Replication Data for Exploring an extinct society through the lens of Habitus-Field theory and the Tocharian text corpus

The files and workflow will allow you to replicate the study titled "Exploring an extinct society through the lens of Habitus-Field theory and the Tocharian text corpus". This study aimed at utilizing the CEToM-corpus (https://cetom.univie.ac.at/) (Tocharian) to analyze the life-world of the elites of an extinct society situated in modern eastern China. To acquire the raw data needed for steps 1 & 2, please contact Melanie Malzahn [e-mail]. We conducted a mixed methods study, containing of close reading, content analysis, and multiple correspondence analysis (MCA). The excel file titled "fragments_architecture_combined.xlsx" allows for replication of the MCA and equates to the third step of the workflow outlined below. We used the following programming languages and packages to prepare the dataset and to analyze the data. Data preparation and merging procedures were achieved in python (version 3.9.10) with packages pandas (version 1.5.3), os (version 3.12.0), re (version 3.12.0), numpy (version 1.24.3), gensim (version 4.3.1), BeautifulSoup4 (version 4.12.2), pyasn1 (version 0.4.8), and langdetect (version 1.0.9). Multiple Correspondence Analyses were conducted in R (version 4.3.2) with the packages FactoMineR (version 2.9), factoextra (version 1.0.7), readxl version(1.4.3), tidyverse version(2.0.0), ggplot2 (version 3.4.4) and psych (version 2.3.9). After requesting the necessary files, please open the scripts in the order outlined bellow and execute the code-files to replicate the analysis: Preparatory step: Create a folder for the python and r-scripts downloadable in this repository. Open the file 0_create folders.py and declare a root folder in line 19. This first script will generate you the following folders: "tarim-brahmi_database" = Folder, which contains tocharian dictionaries and tocharian text fragments. "dictionaries" = contains tocharian A and tocharian B vocabularies, including linguistic features such as translations, meanings, part of speech tags etc. A full overview of the words is provided on https://cetom.univie.ac.at/?words. "fragments" = contains tocharian text fragments as xml-files. "word_corpus_data" = folder will contain excel-files of the corpus data after the first step. "Architectural_terms" = This folder contains the data on the architectural terms used in the dataset (e.g. dwelling, house). "regional_data" = This folder contains the data on the findsports (tocharian and modern chinese equivalent, e.g. Duldur-Akhur & Kucha). "mca_ready_data" = This is the folder, in which the excel-file with the merged data will be saved. Note that the prepared file named "fragments_architecture_combined.xlsx" can be saved into this directory. This allows you to skip steps 1 &2 and reproduce the MCA of the content analysis based on the third step of our workflow (R-Script titled 3_conduct_MCA.R). First step - run 1_read_xml-files.py: loops over the xml-files in folder dictionaries and identifies a) word metadata, including language (Tocharian A or B), keywords, part of speech, lemmata, word etymology, and loan sources. Then, it loops over the xml-textfiles and extracts a text id number, langauge (Tocharian A or B), text title, text genre, text subgenre, prose type, verse type, material on which the text is written, medium, findspot, the source text in tocharian, and the translation where available. After successful feature extraction, the resulting pandas dataframe object is exported to the word_corpus_data folder. Second step - run 2_merge_excel_files.py: merges all excel files (corpus, data on findspots, word data) and reproduces the content analysis, which was based upon close reading in the first place. Third step - run 3_conduct_MCA.R: recodes, prepares, and selects the variables necessary to conduct the MCA. Then produces the descriptive values, before conducitng the MCA, identifying typical texts per dimension, and exporting the png-files uploaded to this repository.