Fractal Illusions: An Experimental Study of Long-Range Sentence-Length Correlations in Randomly Generated Natural Language Texts

TL;DR

This study demonstrates that long-range correlations in sentence length can occur in randomized texts, challenging the idea that such correlations are unique markers of literary style or structural depth in natural language.

Contribution

It provides evidence that long-range sentence length correlations are not exclusive to literary texts and can arise in randomized sequences, questioning their use as stylistic indicators.

Findings

Randomized texts show long-range correlations similar to literary works.

Punctuation alone does not account for long-range correlations in sentence length.

Long-range correlations are not reliable markers of literary style or authorial intent.

Abstract

This study re-evaluates the assumption that long-range correlations in sentence length are a fundamental feature of natural language and a marker of literary style. While previous research has suggested that punctuation marks--particularly full stops--generate structural regularities in narrative texts, our experiments challenge this view. Using Chinese as the primary language, supplemented with English, we constructed randomized linguistic sequences through three distinct methods. Surprisingly, these randomized texts also exhibit long-range correlations in sentence length, some even with stronger fractal characteristics than those found in canonical literary works. These findings suggest that the presence of long-range correlations in sentence length is not sufficient to indicate authorial intention, structural depth, or literary value. We argue that punctuation-induced long-range correlations have limited value for literary stylistic analysis and cannot be considered a fundamental characteristic of human writing.