto-Lezgian situation. In many other lects *ʔiƛʼe acquired both singular and plural functions.

    Some of the lects have lost the original lability. Thus, in modern Udi 'to kill' is the synchronic causative from 'to die'. In Archi, 'to kill' is euphemistically expressed by the verb 'to perform an action most typically associated with the given object in the current situation' (*ʔarčːa-), although the synchronic causatives from 'to die (sg.)' and 'to die (pl.)' are also used for 'to kill'. In Tsakhur, 'to die' and 'to kill' are distinguished by means of different fossilized prefixes.

    In Budukh, 'to kill' is expressed by *ʔatʼʷɨ-, whose original meaning was 'to cut' [NCED: 271]. Independently, the same root shifted to plural 'to die, kill' in Tsakhur.

    In Rutul dialects, the verb 'to strike, hit' (*yirχˤa- [NCED: 581]) can acquire the basic meaning 'to kill'.

    Consequently, we reconstruct two Proto-Lezgian labile verbs: *ʔiƛʼe 'to die / to kill' (sg.) and *ʔilχʷe 'to die / to kill' (pl.).

    Replacements: {'to cut' > 'to die / to kill'} (Tsakhur).

    Reconstruction shape: Correspondences seem regular up to the fossilized class prefix in Caucasian Albanian-Udi.

    Semantics and structure: Primary verbal root.

    Distinct from the more specific term tʼartʼa {тIартIа} 'short-haired dog' [Meylanova 1984: 137].

    Tsakhur-Kum Tsakhur: χʷeː [Kibrik & Kodzasov 1990: 73].

    In [Comrie & Khalilov 2010: 147], a second word for 'dog' is also quoted: qʼatʼay {кьатIай}. This is actually the adjective qʼätʼ-äy {кьаьтIаьй} 'tailless, short-tailed (said of animals)' [Dzhamalov & Semedov 2006: 162] (from qʼatʼ 'cut-off fragment').

    The same in other subdialects: Kurag, Duldug χuy 'dog' [Magometov 1970: 41; Shaumyan 1941: 191]; Tsirkhe χːuy 'dog' [Magometov 1970: 215 sentence 14; Shaumyan 1941: 191].

    The same in the Khanag subdialect: χːuy 'dog' [Uslar 1979: 958, 1006; Dirr 1905: 216, 242]. The same in other subdialects: Khyuryuk χːuy {ххуй}, Kumi χuy {хуй} 'dog' [Genko 2005: 166, 168].

    The same in the Khiv subdialect: χu {ху} 'dog' [Genko 2005: 166]. The same in Literary Tabasaran: χu {ху} 'dog' [Khanmagomedov & Shalbuzov 2001: 317].

    The same in Literary Lezgi: kicʼ {кицI} 'dog' [Talibov & Gadzhiev 1966: 160; Gadzhiev 1950: 791; Haspelmath 1993: 494, 518].

    The same in the Akhty dialect: Khlyut kicʼ 'dog' [Kibrik & Kodzasov 1990: 73].

    Superseded with loanwords in Archi (< Lak), Rutul and Keren Aghul (< Azerbaijani or Iranian).

    Replacements: {'puppy' > 'dog'} (Lezgi).

    Reconstruction shape: Correspondences seem regular.

    Semantics and structure: Primary substantive root. The oblique stem seems to have been *χːʷäy-rV-.

    In [Mobili 2010: 151-152] χupː, χupː-sun are also translated into Russian as 'to drink', 'to drink water', but their Azerbaijani glosses rather mean something like 'to suck up (water) slurping'.

    It must be noted that in [Kibrik et al. 1999: 878, 897] this root is quoted with a labialized uvular as ilʸ=oʁʷ-, cf. imperf. class 4 ilʸ-oʁʷ-a, fut. class 4 ilʸ-oʁʷ-as (also perf. class 4 ilʸ-o-d-ʁʷ-u). This labialization contradicts data from other sources and looks strange from the synchronic viewpoint: according to Tsakhur morphophonology, Cʷ normally dissimilates into C after labial o, u, b, p, w, cf., e.g., Mishlesh imperf. 2 hē-čʼʷan ← {h-o-y-y-čʼʷan}, but 3 hōčʼan < *hōčʼʷan ← {h-o-w-y-čʼʷan}, 1/4 hoyčʼan < *hoyčʼʷan ← {h-o-y-čʼʷan} 'to press' [Kibrik et al. 1999: 72, also 70], see the same statement for Mikik in [Kibrik & Kodzasov 1988: 41 fn. 116]. It might be suggested that Kibrik et al.'s transription ilʸoʁʷa- reflects some sporadic and irregular progressive assimilation, e.g. -oʁa- > -oʁʷa-?

    The same in the other subdialects: Tsirkhe, Duldug, Kurag uχ-a- 'to drink' [Shaumyan 1941: 145; Magometov 1970: 151].

    The same basic verb in the Khanag subdialect: wuqː-ˈ 'to drink' [Uslar 1979: 627, 1001]. It should be noted that in [Dirr 1905: 212, 237], this verb is transcribed as wuʁ- or uʁ- {у̨ҕ-} 'to drink' - the form either actually represents some specific Southern Tabasaran subdialect or the beginning of the phonetic process qː > ʁ in Khanag during the 2^nd half of the 19^th century between Uslar's and Dirr' records.

    The same in the Khyuryuk subdialect: wuqː-ˈ {вубкъув} 'to drink' [Genko 2005: 33].

    The same in the Khiv subdialect: uχ-ˈ [imperf., perf., inf.] / iχ [imv.] {ухуб} 'to drink' [Genko 2005: 157].

    In some subdialects another root is attested: Chara ˈuq- {ухъуб}, Sirtych üq-ˈ {юхъуб} 'to drink' [Genko 2005: 157, 198]. The same in Literary Tabasaran: uq-ˈ {убхъуб} 'to drink' [Khanmagomedov & Shalbuzov 2001: 291].

    The same in Literary Lezgi: qʷa- {хъун} 'to drink' [Talibov & Gadzhiev 1966: 354; Gadzhiev 1950: 539; Haspelmath 1993: 504, 518; Gyulmagomedov 2004, 2: 358].

    Three other roots, attested with the generic meaning 'to drink' in Lezgian lects, are clear innovations from the distributive point of view.

    In Archi, 'to drink' is expressed with *ʔVcʼV (~ -cʼː-). This root seems isolated within Lezgian, but external North Caucasian comparison seems to point to the original meaning 'to gulp (vel sim.)' [NCED: 1017].

    The root *ʔoχʷa has acquired the basic meaning 'to drink' in Aghul and some Southern Tabasaran subdialects (an areal isogloss); its original Proto-Lezgian meaning is likely to have been 'to gulp' (further to North Caucasian 'to suck') [NCED: 1027].

    Similarly, *ʔoqʷa- has become the basic root for 'to drink' in Lezgi and some Southern Tabasaran subdialects (an areal isogloss); its original Proto-Lezgian meaning was 'to suck' [NCED: 222].

    Replacements: {'to gulp' > 'to drink'} (Archi, Aghul, Southern Tabasaran), {'to suck' > 'to drink'} (Lezgi, Southern Tabasaran), {'to drink' > 'a gulp'} (Archi).

    Reconstruction shape: Correspondences seem regular, except for vocalic fluctuations.

    Semantics and structure: Primary verbal root.

    Distinct from qːurah {къурагь} 'dry (unknown application); dry (of weather)' [Meylanova 1984: 91].

    Tsakhur-Kum Tsakhur: ɢuru-n [Kibrik & Kodzasov 1990: 242].

    A second candidate is s=uqu-d or s=uqʼu-d 'dry', quoted in [Dirr 1912: 172, 201] with the example "dry firewood" (note that Dirr's notation rather speaks in favour of s=uqʼu-d with ejective -qʼ-). This is the participle from the verb 'to get dry', which is quoted in [Makhmudova 2001: 245] as s=uɢ- {сукъас} with -ɢ- (sic!). Its counterpart in the Borch-Khnov dialect sounds as su=q=uq- {сухъухъури} 'to get dry' [Ibragimov 1978: 268, 272], with -q- in the root. The exact phonetics, as well as the etymology of this Rutul verb is unclear.

    For the Khudig subdialect, the unclear form arka-d 'dry' is quoted in [Shaumyan 1941: 160]; apparently a corrupt spelling.

    Distinct from Tpig qːurah {къурагь} 'drought; arid' [Suleymanov 2003: 118], borrowed from Azerbaijani gurag 'drought; arid'.

    The same root in the Kumi subdialect: qː=ˈi-r-cːi {къирцци} 'dry' [Genko 2005: 101] (with the class infix -r- and the verbal perfective exponent qː=).

    Differently in other subdialects: Khanag qːurˈi 'dry' [Dirr 1905: 189, 244] (not found in [Uslar 1979]), Khyuryuk qːurˈi {къури} 'dry' [Genko 2005: 102].

    The same in Literary Tabasaran: ʁe=yˈe-r-cːu {гъеерццу} ~ ʁ=ˈe-r-cːu {гъерццу} 'dry', found in such examples as "withered flowers" [Khanmagomedov & Shalbuzov 2001: 155a], "dry leaves rustled in the breeze" [Khanmagomedov & Shalbuzov 2001: 349a] (with the fossilized class infix -r-). Distinct from Literary Tabasaran qːurˈah {къурагь} 'dry (of climate)' [Khanmagomedov & Shalbuzov 2001: 207], borrowed from Azerbaijani gurag 'drought; arid'.

    Three words for 'dry' are quoted for the Khiv subdialect, with no known difference: qːurˈu {къуру} 'dry' [Genko 2005: 102], ʁurˈu {гъуру} 'dry' [Genko 2005: 45] and ʁ=ˈe-r-cːu {гъерццу} 'dry' [Genko 2005: 42] (with the fossilized class infix -r-). The Khiv opposition qːurˈu / ʁurˈu is very suspicious (Genko's error?); it should be noted that the etymologically correct variant is qːurˈu.

