Dr. Gregor Wiedemann and Cornelia Fedtke illustrate in their contribution to the "Handbook of Computational Social Science" the semantic shift of the term "gold pieces" in the context of hate speech in the German Facebook discourse.
The Handbook is published by Routledge and is partly available as an open access publication.
Abstract
For automatic content analysis, topic models became an increasingly popular research method to reveal thematic structures of large document collections. However, research interests often go beyond topics that are limited to broad discourse-level semantics. On a more fine-grained level, it is also of interest that arguments, stances, frames, or discourse positions are expressed in what specific contexts and how they emerged in the first place. We argue that recent advancements in natural language processing based on deep learning for distributional semantics enable researchers to explicitly combine word-, sentence-, and discourse-level meaning for computational content analysis. In the chapter, we elaborate on how basic text mining and new neural network-based embedding technologies such as BERT relate to linguistic structuralism as the theoretical and methodological foundation of discourse analysis and many other qualitative research methods. We illustrate our argumentation along with an exemplary study of the semantic shift of the term “Goldstücke” (English: gold pieces) in the context of hate speech in the German Facebook discourse.
Wiedemann, G.; Fedtke, C. (2021): From Frequency Counts to Contextualized Word Embeddings: The Saussurean Turn in Automatic Content Analysis. In: Engel, U.; Quan-Haase, A.; Liu, S. X.; Lyberg, L. (eds.), Handbook of Computational Social Science, Volume 2. Routlegde. https://doi.org/10.4324/9781003025245