Volume 37, Issue 1 - Special Issue: Japanese Syntax in Minimalist Perspective, Guest Editor: Mamoru Saito

This paper explores Merge and proposes a new form of sideward movement ( double sideward movement ) carried out by a new application of External Merge. Double sideward movement occurs in the following way: given a syntactic object S containing XP and YP, Merge applies to XP and YP, and creates {XP, YP} outside S, thus causing XP and YP to undergo sideward movement at the same time. It is argued that multiple clefts (cleft sentences with multiple phrases in the focus position) in Japanese are derived by double sideward movement of the multiple focus phrases and that this derivation is responsible for certain surprising properties of Japanese multiple clefts, some well known and others newly discovered, including the lack of island effects and the presence and absence of clausemate effects. Other consequences are discussed for the nature of operator movement and scrambling as well as for restrictions on the application of Merge.

The present article is concerned with two important observations about argument ellipsis. One is that it may apply to arguments that do not enter into agreement relationship with functional categories but not to those that do; the other is that extraction out of elided arguments is possible in Japanese. In order to account for these properties, I adopt the theory of derivational ellipsis, according to which elliptic constituents are marked as such as early as in the syntactic component, and the hypothesis that v , V, and an internal argument can be combined in different ways in different languages. The proposed analysis does not only account for the relevant facts about argument ellipsis in Japanese, but also extends to other languages that exhibit slightly different behaviors with regard to null arguments.

The primary goal of the present paper is to argue for the hypothesis that labeling is required for linearization, which is called Labeling for Linearization ( LfL ). To achieve this goal, it is first argued that labels are not necessary for semantic interpretation. It is then proposed that labels are necessary for linearization at the PF-interface in that they serve as a device to encode structural asymmetries that are employed to determine precedence relations, which are asymmetric as well. It is also shown that LfL can remove several problems of the original labeling framework. Building on the idea that Spell-Out applies to the whole phase but not its subpart, it is illustrated that the LfL -based analysis can solve the problem concerning the variable ways of applying Spell-Out, which arises in the standard phase theory. Extending the LfL -based framework to Japanese, a novel analysis of particle-stranding ellipsis is also proposed. Incorporating some insights of recent approaches that particle-stranding ellipsis arises through a PF-deletion process, it is shown that the proposed analysis based on LfL offers a theoretically more suitable characterization of the PF-deletion process. In this way, the present article contributes to not only sharpening the core theoretical notions regarding structure building and linearization in terms of labeling but also deepening our understanding of the structure of Japanese.

It is challenging to make empirical arguments either for or against the existence of verb-raising in head-final languages like Japanese since word order facts are not informative in such languages unlike in head-initial languages such as English and French. This article aims to make a novel argument for the existence of verb-raising in Japanese, based on facts about VP-fronting.

This paper examines very early child grammars from the minimalist perspective. It discusses the well-known erroneous strings very young children produce such as Root Infinitives in English and their Japanese counterparts, preverbal object sentences in English, and sentences without Case markers in Japanese. The main question to be addressed is whether those sentences children produce are labeled, and if so, how the labeling takes place. Assuming that ϕ-feature agreement and suffixal Case markers play crucial roles for labeling in English and Japanese respectively (Chomsky 2013; Saito 2016), I consider two possibilities. One is that children are equipped with those almost from the outset although they are not phonetically realized. This means that even the erroneous strings children produced are properly labeled. The other is that those strings are not labeled in the adult way and that children at the relevant stage are still in the process of figuring out how the {XP, YP} structure is labeled in their respective languages. I argue that the latter is a viable possibility, given the parameterization in the labeling mechanism, and receives support from the child data as well. This conclusion implies that a main part of the acquisition of syntax is for a child to discover how her/his target language labels the {XP, YP} structure.