    The same in Literary Lezgi: qʼurˈa-y {кьурай} 'dry' [Talibov & Gadzhiev 1966: 208; Gadzhiev 1950: 831; Haspelmath 1993: 504, 518], participle from the verb qʼurˈa- {кьурун} 'to dry (intrans.)' [Talibov & Gadzhiev 1966: 208]. A second literary term for 'dry' is the paronymous qʼurˈu {кьуру} [Talibov & Gadzhiev 1966: 208; Gadzhiev 1950: 831; Haspelmath 1993: 504] - an inherited form, whose vocalism may have been influenced by the Azerbaijani word guru 'dry'. The difference between qʼurˈa-y and qʼurˈu is unclear. Distinct from literary qʼurˈah {кьурагь} 'dry (of climate)' [Talibov & Gadzhiev 1966: 208; Haspelmath 1993: 504], ultimately borrowed from Azerbaijani gurag 'drought; arid', but influenced by the inherited qʼura-. For morphology cf. also the literary substantive qʼur {кьур} 'dryness' [Talibov & Gadzhiev 1966: 208].

    Morphologically different in the Akhty dialect: Khlyut qːur-qːˈur 'dry' [Kibrik & Kodzasov 1990: 242]; this is a reduplicated formation from the same Lezgi root.

    In South Lezgian (Kryts, Budukh), 'dry' is a participle from the etymologically obscure verbal root *saʔV- or *aʔV-, unattested elsewhere [LEDb: #251].

    In many Tabasaran subdialects, the original participle has been superseded with the participle from another verb for 'to dry (intrans.)': ecː-, see notes on Common Tabasaran.

    Reconstruction shape: Correspondences seem regular up to Azerbaijani influenced forms.

    Semantics and structure: Primary verbal root 'to be dry, get dry'.

    Tsakhur-Kum Tsakhur: kʼɨrɨ [Kibrik & Kodzasov 1990: 12].

    The same in the Khanag subdialect: ew 'ear' [Uslar 1979: 677, 1009; Dirr 1905: 166, 245]. The same in the Kumi subdialect: iw {ив} 'ear' [Genko 2005: 73].

    The same in other subdialects: Khiv eb {эб}, Chara ib {иб} 'ear' [Genko 2005: 73, 193]. The same in Literary Tabasaran: ib {иб} 'ear' [Khanmagomedov & Shalbuzov 2001: 172].

    The same in Literary Lezgi: yab [abs.] / yapː-ˈu- [obl.] {яб} 'ear' [Talibov & Gadzhiev 1966: 399; Gadzhiev 1950: 897; Haspelmath 1993: 493, 518].

    The same in the Akhty dialect: Khlyut eb [abs.] / epː-ˈeni- [obl.] 'ear' [Kibrik & Kodzasov 1990: 12].

    In many languages, synchronic forms are modified with fossilized plural suffixes.

    Reconstruction shape: Correspondences seem regular.

    Semantics and structure: Primary substantive root.

    Distinct from kːul {кIул} 'earth' [Gukasyan 1974: 146; Kibrik & Kodzasov 1990: 199; Comrie & Khalilov 2010: 29; Mobili 2010: 178].

    All sources quote kːul and očˤal as synonyms for 'earth, soil', except for [Kibrik & Kodzasov 1990]. The latter dictionary gives only kːul as the translation for 'earth, soil' (with polysemy: 'earth, soil / territory, plot'). On the contrary, text evidence from [Ganenkov et al. 2008: 240 (8), 254 (34, 41)] confirms očˤal as the most generic and common word for 'earth, soil', as well as 'land': "There is a place named so-and-so ... which existed on the Nidzh land (očˤal) as early as 2400-2500 years ago", "The soil (očˤal) cleans it (just prepared vodka) of its odours ... We take it (a buried jug with vodka) out from the ground (očˤal)".

    According to Dm. Ganenkov's p.c., in Nidzh kːul indeed means 'soil', but this word is very rare in the collected corpus. The default expression for all the meanings is currently očˤal.

    Distinct from the term očˤal {очъал} 'earth' [Gukasyan 1974: 185; Fähnrich 1999: 25; Dirr 1903: 19, 26; Schiefner 1863: 78; Schulze 2001: 303; Starchevskiy 1891: 499]. In [Gukasyan 1974: 185] this is quoted as očːˤal {оч́ал} (apparently an error, see notes on Nidzh Udi); in [Fähnrich 1999: 25] the word is quoted as simply očal - apparently for očˤal; also quoted as očˤal in [Schulze 2001], despite the fact that Bežanovs' {ч̆} may cover čː, čˤ and čːˤ; graphical {c} in [Schiefner 1863], {ч} in [Dirr 1903] and {ц} in [Starchevskiy 1891] can hardly clarify the phonetical nature of the sibilant. It should be noted that normally this term is graphically opposed to očːˤi 'dirt, mud' (e.g., [Fähnrich 1999: 25; Dirr 1903: 18; Schiefner 1863: 78]).

    Vartashen očˤal means 'earth, ground', not 'soil', cf. some contexts: "to the ground", "he saw, as a mouse got out of the ground", "the earth quakes" (= 'earthquake') [Dirr 1903: 19, 64], "the silver ingot melted and spilt on the ground" [Schiefner 1863: 68]. In [Bezhanov & Bezhanov 1902] {оч̆ал} normally means 'earth' as opposed to 'heaven' ("I thank you, Father, Lord of heaven and earth", etc.) or 'ground' (cf. above Mt. 13.5, where two terms are opposed).

    Distinct from the adverbs qʼˈa-tːu 'on the ground', qʼˈa-tːu-k 'to the ground' [Chumakina et al. 2007; Kibrik et al. 1977b: 302] (from *qʼa- 'down' [NCED: 616] with the adjective suffix -tːu) and dunˈil with polysemy 'sky / world / life' [Chumakina et al. 2007; Kibrik et al. 1977b: 224] (the latter was borrowed from Avar dunˈiyal 'world / life / weather', ultimately from Arabic dunyaː 'earth, world').

    A second term for 'earth, soil' is qʼum {кьум} [Meylanova 1984: 96]. Semantic or pragmatic details are unknown.

    A second candidate is the more marginal term torpaχ [Kibrik et al. 1999: 888, 894; Ibragimov & Nurmamedov 2010: 342], which means both 'earth, plot' (cf. Ibragimov & Nurmamedov's example "ground landlord") and 'earth, soil' (cf. torpaχ-šunas 'soil scientist, pedologist' [Ibragimov & Nurmamedov 2010: 342]). Borrowed from Azerbaijani torpag 'earth (soil, plot, land, ground)'.

    Distinct from ǯilʸ {джилʹ} 'earth floor' [Kibrik et al. 1999: 890; [Ibragimov & Nurmamedov 2010: 160].

    Distinct from dʸunʸye 'earth, world' [Kibrik et al. 1999: 872, 894; Ibragimov & Nurmamedov 2010: 158], borrowed from Azerbaijani dünya 'earth, world' (ultimately from Arabic dunyaː 'earth, world').

    The old term nʸaqʼʷ [abs.] / nʸuqʼ- [obl.] is retained in the meaning 'grave, tomb' [Kibrik et al. 1999: 883]; in [Ibragimov & Nurmamedov 2010: 277] this is quoted with polysemy: 'clay, earth / grave, tomb'.

    In [Comrie & Khalilov 2010: 29], Literary Tsakhur and Gelmets Tsakhur forms for 'earth' are interchanged: naqʼʷ (an error for nʸaqʼʷ) is quoted for the literary language, čʼiye - for Gelmets.

    Tsakhur-Kum Tsakhur: nʸaqʼʷ [Kibrik & Kodzasov 1990: 199]. Polysemy: 'earth / grave, tomb'.

    A second candidate is čʼiye [Dirr 1913: 215, 226], but this rather means 'earth, ground', cf. an example: "Underground there is a jug with oil (= badger)" (a riddle) [Dirr 1913: 119].

    Distinct from ǯil-a 'floor' (i.e. 'earth floor'?) [Dirr 1913: 158].

    In [Comrie & Khalilov 2010: 29], a second term for 'earth' is quoted: čʼir {чIир}, which actually means 'grass-covered place' [Dirr 1912: 181] (cf. its Ixrek Rutul counterpart čʼir 'pasture, common' [Dzhamalov & Semedov 2006: 297]).

    Distinct from ǯil ‘earth floor’ [Ibragimov 1978: 117].

    Distinct from ǯil {джил} ‘earth floor’ [Dzhamalov & Semedov 2006: 108].

    Distinct from the more specific Burshag term neqʼʷ with polysemy: 'black soil / grave, tomb' in [Kibrik & Kodzasov 1990: 190, 199] (the two meanings are formally discriminated in the ergative form).

    Distinct from Richa rug 'dust' [Kibrik & Kodzasov 1990: 200].

    Distinct from ǯil 'earth, ground; floor' [Dirr 1907: 115].

    A second candidate is Tpig ǯil, but the gloss 'earth, soil' in [Suleymanov 2003: 81] seems an inaccuracy; one would expect the meaning 'ground'.

    Distinct from Tpig qʼuramat 'land (opposed to water)' [Suleymanov 2003: 123].

    The same in the Khanag subdialect: rug 'earth, soil' [Uslar 1979: 898, 994; Dirr 1905: 203, 229] (specified as 'soil' by both authors). Distinct from Khanag yišʷ 'place; land, region' [Uslar 1979: 755; Dirr 1905: 180].

    The same in the Khyuryuk subdialect: rug {руг} 'earth' [Genko 2005: 133], distinct from yišʷ {йишв} 'place; land, region' [Genko 2005: 81].

    The same in the Khiv subdialect: rug ~ rugʷ {руг(в)} 'earth, soil' [Genko 2005: 133] (specified by Genko as 'soil'; the variant with gʷ is etymologically unclear). Distinct from Khiv ǯil {жжил} 'ground; floor; field' [Genko 2005: 68] and from Khiv yišʷ {йишв} 'place; land, region' [Genko 2005: 81].

