Re: [tied] Grimm and Verner

It is not _impossible_ to date Verner's Law (VL) before Grimm's Law (GL), and thought-experiments of this kind (including attempts to combine both laws into one) aren't all that new. The question is whether we can gain anything by doing so. Let's call the hypothetical pre-Grimm lenition "Revised Verner's Law" (RVL). The new formulation would be like this:

Pre-Germanic voiceless stops (*p, *t, *k) remain voiceless if initial or immediately preceded by a stressed syllable, or forming a cluster with another voiceless obstruent (e.g. *ks, *ps, *pt); otherwise they become breathy-voiced (> *bH, *dH, *gH). Under the same conditions pre-Germanic *s > *z.

The problem here is that we turn the neat and parsimonious formulation of VL into a disjunctive statement: instead of applying to a natural class of segments (voiceless fricatives) and modifying them in the same way (voicing), RVL has to handle voiceless stops and *s separately. In fact, VL breaks down into two independent changes that mysteriously happen to operate in the same non-trivial environment.

One can try to eliminate this particular hitch e.g. by positing *s

> *zH (different from plain [z]) instead of simple voicing (voiceless

obstruent > breathy-voiced obstruent), with the subsequent change of *zH > *z by "generalised" GL, but introducing an otherwise unknown "transitional" phoneme completely ad hoc, just to force a preconceived idea, is also inelegant. Besides, I have yet to see /t/ "lenited" to /dH/ before I believe that such a process is possible, while the voicing of intersonorant fricatives is one of the most natural types of lenition.

In brief, if VL is ordered chronologically after GL and is for the most part fed by it (that is, applies to the outcome of GL), we get a simple and natural generalisation, while the ordering of RVL before GL leaves us with a less natural-looking process and makes us resort to fishy formal tricks in order to maintain the same level of generalisation. In such cases linguists do not hesitate to prefer the more elegant solution.

Pre-Germanic *wúlkW-o- and *wulkW-í:- were independent lexemes and the irregular change of *kW > p (or *xw > *f after GL) need not have affected both of them. It would be naive to argue that whatever happens to the he-wolf must happen to the she-wolf (or the other way over) by analogy. Old English had masculine <fox> and feminine <fyxe> (ME fixen), transparently related at that time; now we have standard English <fox> and <vixen>, the latter -- _only_ the latter -- showing sporadic (West Country) voicing.

Piotr

----- Original Message -----
From: Giuseppe Pagliarulo

To: cybalist@yahoogroups.com

Sent: Sunday, December 16, 2001 8:58 PM

Subject: Re: [tied] Grimm and Verner

Good day, Piotr and all.

Piotr wrote:

(2) VERNER'S LAW

A voiceless fricative (*f, *þ, *x or *s) remains voiceless if it is initial or immediately preceded by a stressed vowel, or if it forms a cluster with another voiceless obstruent (e.g. *xs, *fs, *ft); otherwise it becomes voiced:

This is the classical formulation of the law and the one I've been taught myself. It implies that this voicing phenomenon took place after the operation of Grimm's law. However, I remember reading in P. Ramat's _Einfuehrung in das Germanische_ that there is a new theory proposing that Verner's law is an independent lenition process predating Grimm's law. This seems quite uneconomic to me, but then I think of Old Norse _ylgr_ "she-wolf" where the voicing of IE *kW seems to have taken place before the *kW > *p (> *f) shift. I'd like to know your opinion about it.

Thanks in advance,

G. P.