Re: Germanic

From: Piotr Gasiorowski
Message: 264
Date: 1999-11-14

cybalist message #142cybalist: Odp: Cowboys on Horseback
Dear Piotr,
You have demonstrated with a high convincingness that all my attempts to reconcile the Germano-Balto-Slavic kinship hypothesis with the phonetical evidences were too weak. Now, on the ruins of my previous notions, I sincerely wish to see the reconcilability of the Germano-Balto-Slavic Sprachbund hypothesis with the morphological evidences. They are the following (I have taken them from the Cyril's site - Archive 13. Indo-European Proto-dialects):
1. Genitive sg. -eso in Germanic and Baltic, -ó in Baltic and Slavic, -osyo in Indo-Iranian, Greek, Armenian vs. -í in Tocharic, Italic, Celtic, Venetic, Illyrian
2. Instrumental masculine sg. -ó  in Germanic, Baltic and Indo-Iranian
3. Indirect cases sg. & pl. -m- in Germanic and Balto-Slavic
4. Locative pl. -su / -si   in Germanic, Balto-Slavic and Indo-Iranian
5. Degrees of comparison in -tero-, -isto- in Germanic, Indo-Iranian and Greek vs. -samo- in Celtic, Italic
6. Medium voice in -oi / -moi  in Germanic, Balto-Slavic, Indo-Iranian and Greek vs. -r  in Anatolian, Tocharic, Italic, Celtic, Phrygian
1. When the PIE case system became more or less fixed, the genitive ending was *-os (*-es), also in  thematic o-nouns; this made the Gen. sg. of the latter identical with Nom. sg. (as in Hittite). Various daughter languages innovated in various ways to give the genitive (a very important category, after all) a less ambiguous form. Some of the attested variants probably contain an enclitic pronoun (*-jo) added to the old form. In Germanic, the reconstructed ending is *-eso, taken from the declension of pronouns (*kweso 'whose' etc.). In West Baltic the most common ending of the o-stem genitive is -as (sometimes also -es/-is), which some equate with the Germanic ending for no good reason. The phonetic correspondence is weak at best, and the Old Prussian form is most likely just the generalised genitive ending of feminine nouns (root nouns and i-stems providing the other variants). East Baltic and Slavic both have o-stem genitives in *-o: (> Slavic a), which are generally assumed to be old ablatives in *-o:t. Note, by the way, that the forms of Gen. pl. are different in Balto-Slavic (*-oN) and Germanic (*-e:n < [??] *-e-om). 
As for the other branches you mention, of Venetic and Illyrian (I suppose you really mean Messapic) so little is known that they do not contribute anything of value to this discussion. It was once thought that the i:-genitive was an important Italo-Celtic innovation. However, Faliscan and archaic Latin had genitives in *-osjo (just like Greek, Aryan and Armenian), and Hispano-Celtic used the old ablative in -o (< *-o:t) as Gen. sg (just like Balto-Slavic). Since the origin of i:-genitives is unclear, the possibility that they are coincidentally similar forms of different origin is difficult to exclude. Tocharian has various genitive endings, the predominant ones being Toch. B -(e)ntse, Toch. A -es/-is.
On the whole it doesn't seem as if a classification  of IE languages on the basis of Gen. sg. endings were possible. *-osjo is certainly too archaic to be indicative of Graeco-Aryo-Armenian unity, and Germanic, Balto-Slavic and West Slavic innovations are independent phenomena.

2. The Instr. sg. in *-e, which gives contracted *-o: < *-o-e with o-stems, is the best candidate I know for the status of a PIE Instr. ending (its lexicalised traces are also visible in Greek). Shared archaisms (as opposed to shared innovations) cannot be taken as evidence of close genetic relatedness.

3. This is the best single argument.  The *m/*bh division neatly separates Gmc. and B.-Sl. from the rest of the family. To be sure, the only Gmc. form that has this -m- is Dat. pl. (Instr.  and the whole dual declension being lost there), so it's difficult to compare the two branches case for case, but the form of the suffix still looks unique. Whether a case for Gmc./B.-Sl. can be built on this similarity depends on how plausibly it could be regarded as due to convergence (independent development seems hardly likely). In either case the territorial proximity of the branches must be assumed. I wouldn't exclude the possibility of borrowing -- probably with Gmc. as the source. As *bh is a very unusual sound in IE suffixes, its change into *m could be regarded as a kind of phonetic lenition producing a more acceptable consonant. I know examples of derivative suffixes borrowed from Germanic into Baltic and Slavic -- why not a single inflectional element?

4. The Loc pl. endings *-su (athematic nouns) and *-o-isu (thematic nouns, perhaps influenced by pronominal declension) are in all likelihood common non-Anatolian IE, so the same comments as in 2. (above) apply.

5. The situation is more complex than that. The most widespread suffixes (traceable back to PIE) are *-jo:s/-jes-/-is- (comparative) and *-isto- (superlative). In addition, we find *-tero-, *-mo-, *-t@...- (with Italo-Celtic byforms of the *-s@...-/-ismo- type). They probably had their own uses in PIE, but in several branches they compete for survival with the former set. Neither *-tero- nor *-t@...-, which predominate in Sanskrit, are characteristic of Germanic, where the regular suffixes contain *-iz- and *-ist-. Traces of *-jes-, *-isto- can be found practically everywhere in IE as shared archaisms. Except for the Italo-Celtic innovations, the suffixes of adjectival gradation don't tell us much about the relatedness of the branches.

6. Balto-Slavic languages do not have real mediopassives; they just display some active-voice forms interpreted (justifiably, in my opinion) as old mediopassives. In Germanic, the existence of passive-voice inflections is guaranteed by Gothic. There is no generally accepted theory of middle-voice endings in PIE. I think most Indo-Europeanists would subscribe to the view that both r-mediopassives and endings with *-oi (*-ai) existed in PIE, but their distribution and/or function underwent serious changes in the descendent languages. The relevant fact are so obscure that no convincing argument can be built on them.

To sum up: Of the cases you quote only one (*m/*bh) can be regarded as convincing. Another possible indication of Gmc./B.-Sl. unity (not in your list) is the development of the so-called "strong" adjectival declension with pronominal endings (as in Russian milyj 'nice' < *mil@-j@, declined like a pronoun, as opposed to mil < *mil@, formerly declined like a noun). Since the details are quite different in all three groups, it's probably a common areal tendency rather than a shared innovation, but still the weakest conclusion that can be drawn is that Germanic, Baltic and Slavic can't have been spoken very far from one another.