    Two Literary Tabasaran terms for 'soil' are known: rug {руг} with polysemy: 'soil / dust, litter' [Khanmagomedov & Shalbuzov 2001: 262] and ǯil {жил} with polysemy: 'soil / ground / land / Earth' [Khanmagomedov & Shalbuzov 2001: 162]. Distinct from Literary Tabasaran yišʷ {йишв} 'place' [Khanmagomedov & Shalbuzov 2001: 186].

    Distinct from Gyune čːil [Uslar 1896: 594, 613], which means 'earth, ground, earth floor' rather than specific 'soil', according to Uslar's examples: "The earth trembled (= earthquake)", "The earth is covered with grass", "He has the earth as his mattress, the sky for cover" [Uslar 1896: 594], "I have smoothed the earth" [Uslar 1896: 411], but also "The earth of this village is fertile" [Uslar 1896: 594].

    The expected Gyune word **naqʼʷ is not documented by Uslar.

    Differently in Literary Lezgi: naqʼʷ {накьв} 'earth, soil' [Talibov & Gadzhiev 1966: 250; Gadzhiev 1950: 245; Haspelmath 1993: 499, 518]. A second candidate is čːil {чил, ччил} with polysemy: 'ground / soil' [Talibov & Gadzhiev 1966: 370; Gadzhiev 1950: 245; Haspelmath 1993: 484, 518]. A third term is rug [abs.] / rukːʷ-ˈadi- [obl.] {руг} with polysemy: 'dust / soil / litter' [Talibov & Gadzhiev 1966: 280] (only as 'dust' in [Haspelmath 1993: 505]). The pragmatic or semantic difference between three words for 'soil' is unclear, although naqʼʷ seems to be the most basic term in this meaning.

    In the Akhty dialect: Khlyut naqʼʷ 'earth, soil' [Kibrik & Kodzasov 1990: 199]. Distinct from Khlyut rug [abs.] / rukː-ˈadi- [obl.] 'dust' [Kibrik & Kodzasov 1990: 200] and čːil 'earth floor' [Kibrik & Kodzasov 1990: 176].

    The external comparison strongly suggests that naqʼʷ was the Proto-Lezgi term for 'earth, soil'. The word rug originally meant 'dust', but has acquired the additional meaning 'soil' - an areal isogloss shared with Aghul and Tabasaran (q.v.). Lezgi čːil denoted 'ground', but currently tends to broaden into the meaning 'soil' under the influence of similar polysemy in Azerbaijani and Russian.

    In the Udi branch, however, *näqʼʷ has been lost, and 'earth, soil' is now expressed with the etymologically obscure form kːul. Furthermore, in Nidzh Udi, the word očˤal, whose Proto-Udi meaning was 'earth, ground', has acquired the second meaning 'earth, soil'.

    In Mishlesh Tsakhur, *näqʼʷ was superseded with *čʼura (~ -o-) [NCED: 555], accompanied with the shift 'clearing, uncultivated land' > 'earth, ground' > 'earth, soil'.

    In East Lezgian, *näqʼʷ tends to be superseded with *rukː, whose original meaning was 'dust' [NCED: 603]. This is not a Proto-East Lezgian replacement, but a late areal isogloss (probably Tabasaran-induced): *rukː became the default root for 'earth, soil' in Tabasaran, many Aghul dialects (Koshan, Gequn, Fite, proper Aghul) and some Lezgi dialects (Gyune).

    Reconstruction of the Proto-Lezgian term for 'earth, ground' is less obvious.

    The root *čːil [NCED: 342] can be reconstructed with the Proto-Nuclear Lezgian meaning 'earth floor' (thus Tsakhur, Rutul, Aghul, Tabasaran, Lezgi). In East Lezgian (Aghul, Tabasaran, Lezgi), however, this word also denotes 'earth, ground', as opposed to various terms for 'soil'. External North Caucasian comparison suggests that *čːil could theoretically be reconstructed as the Proto-Lezgian root for 'earth, ground'.

    On the other hand, in Proto-Caucasian Albanian-Udi, the word for 'earth, ground' originates from Lezgian *yo(m)čV (~ ʔ-) [NCED: 684]: Proto-Udi očˤa-l 'earth, ground', Caucasian Albanian ašˤa-l 'earth, ground' (the correspondence Udi čˤ ~ CA šˤ is unique, but may represent different treatments of the Proto-Lezgian consonant cluster). In the rest of Lezgian, the root got lost. Lezgian *yo(m)čV (~ ʔ-) possesses good external comparanda with the semantics of 'earth' [NCED: 684].

    Thus, the easiest solution is to reconstruct *yo(m)čV (~ ʔ-) with the Proto-Lezgian meaning 'earth, ground' and *čːil with the Proto-Lezgian meaning 'earth floor'.

    If so, *yo(m)čV 'earth, ground' must have been lost in Archi (where there is only a new adverbial formation 'on the ground') and in Nuclear Lezgian. Various terms for 'earth, ground' are attested in Nuclear Lezgian: in Proto-Tsakhur, *čʼura [NCED: 555] (with the semantic shift 'clearing, uncultivated land' > 'earth, ground'); in South Lezgian (Kryts, Budukh), etymologically obscure qʼum; in East Lezgian (Aghul, Tabasaran, Lezgi), *čːil (with the shift 'earth floor' > 'earth, ground').

    It should be noted that in some East Lezgian lects (Literary Tabasaran, Literary Lezgi), *čːil has further shifted from 'earth, ground' > 'earth, soil'.

    Some additional meaning shifts to be noted are: *näqʼʷ 'earth, soil' > 'land' in Archi; *näqʼʷ 'earth, soil' > 'grave, tomb' in Tsakhur and Aghul (Koshan, Gequn, Fite, proper Aghul); *näqʼʷ 'earth, soil' > 'earth, ground / land' in Rutul; *näqʼʷ 'earth, soil' > 'clay' in Ixrek Rutul.

    In all the aforementioned cases, the bidirectional shifts between 'earth, soil' and 'earth, ground' can be explained as influenced by Azerbaijani or Russian polysemy.

    Replacements: {'earth, ground' > 'earth, soil'} (Nidzh Udi, Literary Tabasaran, Literary Lezgi), {'clearing, uncultivated land' > 'earth, ground' > 'earth, soil'} (Mishlesh Tsakhur), {'dust' > 'earth, soil'} (Tabasaran, Aghul, Lezgi), {'earth, soil' > 'grave, tomb'} (Tsakhur and Aghul), {'earth, soil' > 'earth, ground / land'} (Rutul), {'earth, soil' > 'land'} (Archi), {'earth, soil' > 'clay'} (Ixrek Rutul), {'earth floor' > 'earth, ground'} (Aghul, Tabasaran, Lezgi).

    Reconstruction shape: Correspondences seem regular.

    Semantics and structure: Primary substantive root. The oblique stem is *niqʼʷɨ-.

    Synchronically distinct from the suppletive verb 'to eat (subj. = animal)': =kˈukin- [imperf., imv.] / =kˈun- [inf., perf.], see [Kibrik et al. 1977a 2: 72; Kibrik et al. 1977a 3: 242; Kibrik et al. 1977b: 259; Chumakina et al. 2007] (perfective =kˈunni probably < kun-tːˈe, for the nasal sandhi see [Kibrik et al. 1977a 1: 304; Kibrik et al. 1977a 2: 69]). As correctly proposed in [Kibrik et al. 1977a 2: 73 fn. 51] and [NCED: 207], all the stems of both verbs originate from the "Proto-Archi" root kʷan (or rather kʷan / kʷen).

    Distinct from lˈah-bo- 'to want to eat smth.' [Chumakina et al. 2007; Kibrik et al. 1977b: 271] (complex verb, formed with the suppletive light verb -bo- 'to say').

    In [Comrie & Khalilov 2010: 271] 'to eat' is glossed as čʼatʼ yixǝr, lit. 'flat cake, scone' + 'to be, exist' apparently due to an inaccurate Russian question to a Budukh informant: Russian collocation of the type есть хлеб means both 'to eat bread' and 'there is bread'.

    Synchronically, a suppletive paradigm: class=ʔil-ä-r- [imperf.] / liʔ-i-r [perf.] / class=il-ä [imv.]. In [Makhmudova 2001: 70, 94, 98, 114, 147, 158, 209, 247], the perfective and imperative stems are consistently quoted as ul-e- - contracted forms of class 3 (*w=)? In [Comrie & Khalilov 2010: 271], the perfective stem is quoted as luʔˤ-u- with unclear pharyngealization.

    Distinct from y=ɨx- 'to eat (subj. = animal)' [Makhmudova 2001: 94] (not found in other Rutul sources).

    Synchronically, a suppletive paradigm: il-e-r- [imperf.] / liʔ-i-r [perf.] / il-e [imv.].

    Synchronically, a suppletive paradigm: class=il-e-r- [imperf.] / liʔ-i-r [perf.] / class=il-i [imv.].

    The same verb in other subdialects: Kurag, Tsirkhe ʡutʼ-a- 'to eat' [Magometov 1970: 142, 206 sentence 11, 215 sentence 22]; Duldug Hutʼ-a- 'to eat' [Shaumyan 1941: 181]. It should be noted that for the Tsirkhe subdialect the infinitive form Hal-a-s 'to eat' is quoted in [Shaumyan 1941: 181].

    Similar paradigm in the Khanag subdialect: ü-l-d- [imperf.] / uˤpʼ- [perf., imv., inf.: class 2 sg.] / uˤtʼ- [perf., imv., inf.: class pl.] 'to eat' [Uslar 1979: 938; Dirr 1905: 211, 247]. Some interesting Khanag aorist forms (class 2) are also quoted by Dirr: el-, ül-, where medial -l- is the imperfective prefix.

    The same in the Khyuryuk subdialect: ˈü-l-d- [imperf.] / uˤpʼ-, uˤtʼ- [perf., inf.] {уьпIуьв} 'to eat' [Genko 2005: 159].

    The same in other subdialects: Khiv, Chara ipʼ-ˈ {ипIуб} 'to eat' [Genko 2005: 76] (regular paradigm).

