Re: [tied] alb. gji (breast)

----- Original Message -----
From: <jer@...>
To: <cybalist@yahoogroups.com>
Sent: Saturday, February 08, 2003 2:02 AM
Subject: Re: [tied] alb. gji (breast)


> I am not so sure gji is a loanword.

Jens, I'm not so sure of it either. I've never thought of <gji> at all until Alvin asked about its etymology. One thing I _am_ sure of is its connection with <sinus>. The formal correspondence in unimpeachable and the 'breast/gulf' semantics is too idiosyncratic to be the result of independent development.

> A form *sino- (*sH1i-no-) derived from the root of Lat. sino: 'give way, lag'

It's a u-stem in Latin , isnt't it?

> would make the meaning 'bay' parallel with Old Norse vík 'inlet' from víka 'give way, weichen'. The semantic development must then be from 'Meerbusen' to 'Busen' of the body; I consider it too odd for a word meaning 'giving way, sagging' to be used in the meaning 'breast', since that can hardly be the *defining* characteristic, whereas a 'curve' of the body may well be named after a curve of the landscape. German <Wange> may offer a parallel for that. The 'inlet' itself may have been so named in the first
place because the coast line gives way to the sea. Then,
secondarily, the concave bend may have been seen as a convex bend, a protrusion, of the sea, not the coast. All this makes the correspondence between gji and sinus remarkable, sure, but it could all belong to IE already - in which case it need have happened only once.

The anatomical meaning seems to be the latest. Class. Lat. sinus meant 'bosom' as the part of the garment that covers the breast, especially the fold where you can hide things (hence also 'heart, secret feelings'). It became an anatomical term (<sân>) in Romanian (which shares so many other things with the Balkan Latin substrate in Albanian), whereas e.g. It. seno retains the 'gulf' meaning (as for the rest of Romance, note Fr. sein, Sp. seno 'breast' etc.). What seems to have been lost everywhere is the meaning 'curve, concavity'. As a result, Alb. gji looks like a word with typically Romance semantics rather than something independent.

To sum up, we have an Albano-Romance semantic conspiracy, absence of *sinu- elsewhere in IE (unless I'm missing something), and the fact that the Latin/Romance substrate in Albanian is very thick and involves several chronological layers. For these reasons I believe the case for an early loan from Balkan Latin is fairly strong, even if it's hard to be certain.

> I know of no examples at all of Latin loanwords with s- turning up in Albanian with gj-. Are there appealing candidates?

My other candidate is <gjarpër>, which you analyse rather satisfactorily in PIE terms (but I wonder if you can acount for <shtërpinj> equally well). Until yesterday I had little doubt that the word was inherited, but the parallel case of <gji> and the strange incident of Rom. $arpe (< *sErpen) have undermined my certainty. Here we have another case of Romanian/Albanian agreement, and the likelihood that <$arpe> is a substrate word is practically zilch (see Miguel's note about the treatment of consonantal stems in Romanian). If so, why not derive both words from Balkan Latin *sErpen, "thematised" in Proto-Albanian? That would save us the trouble of explaining the Albanian form independently of <serpent->. I have no other candidates yet, but it's a fresh idea. Other examples may be lurking somewhere.

Piotr


> Now that I'm at it: I would derive Alb. gjarpër from *sérp-n.no- (dissimilated from *sérp-m.no-), a thematicized derivative from a verbal noun in *-men- like Greek stémma => stéphanos 'wreath', Latin vi:men 'osier' (*wéyH1-mn.) => Hitt. wiyana- 'vine' (*wéyH1-n.no-). The two post-consonantal allomorphs seem to be distributed according
to root structure, *-n.no- being used after roots containing a labial, while *-m.mo- is unmarked, and the full form *-mno- is used after a vowel. Since a sonant nasal become Albanian /a/, the first change is to something like *z^érpana which would certainly produce gjarpër//gjarpën. Latin serpent- is no sufficient basis for positing IE *-eno-.