'The Order in the Horse's Heart': A Case Study in LLM-Assisted Stylometry for the Discovery of Biblical Allusion in Modern Literary Fiction

2026-04-21 · Source: Computation and Language · Field: Science & Research — Artificial Intelligence & Machine Learning, Research Methodology & Innovation · Depth: Advanced, quick

Summary

A novel dual-track pipeline has been developed for detecting biblical allusions in literary fiction, specifically applied to Cormac McCarthy's novels. The "Bottom-Up Embedding Track" identifies rare vocabulary shared with the King James Bible (KJV), embeds these occurrences for sense disambiguation, and uses cascaded LLM review for candidate passage pairs. Concurrently, the "Top-Down Register Track" employs an LLM to analyze McCarthy's prose for allusions without direct biblical passage comparison, capturing those not based on word rarity. Both tracks are cross-validated by a long-context model comparing entire novels with the KJV, and findings are verified against published scholarship. The pipeline identified 349 allusions, recovering 62 (54% recall) of 115 previously documented allusions, with recall varying from 30% for transformed imagery to 80% for register collisions.

Key takeaway

For literary scholars or computational linguists analyzing intertextuality, this pipeline offers a robust method to identify subtle allusions. You should consider integrating similar dual-track LLM approaches, combining both specific textual echoes and broader stylistic registers, to enhance recall and precision in large-scale literary corpus analysis. This can significantly augment traditional stylometric methods and accelerate the statistical study of intertextuality.

Key insights

A dual-track LLM pipeline effectively detects biblical allusions in literature, cross-validated against scholarship.

Principles

• Combine bottom-up and top-down analysis.
• Cross-validate findings with long-context models.
• Distinguish literary allusions from direct references.

Method

The pipeline uses inverse document frequency for rare vocabulary, local context embedding, cascaded LLM review, and undirected LLM analysis for register, all cross-validated by a long-context model.

In practice

• Apply LLMs for intertextuality studies.
• Use dual-track approach for nuanced detection.
• Validate LLM outputs against human scholarship.

Topics

• LLM-Assisted Stylometry
• Biblical Allusion Detection
• Cormac McCarthy
• King James Bible
• Intertextuality Studies

Best for: NLP Engineer, AI Scientist, Research Scientist

Related on AIssential

• MiqraBERT: Regression-Based Sentence-BERT Finetuning for Biblical Hebrew Parallel Detection — Finetuning a BERT model with regression for semantic similarity effectively detects textual reuse in ancient languages beyond lexical overlap.
• From 50K to 8.2 Million in 24 Hours: Vozinha's Algorithmic Consecration and the Multilingual Making of World Cup Visibility — Algorithmic visibility for public figures is shaped by distinct cross-lingual narrative frames and platform metrics.
• ‘But this one was so . . . male.’ A Corpus-Based and LLM-Augmented Analysis of Language and Gender Bias in Barbie — Combining corpus linguistics with LLMs effectively reveals subtle gender biases in film dialogue.
• Probing Semantic Alignment, Lexical Invariance, and Syntactic Influence in LLM Metaphor Processing — LLMs struggle with deep metaphor comprehension, often misinterpreting concepts and relying on lexical associations over context or full syntactic understanding.
• AI-assisted cultural heritage dissemination: Comparing NMT and glossary-augmented LLM translation in rock art documents — Glossary-augmented LLMs significantly improve terminology accuracy in specialized translation without sacrificing overall quality.
• The Shape of Testimony: A Scalable Framework for Oral History Archive Comparison — Computational analysis reveals nuanced "structuredness" in oral histories, challenging binary classifications.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computation and Language.