Dialectograms: Machine Learning Differences between Discursive Communities
This work addresses the need for more nuanced analysis of word usage differences in communities like political discourse, though it is incremental in leveraging existing embedding methods.
The authors tackled the problem of capturing the full richness of word embedding differences between discursive communities by introducing dialectograms, an unsupervised visual method that provides a new measure to identify words used differently, overcoming biases in existing measures. They applied this to US political subreddits, revealing affective polarization and disagreements on political issues.
Word embeddings provide an unsupervised way to understand differences in word usage between discursive communities. A number of recent papers have focused on identifying words that are used differently by two or more communities. But word embeddings are complex, high-dimensional spaces and a focus on identifying differences only captures a fraction of their richness. Here, we take a step towards leveraging the richness of the full embedding space, by using word embeddings to map out how words are used differently. Specifically, we describe the construction of dialectograms, an unsupervised way to visually explore the characteristic ways in which each community use a focal word. Based on these dialectograms, we provide a new measure of the degree to which words are used differently that overcomes the tendency for existing measures to pick out low frequent or polysemous words. We apply our methods to explore the discourses of two US political subreddits and show how our methods identify stark affective polarisation of politicians and political entities, differences in the assessment of proper political action as well as disagreement about whether certain issues require political intervention at all.