Re: [tied] Lat. niger once again

On 2006-01-13 13:24, Kim Bastin wrote:

> That's very ingenious, and I'd certainly like it to be true. A
> possible problem: if */regnos/ = *[reNnos] > *[riNnos], the immediate
> outcome of n/r metathesis would be *[niNros]. The outcome of *[Nr] is
> uncertain, but would one not expect *[niNgros] */ningros/ rather than
> */nigros/?

Since [N] was merely an allophone of /g/, its phonetic value may have
been automatically adjusted to the new environment created by the
metathesis. On the other hand, the earlier change of *e > i was
_phonemic_ and therefore irreversible (since /i/ was perfectly
acceptable before /gr/).

Piotr