Referential Structure in Japanese: a Diachrony

在这里，“指称”被解释为语言表达和“指称”之间的关系。后者被理解为说话人头脑中的形象，既可以代表真实的实体，也可以代表抽象的实体。完整的名词短语(专有的和常见的名称)和简化的名词短语(代词和零式)构成了语言中主要的“指称手段”。在大多数情况下，两者之间的选择取决于说话者头脑中所指对象被激活的程度：激活程度越高，使用简化名词短语的可能性就越大。关于决定所指对象激活程度的因素，人们提出了许多建议(Kibrik 2011：390-393)。当其他学者单独考虑这些因素时，Kibrik(2011：60-61)认为激活程度是由所有因素共同决定的，可以用他的认知多因素(CMF)方法来定量测量。我将把CMF方法应用于日语的几个历时阶段。在此之前，Efimova(2006)曾将CMF方法应用于现代日语。虽然她的研究在这一领域取得了突破性的进展，但仍然存在严重的局限性。Efimova在她的统计分析中使用了x2检验，该检验只允许在两个参数之间建立相关性：在她的情况下，是在每个激活因子和一个参考装置之间。换句话说，这种统计方法不适合她的多因素分析，因为它不能反映激活因素之间的相互作用。关键研究问题：1)日语书面叙事中指称选择的策略是否历时变化？2)如果是，这种变化的形态合成机制是什么？3)哪些激活因素在不同的时代和体裁中影响指称选择？4)第三人称代词和被动化(而不是主题化)的使用在现代日语中是否在欧洲语言的影响下变得更加广泛(Li 2000)？本研究的理论意义是双重的。首先，据我所知，还没有对前现代日语共时阶段的指称选择进行研究。同时，这样的研究可能会为一些关于空成分和代词使用的问题提供新的线索。其次，这项研究将有助于语篇语言学的一般理论，因为目前还没有对任何一种特定语言中指称结构的历时进行广泛的研究。在这个方向上，唯一的尝试是由Song(2021)进行的，他对英语的三个阶段(旧的、中的和现代的)进行了“指称密度”测试，并证明了“指称密度”的稳定增长反映了论元结构、配价和及物性的重大变化。我还将把“参照密度”测试应用于日语的不同阶段。从实践意义上讲，本研究的成果将有助于解决机器翻译中长期存在的问题，如零成分的消解、代词指称等。

Here, "reference" is construed as the relation between a linguistic expression and a "referent". The latter is understood as an image in the speaker's mind, which can represent either a real or an abstract entity. Full noun phrases (proper and common names) and reduced noun phrases (pronouns and zero forms) constitute the main "referential devices" in a language. In most cases, the choice between them depends on how much a referent is "activated" in the speaker's mind: the higher the degree of activation, the higher the possibility that a reduced noun phrase will be used. There have been numerous suggestions regarding the factors, which determine the degree of a referent's activation (Kibrik 2011: 390-393).While other scholars considered these factors separately, Kibrik (2011: 60-61) suggested that the degree of activation is determined by all of them collectively and can be measured quantitatively with his cognitive multi-factorial (CMF, henceforth) method. I will apply the CMF approach to several diachronic stages of the Japanese language. The CMF approach has been previously applied to Contemporary Japanese by Efimova (2006). While her study has achieved a breakthrough in the field, it still suffers from a serious limitation. Efimova uses the x2 test for her statistical analysis which only allows to establish a correlation between two parameters: in her case, between each activation factor and a referential device. In other words, this statistical method is not adequate to her multi-factorial analysis, since it cannot reflect the interaction between activation factors. In my research, the multinomial (polytomous) logistic regression will be used.Key research questions:1) Have the strategies of referential choice in Japanese written narratives changed diachronically? 2) If so, what are the morphosyntactic mechanics of this change? 3) Which of the activation factors affect referential choice in different epochs and genres? 4) Has the use of 3rd person pronouns and passivisation (instead of topicalisation) actually become more extensive in Modern Japanese under the influence of European languages (Li 2000)?The theoretical significance of the proposed research is two-fold. Firstly, to my knowledge, no investigation of referential choice at any synchronic stage of pre-Modern Japanese has been conducted before. Meanwhile, a study like this might shed new light on some issues, concerning the use of null elements and pronouns. Secondly, the proposed research will contribute to the general theory of discourse linguistics, since there has been no extensive study on the diachrony of referential structure in any given language. The only attempt in this direction has been undertaken by Song (2021), who applied "referential density" tests to three stages of the English language (Old, Middle, and Modern) and demonstrated that the stable increase of "referential density" (number of overt arguments over the number of predicates) reflects major changes in argument structure, valency, and transitivity. I will also apply the "referential density" tests to different stages of Japanese as well. As for practical significance, the findings of this research will be useful in solving the perennial problems of machine translation such as the resolution of zero elements and pronoun reference.