A Mixed Methods Design for Computational Genre Stylistics and Unstructured Genres. Towards a Functional History of 19th Century German Novellas.

德国中篇小说实际上分为两个流派，即《中篇小说》和《埃尔扎隆格》，是一个极具争议的流派。一个学术团体认为，小说是最严格的散文小说形式。这一派主张，必须承认严格定义的小说和较为松散的《厄策伦根》之间有着根本的区别。与第一派观点相反，第二派认为，《小说》和《童话》不是不同的体裁，而是一个无定形的整体。本项目的总体目标是理解和澄清这一矛盾。到目前为止，研究忽略了一个有助于理解这一矛盾的核心方面：关于诗学与阅读实践之间的关系，《小说》和《二重性》概念的核心是一个辩证结构。这种结构是辩证的，因为在阅读经验和诗学概念之间并不匹配。在19世纪，作家和读者对这种辩证法有一种感觉：他们对《诺韦伦》在诗学话语中的审美一致性抱有强烈的期望，但同时也承认，这些期望在实践中是失望的。“二次律”的概念缺乏这种辩证的结构。该项目将描述和解释这种辩证法，从而对历史体裁的文化结构有更深的理解。我将尝试通过两个任务来实现这一目标：第一个任务是阐述一个混合的方法设计，允许连接三个维度：体裁标签的使用，文本特征，以及媒体和社会背景因素。第二个任务是开发必要的数据建模方法。到目前为止，计算体裁文体学主要是基于词袋模型对文本特征进行建模。然而，这种模式并不适合测试的假设，德国中篇小说的辩证结构，因为词袋模型并不透明地代表结构，内容为基础的，和诗学体裁特征。因此，被认为与《小说》和《童话》的语义相关的体裁特征必须被建模。在语类特征的模型化和操作化的基础上，对语类概念的辩证结构进行了实证检验。最后，本文梳理了德国中篇小说体裁的概念史，即体裁语义和体裁实践的变化。这最后一步是基于监督机器学习的改编，也被称为“透视建模”（perspectival modeling）。通过这种方式，该项目将有助于更好地理解文学史中体裁概念使用的实际规则，也将提供必要但尚未提供的方法论，即对松散结构的体裁进行计算、面向上下文和历史分析。

The German novella, which is actually split up into two genres, that of the ›Novelle‹ and that of the ›Erzählung‹, is a highly controversial genre. One academic group believes that the Novelle is the most stringent form of prose fiction. This group claims that there has to be acknowledged a fundamental difference between the strictly defined ›Novelle‹ and the more loose field of the ›Erzählungen‹. In contradiction to the first group, the second academic group believes that ›Novelle‹ and ›Erzählung‹ are not different genres but one amorphous mass. The general goal of the project is to understand and to clarify this contradiction. Up to now, research has overlooked one central aspect that helps to understand this contradiction: There is a dialectical structure at the core of the concepts of ›Novelle‹ and ›Erzählung‹ with regard to the relationship between poetics and reading practice. The structure is dialectical, because the poetological concept and reading experience do not match in the case of ›Novellen‹. In the 19th century, writers and readers had a sense of this dialectics: They had strong expectations regarding the aesthetic uniformity of ›Novellen‹ within poetological discourse, but acknowledged at the same time that these expectations were disappointed in practice. The concept of ›Erzählung‹ lacks this dialectical structure. The project will describe and explain this dialectics and thus yield a deeper understanding of the cultural structure of historical genres. I will try to reach this goal by two tasks:The first task is to elaborate a mixed methods design that allows to connect three dimensions: the use of genre labels, textual features, and medial and social context factors. The second task is to develop the requisite methodology of data modeling. Up to now, computational genre stylistics has been modeling textual features mostly as document-term-matrices based on bag of words models. However, this modeling does not suit for testing hypotheses on the dialectical structure of German novellas, because the bag of words model does not transparently represent structural, content based, and poetological genre features. Therefore, genre features which are presumed to be relevant to the semantics of ›Novelle‹ and ›Erzählung‹ have to be modeled. Based on the modeling and operationalization of genre features, empirical hypotheses on the dialectical structure of the relevant genre concepts are tested. Finally, the change of the genre semantics and practices is worked out as the conceptual history of the German novella genres. This final step is based on an adaption of supervised machine learning, which is also called ›perspectival modeling‹.In this way, the project will help to better understand the actual rules of the use of genre concepts in literary history, and it will also offer the requisite but not yet available methodology of a computational, context oriented, and historical analysis of loosely structured genres.