    Suppletive paradigm in Literary Tabasaran: ipʼ-ˈ [sg. obj.] / itʼ-ˈ [pl. obj.] {ипIуб, итIуб} 'to eat' [Khanmagomedov & Shalbuzov 2001: 179].

    The same in Literary Lezgi: ne- [imperf.] / tʼü- [perf.] / neʔ [imv.] {тIуьн} 'to eat' [Talibov & Gadzhiev 1966: 326; Gadzhiev 1950: 194; Haspelmath 1993: 509, 518; Gyulmagomedov 2004, 2: 289].

    In Proto-Nuclear Lezgian, *ʔikʷVn- was superseded by the verbal root *ʔiʔʷäl- [NCED: 625], whose original meaning is not entirely clear (cf. its Archi cognate with the meaning 'to want to eat smth.', and further North Caucasian comparanda which mean 'to bite' or 'to feed on smth.').

    Additionally, in Tsakhur, *ʔiʔʷäl- was superseded by *ʔiɬʷV(n)- [NCED: 516], which originally meant 'to graze, pasture' in Proto-Lezgian.

    Replacements: {'to graze, pasture' > 'to eat'} (Tsakhur).

    Reconstruction shape: Correspondences seem regular.

    Semantics and structure: Primary verbal root with several Ablaut grades.

    Dictinct from Vartashen qːˤaqːapun {къаъкъаъпун} 'fried eggs' [Gukasyan 1974: 156]; according to [NCED: 932], qːˤaqːa-pun with an unclear suffix -pun, but such an analysis is uncertain.

    Tsakhur-Kum Tsakhur: qʼuqʼ [Kibrik & Kodzasov 1990: 123].

    Differently in the Khanag subdialect: pelˈinʒuw ~ pelunʒuw ~ penunʒuw 'egg' [Uslar 1979: 874, 1010; Dirr 1905: 199, 247]. According to [Uslar 1979: 875], the non-compressed expression peʔlˈin liʒˈuw 'egg' was occasionally used as late as the 2^nd half of the 19^th century - literally 'hen's white' (peʔ, gen. peʔ-eli-n ~ peʔ-li-n 'hen'; liʒi, class 2 liʒu-w 'white' q.v.).

    The same in other subdialects: Khyuryuk pelinʒuw {пелинззув} ~ peʔlˈuʒuw {пеълуззув} 'egg' [Genko 2005: 128, 129] (the first variant is erroneously quoted as peliʒuw {пелиззув} in the head of the entry), Khapil peluˤʒuw {пелюззюв} 'egg' [Genko 2005: 128].

    Superseded with the loanword in the Kumi and Chuvek subdialects: murtˈa {мурта} 'egg' [Genko 2005: 123].

    The same loanword in the Tinit subdialects: yumurtˈa {юмурта} 'egg' [Genko 2005: 197], and in Literary Tabasaran: murtˈa {мурта} 'egg' [Khanmagomedov & Shalbuzov 2001: 234].

    An inherited word in the Khiv subdialect: gugˈu {гугу} 'egg' [Genko 2005: 37].

    The same in Literary Lezgi: kːakːˈa {кака} 'egg' [Talibov & Gadzhiev 1966: 148; Gadzhiev 1950: 962; Haspelmath 1993: 494, 519].

    The same in the Akhty dialect: Khlyut kːakːˈa 'egg' [Kibrik & Kodzasov 1990: 123].

    The first one is reconstructed as *qːVlVqː in [NCED: 906]. This stem means 'egg' in Udi, on the one hand, and Rutul and Aghul, on the other. In the rest of Lezgian, the root has been lost.

    The second one is reconstructed as *qʼoloqʼ in [NCED: 932]. This stem means 'egg' in Tsakhur (qʼuqʼ), 'testiculus' in Budukh, Rutul and Tabasaran (these three forms have only been found in [NCED]), 'fried eggs' in Vartashen Udi.

    In three languages, both of the roots have survived. The reflexes are opposed as follows:

    Udi: qokːla ~ qoqːla ~ qːoqla ~ χoχla ~ qoqla ~ qːoqːla ~ kokla ~ koχla 'egg' / qːˤaqːapun (qːˤaqːa-pun?) 'fried eggs'.

    Rutul (Khnyukh): ʁɨlɨʁ 'egg' / qʼɨlɨqʼ 'testiculus'.

    Tabasaran (Dyubek): ʁɨlɨʁ 'egg' / qʼurqʼ-ur 'testiculus' (with the fossilized plural suffix).

    It should be noted, however, that Vartashen Udi qːˤaqːapun is morphologically obscure (-pun is a unique suffix) and too irregular phonetically (normally Lezgian *qʼ yields a zero reflex in Udi). It seems better to separate the Udi form from these roots: we prefer to treat qːˤaqːapun as a word of unknown origin.

    If so, the descendants of the hypothetical *qʼoloqʼ mean 'testiculus' in all three Nuclear Lezgian subbranches, but 'egg' in Tsakhur. The semantic derivation 'testiculus' > 'egg' is extremely rare cross-linguistically; thus, Tsakhur qʼuqʼ would rather seem to originate from *qːVlVqː 'egg', but it must have been influenced by *qʼoloqʼ. The second solution is to unite all the aforementioned forms (excepting the Udi 'fried eggs') under one proto-root *QVlVQ 'egg' with very irregular reflexes. If so, in Proto-Nuclear Lezgian, *QVlVQ divided into two phonetic shapes - one retained the meaning 'egg', the other acquired the meaning 'testiculus' (the shift 'egg' > 'testiculus' is normal).

    It is also likely that the discussed words for 'egg' have been influenced by various onomatopoeic forms for hen cackling, cf., e.g., Budukh qʼaɢɨldamˈi 'cackling, clucking' [Meylanova 1984: 94], etc.

    In Proto-Tabasaran and Lezgi, 'egg' is expressed by *kːakːay [NCED: 429], not observed in other languages. The primary meaning of this root is unclear. It cannot be posited as the Proto-East Lezgian term for 'egg', since the Aghul language retains *qːVlVqː for this meaning. In fact, however, both Khiv Tabasaran gugˈu and Lezgi kːakːˈa may represent independent introductions of onomatopeic nature.

    In Northern Tabasaran, two recent compounds are attested with the meaning 'egg', both with *paʡ 'hen' [NCED: 865] as the first element (one of them literally means 'white of hen').

    In South Lezgian (Kryts, Budukh), the obscure forms kusuntʼ ~ kusχuntʼ ~ kusxud 'egg' are attested. Their first element kus- also looks similar to the word for 'hen'.

    Superseded with loanwords in Archi (< Lak) and some Tabasaran dialects (< Azerbaijani).

    Replacements: {'white of hen' > 'egg'} (Northern Tabasaran), {'X of hen' > 'egg'} (Northern Tabasaran).

    Reconstruction shape: Correspondences are rather irregular, due to various assimilative/dissimilative processes and vowel syncope.

    Semantics and structure: Primary substantive root, maybe of onomatopoeic origin.

    Tsakhur-Kum Tsakhur: ulʸ [Kibrik & Kodzasov 1990: 11].

    The same in the Khanag subdialect: ul 'eye' [Uslar 1979: 927, 991; Dirr 1905: 209, 226]. The same in the Khyuryuk subdialect: ul {ул} 'eye' [Genko 2005: 152].

    The same in the Khiv subdialect: ul {ул} 'eye' [Genko 2005: 152]. The same in Literary Tabasaran: ul {ул} 'eye' [Khanmagomedov & Shalbuzov 2001: 294].

    The same in Literary Lezgi: wil {вил} with polysemy: 'eye / window framing / mesh (of net) / pore' [Talibov & Gadzhiev 1966: 75; Gadzhiev 1950: 141; Haspelmath 1993: 511, 519].

    The same in the Akhty dialect: Khlyut ul 'eye' [Kibrik & Kodzasov 1990: 11].

    Reconstruction shape: Correspondences seem quite regular.

    Semantics and structure: Primary substantive root. The oblique stem is *ʔʷile-.

    Distinct from the specific term pi 'goat's fat' [Dzheiranishvili 1971: 217, 255; Fähnrich 1999: 26] (glossed simply as 'fat' in [Schiefner 1863: 98] and [Starchevskiy 1891: 499]), which is borrowed from Azerbaijani piy 'fat, lard' or directly from Persian piːh 'fat'.

    Distinct from zeyt ~ zet 'vegetable oil', ʒet 'olive oil' [Schiefner 1863: 90, 93] (an Arabic loanword, see notes on the Nidzh dialect).

    In [Comrie & Khalilov 2010: 310] the form šːelekul {щелекул} is also quoted for 'fat, grease' - a corrupted spelling of the masdar in -kul from the stative verb šːeˤlˈe 'to be fat (normally of food)' [Kibrik et al. 1977b: 320; Chumakina et al. 2007].

    Distinct from tennˈe 'layer of fat on soup; ointment' [Kibrik et al. 1977b: 322; Chumakina et al. 2007] (quoted for 'dissolved grease' in [Kibrik & Kodzasov 1990: 121]).

    Distinct from yɨχˤ 'dissolved grease (Russian: сало)' [Ibragimov & Nurmamedov 2010: 184] and čoz ~ čaz 'visceral fat, suet' [Ibragimov & Nurmamedov 2010: 399].

    Tsakhur-Kum Tsakhur: maʔ [Kibrik & Kodzasov 1990: 122]. Distinct from yɨχˤ 'dissolved grease' [Kibrik & Kodzasov 1990: 122].

    Distinct from yɨχˤ 'dissolved grease' and čoz 'visceral fat, suet' [Kibrik & Kodzasov 1990: 122].

    Distinct from yɨχˤ 'fat on meat' (an error for 'dissolved grease'?) and čoz 'visceral fat, suet' [Kibrik & Kodzasov 1990: 122].

    A second candidate is yɨχˤ {йыIх}, which is glossed as generic 'rendered fat, fat (Russian: сало, жир)' in [Dirr 1912: 148, 190], but specifically as 'dissolved grease (Russian: топленое сало)' in [Ibragimov 1978: 224]. In [Comrie & Khalilov 2010: 310], this word is erroneously quoted in the generic meaning 'fat' as yɨχ {йых}.

