Methods & sources

Scope

The Romantic Literature Corpus currently contains more than one hundred works totaling millions of words. It brings together British, Irish, and American writing from the late eighteenth and early nineteenth centuries.

Poetry, fiction, letters, journals, essays, autobiography, and other prose forms appear together so that readers can move between close reading and corpus-scale comparison.

Corpus-level patterns should therefore be interpreted as patterns in this collection—not as direct measurements of an entire period, readership, or literary culture.

Editorial method

Texts are derived from documented public-domain or openly licensed digital sources. Source boilerplate is removed, line and paragraph divisions are retained where practicable, and each record carries its source, reuse statement, and genre. First-publication year and national tradition are supplied where documented.

People, works, editions, sources, files, places, and artworks receive permanent archive identifiers. Those identifiers remain stable if a title, spelling, or attribution is later corrected; established public text links continue to use their existing readable addresses.

The downloadable files contain the literary text without the website interface. They are intended for reading, classroom analysis, and reproducible computational work.

Search method

Search is case-insensitive and literal. “All terms” requires every entered term to appear on the same line; “Any term” accepts a line containing at least one term; “Exact phrase” preserves the entered word sequence. Search does not currently stem words, regularize spelling, or infer semantic similarity.

The catalogue loads first. The larger full-text index is fetched in compact sections only when a search is submitted, keeping ordinary browsing lightweight.

Each result supplies one line of surrounding context and links to the corresponding place in the reading text. Counts are evidence for investigation, not interpretations by themselves.

Visualization method

The Visualization Lab calculates term counts from the same full-text sections used by corpus search. Comparative frequencies are normalized by text length where indicated. Timelines omit works whose dates have not been documented rather than estimating them.

Geographic views use exact-name matching against a small curated gazetteer. A recognized place name is a lead for close reading, not proof that a work is set, composed, or published there. Similarity and vocabulary comparisons use a reduced content-word representation and should be treated as navigational aids rather than claims of influence.

Author atlas

The Author Atlas joins documented biographical events to permanent person, place, map, event, route, and source records. Each point states its coordinate precision and evidentiary source. Where a location is debated, the interface labels it as contested rather than resolving the question silently.

Street and terrain maps support continuous zoom. Selected views also include georeferenced public-domain period maps. Lines connect events in documented chronological sequence; they do not reconstruct an author’s exact roads or travel path.

Sources

Every reading page identifies its immediate transcription source. The table below summarizes the sources represented in this release.

Citation

Each reading page includes a button that copies an MLA-style citation. When citing a search, include the query, search mode, filters, stable results URL, and access date.