Building a Franco-British Historical Newspaper Corpus on the Greek War of Independence, 1821–1832: Corpus Design, Annotation and Transnational News Circulation

Abstract The study of nineteenth-century transnational news circulation is often hindered by fragmented digital archives, uneven OCR quality, and sparse or inconsistent metadata. This article presents the design, annotation, and application of a large-scale Franco-British historical newspaper corpus on the Greek War of Independence (1821–1832). Starting from a raw retrieval universe of more than 200,000 records, the workflow produced an analysis-ready corpus of 80,507 records, from which 55,588 were classified as directly relevant to the Greek case. The article details a reproducible computational workflow combining OCR-aware cleaning, metadata recovery, conservative country/title normalization, and human-in-the-loop AI-assisted annotation. Large language models (LLMs) were used to support relevance classification, topic tagging, stance analysis, named entity extraction, and the identification of declared news origins. By extracting and normalizing news_origin information, the corpus moves beyond keyword frequency analysis and enables the reconstruction of transnational news routes between the Eastern Mediterranean and major European publication centres. As a proof of concept, the article uses the annotated corpus to map origin-to-publication flows and to estimate dateline-to-publication media lag, demonstrating how corpus design and annotation choices make visible the structural hubs, temporal delays, and archival biases that shaped European press coverage of the Greek War of Independence.