01-11-03 22:43, Richard Wordingham wrote:

>> The step /ft/ > /pt/ in
>> Romanian also offers no phonetic problems, although /f/ does not
> otherwise
>> yield /p/ in Romanian (only in the clusters /pt/, /ps/ < */ft/,
> */fs/).
> Are there other examples of this limited change? I couldn't think
> of any, which is what bothers me about the idea.

It's similar to the extremely limited change of *fs > ps in Old English
(*wafsa- > wæps and *rafs- > ræfs-/refs- (Early Mercian) > ræps-), which
parallels the much more frequent "hardening" of *xs > ks, as in <fox,
oxa, six, weaxan> etc.