On 2009-02-03 11:44,
the_black_sheep@... wrote:
> I'm unclear as to what happened in *lið-a- and *swaB-ja-:
> *lið-a- < *leiþ-a- would be:
> voicing of *þ - (root-final fricative in derivation) Verner's Law
Yep. Same with *p > *f (Grimm's Law) > *B (Verner's Law).
> *ei > *i: - why the short vowel?
The present stem vocalism was PIE *ei (> OE i:), alternating with PIE
*oi (> PGmc. *ai > OE a:) in the strong stem of the Gmc. preterite and
short *i (the zero grade). It's the same alternation that you find in OE
wri:tan 'write' vs. writ (n.) 'writ, scripture'.
> *swaB-ja-
> does it come directly from PIE or is it derived from PGmc *swef- ?
> it would seem then that the same rule as above applies (Ringe 2006:
> 217)...
> on the other hand *swop- > *swaB- also explains the vowel, so I'm
> just wondering...
The _type_ goes back to PIE. Most verb stems formed PIE causatives
(and/or iteratives) by adding the formative suffix *-éje- to an
unaccented o-grade of the root. Compare *nes- 'return home, survive'
(cf. OE (ge)nesan) --> *nos-éje/o- 'cause to survive' = 'save' (hence
PGmc. *nazja/i- > OE nerian). There are also lots of examples in other
IE branches, e.g. Pol. mrze 'dies' < PSl. *mertI < PIE *mér-ti vs. morzy
'causes to die' < PSl. *moritI < PIE *mor-éje-ti.
Thus, the relationship between *swef-a-/*swif-i- (OE swefan 'sleep'
[3sg. swif(e)þ]) and *swaB-ja/i- (OE swebban 'send to sleep, put to
death' [3sg. swefeþ]) was historically regular and productive. This
means that the formation of the causative is hard to date; it could have
happened at any time between PIE and early Germanic.
By contrast, the type of *swó:p-je/o- (> PGmc. *swo:f-ja/i- > ON sø:fa)
is archaic, residual and anything _but_ productive in Germanic, so it
must be old. There are very few causatives of this kind attested anywhere.
Piotr