Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Estranged Predictions: Measuring Semantic Category Disruption with Masked Language Modelling

Yuxuan Liu, Haim Dubossarsky, Ruth Ahnert

TL;DR

This study addresses how science fiction reconfigures ontological boundaries by measuring semantic permeability among human, animal, and machine terms. It develops a computational pipeline using masked language modelling (RoBERTa) and a Gemini classifier to generate and categorize top-5 token substitutions in two corpora (Gollancz SF Masterworks vs NovelTM), quantified by retention, replacement, and entropy. Across metrics, science fiction shows greater permeability, especially for machine referents, revealing genre-conditioned shifts in semantic space and a tendency toward anthropomorphism and higher predictive dispersion. The work demonstrates how masked language models, employed critically and with human-in-the-loop interpretation, can surface linguistic infrastructures underlying science fiction's estrangements and advance computational literary analysis.

Abstract

This paper examines how science fiction destabilises ontological categories by measuring conceptual permeability across the terms human, animal, and machine using masked language modelling (MLM). Drawing on corpora of science fiction (Gollancz SF Masterworks) and general fiction (NovelTM), we operationalise Darko Suvin's theory of estrangement as computationally measurable deviation in token prediction, using RoBERTa to generate lexical substitutes for masked referents and classifying them via Gemini. We quantify conceptual slippage through three metrics: retention rate, replacement rate, and entropy, mapping the stability or disruption of category boundaries across genres. Our findings reveal that science fiction exhibits heightened conceptual permeability, particularly around machine referents, which show significant cross-category substitution and dispersion. Human terms, by contrast, maintain semantic coherence and often anchor substitutional hierarchies. These patterns suggest a genre-specific restructuring within anthropocentric logics. We argue that estrangement in science fiction operates as a controlled perturbation of semantic norms, detectable through probabilistic modelling, and that MLMs, when used critically, serve as interpretive instruments capable of surfacing genre-conditioned ontological assumptions. This study contributes to the methodological repertoire of computational literary studies and offers new insights into the linguistic infrastructure of science fiction.

Estranged Predictions: Measuring Semantic Category Disruption with Masked Language Modelling

TL;DR

This study addresses how science fiction reconfigures ontological boundaries by measuring semantic permeability among human, animal, and machine terms. It develops a computational pipeline using masked language modelling (RoBERTa) and a Gemini classifier to generate and categorize top-5 token substitutions in two corpora (Gollancz SF Masterworks vs NovelTM), quantified by retention, replacement, and entropy. Across metrics, science fiction shows greater permeability, especially for machine referents, revealing genre-conditioned shifts in semantic space and a tendency toward anthropomorphism and higher predictive dispersion. The work demonstrates how masked language models, employed critically and with human-in-the-loop interpretation, can surface linguistic infrastructures underlying science fiction's estrangements and advance computational literary analysis.

Abstract

This paper examines how science fiction destabilises ontological categories by measuring conceptual permeability across the terms human, animal, and machine using masked language modelling (MLM). Drawing on corpora of science fiction (Gollancz SF Masterworks) and general fiction (NovelTM), we operationalise Darko Suvin's theory of estrangement as computationally measurable deviation in token prediction, using RoBERTa to generate lexical substitutes for masked referents and classifying them via Gemini. We quantify conceptual slippage through three metrics: retention rate, replacement rate, and entropy, mapping the stability or disruption of category boundaries across genres. Our findings reveal that science fiction exhibits heightened conceptual permeability, particularly around machine referents, which show significant cross-category substitution and dispersion. Human terms, by contrast, maintain semantic coherence and often anchor substitutional hierarchies. These patterns suggest a genre-specific restructuring within anthropocentric logics. We argue that estrangement in science fiction operates as a controlled perturbation of semantic norms, detectable through probabilistic modelling, and that MLMs, when used critically, serve as interpretive instruments capable of surfacing genre-conditioned ontological assumptions. This study contributes to the methodological repertoire of computational literary studies and offers new insights into the linguistic infrastructure of science fiction.

Paper Structure

This paper contains 21 sections, 2 equations, 6 figures.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Approach
  4. Materials and methods
  5. Corpus Selection
  6. Contextualised Model
  7. Experimental Procedure
  8. Sentence Extraction and Masked Language Modelling
  9. Semantic Classification via Gemini
  10. Metrics of Analysis
  11. Retention Rate
  12. Replacement Rate
  13. Entropy
  14. Aggregation Schemes
  15. Statistical Tests
  16. ...and 6 more sections

Figures (6)

  • Figure 1: Cross-genre conceptual permeability detection model
  • Figure 2: Three-stage Gemini-based classification workflow
  • Figure 3: Cross-Entity Replacement Patterns in Gollancz SF and NovelTM Corpora
  • Figure 4: Mean Entropy by Entity Type and Corpus (with Standard Errors)
  • Figure 5: Top 10 Predicted Categories for Masked Entity Terms in Gollancz SF and NovelTM Corpora
  • ...and 1 more figures