2 Feature-equilibria in syntax*
Hiroki Narita and Naoki Fukui
1 Introduction
Linguistic expressions exhibit a variety of formal and interpretive properties at
the SEM(antic) interface with the C(onceptual)-I(ntentional) system, depending
on how they are structurally organized. In particular, it is known that for each
syntactic object (SO) Σ, a certain designated lexical item (LI) – called the head
of Σ – plays a central role in determining the formal/interpretive properties of Σ.
For instance, major properties of a verbal phrase read a book are determined by
its head verb read, yielding its event-predicatehood and θ-roles, among others.
In contrast, nominal phrases like a book show properties different from verbal
phrases, such as being referential, indefi nite, third person, etc., each provided by
their head noun (or perhaps determiner). This traditional observation is com-
monly referred to as the endocentricity (headedness) of phrase structure:
(1) Endocentricity/headedness of phrase structure:
Interpretive properties of an SO Σ are determined largely by the features
of a unique head LI within Σ.
The proper analysis of endocentricity is one of the most important research
topics in generative linguistics. Essentially since Chomsky’s (1970, 1981) intro-
duction of “X-bar theory,” it has been predominantly stipulated that endocen-
tricity is characterized by the notion of “projection.” Projection can be
understood as a relation in which a certain LI transmits its lexical features to
nodes that “dominate” (i.e., contain) it – a very strong form of (total)
feature-percolation. An X-bar-theoretic characterization of endocentricity typi-
cally consists of the following assumptions.
(2) Universal projection:
All SOs are not “bare” but always associated with nonterminal symbols (or
“labels”) such as NP, C’, etc., namely “projections” of some LIs (features
of a head LI, perhaps accompanied by bar-level indices).
(3) Projection = endocentricity:
Projection is the device to encode endocentricity.