From: Piotr Gasiorowski
Message: 18419
Date: 2003-02-04
----- Original Message -----
From: <x99lynx@...>
To: <cybalist@yahoogroups.com>
Sent: Tuesday, February 04, 2003 3:06 AM
Subject: [tied] Latin versus *Proto-Romance
> Piotr, did you see the trees on this website? It has simply "Latin" as the single parent giving rise to two branches -- "Proto-Romance" and "Classical Latin".
Yes, but it's "Latin sensu lato", not to be identified with the written language of the Classical period (though including it together with numerous other varieties). The common ancestor of all those Latins was the Italic language of Latium (Old Latin). It was certainly non-homogeneous either (though presumably less variable than Imperial Latin, since its geographical range was rather small), and it was surrounded by its close cousins, later conquered and absorbed by Latin.
> "Vulgar Latin" is identified as a lower node, "Proto-Western
Romance", with all of Romance deriving above from "Proto-Romance", and
"Classical Latin" being an isolated branch-off.
> In any case, reconstructing a parent from modern Romance languages without using knowledge of either "Latin" or "Classical Latin" would seem to be a dandy reality check for the comparative method, if it were possible. I would think that a lot of the changes that are apparent with hindsight would not be so obvious in a blind-fold test. The accusative nominative displacement would not have been easy to detect, for example, and I suspect it would have been difficult to reconstruct the nominative in Classical Latin or its presumably similar "Latin" parent.
It's been done, and the results are often quoted in handbooks. Many familiar Classical words and some grammatical features are virtually unreconstructible from Romance evidence (e.g. <equus>, though its feminine counterpart <equa> is reconstructible; the neuter gender, OV syntax, etc.), and there are reconstructed Proto-Romance words unattested in written Latin. If the Romance languages were the only IE languages known to us, we'd never learn that the old m./f. accusative ending was once -m.
> Let me put a slightly different scenario to you. The early loan.
> We've reached the stage where PIE's dialects have lost mutual
comprehensibilty and therefore might be called THREE distinct languages.
PIE Jr.(A) is essentially unchanged from PIE. PIE Jr(B) is changed a bit. PIE Jr(C)is changed a bit more. A new word is innovated or borrowed from non-IE by PIE Jr(C). This innovation is a word for a horse -- lets call it <ekWos>. The word is borrowed by PIE Jr(A) and (B). Let say there was nothing in the sounds of the word that would cause it to change during borrowing. Let's say all the specific sound changes that might affect the word in the individual languages happen after the borrowing. And that the word in all its later expected forms survives in ALL the attested daughter languages
> Am I wrong in assuming that this word would show up comparatively as being from the PIE parent?
A minor correction. It was *(h1)ek^wos, not *ekWos (*k^w is a cluster, *kW is a unitary phoneme). Other than that, you are not wrong at all, and it's indeed something I often argue myself. It would be easy to give examples of words borrowed rather late but still looking like genuine proto-stuff -- like the word for 'king' in Slavic (*korljI), which almost certainly is to be identified with Charlemagne's name. Technological vocabulary, such as the "horse-and-wagon" terminology diffuses easily.
One proviso here is that its spread would have had to be really early -- before the Satem innovation, and thus well before 2000 BC. But if it had spread ca. 3000 BC, give or take a little, the results would have been indistinguishable from common inheritance, I suppose. It may be significant that the IE words for 'wagon' and 'wheel' don't occur in documented Anatolian, and the Anatolian 'horse' word seems to be a para-Indic ("Mitanni Aryan") loan. Perhaps Proto-Anatolian was already spoken too far from the bulk of IE when the first horse-drawn vehicles appeared.
Of course, even if *ek^wos is not literally PIE, it may still be IE (an etymological connection with *o:k^u- 'swift' has often been suggested) rather than borrowed from an external source. But there are further complications. Even if it represents a native IE formation, and if it meant 'horse' in early IE, it's hard to prove that the original meaning was 'domesticated horse' rather than 'any horse'. In other words, the use of *ek^wos for the dating of horse domestication relative to IE dispersals is methodologically questionable.
Piotr