Paper short abstract:

Using The Oxford-NINJAL Corpus of Old Japanese, this presentation aims to describe and analyse the distribution of Old Japanese (700—800AD) genitive case particles no and ga with a focus on noun-modifying constructions (NP-ga/no NP).

Paper long abstract:

In “standard” Modern Japanese (1600 onwards), no is the only productive genitive particle. However, in Old Japanese (700—800AD), the oldest attested stage of the language, no and ga were both productive genitives and shared similar main functions: they were both used in adnominalization (NP-ga/no NP) and in subject marking (NP-ga/no VP). Despite their functional similarities, the two particles were distributed differently. In general, the use of ga was far more restricted than that of no. In this presentation I will not consider ga or no used to connect clauses to a following noun or predicate (S ga/no N, or S ga V), or the use of no as an adnominal copula form.

Until now, there have been no exhaustive investigation of genitive no and ga in Old Japanese using modern linguistic frameworks. The principal aim of this presentation is to disentangle the distribution of Old Japanese no and ga with a focus on noun-modifying constructions, such as tuma ga ipye ‘my spouse’s house’ and twomibito no ipye ‘a wealthy person’s house’. The descriptive and analytic work of this presentation is part of an on-going doctoral research project titled No and ga in Old Japanese: a synchronic and diachronic study. The empirical data presented is based on The Oxford-NINJAL Corpus of Old Japanese and includes all examples where the particles, no and ga, are phonographically written. The presentation is split into two main parts. I first provide a comprehensive synchronic description of no and ga’s distribution in noun-modifying constructions. Then, based on the synchronic data, I discuss a number of factors that may influence the choice of no and ga in adnominalization, drawn from past literature on the subject as well as cross-linguistic studies on genitive variation. These factors include animacy, referentiality/specificity, relationality, possessive relations, various pragmatic factors, and so on. The qualitative analysis suggests that multiple factors are at work to condition the distribution of no and ga — in particular, animacy. These results are tested by quantitative analytical methods such as a multivariate statistical analysis to determine whether the different factors interact with each other, and if so, how.