On 2008-09-14 19:37, Brian M. Scott wrote:
> > It's a regular change in WGmc. (via *-lð-, perhaps
> > independently in different WGmc. languages). Typical
> > examples include <fealdan, feld, gold, wilde> etc.
>
> Does <gold> belong here? OHG has <gold>, which, like ON
> <gull> and Goth. <gulþ->, should be from *gulþa-. For what
> it's worth, Ringe includes this one with other neuter
> a-stems in which the daughters show different resolutions of
> a Verner's law alternation in what were originally derived
> collectives.
But there are no unambiguous traces of a PGmc. Vernerian *Gulða-. As the
expected development of *Gulþa- in English is <gold> anyway, parsimony
favours *-lþ- as the only proto-form. Of course WGmc. *-ld- in this
lexical set represents a dialectal voicing which has nothing to do with
VL. OHG -ld arose word-finally from unvoiced *-lþ- (the thematic *-a-
was lost very early). In OE, -ld- was generalised, spreading
analogically from inflected forms (where it was intervocalic) and
ousting -lþ. There are a few examples of <-lth> in very early OE
(<spilth, felth>) and occasional instances of reverse analogy (<feltha>
for <felda>, <halði> for <healdðe>). Hogg attributes them to "confusion
between the spellings <d> and <ð> and hence <th>", but the marginal
survival of archaic /þ/ at the end of words is more likely, IMO; Hogg
himself assumes it in compound names before a second element beginning
with /h/ (<Balth-, Balð-> as opposed to <b(e)ald>).
Piotr