Advances in Biolinguistics - The Human Language Faculty and Its Biological Basis

(Ron) #1

valuation enables us to formulate a generalized system of case valuation by
reducing the two distinct operations postulated in the previous system into a
single operation of the composite of Merge 0 and Search 0.


2.3 The role of formal features: their contribution
to determining labels


Let us consider what role formal features play in narrow syntax – and on what
grounds feature valuation needs to occur in the first place. It is generally assumed
that φ-features of nominal expressions enter into interpretation at the CI interface
and should not be eliminated at that level. On the other hand, φ-features of
v/C-T have no role in interpretation, and should be eliminated before the CI
interface (Chomsky 1995). Their role in syntax is solely to initiate the operation
of Agree to establish an agreement relation (Chomsky 2001). Although the
φ-features play a key role in computation, the status of case features in syntax
is rather obscure. Chomsky (1995) argues that case features do not enter into
interpretation and therefore they are formal features par excellence. As a theory
of case and agreement develops, however, the status of case in syntactic theory
is being threatened with the theory of Agree put forth by Chomsky (2001).
From the viewpoint of case valuation via Agree, the case valuation is seen as a
side effect of the primary operation of Agree. The role of case features is merely
restricted to making a nominal expression visible for the probe-goal agreement.
This and other considerations led some researchers to argue that there is no
role for case features in narrow syntax, and that their contribution is limited to
determining the morphological form of a particular noun at the Sensorimotor
interface (Bobaljik 2008, Marantz 1991).
Contrary to such a view, this chapter suggests that case should play a more
active role in syntax, following the basic insight of Keenan (1999) and Keenan
and Stabler (2003) (see also Baker and Vinokurova 2010). In particular, we
propose that case features of nominal elements, as well as φ-features of verbal
elements, contribute to determining labels of SOs.
In order for a SO to be interpreted at CI interface, the SO must have a label.
Therefore, there must be some syntactic mechanism to determine SO labels.
Chomsky (2013) claims that an operation of minimal search serves to provide
labels to SOs. Elaborating on this idea, Kato et al. (this volume) argue that
the process involved in labeling is to establish a relation of “is headed by”
between a SO and its label. They propose that the application of the two primi-
tive operations, Merge 0 and Search 0 , gives the result of forming a set consisting
of these two elements, and that a semantic operation establishes the relevant
relation based on the set. For example, Search 0 takes a SO, say read books, and
seeks its head read within the domain, and Merge 0 forms a set {{read books},
read}. The application of the semantic operation yields the interpretation of
“read books is headed by read” at the CI interface.
However, it remains unclear in this proposal (and in Chomsky’s labeling
algorithm) how a LI is chosen as a head from the domain.^5 What property helps


Case and predicate-argument relations 51
Free download pdf