    A second term, incorrectly quoted in the generic meaning 'fat' in [Comrie & Khalilov 2010: 310], is kuk-dɨ yak - actually a normal attributive construction with the direct meaning 'fatty meat' [Dirr 1912: 151].

    In [Comrie & Khalilov 2010: 310], qʼɨš is quoted in the generic meaning 'fat'.

    Distinct from mutʼulay 'dissolved grease' [Kibrik & Kodzasov 1990: 121] of Arabic origin.

    Distinct from mutʼula 'dissolved grease' of Arabic origin and inherited yaχ 'fat on meat' [Kibrik & Kodzasov 1990: 122].

    In the Usug subdialect superseded with yaχ 'fat (in general)' [Shaumyan 1941: 147]. The old term is retained as maw 'visceral fat, suet' [Shaumyan 1941: 150].

    Distinct from mutʼulay 'dissolved grease' of Arabic origin and inherited yaχʷ 'fat on meat' [Kibrik & Kodzasov 1990: 122]. Labialized uvular in abs. yaχʷ is apparently secondary, having appeared under the influence of the ergative form yaχ-u (as in some similar Burkikhan cases, e.g., *kerk > kerkʷ, erg. kerk-u 'nail' q.v.).

    Distinct from darc 'dissolved grease' [Kibrik & Kodzasov 1990: 122].

    Distinct from Tpig maw 'visceral fat, suet' [Suleymanov 2003: 130; Shaumyan 1941: 150].

    The same in the Khanag subdialect: χˤul 'fat' [Uslar 1979: 955, 993; Dirr 1905: 215, 229]. The same in the Khyuryuk subdialect: χˤul {хюл} 'fat' [Genko 2005: 167]. Distinct from Khanag, Khyuryuk čːem {ччем} 'melted butter, dissolved grease' [Uslar 1979: 970; Dirr 1905: 217; Genko 2005: 182].

    The same in the Khiv subdialect: χˤul {хюл} 'fat' [Genko 2005: 167]; distinct from Khiv čːem {ччем} 'melted butter, dissolved grease' [Genko 2005: 182].

    The same in Literary Tabasaran: χˤul {хюл} 'fat' [Khanmagomedov & Shalbuzov 2001: 319]; distinct from Literary Tabasaran darcː {дарцц} 'visceral fat, suet' [Khanmagomedov & Shalbuzov 2001: 142], χucʼˈur {хуцIур} 'visceral fat, suet' [Khanmagomedov & Shalbuzov 2001: 318], čːim {ччим} 'animal oil, butter; vegetable oil' [Khanmagomedov & Shalbuzov 2001: 340].

    In Literary Lezgi the same two terms compete with each other: the inherited maqʼ {макь} 'fat (not dissolved)' [Talibov & Gadzhiev 1966: 229; Gadzhiev 1950: 201] (glossed as 'fat' in [Haspelmath 1993: 498, 519]) and the borrowed pi {пи} 'animal fat' [Talibov & Gadzhiev 1966: 264; Gadzhiev 1950: 201] (glossed as 'fat' in [Haspelmath 1993: 500, 519]).

    Distinct from several specific Literary Lezgi terms: ʁerˈi {гъери} 'butter, animal oil, fat (i.e. dissolved fat?)' [Talibov & Gadzhiev 1966: 93; Gadzhiev 1950: 201] (glossed as 'fat, grease' in [Haspelmath 1993: 490, 519]), čʼem {чIем} - a synonym of ʁeri [Talibov & Gadzhiev 1966: 378; Gadzhiev 1950: 201]; qucʼur {хъуцIур} 'visceral fat, suet' [Talibov & Gadzhiev 1966: 354; Gadzhiev 1950: 201]; yaʁ {ягъ} 'machine oil, lube' [Talibov & Gadzhiev 1966: 400] (glossed as 'oil, grease' in [Haspelmath 1993: 493, 519]; borrowed from Azerbaijani yaɣ 'fat (in general)').

    For the Akhty dialect two terms are quoted as synonyms for 'fat (in general)': Khlyut maʔ and qʼɨš [Kibrik & Kodzasov 1990: 122]. Distinct from Khlyut qʼacʼˈir 'dissolved grease' [Kibrik & Kodzasov 1990: 122].

    The form maqʼ / maʔ must be posited as the Proto-Lezgi term for 'fat'.

    In Udi, *maʔˤ has been lost, superseded by *čʼːäˤm [NCED: 624], whose original meaning was 'butter', as proven by both the Nuclear Lezgian and external North Caucasian cognates, as well as the synchronic Udi polysemy 'fat / butter'.

    In Keren Aghul (Usug) and Aghul proper, *maʔˤ shifted to the specific meaning 'visceral fat, suet', superseded as the generic term by *yimχː [NCED: 948]. The original Proto-Lezgian meaning of *yimχː is likely to have been 'butterfat' vel sim. (cf. its Archi and South Lezgian cognates with the meanings 'butter' and 'milk'), but for Proto-Aghul *yimχː can be reconstructed with the specific meaning 'fat on meat'.

    In Tabasaran and, secondarily, in the neighboring Koshan Aghul dialect, *maʔˤ has been lost, superseded by *χˤul (χːˤ-, -o-, -lː) [NCED: 1081]. The root *χˤul apparently denoted a specific kind of fat in Proto-Lezgian, but exact semantic reconstruction is impossible (Lezgian *χˤul seems unattested outside Tabasaran-Aghul area, but external comparison points to the semantics of 'fat').

    Replacements: {'butter' > 'fat'} (Udi), {'butterfat' > 'fat on meat' > 'fat'} (Keren Aghul).

    Reconstruction shape: Correspondences seem regular.

    Semantics and structure: Primary substantive root.

    A second Vartashen term for 'feather', which is currently used in the Zinobiani (Oktomberi) village, is posposkːal [Fähnrich 1999: 26], confirmed by Yu. Lander's field records on 2011. In [Gukasyan 1974: 189] and [Mobili 2010: 235], however, this word is explained as a kind of wild plant (Azerbaijani: xumxuma, Russian: перчий).

    In [Schulze 2001: 312] the word qːänäd 'wing; feather' is quoted (an Azerbaijani loanword, see notes on Nidzh Udi), although in Bežanovs' text it is attested only with the meaning 'wing'.

    In [Comrie & Khalilov 2010: 216] the term qum {хъум} is also quoted for 'feather', but its actual meaning is 'down' [Kibrik et al. 1977b: 298; Chumakina et al. 2007].

    Distinct from pisi {писи}, which is glossed in [Meylanova 1984: 119, 230] as 'feather, small feather' (repeated in [Comrie & Khalilov 2010: 216] as a basic term for 'feather'), but Meylanova's examples rather suggest the specific meaning 'down': "downy pillow", "down appears on chickens".

    Tsakhur-Kum, Mukhakh-Sabunchi & Suvagil Tsakhur: kuk-ra [Kibrik & Kodzasov 1990: 45; Ibragimov 1990: 162, 175]. Final -ra is a fossilized plural marker.

    A second term is pʼeru {пIеру}, glossed simply as 'feather' in [Dzhamalov & Semedov 2006: 211, 372], but we suspect that pʼeru rather means 'feather pen', borrowed from Russian pʸerˈo 'feather (anatomic); feather pen'.

    In [Comrie & Khalilov 2010: 216], 'feather' is apparently erroneously glossed as maɢˤaɢ {макъаIкъ}.

    In [Suleymanov 2003: 136], two additional Koshan words for 'feather' are also quoted: murcʼ and cʼurcʼum (both without exact dialectal prevenance).

    In the Usug subdialect: purcʼ 'feather' [Shaumyan 1941: 152].

    Differently in the Khanag subdialect: χil 'feather' [Uslar 1979: 951, 1001] (missing from [Dirr 1905]); distinct from Khanag zik 'down' [Uslar 1979: 701, 1003]. It should be noted that in [Dirr 1905: 171, 237] zik is glossed as 'feather' - apparently an inaccuracy.

    The same in the Khyuryuk subdialect: χil {хил} 'feather' [Genko 2005: 165]; distinct from Khyuryuk zik {зик} 'down' [Genko 2005: 71].

    The same in the Tinit subdialect: zik {зик} 'feather; down' [Genko 2005: 71].

    Differently in the Khiv subdialect: cʼupˈur {цIупур} 'feather' [Genko 2005: 178; Khanmagomedov & Shalbuzov 2001: 334]; distinct from Khiv zük ~ ʒük {зюк, ззюк} 'down' [Genko 2005: 72, 73].

    Differently in Literary Tabasaran: zikʷ {зикв} 'feather' [Khanmagomedov & Shalbuzov 2001: 169]; 'down' is expressed as žʷilːˈi zikʷ {жвилли зикв} [Khanmagomedov & Shalbuzov 2001: 169], literally 'fine feather'.

    There are two words for 'feather' in Literary Lezgi. The most frequent one is cʼak-ˈul {цIакул} 'feather' [Talibov & Gadzhiev 1966: 360; Gadzhiev 1950: 534; Haspelmath 1993: 484, 519]. The second, more marginal, term is cʼap-ˈur {цIапур} with polysemy: 'feather / propeller blade' [Talibov & Gadzhiev 1966: 361]. Distinct from literary tük {туьк} 'down' [Talibov & Gadzhiev 1966: 319; Gadzhiev 1950: 674], borrowed from Azerbaijani tük 'hair, fur, down'.

    In the Yarki dialect (the same Kyuri group): Nyutyug cʼap-ur 'feather' [Meylanova 1964: 79].

    In the Akhty dialect (Samur group): Khlyut čʷkʷ-al 'feather' [Kibrik & Kodzasov 1990: 45] (regularly < čʼʷkʷ-), Khuryug čʼakʷ-al 'feather' [Meylanova 1964: 281]. The same in some other dialects of the Samur group: Jaba čkː-ol 'feather' [Ganieva 2007: 22], Qurush čkːʷ-al 'feather' [Ganieva 2008: 33].

    But in the Fiy dialect of the Samur group: cʼuw-ur with polysemy: 'feather; down' [Meylanova 1964: 394].

