On 2006-01-27 10:36, tgpedersen wrote:
> Also, in those stems where the initial consonant C1 was subject to
> Grimm, the reduplicated forms C1eC1C2- the second occurence of C1
> would be subject to Verner, which made the stem unrecognizable. Then
> better leave it out altogether, with compensatory lengthening of the
> vowel (perhaps via a diphthong).
But the de-reduplication of most verb classes must have happened before
Verner's Law; otherwise VL (where applicable) would have affected the
initial consonant of the preterite singular, e.g. *se-sókW-e > *sezaxW >
*zax(w). This is actually what some relict forms show in Class VII with
reduplication still preserved, e.g. Goth. sai-zle:p 'slept', OIcel.
se-ra 'sowed'. In most cases Gothic solved the problem by cancelling
Verner's Law through paradigmatic levelling (--> sai-sle:p-), while the
NW Germanic languages either "recycled" the restructured reduplication,
generalising the same allomorph for both numbers in the preterite, or in
fact abandoned reduplication altogether in favour of a long vowel. The
first option (as in Angl. heht, hehton) was extremely rare because in
such cases the conjugation became practically suppletive. Some of the
original reduplications are so rare and so odd that their isolated
occurrences have often been dismissed as scribal errors, e.g. the OE
hapax <blefla>, the pret. of <bla:wan> -- a beautiful reflex of
*b(l)e-blo:w (with the first /l/ analogically restored in the
reduplication syllable)
Piotr