The discussion was based on the fundamental hypothesis that Merge-based
syntax is essentially “symmetric,” free from any asymmetric representations like
left-to-right ordering or projection, and further driven by the need for feature-
equilibria. Note that this is in signifi cant contrast to the still-dominant hypothesis
that “asymmetry” of some sort is the norm for syntax. The traditional stipula-
tions of universal endocentricity/projection (2)-(4) are examples of such an
asymmetry-oriented approach, and so are various versions of Kayne’s (1994
et seq.) “antisymmetry” approach (see also Moro 2000), all of which clearly fall
short of capturing the fundamental transitional directionality from F-asymmetry
to F-symmetry in (24). Although a number of important implications of the
proposal remain to be explored and examined carefully, we hope that the analysis
presented in this chapter provides a promising insight into the nature of linguistic
computation and the syntax-semantic interface of human language.
Notes
∗ Part of this research is supported by the Japan Science and Technology Agency
(CREST) and by the Japan Society for the Promotion of Science (Grant-in-Aid
for Scientifi c Research, Scientifi c Research (A) (General) #23242025, and Chal-
lenging Exploratory Research #25580095).
1 Here and in what follows, we will freely use notations like XP, YP to refer to
phrasal constituents. However, it should be noted that we are not granting any
ontological status to nonterminal label-symbols such as XP, or to any label-based
structural notions like Spec or complement. That is to say, no notion of labeling
or projection is implied in our informal usage of these terms. Thus, notations
like TP, vP, and so on, refer to nothing more than phrasal constituents containing
T, v, etc.
2 We may say that nominals are headed by D(eterminer) (Fukui and Speas 1986,
Abney 1987, and Fukui 1986/1995, 1988) or by the categorizer head n into
which D incorporates (Chomsky 2007). We refrain from making specifi c assump-
tions regarding the exact nature of the nominal head, for want of better under-
standing of nominal-internal syntax.
3 This systematic transitional pattern cannot be captured by, e.g., Moro’s (2000)
approach, which holds that movement is instead a “symmetry-breaking” (i.e.,
asymmetrizing) phenomenon serving for linearization (or labeling, see also Chom-
sky 2008). We will provide various arguments bearing on Moro’s and others’
hypotheses that asymmetrization is the norm of syntactic computation.
4 When we discuss the X^0 -Y^0 -merger, it is not accurate to provide an asymmetric
characterization like “X^0 moves to Y^0 .” In the symmetric {X^0 , Y^0 } structure, the
two LIs are equally prominent, and neither is a “target” or “adjunct” of the other.
In the traditional description, the X^0 -Y^0 -merger has been assumed to feed
“upward head-movement” at the PHON(ological) interface, ensuring that {X^0 ,
Y^0 } is pronounced at the higher position of Y^0. However, since X^0 is no less
prominent than Y^0 in {X^0 , Y^0 }, there is no reason to exclude the possibility that
{X^0 , Y^0 } instead gets pronounced at X^0 , resulting in a “downward head-movement”
form at PHON. For example, Pollock (1989) argues that the V^0 /v^0 -movement
in Romance languages gets pronounced at T^0 , whereas its counterpart in English
yields T^0 ’s downward attachment to V^0 /v^0. Under our analysis, we may suppose
that one and the same SO {v^0 , T^0 } may underlie these two different phonetic
forms. We will leave further exploration of this approach for future research.
Feature-equilibria in syntax 25