Show the evidence, and show its limits.

Source hierarchy

Period orthoepists and rights-cleared primary texts provide contemporaneous evidence. Mainstream specialist scholarship—including work by Helge Kökeritz, Fausto Cercignani and David Crystal—guides the reconstruction. The engine uses cited-per-rule synthesis and a limited exception set; it does not bulk-copy an in-copyright pronunciation dictionary.

The checked-in Shakespeare corpus comes from a pinned public-domain MIT source and is transformed at build time. The deployed tool requires no third-party corpus API. Full provenance and licence decisions are recorded in the project methodology and licensing documents.

Fact, inference and uncertainty

Dates, source editions, code versions and measured test results are treated as verifiable facts. A proposed pronunciation, rhyme equivalence or performance implication is an interpretation. Where sources permit more than one reading, the interface uses uncertainty notes rather than hiding the choice.

Original Pronunciation is not presented as one uniform national accent. Regional, social, individual and theatrical variation remain part of the historical reality.

AI assistance

AI tools may assist with code, structured drafting, test generation and comparison. They are not accepted as authorities. Public claims must be traceable to cited scholarship, rights-cleared primary evidence or reproducible project tests, and the named editor remains accountable for publication.

Corrections

Corrections should identify the page or word, the disputed output and a source where possible. Send them to [email protected]. Material corrections are tested, versioned and disclosed; silent retroactive changes to the engine are avoided.

Read how the engine works, the educational limits, or who is accountable.