The Narrative Composition Analyzer (NCA) is a deterministic framework for the structural analysis of narrative prose. The system models narrative text as a sequence of functional sentence roles and evaluates how those roles combine to produce measurable patterns of narrative composition. Rather than interpreting semantic meaning or literary quality, NCA focuses on observable structural properties of prose, including sentence-function distribution, narrative grammar conformance, and localized structural constraints. Within the NCA framework, each sentence in a text is assigned a functional tag drawn from a finite alphabet representing common narrative roles, including environmental description, action, dialogue, internal reflection, institutional framing, and figurative compression. These tags serve as the terminal symbols of a narrative grammar that operates over sequences of sentence functions rather than lexical tokens. By representing prose as a structured sequence over this functional alphabet, the analyzer can evaluate whether passages conform to a declared narrative grammar and whether the distribution of sentence roles remains balanced relative to a specified composition profile. The framework introduces a dual-layer structural model. A grammar layer defines permissible arrangements of sentence functions using a BNF-inspired rule system, while a distribution layer measures proportional balance among sentence roles within spans, scenes, or entire documents. Deviations from target distributions are quantified as distribution drift, and additional constraint rules allow the detection of localized structural patterns such as clustering of figurative language, extended runs of reflection, or prolonged absence of dialogue. These structural measurements are combined to produce interpretable diagnostic reports and composition stability scores. NCA operates as a static analysis instrument for narrative prose. The system does not generate text, rewrite sentences, or attempt to infer narrative meaning. Instead, it exposes compositional patterns that are often difficult to detect during manual revision, enabling authors, editors, and researchers to observe how narrative functions evolve across a manuscript. By translating prose into measurable structural signals, the framework enables reproducible analysis of narrative composition across individual texts, genres, and large corpora. This document defines Version 1.0 of the Narrative Composition Analyzer specification, including the formal grammar model, sentence tagging system, distribution analysis methods, constraint framework, diagnostic reporting system, and reference implementation architecture. The specification establishes a transparent and extensible analytical foundation for the systematic study of narrative composition.
Corey Mock (Tue,) studied this question.