Syntactic patterns predict style in narrative prose with high accuracy
Syntactic Patterns and Stylistic Functions in Narrative Prose: A Rule-Based and Machine-Learning Approach
Computation and Language
Summary
It can be hard to tell the style or tone of a written sentence just by looking at its words. This paper shows that by looking closely at grammar and the way words depend on each other, computers can guess the style of sentences in stories, like whether they describe something, explain causes, or reflect feelings. The authors used rules to label sentences and then taught a machine to recognize these styles using patterns of grammar. Their computer model was very good at guessing styles, showing grammar patterns do help understand writing style.
What this means in practice
- •For literary software developers: Create tools that automatically identify sentence styles in stories to assist writers and editors in stylistic analysis.
- •For natural language processing engineers: Enhance text classification systems by including syntactic pattern features to improve narrative style detection accuracy.
Authors
Stefana Janicijevic
Abstract
This paper presents a small-scale quantitative experiment that links syntactic structure to stylistic functions in narrative prose. Starting from a dependency-parsed corpus of 3,300 sentences, we derive sentence-level stylistic labels across five categories --- descriptive, introspective, causal, ideological, and neutral --- using a transparent rule-based procedure that inspects lemmas, universal part-of-speech tags, and syntactic relations. For each sentence we construct a compact representation of its syntactic profile as a sequence of linearised triples combining lemma, POS tag, and dependency relation. These patterns serve as input to standard machine-learning classifiers trained to predict sentence-level style. The best-performing model achieves a macro-F1 of 0.948 under 10-fold cross-validation. The experiment is implemented entirely in Python using open-source tools. Our goal is not to propose a fully fledged stylistic theory, but to offer a reproducible and extensible workflow for exploring how grammatical structure contributes to narrative interpretation.