*d^héng^u̯-i-/d^hn̥g^u̯-éi̯-

The equation between Hittite dankui- and Luwian dakkuwa/i- suggests the reconstruction of Proto-Anatolian adjective *d^héng^u̯-i-/d^hn̥g^u̯-éi̯- meaning ‘dark’; cf. Melchert 1994a:232, Starke 1995a:120 n. 243, HEG T:109f. and EDHIL:829 both with further literature. For the special development of a voiced labio-velar next to a nasal in Luwian, see *g^u̯hén-/*g^u̯hn-´.

[D.S.]

The Anatolian, Germanic, Celtic, and – if it belongs here (see below) – Greek evidence permit the reconstruction of a root *d^heng^u̯h- or *d^hneg^u̯h- ‘dark’. As proposed by Kümmel 2016a:221 (see also LIV²+ s.v. *d^heng^wh-), this meaning may ultimately derive from a verbal root ‘to cover’ reflected in Lithuanian deñgti ‘to cover’, dangùs ‘sky’, and related forms, with the semantic development ‘covered, hidden’ > ‘dark’. This scenario is, however, open to some doubt since verbal forms are confined to Balto-Slavic, whereas Anatolian already attests the adjectival meaning ‘dark’. Moreover, Proto-Germanic *dunga- m. ‘cellar, groundhouse’ and ‘dung, manure’ (< ‘covering’ (?)) points to a plain velar rather than a labiovelar root-final consonant. It therefore seems preferable to distinguish a verbal root *d^heng^h- ‘to cover’ (thus Petit 2021a, ALEW 2.0:217, Smoczyński 2018a:211) from an adjectival (“Caland”) root *d^heng^u̯h- or *d^hneg^u̯h-  ‘dark’. For the pre-Proto-Germanic merger of *T, *D, *D^h after homorganic nasal in and this cluster’s subsequent conditioned split into Proto-Germanic *NT and *ND see Kümmel 2016a.

The adjectival root *d^heng^u̯h- ‘dark’ is reflected in a variety of expected adjectival formations: an *-i-stem adjective in Anatolian, a substantivized reflex of an *-nt- adjective in Celtic (Middle Welsh deweint), a *-ro- and a *-lo- adjective in Germanic (Dutch donker; Old High German tunkal), as well as two simple thematic adjectives, one with e-grade (Old Frisian diunk, Old Swedish dionker, diunker, Old Norse døkkr) and one with o-grade (Old Norse dǫkkr). Not all formations attested in Germanic are likely to go back to Proto-Indo-European. Rather, their geographical distribution suggests that only one or two are inherited (e.g., *dunk^w(a)laz and *dink^waz), while the remaining forms are later analogical remodelings based on more productive word-formation types, *-ra- being much more common than *-la- (Heidermanns 1993a:44f.) and R(o)-o- more common than R(e)-o- (Heidermanns 1993a:39–41).

Sturtevant 1933a:123f. connected Ancient Greek δνόφος ‘darkness, dusk, gloom’ with the Anatolian adjective, an attractive comparison from the semantic point of view but one that is phonologically less straightforward. Leaving ablaut aside for the moment, a pre-form *d^hnóg^u̯h-o- would first undergo devoicing of the mediae aspiratae to *t^hnók^u̯h-o-, followed by dissimilatory loss of aspiration of the first of two aspirated stops (“Grassmann’s Law”), yielding *tnók^u̯h-o-. In principle, this form would be expected to produce Ancient Greek †τνόφος. Since Greek has no inherited words beginning with *τν-, one might hypothesize that word-initial *τν- was then regularly voiced to δν-, so that δνόφος could ultimately continue *d^hnóg^u̯h-o-. Although no clear counterexamples are known, a sound law supported by a single lexical item should be adopted only with caution.

The ablaut difference is less problematic. Greek δνόφος presupposes *d^hnóg^u̯h-o- (and *d^hnég^u̯h-es- in ἰο-δνεφής ‘purple-dark’), whereas Proto-Germanic *dink^waz points to *d^heng^u̯h-o-. Either ablaut variant may be explained as a secondary full grade based on the better-attested zero grade *d^hn̥g^u̯h-, reflected in Anatolian, Celtic, and Germanic. On distributional grounds, it is perhaps preferable to regard *d^hneg^u̯h- as the original full grade. Proto-Germanic *dink^waz < *d^heng^u̯h-o- could then be explained as an analogical replacement of the expected *dneg^waz < *d^hneg^u̯h-o-, influenced by the position of the vowel in the zero-grade stem *dunk^w- < *d^hn̥g^u̯h- reflected in *dunk^w(a)laz ‘dark’.

[S.H.]

Hittite dankui-, dankuwai- ‘black, dark’, dankuēšš-^(mi) ‘to become black’, Cuneiform Luwian dakkuwa/i- ‘dark’, Luwian in Hittite transmission ta(n)kuwa/i- ‘dark’.Cf., e.g., EDHIL:829 s.v. dankui-, HEG T:109f.

(?) Ancient Greek δνόφος ‘darkness, dusk, gloom’ (Simon., A. in lyr.), δνοφερός ‘dark, murky’ (poet. since Il., Hp.), ἰο-δνεφής ‘purple-dark’ (Od.); later γνόφος, etc.Cf., e.g., GEW I:403 s.v. δνόφος, I:317 s.v. γνόφος.

Middle Welsh deweint m. ‘midnight, darkness’ < *dang^wantī.Cf., e.g., GPC s.v. dewaint, Kümmel 2016a:221.

Proto-Germanic *dunk^w(a)laz ‘dark’ (Old High German tunkal, German dunkel), *dunk^w(a)raz ‘dark’ (Old Saxon dunkar, Dutch donker), *dink^waz ‘dark’ (Old Frisian diunk, Old Swedish dionker, diunker, Old Norse døkkr), *dank^waz ‘dark’ (Old Norse dǫkkr).

(?) Proto-Germanic *dunga- m. ‘cellar, groundhouse’ (Old English dung, Old High German tung) and ‘dung, manure’ < *‘covering’ (?) (English dung, German Dung, etc.).Cf., e.g., Heidermanns 1993a:167, 152, 146, EDPG:109, 96, Kümmel 2016a:221.

(?) Lithuanian inf. deñgti, prs. deñgia, prt. deñgė ‘to cover’, inf. diñgti, pres. diñgsta, pret. diñgo ‘to disappear’ (< *‘be covered’?), dangùs m., Old Prussian dangus ‘sky’, dangà f. ‘cover, blanket’.Cf., e.g., LIV2+ s.v. *d^heng^wh-, Petit 2021a, ALEW 2.0:217f., 195–197, Smoczyński 2018a:211f., 194–196.

(?) Ukrainian o-djahnúty, o-djahtý ‘to clothe’, Proto-Slavic *dǫga f. ‘arc, arch, barrel stave; rainbow’ (Russian dugá, Czech duha, Serbo-Croatian dúga, etc.).Cf., e.g., LIV2+ s.v. *d^heng^wh-, EDSIL:114.