Notice that the occurrence of Y^0 is obviously not a term of X^0 (and vice versa).
Then, such instances of Merge should count as EM under Chomsky’s defi ni-
tion (8). However, head-movement is clearly a movement operation just like
IM, and it yields copies of X^0. The informal tree notation in (14b) may look
as though the operation counts as a kind of “sideways remerge,” yielding
“multidominance” structures as in (14b). No matter how we grasp the intuition
behind (14), however, the point is that the replacement of Y^0 in an SO with
a Y0max category is beyond the generative power of Merge. Traditional examples
of head-movement, such as T^0 -to-C^0 and V^0 -to-v^0 , should be reanalyzed along
these lines of reasoning.^4
Now, note that head-movement as depicted in (14) yields another kind of
“symmetric,” balanced branching, in this case of the form {X^0 , Y^0 }. It differs
from core cases of IM only in the “size” of the relevant constituents ({X^0 , Y^0 }
vs. {XP, YP}), so it is reasonable to suppose that this movement operation also
satisfi es Fukui’s (2011) generalization in (11), serving to form a symmetric
structure.
In this manner, Fukui’s (2011) generalization in (11) can unify the class
of possible movement operations, XP-movement/IM and head-movement,
under the category of symmetry-formation. Notice that in earlier theories
of X-bar-theoretic syntax, it is plainly stipulated that XP-movement must
target some “Spec” position (the sister of an X’-phrase), while X^0 -movement
must target another X^0 position. This restriction does not rest on principled
grounds, and, what is worse, its formulation makes heavy recourse to
projection-based notions like “Spec” and “X0(max),” hence it has lost its basis
in Merge-based syntax without projection. However, (11) naturally provides
a simpler, projection-free characterization of the two positions, a highly
desirable result.
3 Dynamic Symmetrization Condition
In the preceding section, we introduced Fukui’s (2011) generalization in (11)
and argued that it can naturally unify the target positions of XP-movement/
IM and head-movement, i.e., {XP, YP} and {X^0 , Y^0 }, respectively. Fukui’s notion
of symmetry is simply defi ned in terms of the LI vs. non-LI/phrase distinction,
but we will see in this section that this is insuffi cient and should be modifi ed
by referring to the distribution of formal features within relevant SOs
Under Fukui’s (2011) proposal, all SOs of the form {XP, YP} are regarded
as symmetric and stable. However, there are ample cases in which XP moves
out of {XP, YP}, breaking the “symmetry” in Fukui’s sense. For example,
consider (15a), in which the external argument nP, once externally merged
with {v, VP}, is required to move out of this SO into Spec-T (the effect of
the so-called “Extended Projection Principle,” EPP). Another example is the
wh-phrase at the edge of (15b), which does not match with [–WH] C and
hence is required to undergo successive cyclic movement out of this SO.
Feature-equilibria in syntax 15