Null Arguments in EA Languages Revisited

The main question this paper addresses is what makes argument ellipsis in Japanese (and Korean) possible. It first presents evidence for the LF copying analysis of the phenomenon. Then, it proposes that Japanese allows ellipsis of all sorts of arguments due to two factors. The first is the absence of j-feature agreement. A DP antecedent copied into an ellipsis site at LF lacks unvalued features and cannot participate in j-feature agreement because of the activation condition. Therefore, the absence of j-feature agreement is a prerequisite for argument ellipsis. The second factor is the presence of suffixal Cases and predicate inflections as weak heads that are unable to provide labels. One of the crucial assumptions in this paper is Norvin Richards’ hypothesis that ellipsis sites count as heads in syntactic structure. It is shown that it makes structures with elided arguments unlabeled in many languages. For example, the straightforward structure with an elided complement would be {H, H}. Then, it is argued that Case markers and predicate inflections as weak heads enable the structures with elided arguments to be properly labeled. Finally, some languages that allow object ellipsis but not subject ellipsis are considered. Chinese, for example, lacks not only j-feature agreement but also Case markers, and allows object ellipsis. It is suggested that Case feature sharing makes the labeling of {H, H} structure possible in this case.