Pragmatic implicatures and particles in Japanese.

摘要

This study investigates the properties of particles in Japanese from a pragmatic viewpoint. The main target of analysis is wa and ga, which are regarded as a topic and nominative marker respectively. The research is based on the pragmatic theory proposed by Levinson (2000).;This dissertation consists of three parts as follows. The first part investigates the contrastive usage of wa/ha from a historical perspective. I elicited examples of wa/ha, ga and no from five Classical literary works and two Modern works. The text data showed that, in Late Old/Middle Japanese, the ha-marking worked as the emphatic marker, but not used for maintaining the topic continuity, while the ga as the nominative marker was scarce and with a limited distribution. I claim that the contrastive property of ha in Classical Japanese was provided by the M-implicature. In Early Modern Japanese, the two works showed different result. Saikaku Shokokubanashi is regarded as a middle step toward Modern Japanese from Middle Japanese while Oku no Hosomichi retained the trend of the Late Old/Middle Japanese.;The second part argues that the contrastive property of wa is given by the non-specificity implicature based on Grice's Maxim of Quantity. I proposed that the two particles form the scalar terms particles form the scalar terms, ga. wa>, which means that the wa is informationally weaker than the ga. The hearer pragmatically infers that if a speaker knew that an NP was a specific item in the discourse, s/he would use ga, but actually s/he did not, and used wa instead, where s/he eliminates the possibility that the NP is the one and only one item in the discourse. In the case where wa appears at the position incompatible with ga, the implicature is given by M-implicature. Experiment was performed to test the hypothesis. A free association task of writing a sentence following a wa-altered perceptual judgment sentence (ga-marked by default) and experiential judgment sentence (wa-marked by default) was given. The participants wrote the sentences in the way that they fulfill the contrastive framework more frequently after the wa-altered perceptual judgment sentences than after the experiential judgment sentences. The result exhibited the effect of the stronger implicature in the wa-altered perceptual judgment sentence.;The third part examines the effect of particles attached to the quantifier terms to scalar implicatures in Japanese. A neo-Gricean view on scalar implicatures argues that quantifiers represent the lower bound in a default, i.e., "at least" interpretation, and pragmatics provides them with upper bound scalar implicatures. I hypothesized the particle ga attached to quantifiers strengthens the implicatures of the sentence, while the particle wa weakens them and provides the "at least" interpretation. Zero marking cancels both effects. O has the similar but not so strong effect as ga. To prove this, two Experiments were performed based on the concept in Papafragou & Musolino (2003). A truth value judgment task was given to test sentences involving weaker terms, 'eight', 'some', and 'most', marked by the particle ga, wa, o, and zero under the condition involving the strong term 'all ten.' In Experiment I the effect of the particles was tested within-group. In Experiment II their effect were tested between distinct groups. The number term showed lower ratings of truth value judgment with ga than it did with wa in both experiments. In contrast, 'most' showed higher ratings with ga than it did with wa only in Experiment I. The particles attached to 'some' hardly provide any effect. The observed regularities for the number term are explained by the exclusive function of ga and the contrastive usage of wa. The result for 'most' is also explained by the proposal by Ariel (2004).