1 Introduction
Linguistic items (LIs) are bundles of features. Among the features of LIs, the
role of formal features in narrow syntax is very important, as formal features in
their interaction with semantic features crucially characterize human language
in a way that distinguishes it from other computational systems and other types
of symbolic systems. What role do computational atoms like formal features
play in human languages? In particular, what role do formal features, such
as case and φ features, play in syntax? These are among the most contested
issues in contemporary theoretical linguistics. This chapter intends to contribute
to the understanding of these issues by intensively investigating the nature of
formal features with a special focus on case features.
The dominant view on case in current linguistic theory is the one put forth
by Chomsky (2001), according to which a case feature of nominal expressions is
valued by means of Agree. On this view, case valuation takes place contingent on
φ-feature agreement. Although this way of case valuation appears to work well in
agreeing languages, including well-studied Indo-European languages, it faces
serious empirical challenges when our investigation goes beyond such familiar
languages and begins to explore empirically adequate theories of other types of
languages, including Japanese, in which φ-features obviously play no role in syntax.
How do case features of nominal expressions obtain their values without recourse
to φ-feature agreement in non-agreeing languages? How can various peculiar case
phenomena be explained? Several conditions of the application of Agree are seem-
ingly violated in these phenomena. In order to resolve these questions, Zushi
(2014a, b) proposes that case valuation in terms of external Merge, in addition
to Agree, should be available in UG. This chapter presents how this proposal
provides natural accounts of peculiar case phenomena in Japanese, such as multiple
occurrences of identical cases and case alternation. Additionally, we attempt to
reduce the two modes of case valuation into a single operation by incorporating
the proposal made by Kato et al. (this volume) that the alleged basic operations
of Agree and Merge are essentially the same operations.
Furthermore, this chapter addresses the question as to what role formal
features play in human language. It has been generally assumed in current