    An unclear situation, with two terms in competition: cʼap-ˈur (in the Kyuri and Samur dialectal groups) vs. cʼak-ˈul (Samur). Both words are present in Literary Lezgi. The available dialectal data are too scant for a full-fledged distributive analysis, but cʼap-ˈur seems more preferable from the etymological point of view: cf. Aghul *pincʼʷ 'feather', Tabasaran (Khiv) cʼup-ˈur 'feather' q.v.

    The rare substantive suffix with generic semantics -al is both denominal [Haspelmath 1993: 107] and deverbal [Gaydarov et al. 2009: 131]. Final -ur looks like a fossilized plural exponent, but it must be noted that synchronic Lezgi plural suffixes are either -ar or -er [Haspelmath 1993: 71; Kibrik & Kodzasov 1990: 291].

    The modern Vartashen Udi word for 'feather' is, however, posposkːal, which primarily denotes a kind of plant. In Nidzh Udi, *kʷi(m)k was superseded with the etymologically unclear word maǯa, whose original Proto-Udi meaning is likely to have been 'tassel, bunch'.

    In Archi, the root *cːal [NCED: 1090] with polysemy 'wing / feather' is observed. This root is isolated within Lezgian, but if the North Caucasian etymology, proposed in [NCED], is correct, this implies the shift 'fist' > 'wing' > 'feather' in Archi.

    In some Tsakhur dialects (Mishlesh, Gelmets), 'feather' is expressed by the etymologically isolated root *wɨs- (~ -o-) [NCED: 1058]. In Mikik Tsakhur, the neologism 'bird's hair' is also attested for 'feather'.

    In Rutul, the etymologically obscure form maqˤaq 'feather' occurs.

    As for East Lezgian (Aghul, Tabasaran, Lezgi), the best solution here is to postulate *pincʼʷ [LEDb: #145] as the Proto-East Lezgian root for 'feather' (however, without any further etymology). It has been retained in Aghul, Southern Tabasaran (Khiv), Lezgi (Gyune, Yarki, Fiy). In Northern Tabasaran (Khanag, Khyuryuk), *pincʼʷ was superseded by *χäla (~ -lː-) [NCED: 1070], whose local meaning could be 'wing' (thus 'wing' > 'feather'). On the contrary, in Southern Tabasaran (Kondik, Tinit) and some Lezgi dialects (Akhty, Literary Lezgi), the root *cʼːämk / *kämcʼː 'down' [NCED: 1091] has acquired the meaning 'feather' (sometimes with synchronic polysemy 'feather / down').

    Out of a number of etymologically obscure terms for 'feather', attested in Lezgian lects, in several cases the source of borrowing can be established: Budukh (< Azerbaijani), Borch-Khnov Rutul (< Azerbaijani), Ixrek Rutul (< Lak?).

    Replacements: {'a kind of plant' > 'feather'} (Vartashen Udi), {'tassel, bunch' > 'feather'} (Nidzh Udi), {'bird's hair' > 'feather'} (Mikik Tsakhur), {'down' > 'feather'} (Southern Tabasaran, Lezgi), {'wing' > 'feather'} (Northern Tabasaran), {'fist' > 'wing' > 'feather'(?)} (Archi).

    Reconstruction shape: Correspondences seem regular.

    Semantics and structure: Primary substantive root.

    Tsakhur-Kum Tsakhur: cʼa [Kibrik & Kodzasov 1990: 206].

    The same in the Khanag subdialect: cʼaʔ ~ cʼa 'fire' [Uslar 1979: 959, 999; Dirr 1905: 216, 235]. The same in other subdialects: Khyuryuk cʼaʔ {цIаъ}, Kumi cʼa {цIа} 'fire' [Genko 2005: 176, 177].

    The same in other subdialects: Khiv cʼay {цIай}, Eteg, Tinit cʼa {цIа} 'fire' [Genko 2005: 176]. Distinct from Khiv murz {мурз} 'flame, blaze (Russian: пламя)' [Genko 2005: 122].

    The same in Literary Tabasaran: cʼa {цIа} 'fire' [Khanmagomedov & Shalbuzov 2001: 331]. Distinct from literary murz {мурз} 'flame, blaze (Russian: пламя)' [Khanmagomedov & Shalbuzov 2001: 234].

    The same in Literary Lezgi: cʼay [abs.] / cʼ-u- [obl.] {цIай} 'fire' [Talibov & Gadzhiev 1966: 360; Gadzhiev 1950: 448; Haspelmath 1993: 484, 519]. Distinct from two literary terms for 'flame, blaze (Russian: пламя)': cʼun murz {цIун мурз} [Talibov & Gadzhiev 1966: 245; Gadzhiev 1950: 540] (literally 'blade of fire') and yalaw {ялав} [Talibov & Gadzhiev 1966: 402; Gadzhiev 1950: 540] (borrowed from Azerbaijani alow 'flame, blaze').

    The same in the Akhty dialect: Khlyut cʼay [abs.] / yicʼ-ˈa- [obl.] 'fire' [Kibrik & Kodzasov 1990: 206]. The Akhty oblique stem is historically the result of metathesis < *cʼiy-.

    Replacements: {'fire' > 'stone in finger-ring'} (Archi).

    Reconstruction shape: Correspondences seem regular, assuming the metathesis cʼVy(V) > yVcʼ(V) that occurred in several lects (Archi, Akhty Lezgi). The Udi form contains the fossilized plural suffix (which seems somewhat surprising from a semantic point of view).

    Semantics and structure: Primary substantival root. The oblique stems are *cʼoyɨ- and *cʼo(y)-rV- (functional difference is unclear, but the latter is attested in Udi, Archi and Nuclear Lezgian, therefore, is more likely to represent an archaism).

    Tsakhur-Kum Tsakhur: baluʁ [Kibrik & Kodzasov 1990: 87].

    A second (apparently less frequent) term is murǯuχ {мурджух} 'fish' [Dzhamalov & Semedov 2006: 194, 389] (no examples found).

    In [Kibrik & Kodzasov 1990: 87], only the loanword baluʁ is quoted for 'fish'.

    Distinct from the specific Tpig term baluʁ 'large fish, any salt-water fish' [Suleymanov 2003: 38] of Azerbaijani origin.

    In other subdialects, superseded with a loanword: Tsirkhe, Duldug baluʁ 'fish' [Shaumyan 1941: 167].

    For the Khanag subdialect two words are known: the borrowing balˈurχː 'fish (in general)' [Uslar 1979: 605, 1004] and the inherited term čečʼ with polysemy: 'small fish / locust' [Uslar 1979: 967, 1004]. Dirr, however, quotes only čečʼ 'fish (in general)' [Dirr 1905: 218, 241].

    Similarly in the Khyuryuk subdialect: balˈurχː {балурхх} 'fish' [Genko 2005: 26]; distinct from inherited čečʼ {чечI} with polysemy: 'small fish / locust' [Genko 2005: 180].

    For the Khiv subdialect two words are known: the inherited term čičʼ {чичI} 'small fish, river fish' [Genko 2005: 181] and the borrowing balˈuʁ {балугъ} 'salt-water fish' [Genko 2005: 26].

    Quite differently in two subdialects of the Eteg cluster: Tinit, Dzhikhtig χˤad {хяд} 'fish' [Genko 2005: 168, 223].

    Two words for 'fish' exist in Literary Tabasaran: the borrowing balˈuʁ {балугъ} 'fish' [Khanmagomedov & Shalbuzov 2001: 76] and the inherited term χˤad {хяд} 'fish' [Khanmagomedov & Shalbuzov 2001: 319] - the difference is unknown.

    In Literary Lezgi the generic term for 'fish' is balˈuʁ {балугъ} 'fish' [Talibov & Gadzhiev 1966: 58; Gadzhiev 1950: 737; Haspelmath 1993: 482, 519], borrowed from Azerbaijani balɨg, dialectal (e.g., Quba) baluɣ 'fish'. The inherited term ʁed [abs.] / ʁetː-rˈe [obl.] {гъед} shifted to the more specific meaning 'large fish' [Talibov & Gadzhiev 1966: 93; Gadzhiev 1950: 737] (glossed simply as 'fish' in [Haspelmath 1993: 490, 519]).

    Only the loanword is attested in the Akhty dialect: Khlyut balˈuʁ 'fish' [Kibrik & Kodzasov 1990: 87].

    From the distributive point of view, three roots enter into competition.

    1) *χːˤanː [NCED: 1078]. This root is retained in several Nuclear Lezgian languages: Proto-Rutul (attested in Shinaz), Proto-Tabasaran (attested in Eteg), Proto-Lezgi (attested in Gyune). Since *χːˤanː is descended from the best candidate for the status of the general North Caucasian term for 'fish', we reconstruct this Lezgian root with the meaning 'fish'.

    2) *χˤawχay [LEDb: #262]. This root denotes 'fish' in Archi, but 'snail' in Tabasaran; no further etymology. In Proto-Lezgian, it could have denoted 'snail' vel sim.

    3) *čʷiˤlä- [NCED: 532]. This root denotes 'fish' in Udi, but 'green', 'blue' and 'wet' in Nuclear Lezgian. It can actually be considered the best candidate for Proto-Lezgian 'green' (the shift 'green, blue' > 'fish' seems more natural than vice versa).

    A fourth inherited root, also attested with the meaning 'fish', is *čʼeƛʼ [NCED: 333] (the variant *čʼeƛʼː is apparently unnecessary, because dialectal fluctuations in Aghul and Kryts seem local and late). It means 'fish' in Aghul and, secondarily, in the neighboring Tabasaran dialects (the original Proto-Tabasaran meaning of this root was, quite likely, 'locust'). Its suffixed Kryts cognate means 'green', whereas external North Caucasian comparanda suggest the meaning 'lizard' or 'frog'. An unclear situation. Maybe the Proto-Lezgian meaning of *čʼeƛʼ was indeed 'a k. of reptile' with a later shift to 'green' in Kryts, 'fish' in Aghul and 'locust' in Tabasaran.

