Traditionally, Japanese linguists have analyzed that postpositional particles are used to mark case relations in Japanese
sentences.  They consist of nominative 'ga', accusative 'o', dative 'ni', and genitive 'no'.  However, some Japanese verbs of possession, capability, or perception, such as 'iru/aru' (have)'dekiru' (can do), 'wakaru' (to understand), and verbs with the auxiliary for ability 'reru/rareru' (be able to), take either an [NP1ga  NP 2ga  VP] or an[NP1ni NP 2ga  VP] structure as in the following sentences.  This is called the dative-subject construction, or the ga/ni alternation. In the previous literature, this alternation was claimed to be simply a syntactic matter, and sometimes treated as the result of an avoidance of double 'ga' construction (Kuno 1972, Shibatani 1987).

 (1) Taroo   ga/ni        oniisan     ga     iru.
  Taro     NOM/DAT      brother    NOM    have
  'Taro has a brother'
 
 (2) Taroo   ga/ni        furansugo  ga       dekiru
  Taro    NOM/DAT      French     NOM be able
  'Taro can speak French.'

Recently, however, based upon Langacker's archetype model, Kabata (1998) has suggested that the ga/ni alternation is a semantic/cognitive phenomenon reflecting the nature of agent/experiencer continuum.  Kabata claims that 'ga' represents the highest 'energy source' cording volitionality and agentivity in participants, while 'ni' illustrates 'energy-sink' coding a lack of volitionality and agentitvity in participants. Kabata conducted a survey study with 112 Japanese high school students who were asked to fill in the blanks by choosing aparticle in clauses with different semantic factors. Her study has shown that semantic characteristics such as the type of verb, type of object,  volitionality, negation, and inchoative are crucial for the selection of ga/ni alternation, and that higher agentivity/volitionality tends to be marked with 'ga'.

Her results thus suggest that transitivity plays a central role in understanding of clause structure in human language (Hopper and Thomson 1980).  Though Kabata's semantic/cognitive treatment is the ga/ni alternation is extremely interesting, it remains to be seen how her findings relate to dicourse-level factors, which motivate the use of 'ga'
as discussed in a number of studies (Kuno 1972, Chafe 1970 & 1987, Maynard, 1980 & 1987).

In this paper, I would like to extend Kabata's study by adding a discourse/ pragmatic factor.  I propose that not only semantic factors, but also a pragmatic factor such as information status is also crucial factors to determine the selection of ga/ni.  I will focus on the Japanese nominative marker 'ga' as a new information marker following the previous
studies such as Kuno (1972), and Chafe (1970 & 1987). To support my claim, I will present introspective evidence from 20 adult native Japanese speakers who were tested for the acceptability of ga/ni markings in clauses with different semantic/ pragmatic factors. Besides the semantic factors that Kabata tested in her survey, I include three different type
of nouns for subject in test sentences which are distinctive in terms of information status.  My survey has provided supporting evidence both for Kabata's main findings as well as my proposal that not only transitivity in verbs and other semantic cues but also the information status of subjects play a crucial role in the selection of 'ga', or 'ni' as subject markers. This study tells us that there is a need toinvestigate specific interactions among transitivity and information
flow factors.

Back to the Conference Schedule