> But the question remains, is the remaining sheva silent or voiced? I
> always assumed that in forms like B.IQ:$W. the sheva was
> silent, and the
> dagesh was dropped to correspond to this lack of
> vocalisation.
According to the usual rules, the shva following a lengthened consonant
could not be silent. Jouon suggests that the dagesh was dropped because
even a vocal shva was too weak to sustain a full lengthening of the
preceding consonant.
> If it is
> voiced, how can I decide whether sheva is voiced or not in
> any word of
> this pattern?
You would have to know whether the consonant should be lengthened. If
it's a verbal inflection, you would know based on parsing. Even if it's
not, you could generally tell by the preceding vowel.