From: Piotr Gasiorowski
Message: 12405
Date: 2002-02-20
----- Original Message -----From: tgpedersenSent: Wednesday, February 20, 2002 1:06 PMSubject: [tied] Re: A "Germanic" query>> Old English lost *x between vowels and also when preceded by a liquid and followed by a vowel (e.g. <seolh> 'seal [Phoca]' vs. pl. <se:olas> from *selxo:s), but not in <-rhl-> (<earhlic> 'cowardly', etc.).
> That doesn't count. Two morphemes, i.e. <-rh-> + <-l->. You'd expect <-x-> to be restored in that case.I wouldn't. <se:ol-as> is also bimorphemic, and yet the <h> is not restored. The boundary between the base and unstressed <-lic> was "close", as in ordinary derivation, not "open" as in compounds: we have <he:alic> from <he:ah>.
> All right. Let's take two mutually exclusive statements
> 1) All <x>'s vanished at the same time
> 2) <x> vanished in some contexts sooner than others
> "Ad hoc pleading" must, as I understand your term, involve assuming an unusual state of affairs. Now which one is the "usual" state of affairs? I have argued for 2). Any German village between Hamburg and Munich in 19th century Germany would have illustrated 2) in the case of the second German sound shift (ik/ich, maken/machen). Your own "seal" example is a case of 2). Why do you then claim that 1) is the usual state of affairs?"Ad hoc" means that you posit a "rule" that produces your example and is not needed for anything else. You claim, in fact, that *x was lost in such a restricted environment that the loss affected the word *erxlaz only -- at any rate, there are no other examples of positional h-loss anywhere in Runic at the time when <erilaz> was attested (not even _gradual_ loss, as in Old English; that happened later). All that in order to save the derivation of <erilaz> via *erxlaz from Pre-Germanic *erklos and to identify the latter with <hercules> or <he:raklee:s>.Piotr