Statistical topic models are increasingly and popularly used by Digital Humanities scholars to perform distant reading tasks on literary data (Navarro-Colorado, 2018), (Hettinger et al., 2016. It allows us to estimate what people talk about. Especially Latent Dirichlet Allocation (LDA), see (Blei et al., 2003), has shown its usefulness, as it is unsupervised, robust, easy to use, scalable, and it offers interpretable results. In a preliminary study, we apply LDA to a corpus of New High German poetry (textgrid, with 51k poems, 8m token) and interpret salient topics, their trend over time (1575-1925 A.D.), and use the distribution of topics over documents for a classification of poems into time periods and for authorship attribution.
ExperimentsWe approach diachronic variation of poetry from two perspectives. First, as distant reading task to visualize the development of clearly interpretable topics