The Development of the Nominative Case Marker GA : A Competing Constraints Approach

The establishment of the particle ga as a nominative case marker in the Muromachi-period is supposed to mark the beginning of Modern Japanese. A theoretical question is whether this directly reflects a drastic change in the syntactic make-up of the language. Researchers in Japanese philology (Kokugo-gaku) have implicitly answered this in the negative for a long time, while generative syntacticians explicitly claim the opposite. Although the generative approach claims to be scientific, and therefore close to the truth, it is still doubtful whether the change in behavior of particle ga directly pertains to structural change. In Old Japanese (OJ), ga is only one of the possible nominative markers; the subject argument can be realized in such forms as ga, no, zero, wa (or mo and other kakari-zyosi), and their distribution is syntactically overlapping. If case markers only indicate the structural position where they occur, as generally assumed by generative syntacticians, all the markers (including zero) in OJ should be syntactically distinguished. However, there appears to be a considerable amount of overlap and optionality in the choice of subject markers. The problem, I believe, is that while ga is a case marker, it is not just a label of a syntactic node, but a morpheme with its own content. Along the lines of Kikuta (2003, 2005), this paper proposes to separate abstract case marking as a type of argument licensing on the one hand, and morphological case marking as a choice of the most adequate phonological/morphological realization of the abstract case on the other. Following the ideas of the lexicalist frameworks HPSG and LFG, I assume that the abstract case marking is done lexically by a head predicate selecting (subcategorizing for) the argument. The abstract case, such as the nominative and accusative, will be realized in some appropriate phonological form. The choice of the appropriate form depends on two major factors: the syntactic-functional profile of each marker and the interaction of constraints affecting the choice. The diachronic change leading to the establishment of ga as nominative marker also results from the interaction of the two factors, both of which have changed over time. The case marker ga becomes by far the best choice to mark the subject argument in Muromachi, when (1) ga becomes compatible with verbal arguments while no retains the nominal character and (2) it becomes more important in Japanese for case to be explicitly marked by appropriate morphological case markers. The proposed analysis has a significant implication in that the apparently complex diachronic change in case marking can be seen as a consequence of interactions of separate, gradual, unidirectional changes in the weight of constraints and slight changes in the profile of case markers. This implication is preferable for the obvious reason that language change within a monolingual community ought to be gradual, and lends support to the approach taken in this paper.