From: dgkilday57
Message: 68245
Date: 2011-11-28
>According to Schrijver, cited by Lubotsky (Reflexes of PIE *sk in In-Ir, Incontri Linguistici 24:25-57, fn. 21, 2001), the Celtic forms (including Middle Welsh <ceneu> 'puppy') reflect Proto-Celtic *kanawon- < PIE *kenh{x}won-. If this suffix *-won- functions like Sanskrit -van- (e.g. <yájvan-> 'worshipping' from <yaj-> 'to worship'), the root can hardly be *kenh1- 'to pinch, compress' vel sim., since *kénh1won- would have an active sense 'pinching, pincher'. Then again, perhaps a puppy was considered a 'little nipper'.
> [...]
>
> Burrow, in "A note on the Indo-Iranian root <kan-> 'small' and on the etymology of Latin <canis> 'dog'" (TrPhS 81:155-64, 1983), rejects the assignment of Sanskrit <kanÃ:nas> 'young, youthful' and related words to Pokorny's *ken-(3) 'frisch hervorkommen' usw. (IEW 563), under which we find Greek <kainós> 'new, fresh'. Noting that Skt. <káni:yas-> and <kánis.t.ha-> mean 'younger' and 'youngest' only when referring to a brother, son, or the like, otherwise 'smaller, less' and 'smallest, least', Burrow deduces that the original sense of In-Ir *kan- was 'small', not 'young'. This is reinforced by Skt. <kanÃ:nika:>, <kánis.t.ha:>, <kanÃs.t.hika:> 'little finger' and Khotanese <kanaiska-> 'id.' Obviously the little finger is no younger than its handmates, only smaller.
>
> Since Burrow has no problem with Proto-Indo-European */a/, he extends *kan- 'small' back to PIE and derives from it Middle Irish <cana>, <cano> 'wolf-cub', Welsh <cenau> 'wolf-cub, dog-whelp', Latin <canis> 'dog' (on the theory, earlier 'whelp'), the first element of Maeonian <Kandaúle:s> 'Dog-Strangler' (epithet of Hermes, Hipponax fr. 3 Masson), and Slavic <konI> 'horse' (on the theory, earlier 'foal'). The semantics are not difficult, with Umbrian <katel> 'dog' against Lat. <catulus> 'young animal, whelp' providing an illustration, but for those of us who lean toward Lubotsky in avoiding PIE */a/, the phonology and morphology pose a challenge. In prevocalic zero-grade, PIE *ken- should yield *kn.nV- by Sievers-Edgerton, whence *kanV- in In-Ir and Italic. (The latter is argued from the P-Italic negative prefix <an->, apparently generalized from prevocalic position while Q-Italic extracted preconsonantal *en-, Lat. <in->.) I will leave the Celtic words aside, since I no longer have access to recent etymological material.