linguistic theory that while φ-features carry out important tasks in narrow syntax,
the role of case features remains rather obscure. This chapter suggests that case
features of nominal expressions, as well as φ-features of verbal elements, do in
fact play an active role in determining labels of syntactic objects (SOs). Two
ways of labeling are available. One way is based on case features of nominal
elements, while the other way is based on φ-features of verbal elements. The
former is typically employed in dependent-marking languages (to use Nichols’s
1986 terms) such as Japanese, whereas the latter is utilized in head-marking
languages like Mohawk. Thus, in dependent-marking languages, case serves as
a key feature in building up the system of predicate-argument relations, which
constitutes a core portion of the human language of thought. The role of case
in this type of language, therefore, actually corresponds to the role of agreement
in head-marking languages.
This chapter is organized as follows. Section 2.1 puts forth a proposal for
case valuation in terms of Merge, in contradistinction to the widely assumed
Agree-based case valuation. Section 2.2 discusses how the above-mentioned two
modes of case valuation can be reduced to a single operation, by incorporating
the ideas proposed by Kato et al. (this volume). Section 2.3 discusses the theo-
retical status of formal features such as case and φ features in narrow syntax.
Section 3 presents some consequences of the proposed case theory for the
analysis of peculiar case phenomena in Japanese. Section 3.1 deals with multiple
occurrences of identical cases, and Section 3.2 discusses the phenomenon of
case alternations, particularly nominative-genitive conversion. Finally, Section 4
presents the conclusion.
2 The case system
2.1 Two modes of case valuation
The standard view of case valuation in recent generative literature is to relate
case valuation with φ-feature agreement (Chomsky 1995, 2001). This view faces
empirical challenges from the type of language that does not exhibit φ-feature
agreement. How does case valuation take place in such non-agreeing languages?
How does it account for various peculiar case phenomena in those languages?
In such phenomena, the conditions for Agree are seemingly broken. In order
to resolve these questions, Zushi (2014a, b) revives and develops the view of
case assignment on a structural basis, drawing on “classical” works such as
Kuroda (1965, 1978, 1983, 1986, 1988), Fukui (1986, 1988), and Saito (1982),
in a way that accommodates minimalist conceptions. Central to these classical
works is the fact that the case assigned to a nominal depends on its position in
the phrase structure. Therefore, Zushi reformulates the structural case assign-
ment view in terms of external Merge, a structure-building operation in the
minimalist program. Zushi claims that a case feature is valued by the following
mechanisms in non-agreeing languages such as Japanese.^1
Case and predicate-argument relations 47