Code to reproduce the models and analysis in the paper "Leveraging the Alignment between Machine Learning and Intersectionality: Using Word Embeddings to Measure Intersectional Experiences of the Nineteenth Century U.S. South", authored by Laura K. Nelson and forthcoming in Poetics.
The data folder contains a copy of the collection "First Person Narratives of the American South" and "North American Slave Narratives", from the library Documenting the American South, housed at Duke University.
It also contains word embedding model produced on this collection (word2vec_all_clean.txt) and 40 constructed word embedding models (word2vec_robust) created via a radom sample (with replacement) of the sentences in the corpus, for use in producing confidence intervals.
00_measuringintersectionality_constructmodels.ipynb contains the code to reconstruct the main word embedding model, plus the 40 random models.
01_measuringintersectionality_reproduceanalysis.ipynb contains the code to reproduce the analysis, table, and five figures presented in the paper.