A Python package for analyzing writing style through quantitative metrics.
Please read: This is still in alpha. Analytics are fine but a ways to go before releasing a functional library.
Update 2/27: First features working at basic level
Update 3/26: Revision to methodology. What worked, what didn't, and what next
Focusing now on grammatical structures rather than traditional word frequency analysis.
When applied to Milton's Paradise Lost, the framework reveals:
- Significantly higher use of periodic sentences compared to his prose work Areopagitica (19.0% vs 10.7%)
- Strategic deployment of periodic style for narrative emphasis and thematic development
- Variations in periodic sentence usage across different books, with notable decreases in Books 1 and 7
- Presence of distinctive Latinate constructions such as ablative absolutes
- Complex sentence structures that challenge both human readers and computational tools
Next steps: Focus on how to identify biblical reference before coding. We'll start with Blood Meridian, difficult, but if done right, easier on Milton
| Biblical Reference Category | Sub-category | Description | NLP Technique |
|---|---|---|---|
| Direct Quotations | With attribution | Explicitly identified as biblical | Pattern Matching & Named Entity Recognition (NER) |
| Without attribution | Implicit biblical knowledge assumed | Text Similarity Analysis | |
| Paraphrases | Close paraphrase | Maintaining most semantic content | Semantic Similarity Metrics (e.g., cosine similarity on embeddings) |
| Loose paraphrase | Preserving core meaning with significant reformulation | Paraphrase Detection Algorithms combined with Topic Modeling | |
| Allusions | Nominal allusions | References to biblical names, places, events | Named Entity Recognition (NER) to extract proper nouns and link them to biblical contexts |
| Verbal allusions | Echoes of distinctive biblical phrasing | Keyphrase Extraction and Text Pattern Matching | |
| Thematic allusions | Invocation of biblical themes or motifs | Topic Modeling (e.g., LDA) | |
| Inversions | Moral inversions | Biblical moral principles deliberately reversed | Sentiment Analysis and Contrastive Learning |
| Narrative inversions | Biblical story patterns upended | Sequence Modeling (e.g., RNNs or Transformers) | |
| Character inversions | Biblical figures recast in antithetical roles | Entity Sentiment & Role Analysis | |
| Structural Adaptations | Syntactic mimicry | Sentence structures that echo biblical cadences | Syntactic Parsing (using dependency or constituency parsers) |
| Rhetorical patterns | Use of biblical rhetorical devices | Discourse Analysis | |
| Narrative frameworks | Episode structures that parallel biblical narratives | Narrative Segmentation and Sequence Analysis | |
| Philosophical Engagement | Epistemological challenges | Questioning biblical claims to truth | Argument Mining |
| Ethical critiques | Challenging biblical moral frameworks | Sentiment and Stance Detection | |
| Ontological explorations | Engaging with biblical concepts of being | Concept Extraction and Ontology-Based Modeling | |
| Apocalyptic Imagery | End-times references | Imagery from Revelation and other apocalyptic texts | Image and Metaphor Detection using Semantic Role Labeling |
| Judgment imagery | Scenes that evoke divine judgment | Sentiment Analysis combined with Imagery Recognition | |
| Cosmic warfare | Depictions that echo biblical spiritual conflict | Event Extraction and Thematic Analysis |