Re: [tied] Re: Ingvar and Ivar

From: Piotr Gasiorowski
Message: 6207
Date: 2001-02-26

 
----- Original Message -----
From: tgpedersen@...
To: cybalist@yahoogroups.com
Sent: Monday, February 26, 2001 12:43 PM
Subject: [tied] Re: Ingvar and Ivar

> Inneresting, as you would say. Does that mean that a proper linguist should distinguish rigidly between two kinds of rules, normal and banal/non-puzzling/trivial?
Nearly all "normal" rules are rather banal. A proper linguist may distinguish between phonologically regular and sporadic changes, and accept the commonplace fact that phonological changes of a trivial "unmarked" type, as typologists put it, may apply irregularly if there is a natural motivation for them. Frequently used words are a natural target for subtractive processes (or the "erosion" of phonological substance), since they can be easily recognised in the right context even if there is some transmission noise. This is why English "head" is not "heeved" and why "mistress" before a woman's name has been shortened to [mIs@...].

> The comparative evidence for the prothetic H1, AFAIK, comes from the desire to see the "tooth" as an "eater", thus a -nt- participle of *H1-d- 'eat', and another desire to account for the Greek prothetic vowel o-. This means trouble trying to derive H1 > o (bad!). How about NomSg *H1édon-, AccSg *H1edént- (or even *H1odént-?) (in conformity with the priciple of stressed /e/ and unstressed /o/) and then the general ensuing paradigmatic confusion providing Greek o- and also (by paradigm levelling) accounting for Greek -o- and Latin -e- in the stem?
 
If anything, *o was more common under stress and *e in unstressed syllables unless levelling out or vowel-colouring occurred. IE *e was also the "dummy vowel" used in various late derivational processes that didn't obey the stress-based ablaut of older formations such as root nouns. A prothetic vowel in the "tooth" word occurs also in Armenian, and Hittite has the actual participle adant- 'having eaten, full', which can only be *h1d-ont-, not *h1ed-ont- or *h1od-ent- (?!). The type of stress movement you'd like to suggest is not supported by any real data (Greek accentuation or anything else). Ask any IEist, if you don't like "my opinion", and they will most likely recommend Acc.sg. h1d-ónt-m, Gen.sg. h1d-nt-ós, Nom.pl. h1dónt-es, etc. as reconstructions that best account for the observable reflexes. Latin -en- reflects the syllabic *n of the nil grade, generalised there as it was in Gothic. Greek o- in odous is only "wrong" if you insist that *h1 must be reflected as Greek e-, which doesn't happen to be my view. Greek had its own umlaut-like processes that sometimes affected the prothetic vowel.

> As I understand you the down-to-earth explanation *tans > *tanTs (*tanTus?) > zand > zan should be preferred to *tans > zan because the latter smacks of fantasy?
 
No -- the latter smacks of fantasy BECAUSE the former (*tanTuz > *zand > *zan), though more complex, is supported by good EVIDENCE (Gothic tunTu-, OIcelandic tönn, parallel examples of other root nouns), while *tans is just a conjecture about what PIE *h1donts would have looked like in Germanic if its development had been phonologically regular. It was not, since morphological processes intervened. To tell the truth, I'm slightly fed up with this hairsplitting debate. So much has been said about dental problems that we might just as well pick a new root to discuss.
 
> BTW, is this the split-up of ProtoGermanic you advocate: ProtoGermanic > Proto-NW-Germanic and Proto-East-Germanic?
Splits are never 100% neat if the post-split groups remain within contact range. Splits and areal effects are complementary processes and any view of linguistic evolution that dogmatically ignores either type of "language transmission" (genetic vs. diffusional) is necessarily simplistic. To the extent that the family tree metaphor is applicable to Germanic, I'd say that North and West Germanic languages cluster together in terms of shared innovations, with Gothic being an outgroup (I don't like the arbitrary term "East Germanic", which in practice means "Gothic" anyway, since the other "EG" languages died out unrecorded). This doesn't mean that "Proto-NW-Germanic" was ever completely homogeneous, or that it split neatly into North and West Germanic. "West Germanic" is another artificial unit -- a cover term for at least three groups that don't seem to be more closely related to one another than any of them is to Scandinavian.

> Actually, what I wanted to do was to argue for the existence of "difficult" alternating stems in ProtoGermanic (in Polish they have <ruka>/<w ruce>, in case you didn't know:) Most sensible Slav languages have obliterated that). This would make life easier for my
proposed Nom. *dan-/Obl. *tan- stem. But you have admittted the existence of such stems so that obstacle seems out of the way now.
 
There is no <ruka> in Polish (it's <re,ka>, with a nasalised vowel). It isn't a difficult alternation but an old stem-final palatalisation, completely predictable in *all* Polish feminines in -ka (we also have -ga : -dze). Russian eliminated this type of alternation (or maybe didn't carry it out consistently) but it survives even in Ukrainian and Belarusian, not to mention the rest of Slavic, so your remark about "most sensible Slavic languages" is factually wrong. Nobody denies the existence of alternations, but as for *dan-/*tan- -- the devil's in the details. Only a definite etymological proposal can be evaluated, not a loose idea.
 
Piotr