08-08-03 14:57, Miguel Carrasquer wrote:
>>Leaving this marginal problem apart, the question remains why we don't
>>get *ber-/*bra-/*bur- (no matter if the initial was phonetically *[b-]
>>or *[B-]) rather than *fer-/*fra-/*fur- in unstressed prefixes.
>
> Yes. Is *ga- the only preverb affected by Verner's law?
I think so. Interestingly, *ga- is also exceptional in that it has no
free-standing counterpart and that it's _always_ unstressed, not only in
verbs but also in nouns and adjectives. Thus, while hypothetical *bur-'
would have alternated with *'fur- and with the free adprep that yields
Eng. <for>, *ga- did not alternate with anything. Isolated archaic forms
like *xanso: < *xansso: < *kom-dH(h1)t-ah2 were synchronically obscure
and didn't count morphophonologically. What I'm suggesting is of course
that the effect of Verner's Law was _eliminated_ very soon in the case
of the alternating suffix but was retained in *ga- because all evidence
of the original form had been lost.
Piotr