Return to search

Computational Literary Biodiversity Studies: Closing the Gap between Digital Humanities, Environmental Humanities and Ecology

Computational Literary Biodiversity Studies (CoLiBiS) is a proposed field that engages in elucidating and understanding the relationship between humanity and non-human living beings as reflected in paradigms, such as the nature-culture-entanglement and nature’s contributions to people. One primary goal of this field is to provide extensive quantification to typically intangible concepts concerning the relationship between living nature and culture, thereby challenging the current anthropocentric world view with factual data. Synthesising methodology and concepts from the fields of computer sciences, literary studies and biodiversity studies, several experiments and concept studies demonstrate the feasibility of the CoLiBiS approach, determine noteworthy results with computational methods, show and explore numerous research avenues for research and showcase the potential of cross-study comprehension. The project is characterised by elaborate reflections on the limitations, perspectives, feasibility and (re-)interpretations of concepts, resources and observations.
In the first experiment, animal and plant terms in literature, being the expression of the authors’ thoughts and thereby reflecting on their situational awareness and valuation, are determined over nearly three centuries. These results exhibit a peak after an initial rise during the 18th century, followed by the decline during and after the industrialisation, a particularly crucial phase in the development of human society. In my interpretation, the observations show strong correlations of biodiversity awareness to the detachment from nature as a result of the industrialisation, particularly manifested in the processes of land-use change and urbanisation. The second experiment explores the author- and work-related parameters that are most probable to be associated with the extent of biodiversity within literature, in order to facilitate our understanding of possible causes of the declining awareness for biodiversity during the industrialisation. The findings reveal a correlation of literary biodiversity with the parameters for gender, age, and region, show that authors from villages appear more sensitive towards living nature, and acknowledge a connection to the extent of the vocabulary used as well as genre and literature form of the work.
The data and approaches created are employed further to exemplarily showcase the potential and feasibility of topic modelling, data presentation, and the reuse of resources, concepts and methods. Within this project two promising and comprehensive resources were produced, a comprehensive database for animal and plant terms and a database for work- and author-related parameters, and are openly available for future investigations. It demonstrates the possibility of promoting our comprehension of the nature-culture-entanglement with an interdisciplinary approach that aspires to encourage scholars from digital humanities, literary studies, environmental humanities, ecology, gender studies and other adjacent fields to join the research approach of CoLiBiS.:TABLE OF CONTENTS

List of Figures 16
List of Tables 18
1. Introduction 19
1.1. The multiple perspectives on the environmental crisis 19
1.2. Ecological approaches towards the nature-culture-entanglement 21
1.3. Humanistic approaches towards the nature-culture-entanglement 24
1.4. Computational and Literary Biodiversity Studies (CoLiBiS) 26
1.5. Research agenda 32
2. Experiment 1: The rise and fall of Biodiversity in Literature (BiL) 39
2.1. Background and rationale 39
2.2. Methods 42
2.2.1. Corpus 42
2.2.2. Label database 46
2.2.3. Search method 47
2.2.4. Ecological methods transfer & analysis 49
2.3. Results 52
2.3.1. Occurrence statistics of taxon labels 52
2.3.2. Biodiversity measures 57
2.4. Interpreting the results – perspectives on the nature-culture-entanglement 61
2.4.1. BiL rising in the 18th century – the influences of the Enlightenment 61
2.4.2. BiL falling after 1830 – insights from a comprehensive approach 63
2.4.3. The relation of BiL to awareness for biodiversity 66
2.4.4. Further limitations and the future potential of this approach 67
2.5. Summary 70
3. Experiment 2: Using random forest regression to relate BiL with social and spatial situation of authors 71
3.1. Background and rationale 71
3.2. Methods 77
3.2.1. Sensitivity parameters 78
3.2.2. The random forest analysis 82
3.3. Results 84
3.3.1. Results for the complete corpus 84
3.3.2. Results for the subcorpora of the three phases 89
3.3.3. Correlation matrix based on predictor interaction 92
3.4. Interpreting the results – relating living situation to awareness for biodiversity 94
3.4.1. Limitations to interpreting the results 94
3.4.2. Sensitivity parameters 96
3.4.3. Relevance of the experiment 102
3.4.4. Methodological considerations in resource creation 102
3.5. Summary 104
4. Resource creation and improvement 105
4.1. Background and rationale 105
4.2. Constructing a dictionary of biological taxon labels 108
4.2.1. Methodology and rationale 108
4.2.2. Database structure and usage 114
4.2.3. Quality measures 117
4.2.4. Outlook and applicability 118
4.3. Constructing a metadata database for authors and their works 120
4.4. Résumé and future work 123
5. Exploring the CoLiBiS potential: Three Concept Studies 125
5.1. Biodiversity in academic journal articles 126
5.2. Correlating topics with incorporated BiL 131
5.2.1. Methods and results 131
5.2.2. Animal and plant profiles, and their implications 133
5.2.3. Further potential of topic modelling 135
5.3. The BiL Explorer – prototyping fast discovery of BiL 137
5.4. The feasibility of further CoLiBiS components 145
6. Reflection and conclusion 147
6.1. Summary of the overall contribution 147
6.2. Challenges and limitations 148
6.3. The significance of overarching and cross-study contributions 150
6.4. Outlook into future work 155
7. Data availability 159
8. References 161
Appendix 177
Blacklist compiled for the current extended taxon database 178
Topic modelling – most probable terms 180
Topic modelling – probability / coherence indicators 192
The Author 195

Identiferoai:union.ndltd.org:DRESDEN/oai:qucosa:de:qucosa:92646
Date17 July 2024
CreatorsLanger, Lars
ContributorsUniversität Leipzig
Source SetsHochschulschriftenserver (HSSS) der SLUB Dresden
LanguageEnglish
Detected LanguageEnglish
Typeinfo:eu-repo/semantics/publishedVersion, doc-type:doctoralThesis, info:eu-repo/semantics/doctoralThesis, doc-type:Text
Rightsinfo:eu-repo/semantics/openAccess

Page generated in 0.003 seconds