Divergent Discourses Corpus: E-texts of Early Tibetan Newspapers, 1950–1965

The Divergent Discourses Corpus is a machine-readable corpus of early Tibetan-language newspapers from the 1950s and 1960s. The corpus was compiled by the Divergent Discourses project, a joint SOAS University of London–Leipzig University project with collaboration from Trinity College Dublin, to support the study of narrative construction and competing accounts of Tibet during the formative period of the Sino-Tibetan conflict.</p>

This Zenodo record deposits the e-text CSV release of the corpus. The uploaded CSV contains text-region level records extracted from page images of Tibetan-language newspapers. Each row represents a text region or paragraph and includes the original extracted text, a normalised form, reading-order metadata, a text-region type, the source image filename, newspaper code, issue date, and page number.</p>

The inspected deposit file is <code>divergent_discourses_corpus.csv.zip</code>, a ZIP archive containing <code>divergent_discourses_corpus.csv</code>. The uncompressed CSV has 523,215 text-region rows, 11 columns, and 15,033 unique page-image filenames. Its issue dates range from 1950-01-01 to 1965-12-31. The CSV contains 18 newspaper codes. The published project description gives the core DD Corpus as 16 newspapers; the present CSV also contains two additional codes, NSN and RNB, which should be checked before final publication metadata is frozen.</p>

Users should be aware that this is an OCR/HTR-derived e-text corpus, not a diplomatic edition. The project documentation discusses the biases introduced by survival, collection history, microfilm quality, scan quality, page layout, and OCR/HTR processing. The corpus is not a complete run of any single newspaper and should not be treated as a statistically representative sample of all Tibetan newspapers published in the period.</p>