Extracting narrative signals from public discourse: a network-based approach

TL;DR

This paper tackles the challenge of empirically analyzing political narratives in large digital text corpora by introducing a graph-based, meaning-centric pipeline built on Abstract Meaning Representation (AMR). It operationalizes narrative signals—such as actors, events, and perspectivization—via AMR-derived event and actor subgraphs that feed into a narrative trace table and subsequent narrative and actantial networks, enabling guided close reading of public discourse. The authors implement an open-source pipeline and demonstrate its utility with a case study on State of the European Union addresses from 2010–2023, surfaceing genre-specific narrative signals and illustrating shifts across presidents. The approach offers a robust, replicable framework that links narratology concepts with network science and NLP, supporting large-scale, interpretable analyses of political narratives and enabling more nuanced insights into polarization and discourse dynamics.

Abstract

Narratives are key interpretative devices by which humans make sense of political reality. As the significance of narratives for understanding current societal issues such as polarization and misinformation becomes increasingly evident, there is a growing demand for methods that support their empirical analysis. To this end, we propose a graph-based formalism and machine-guided method for extracting, representing, and analyzing selected narrative signals from digital textual corpora, based on Abstract Meaning Representation (AMR). The formalism and method introduced here specifically cater to the study of political narratives that figure in texts from digital media such as archived political speeches, social media posts, transcripts of parliamentary debates, and political manifestos on party websites. We approach the study of such political narratives as a problem of information retrieval: starting from a textual corpus, we first extract a graph-like representation of the meaning of each sentence in the corpus using AMR. Drawing on transferable concepts from narratology, we then apply a set of heuristics to filter these graphs for representations of 1) actors and their relationships, 2) the events in which these actors figure, and 3) traces of the perspectivization of these events. We approach these references to actors, events, and instances of perspectivization as core narrative signals that allude to larger political narratives. By systematically analyzing and re-assembling these signals into networks that guide the researcher to the relevant parts of the text, the underlying narratives can be reconstructed through a combination of distant and close reading. A case study of State of the European Union addresses (2010 -- 2023) demonstrates how the formalism can be used to inductively surface signals of political narratives from public discourse.

Figures (7)

• Figure 1: Illustration of AMR representation of the sentence "Emmanuel Barroso wants the European Union to invest more in innovation, technology and the role of science." The AMR-parsed input sentence (left) is represented as a rooted, directed, acyclical graph (right), where nodes are AMR frames and directed edges represent the semantic relationships between them. Name-nodes are labeled by concatenating their op-nodes.
• Figure 2: Translation of the AMR graph into structured output from which narrative signals can be extracted.
• Figure 3: Translation of the narrative trace table into narrative and actantial networks.
• Figure 4: Pipeline for narrative signal extraction using AMR. First, the text is segmented into sentences and parsed to obtain one AMR string per sentence. Then, these strings are processed and transformed into AMR graphs, extracting event and actor subgraphs. These subgraphs are transformed to a tabular representation we call narrative trace table, where every row corresponds to one event in the corpus. Finally, the resulting table is analyzed to extract narrative signals.
• Figure 5: Subset of actantial network of SOtEU speeches. Nodes were selected based on their degree and betweenness centrality. Blue/red/grey links correspond to supportive/conflictive/neutral relationships. The inset demonstrates the guided close reading approach for the selected link from "Europe" to "rule of law".