[Paper Review] Beyond Discrete Genres: Mapping News Items onto a Multidimensional Framework of Genre Cues

Zilin Lin, Kasper Welbers+2 more|arXiv (Cornell University)|Dec 8, 2022

Computational and Text Analysis Methods4 citations

TL;DR

This paper proposes a multidimensional, non-discrete framework for mapping news items based on genre cues—specifically factuality and formalness—using fine-tuned BERT models to predict these dimensions at the sentence level. The approach enables systematic comparison of diverse news formats across media platforms, revealing fluid genre boundaries in contemporary journalism.

ABSTRACT

In the contemporary media landscape, with the vast and diverse supply of news, it is increasingly challenging to study such an enormous amount of items without a standardized framework. Although attempts have been made to organize and compare news items on the basis of news values, news genres receive little attention, especially the genres in a news consumer's perception. Yet, perceived news genres serve as an essential component in exploring how news has developed, as well as a precondition for understanding media effects. We approach this concept by conceptualizing and operationalizing a non-discrete framework for mapping news items in terms of genre cues. As a starting point, we propose a preliminary set of dimensions consisting of "factuality" and "formality". To automatically analyze a large amount of news items, we deliver two computational models for predicting news sentences in terms of the said two dimensions. Such predictions could then be used for locating news items within our framework. This proposed approach that positions news items upon a multidimensional grid helps in deepening our insight into the evolving nature of news genres.

Motivation & Objective

• To address the challenge of comparing diverse news formats—such as YouTube interviews and print editorials—within a unified, standardized framework.
• To move beyond static, discrete genre classifications by modeling news genres as fluid continuums shaped by audience perception.
• To operationalize genre perception through measurable linguistic cues, focusing on factuality and formality as core dimensions.
• To develop computational models that predict these genre cues automatically, enabling large-scale analysis of news content.
• To provide a foundation for understanding how news genres evolve and blend in the contemporary media ecosystem, especially across platforms and audiences.

Proposed method

• Proposes a two-dimensional framework based on genre cues: 'factuality' (degree of factual vs. opinionated content) and 'formality' (linguistic formality level).
• Fine-tunes two BERT-based models on a large, diverse Dutch news corpus spanning newspapers, blogs, podcasts, public broadcasts, satire, and YouTube content.
• Uses sentence-level annotation of factuality and formality to train classification models, enabling prediction of genre cues per sentence.
• Aggregates sentence-level predictions to position entire news items within the multidimensional genre framework.
• Validates the framework using large-scale data showcases, demonstrating its ability to capture genre evolution and blending across media types.
• Releases code and corpus to support replication and extension by future researchers, including potential adaptation to other languages or media types.

Experimental results

Research questions

• RQ1 How can news items from diverse media platforms be systematically compared despite their differing formats and styles?
• RQ2 To what extent do perceived genre cues like factuality and formality vary across different types of news content and media outlets?
• RQ3 Can computational models accurately predict genre cues at the sentence level, enabling large-scale mapping of news items in a multidimensional space?
• RQ4 How do the dimensions of factuality and formality reflect the fluid and dynamic nature of contemporary news genres?
• RQ5 In what ways can this non-discrete framework improve the study of media effects and journalistic practices in the digital age?

Key findings

• The proposed two-dimensional framework effectively captures the fluidity and blending of news genres across diverse media platforms, such as traditional journalism and user-generated content.
• Fine-tuned BERT models achieve reliable prediction of factuality and formality at the sentence level, enabling scalable mapping of news items in the genre space.
• The framework reveals that news items are not confined to discrete genres but exist on a spectrum, with significant overlap between formats like news reports and opinion pieces.
• The model outputs demonstrate that genre cues vary meaningfully across media types, with YouTube and satirical content showing lower formality and higher opinionation compared to traditional press.
• The study validates the approach through large-scale data showcases, showing that the framework can detect genre shifts and structural changes in news content over time.
• The released corpus and code enable future research to build specialized models for niche media types, such as fringe news or social media content, improving demographic and platform-specific generalizability.