    In Rutul dialects, the forms murǯuχ ~ miǯuruχ ~ mižruχ 'fish' have no clear origin.

    Replacements: {'green' > 'fish'} (Udi), {'a k. of reptile' > 'fish'} (Kryts), {'snail' > 'fish'(?)} (Archi), {'locust' > 'fish'(?)} (Tabasaran). See also notes on 'star'.

    Reconstruction shape: Correspondences seem regular.

    Semantics and structure: Primary substantive root.

    Distinct from pɨr-pɨr siʔi 'to fly up, take wing', literally 'to make pɨr-pɨr' [Meylanova 1984: 120]. Further cf. pɨr-pɨr 'propeller blade; may-bug' [Meylanova 1984: 120].

    In [Kibrik & Kodzasov 1988: 82], however, only the loanword učmɨš-x- 'to fly' is quoted, borrowed from the Azerbaijani perfect stem uč-muš- (infinitive uč-mag) 'to fly', plus the Tsakhur verb ɨx- 'to become'.

    It must be noted that in [Suleymanov 2003: 62], Koshan ʢ=alga-na- or ʕ=alga-na- {гIалганас} (without subdialectal specification) is translated as 'to fly'. Cf. the same verb, modified with another prefix: Burshag h=alga-na- 'to run away' [Kibrik & Kodzasov 1988: 81; Suleymanov 2003: 62].

    The same in the Khanag subdialect: tʼ=iχ-ˈ 'to fly' [Uslar 1979: 922, 996; Dirr 1905: 206, 232]. The same in the Khyuryuk subdialect: tʼ=iχ-ˈ {тIибхув} 'to fly' [Genko 2005: 148].

    The same in the Khiv subdialect: tʼ=iχ-ˈ {тIирхуб} 'to fly' [Genko 2005: 149]. The same in Literary Tabasaran: tʼ=iχ-ˈ {тIибхуб} 'to fly' [Khanmagomedov & Shalbuzov 2001: 288].

    Cf. Literary Tabasaran pʼurr apʼ- 'to fly up, take wing', literally 'to make pʼurr' [Khanmagomedov & Shalbuzov 2001: 256].

    The same two expressions are used for 'to fly' in Literary Lezgi: luw gu- {лув гун} 'to give wing' [Talibov & Gadzhiev 1966: 225; Gadzhiev 1950: 333; Haspelmath 1993: 519] and cːaw-ay fi- {цавай фин} 'to go across the sky' [Talibov & Gadzhiev 1966: 357; Gadzhiev 1950: 333]. Cf. Literary Lezgi purr awˈu- (or pːurr awˈu-?) 'to fly up, take wing', literally 'to make purr' [Talibov & Gadzhiev 1966: 268].

    The best candidate seems the root *ʔiχV [NCED: 582], which means 'to fly' in two Nuclear Lezgian languages: Tsakhur, Tabasaran, but got lost in the rest of the lects. External North Caucasian comparison may confirm the Proto-Lezgian reconstruction *ʔiχV 'to fly'.

    The second candidate is the expressive root *pVr- [NCED: 874]. It is encountered as part of complex verbs in two outliers: Udi ('to fly') and Archi ('to stream, fly (of flag)'; the generic meaning 'to fly' is expressed by the Avar loanword). Cf. also Nuclear Lezgian expressions for 'to fly up, take wing': Budukh 'to make pɨr-pɨr', Tabasaran 'to make pʼurr', Lezgi 'to make purr', which should rather be analyzed as onomatopoeic. Lezgian *pVr- also possesses external North Caucasian comparanda with the meaning 'to fly', although verbs of the shape pVr are rather frequently attested as expressions for 'to fly' among the world's languages, and normally it is impossible to discriminate between etymological cognates and new onomatopoeic formations. For this reason we prefer to exclude Lezgian *pVr- from the list.

    In Kryts, 'to fly' is expressed by the etymologically isolated =alqʼon- ~ =anqʼʷan- [LEDb: #234].

    In Rutul, the old root was superseded with *ʔäča-, whose Lezgian cognates, discussed in [NCED: 283], are semantically dubious, but the original Proto-Rutul meaning of this root should be 'to jump', judging by the synchronic Rutul polysemy 'to fly / to jump' (i.e. 'to jump' > 'to fly up' > 'to fly').

    In Aghul dialects, the meaning 'to fly' is expressed by various prefixed roots with the original meanings 'to go' or 'to run' (the full collocation 'to go across the sky' = 'to fly' is also attested).

    In Lezgi, two collocations for 'to fly' coexist: 'to give wing' and 'to go across the sky'.

    In Budukh, superseded with the Azerbaijani loanword.

    Replacements: {'to jump' > 'to fly'} (Rutul), {'to go' > 'to fly'} (Aghul), {'to run' > 'to fly'} (Aghul), {'to go across the sky' > 'to fly'} (Aghul, Lezgi), {'to give wing' > 'to fly'} (Lezgi).

    Reconstruction shape: Correspondences seem regular.

    Semantics and structure: Primary verbal root.

    There exists a separate specific Nidzh term for 'foot': čːil {чIил} 'foot' [Comrie & Khalilov 2010: 213] (glossed as 'foot'), [Gukasyan 1974: 256] (glossed as Russian 'нога (i.e. leg with foot)'; missing from the main section of the dictionary). The origin of čːil is unclear.

    Distinct from the bound term moɬː-ˈol 'foot', used in some idiomatic expressions [Kibrik et al. 1977b: 281; Kibrik & Kodzasov 1990: 30; Chumakina et al. 2007]. Etymological evidence [NCED: 309] points to the primary meaning 'support'.

    Distinct from the more specific term čʼerčʼ {чIерчI} 'shinbone; shin; leg' [Meylanova 1984: 158].

    In [Comrie & Khalilov 2010: 213], 'foot' is glossed as mɨkʼlʸi {мыкIли} - an enigmatic form (cf. mɨkʼ 'dance; kick' [Ibragimov & Nurmamedov 2010: 268]).

    Tsakhur-Kum Tsakhur: ɢelʸ [Kibrik & Kodzasov 1990: 30]. Polysemy: 'foot / leg'. 'Foot' can also be expressed as ɢelʸ-in xanʸe, literally 'bottom of ɢelʸ' [Kibrik & Kodzasov 1990: 31].

    In [Comrie & Khalilov 2010: 213], 'foot' is erroneously glossed as kʼutnʸi {кIутни}; in fact kʼutʸnʸi {кIутʹни} means merely 'end, tip', at least in Literary Tsakhur [Ibragimov & Nurmamedov 2010: 205].

    In [Comrie & Khalilov 2010: 212], 'foot' is erroneously quoted as gɨl {гыл}, which actually means 'foreleg (of animal)' [Dzhamalov & Semedov 2006: 58].

    Two additional (more marginal) words are glossed as 'нога' in [Suleymanov 2003]: murkʼ 'нога; hoof' [Suleymanov 2003: 135] and tʼurʡ 'furniture leg; нога (disparaging)' [Suleymanov 2003: 159].

    The same in the Khanag subdialect: lik ~ lek 'foot; leg' [Uslar 1979: 842, 999; Dirr 1905: 193, 235]. The same in the Khyuryuk subdialect: lik {лик} 'foot; leg' [Genko 2005: 114].

    The same in the Khiv subdialect: lik {лик} 'foot; leg' [Genko 2005: 114]. The same in Literary Tabasaran: lik {лик} 'foot; leg' [Khanmagomedov & Shalbuzov 2001: 218].

    The same basic term in Literary Lezgi: kʼʷač {кIвач} 'foot / leg / paw' [Talibov & Gadzhiev 1966: 214; Gadzhiev 1950: 426; Haspelmath 1993: 496, 522, 519]. Distinct from literary qʼül {кьуьл} with polysemy: 'foot / kick / dance, dancing' [Talibov & Gadzhiev 1966: 209] (incorrectly glossed as 'leg; dance' in [Haspelmath 1993: 504]).

    The same in the Akhty dialect: Khlyut kʼʷač 'foot / leg' [Kibrik & Kodzasov 1990: 30]. The meaning 'foot' can also be expressed as kʼʷač-ˈin kʼan 'bottom of kʼʷač' [Kibrik & Kodzasov 1990: 31].

    Three Lezgian roots are equal candidates for the status of the Proto-Lezgian term for 'foot / leg' from the distributive point of view: (1) *ya(m)tur [NCED: 674], meaning 'foot / leg' in Udi; (2) *ʔaqː [NCED: 244], meaning 'foot / leg' in Archi; (3) *qːel [NCED: 455], meaning 'foot / leg' in South and West Lezgian.

    Out of these, *ya(m)tur denotes 'thigh, hip' in Nuclear Lezgian, and its external North Caucasian comparanda also point to the meaning 'thigh, hip' [NCED: 674]. Thus, it is natural to posit *ya(m)tur as the Proto-Lezgian root for 'thigh, hip' and assume the shift 'thigh, hip' > 'foot / leg' in the Udi branch (in Caucasian Albanian this root is attested for 'foot').

    The second root, *ʔaqː [NCED: 244], is attested with the polysemy 'leg (of human) / foot (of human) / hind leg (of animal)' in Archi and as narrower 'hind leg (of animal)' in some Nuclear Lezgian lects. The most economic solution is to reconstruct its Proto-Lezgian meaning as 'hind leg (of animal)' (with further developments in Archi).

    The third competing root is *qːel [NCED: 455], which is retained in South Lezgian (Kryts, Budukh) and West Lezgian (Tsakhur, Rutul), where it means 'foot / leg'. Distributively this should be posited as the Proto-Nuclear Lezgian term for 'foot / leg', and there is no counter evidence for the same semantic reconstruction on the Proto-Lezgian level.

    In East Lezgian, *qːel was probably superseded with other roots. In Aghul and Tabasaran, 'foot / leg' is expressed by *läk [NCED: 755], whose original Proto-Lezgian meaning is likely to have been 'leg bone' (see notes on 'bone'). In Lezgi, the root *kʼʷarč [NCED: 733] is used for 'foot / leg'; the original meaning of *kʼʷarč is not entirely clear, but it could be 'heel, sole' (shifted to 'hoof' in Rutul-Tsakhur), as suggested by the external North Caucasian comparanda.

