From: Tavi
Message: 69359
Date: 2012-04-17
>Sometime ago I had considered this IE etymology, but later I dropped in favour of a "pre-IE" one: Kartvelian *qew- 'ravine'.
> 4. 'Cage': It. <gabbia>, Prov. <gabia>, Cat./Sp. <gavia>, etc. against Ven. <keba>, OFr <chage>, Wall. <c^ef>, etc. Much has been made about the discrepancy within French between <cage> (Lat. <cavea>) and <geôle> 'jail' (OFr jaiole, VL *gaveola). Obviously however <chage> has been replaced by <cage> from a dialect retaining /ka/ unpalatalized, and French has borrowed <gabie> 'ship's lookout perch, crow's-nest' from Prov. <gabia>, so <jaiole> could represent an earlier interdialectal borrowing, given the mobility of these words. More importantly Venetian contrasts <gebo> 'river-bed' with <keba> 'cage'. This suggests that a second root is in play, *gab- or *gaw-, whose semantic overlap with Lat. <cavus> led to confusion and substitution in many of the Romance dialects.
>
> V. Bertoldi, "GAVA e derivati nell'idronimia tirrena" (Studi Etruschi 3:293-320, 1929), extracted an element *gava 'watercourse' vel sim., which he presumed to be pre-Indo-European, from hydronyms, toponyms, and appellatives in the Alps, Liguria, and the Pyrenees. Skeptics are free to dismiss B. as a "substrate romantic", and at times he did go too far. I am not pleased with his attempt to lump *gab- and *gamb- together with *gaw- in hydronyms. Nevertheless he made a good case for a pre-Roman element *gaw- distinct from Lat. <cavus>. He argued that Pliny's <gavia> 'gull' (HN 10:91) is a derivative; VL extensions such as *gavia:nus underlie the Romance words for 'gull', It. <gabbiano>, <gavina>, OFr <gaverial>, Prov. <gavian>, <gabian>, <gabino>, Sp. <gaviota>, Pg. <gaivâo>, <gaivota>, etc.
>
> I am not convinced that Bertoldi's *gava must be pre-IE, since it could represent Illyrian *gawa:- 'outpouring, drainage' vel sim., from *g^Howeh2-, from Proto-IE *g^Heu- 'to pour'. At any rate the semantic development 'watercourse' > 'river-bed' > 'trough' > 'hollow trough-like object' would bring derivatives of *gava into the same semantic field as those of <cavus>. It is not implausible that confusion would arise with certain words in certain dialects, leading at times to /g/-forms substituting for expected /k/-forms.
>