On 2008-03-17 00:18, Patrick Ryan wrote:
> I do not see any Latin <f> in the words we have been discussing...
>
> Or is there another rule to get from <f> to <b>?
Allophonic voicing -- as simple as that.
PIE aspirated stops were spirantised in Italic. *dH and *tH changed into
*þ- initially and *-ð- medially in voiced environments (they were in
complementary distribution and so became allophones of the same
phoneme). Initial *þ- fell together with *f- from *bH (and *xW from
*gWH), yielding Lat. f-. Medial *-ð- became Lat. -d-, except when
adjacent to {*r, *u/*w} or followed by *l, in which case it merged with
*-B- (the voiced allophone of *f) < *-bH- and was then "hardened" into
Lat. b:
*-tHlom > *-ðlom > *-Blom > Lat. -b(u)lum
(an anaptyctic /u/ developed in Classical Latin if the /l/ was dark
[pinguis]).
Piotr