    Cf. also recent semantic developments into 'foot' in individual lects: 'hoof' > 'foot' in Aghul Proper (murkʼ); 'furniture leg' > 'foot / leg' in Aghul Proper (tʼurʡ); 'kick; dance, dancing' > 'foot' in Gyune and Literary Lezgi (qʼül). In many Nuclear Lezgian lects, the collocations 'bottom of leg/foot' (Kryts, Tsakhur, Rutul, Tabasaran, Lezgi) or 'flat leg/foot' (Aghul) can also be used for the specific meaning 'foot'.

    Replacements: {'thigh, hip' > 'leg / foot'} (Udi, Caucasian Albanian), {'hind leg (of animal)' > 'leg / foot'} (Archi), {'leg bone' > 'leg / foot'} (Aghul, Tabasaran), {'furniture leg' > 'foot / leg'} (Aghul Proper), {'hoof' > 'foot'} (Aghul Proper), {'kick; dance, dancing' > 'foot'} (Lezgi), {'bottom of leg' > 'foot'} (Kryts, Tsakhur, Rutul, Tabasaran, Lezgi), {'flat leg' > 'foot'} (Aghul), {'heel, sole' > 'leg / foot'(?)} (Lezgi).

    Reconstruction shape: Correspondences seem regular.

    Semantics and structure: Primary substantive root with polysemy 'foot / leg'. The oblique stem is *qːelɨ-.

    Distinct from dulu {дулу} 'full' [Meylanova 1984: 54], whose application is unknown; borrowed from Azerbaijani dolu 'full'.

    In [Comrie & Khalilov 2010: 612], 'full' is glossed as tamamu {тамаму}, but in fact tamam means 'finished; comprehensive' [Meylanova 1984: 132], borrowed from Azerbaijani tamam 'full (in abstract sense)', ultimately from Arabic.

    Distinct from bɨkɨrɨ-n 'full (in abstract sense), whole; finished' [Kibrik et al. 1999: 870].

    Tsakhur-Kum Tsakhur: cf. gʸ=acʼɨ-n 'satiated' [Kibrik & Kodzasov 1990: 241].

    Suleymanov's ʔ=acʼi-r is a regular derivative from the verbs acʼ-i- ~ ʔ=acʼ-i- 'to fill (intrans.); to satiate oneself' [Kibrik & Kodzasov 1988: 101; Suleymanov 2003: 35, 209; Shaumyan 1941: 139] (in [Kibrik & Kodzasov 1988] and [Shaumyan 1941], only the prefixless stem is quoted, Suleymanov gives both verbs; incorrectly glossed as 'to fill (trans.)' by Shaumyan). Kibrik & Kodzasov's and Shaumyan's acʼ-a-r 'full' is less clear because of the thematic -a-.

    Cf. in the Usug subdialect the regular participle acʼu-f 'full / satiated' [Shaumyan 1941: 139].

    The same in the other subdialects: Tsirkhe, Duldug acʼu-f 'full / satiated' [Shaumyan 1941: 139].

    The proper Kondik term for 'full' is not documented, cf. Kondik acʼˈu 'satiated' [Kibrik & Kodzasov 1990: 241], which apparently possesses the basic meaning 'full'.

    The same in Literary Lezgi: acʼˈa-y {ацIай} 'full, filled' [Talibov & Gadzhiev 1966: 54; Gadzhiev 1950: 583; Haspelmath 1993: 479, 520], participle from the verb acʼˈu- [imperf.] / acʼˈa- [perf.] {ацIун} 'to fill (intrans.)' [Talibov & Gadzhiev 1966: 54; Haspelmath 1993: 479]. Distinct from the Azerbaijani loanword tuχ {тух} 'satiated' [Talibov & Gadzhiev 1966: 318; Gadzhiev 1950: 836].

    For the Akhty dialect (Khlyut), inherited acʼˈa-y is documented only in the meaning 'satiated' [Kibrik & Kodzasov 1990: 241].

    This basic adjective displays the polysemy 'full / satiated' in all or almost all the languages; the same polysemy 'to be full / to be satiated' should be reconstructed for the Proto-Lezgian verb (the Proto-Lezgian development 'full' > 'satiated'). In some Nuclear Lezgian lects the development 'full' > 'thick' is attested as well (Budukh, Tsakhur, Aghul).

    Replacements: {'full' > 'satiated'} (Proto-Lezgian), {'full' > 'thick'} (Budukh, Tsakhur, Aghul).

    Reconstruction shape: Correspondences seem regular, although reconstruction of the initial hˤ- is uncertain (Kryts ʕ- could actually be an old prefix with general semantics, the Archi vowel length can be unrelated). Thus, the reconstruction *acʼɨ- is also probable. In Caucasian Albanian-Udi, the fossilized class prefix b- occurs.

    Semantics and structure: Primary verbal root with polysemy 'to be full / to be satiated'.

    Distinct from esi- (e-l-si-) 'to give back, return' [Meylanova 1984: 166; Kibrik & Kodzasov 1988: 68].

    Suppletive paradigm: class=w=ɨ=l=cʼ-a-r- [imperf.] / class=w-ɨ-r [perf.] / class=w-i [imv.]. It must be noted that according to [Ibragimov 1978: 85], the modern imperative form is class=iy, where the original root consonant -w- was reanalyzed as the class 3 exponent. In [Dirr 1912], the imperfective and perfective stems are quoted with the assimilated sequence -wu- instead of -wɨ-.

    Suppletive paradigm: class=ɨ=r=cʼ-ä-r- [imperf.] / class=w-ɨ-r [perf.] / class=iy [imv.]. In [Ibragimov 1978], the imperf. stem is quoted with regular -a- instead of -ä-. The imperative form is a result of secondary reanalysis, see notes on Mukhad Rutul.

    Suppletive paradigm: class=w=ɨ=l=cʼ-a-r- [imperf.] / class=w-ɨ-r [perf.] / class=w-ɨy [imv.].

    Suppletive paradigm: icʼ-an-di- [imperf.] / i-na-w [perf.] / i-s [inf.] / t-il [imv.] / ma=l=icʼ-a [prohib.]. Final -w in the perfective (aorist) form looks like a fossilized class exponent; l= in the prohibitive form is a spatial prefix [Magometov 1970: 158 ff.].

    A somewhat different paradigm in the Usug subdialect: cʼ-ay- [imperf.] / g=i-ne [perf.] / cʼ-a-s [inf.] / t-in [imv.] 'to give' [Shaumyan 1941: 147]. Initial g= in the perfective form is apparently a rare spatial prefix.

    Similarly in the Khanag subdialect: tː=ˈu=l=ɣ- ~ l=icʼ-ˈ [imperf.] / tːu=ʔˈuɣ- ~ tː=ˈuɣ- ~ ˈaɣ- ~ ɣ- [perf.] / tː=ˈuɣ- ~ ɣ- [inf.] / tː=uɣ ~ ča ~ čaw [imv.] 'to give' [Uslar 1979: 655, 920, 992; Dirr 1905: 162, 194, 205, 217, 227].

    Similarly in other subdialects: Khyuryuk l=icʼ-ˈ [imperf.] / aɣ-ˈ [perf.] / tː=uɣ ~ ča [imv.] / tː=ˈuɣ ~ ɣ- [inf.] {ггув, ттубггув} 'to give' [Genko 2005: 40, 146]; Arkit l=icʼ-ˈ [imperf.] / t=ˈuw- [inf.] {тувув} 'to give' [Genko 2005: 144].

    Similarly in the Khiv subdialect: tː=ˈuw- ~ tː=ˈaw- {ттувуб, ттавуб} 'to give' [Genko 2005: 146]. The same in Literary Tabasaran: t=ˈuw- ~ tː=ˈuw- {тувуб} 'to give' [Khanmagomedov & Shalbuzov 2001: 282].

    The same in Literary Lezgi: gu- [imperf.] / ga- [perf.] / ce [imv.] {гун} 'to give' [Talibov & Gadzhiev 1966: 89; Gadzhiev 1950: 157; Haspelmath 1993: 489, 520; Gyulmagomedov 2004, 1: 194]. Pace [NCED: 626], the imperative stem is ce (i.e. cʰe), not cːe, as proved by both Uslar's and Haspelmath's transcription.

    This root has survived as the basic equivalent for 'to give' in all Lezgian lects. In Caucasian Albanian and especially in modern Udi, the reflexes of *ʔiƛːɨ- are not fully transparent. Following [Schulze 2005: 542 (3.4.2.1 #24); Gippert et al. 2008: II-71], we analyze the Proto-Caucasian Albanian-Udi paradigm *luʁ- [imperf.] / *daʁ- [perf.] 'to give' as *lu=ʁ- / *da=ʁ-, where the original root *ʁ- is modified by two unique prefixes lu= and da= (at least for *lu=ʁ- there is an interesting comparandum in Khinalug: lä=kʼʷi 'to give'; lä= is the Khinalug preverb 'from the speaker'). Due to phonetic mutations and morphological levellings, in modern Udi the paradigm *lu=ʁ- / *da=ʁ- has been transformed into a unified stem tad-, historically ta=d-, where ta= is the rare prefix 'thither'.

    In Proto-Nuclear Lezgian, the paradigm became suppletive, with the additional root *ʔicʼa- involved: *ʔicʼa- [imperf.] / *ʔiƛːɨ- [perf.]. The original meaning of *ʔicʼa- is not clear; external North Caucasian comparison suggests something like 'to compensate' [NCED: 626]. It must be noted that in Tsakhur, Lezgi and some Tabasaran dialects, this suppletive paradigm was levelled backwards in favor of the root *ʔiƛːɨ-.

    Replacements: {'to compensate' > 'to give'(?)} (Proto-Nuclear Lezgian).

    Reconstruction shape: Basic correspondences seem regular, although reflexes in individual lects are rather complicated.

    Semantics and structure: Primary verbal root with the Ablaut grade *ʔoƛːa-.