A Multilingual Corpus of Historical Music Lexicography from the Long Nineteenth Century

Historical music dictionaries are accessible as scans and optical character recognition (OCR) text, but searching across them requires attention to language, edition, entry boundaries and digitisation history. Lexicon C19 provides a provenance-aware index of eight music, dance, biographical and general-reference traditions, processed in English, French, German, Russian and Portuguese. Release v3.2.1 contains 3760 rule-selected records from a pool of 25,186 parsed candidates, with metadata, classification rules, build code and analytical notebooks. English-language witnesses contribute 81.9% of selected records; the corpus is not a balanced cross-language sample. In an author-conducted, label-blinded audit of 200 selected records, 97.0% were judged real and relevant; broad-category agreement with the author’s re-annotation was 89.7% among those records. This is not independent validation. Two reproducible retrieval examples demonstrate browsing and recovery of candidates outside the selected index, while exposing homographs and selection limits. Reversing keyword-rule order changes 5.08% of top-level labels without changing selected membership. The resource supports passage discovery and classification experiments; it is not a gold-standard annotation corpus, a validated translation resource or an OCR-accuracy benchmark.