(A) Case for labeling: labeling in languages without φ-feature agreement

This paper explores the labeling mechanism in Japanese and other languages that lack phi-feature agreement. It is proposed in Chomsky (2013) that phi-feature sharing between the subject and T makes the labeling of finite clauses possible. It has also been assumed that Case is required for and valued through phi-feature agreement. (See, for example, Chomsky 2008.) These raise fundamental questions about the syntax of languages without phi-feature agreement. Adopting Boskovic's (2007) proposal to divorce Case valuation from phi-feature agreement, I argue that suffixal Case and predicate inflection make phrases opaque for search, and serve to aid labeling in Japanese. The hypothesis provides explanations for the outstanding syntactic properties of the language, such as multiple subject sentences, semantically vacuous scrambling, compound verbs with specific constraints, and argument ellipsis. The analysis of argument ellipsis is based on Richards' (2003) hypothesis that elided constituents count as heads in computation. In Section 5, I extend the analysis to other phenomena, including N'-ellipsis in Japanese and object ellipsis in Chinese. I also consider the labeling of finite clauses in Malayalam, which has neither subject agreement nor suffixal nominative Case. I suggest that Case valuation, instead of phi-feature agreement, creates the configuration of feature sharing for labeling.