This discussion originated (IIRC) from me wondering about
sui generis constructions (i.e. constructions with
unique properties, rather than merely unique combinations
of nonunique properties). The discussion of 2BE has
forced me to look at wh-clefts, which I'd always tried
to ignore as a problem too thorny to risk getting
entangled with, and I find that the best initial analysis
I can come up with exhibits a unique property:
1a. [Dress herself]1 is [[what]1 she did _1 next].
1b. [Dress herself]1 is [[the only thing]1 she did _1 next].
2a. [Proud of herself]1 is [[what]1 she is _1].
2b. [Proud of herself]1 is [[the only thing]1 she is _1].
-- the unique property being, two (nonconjoined) fillers
correspond to a single gap. Each filler-gap relation
separately has counterparts elsewhere in the grammar, but
the two together have no counterpart, and indeed violate
an otherwise-robust extraction constraint.
--And.
|