Content and style in the time drift of written text through a century of newspaper editorials

This deposit contains the data and code of the paper "Content and style in the time drift of written text, traced through a newspaper's editorial prose" (E. Estevez-Rams, A. Mesa-Rodríguez and D. Estevez-Moya). The corpus is the editorial page of the Washington Evening Star from 1854 to 1963, built from the article-level OCR text of the American Stories dataset (Dell et al., 2023, CC BY 4.0) and divided into 1290 monthly units, each holding all the prose printed on the editorial page during one calendar month after removal of advertisements, notices, news reports, short items, regular columns and reprinted text. Every unit is given in four representations of equal length across months: the original text reduced to a 27-symbol alphabet, the same text with its words shuffled, the sequence of its content-free (function) words, and the sequence of its punctuation marks. The deposit includes the unit metadata (dates, word counts, OCR quality, shuffle seeds); per-month entropy density and effective measure complexity estimated from Lempel–Ziv complexity, together with Shannon entropies, sentence, punctuation-segment and paragraph lengths, and rates of use of punctuation marks and function words; the Lempel–Ziv information-distance matrices, the symmetrized Kullback–Leibler divergence matrices and the histogram-only Jensen–Shannon matrices used in the Mantel and partial Mantel tests; the result tables reported in the paper and its Supplementary Information; and the Python scripts used to build the corpus and compute every result. A README describes each file and field, and a manifest lists sizes and SHA-256 checksums. Data are released under CC BY 4.0 and scripts under the MIT License.