Narrative Echoes across Time and Space
Comparative Analysis of Structural Similarities in Myth and Folktale Sequences
by Franziska Anne-Katrin Pannach
Date of Examination:2023-11-28
Date of issue:2024-02-16
Advisor:Prof. Dr. Caroline Sporleder
Referee:Prof. Dr. Caroline Sporleder
Referee:Prof. Dr. Ramin Yahyapour
Persistent Address:
http://dx.doi.org/10.53846/goediss-10349
Files in this item
Name:Dissertation.pdf
Size:4.79Mb
Format:PDF
The following license files are associated with this item:
Abstract
English
Originating in mythological research, the Hylistic theory introduces hylemes as a basic plot unit, containing actions, states, or background information derived from a source. These statements are organized in hyleme sequences which are derived by domain experts and concern a narrative variant. By using those sequences, narrative variants can be compared across different source materials, such as text genres or sources in different languages, and subsequently structurally aligned with variants of the same narrative. Establishing appropriate methods for the automation of these hyleme sequene alignments is the central objective of this work. This thesis presents the first approach towards a Digital Hylistic theory. This work is related to the research discipline of Computational Narratology, and its related areas. Two data sets are the basis of the conceptual and exploratory studies undertaken in this work: the German hyleme data sets consists of sequences from different cultural and temporal backgrounds, including Ancient Greece, or Mesopotamia. Those sequences are extracted by domain experts in the fields of Classics, Ancient Near Eastern Studies, and Religious Studies as part of the DFG-funded myth research group 2064 STRATA. They are derived from a multitude of sources, in different source languages, and from different genres and styles of narratives. The second data set is the first ever English hyleme data set, containing sequences describing a set of Zulu folktales. The data set is also the first approach towards a hylistic representation of folkloristic (in contrast to mythological) material. This thesis follows a multi-method approach, grounded in the narrative theory of Hylistics, that carefully models hylemes and their properties as objects that can be processed and analyzed using methods from the fields of Natural Language Processing, Knowledge Engineering, and Formal Languages. This work approaches the comparability of hylemes from a semantic similarity point-of-view. The problem of aligning hyleme sequences is approached from different angles with a focus on the research question of the domain expert, e.g. comparison of variants of the same myth or exploratory alignment with the purpose to discover interesting patterns. This work does not aim to solve the hyleme alignment task, because alignment can be performed for various purposes. All methods in this work are selected based on the appropriateness with respect to the Hylistic theory. States and events, conveyed by durative and single-point hylemes, are modelled and compared using fundamentally different methods. Previous work on story similarity suggests that many salient features contribute to the judgment of how similar two or more narrative variants are. Therefore, the similarity of background information and plot-driving actions is approached from different standpoints and with different methods. In combination, these methods can be used for alignment and the comparison of background information, which yields a holistic measure of the similarity between narrative variants.
Keywords: Hylistics; Digital Mythological Studies; Computational Narratology; Digital Folkloristics; Computational Linguistics
