The WWP encodes dramatic speakers using the <speaker> element wherever they occur, whether in dramatic texts, or in verse or prose dialogues. The only exception is where the reference to the speaker is nested inside a stage direction, in which case the <speaker> element is omitted.
In dramatic texts, <speaker> is always nested within <sp>. Speakers in drama are identified using a unique role ID, which is declared in the cast list for the play and is referenced using the who= attribute on the <sp> element. The value of the who= attribute is required for all <sp> elements. We consider dramatic texts to be those which include a full dramatic apparatus including a cast list, usually act and scene divisions, and possibly stage directions. Texts representing dramatic dialogue which lack this dramatic apparatus are treated as verse or prose dialogues, and are encoded more simply.
In verse dialogues, <speaker> is the first child of <lg>. We do not use <sp> in verse dialogues, and hence there is neither a who= attribute nor a <castlist>.
In prose dialogues, <speaker> is the first child of <p>. Again, we do not use <sp> or <castlist> in prose dialogues.
The WWP does not currently use <abbr> for names of speakers that are abbreviated, but we may add this in the future. The who= attribute can be used to provide an expanded, regularized version of the speaker’s name if that were ever needed for display or retrieval.
We do use <persname> within <speaker> to encode proper names, and when we resume using key= on <persname> we will do so within <speaker> as well. This is because we feel that the key= functions separately from the who= attribute: the key= identifies someone as a historical or literary figure and is unique within the entire corpus, whereas the who= attribute identifies someone as a dramatic character and is unique within the individual play. Thus Julius Caesar would receive the same key= value in several different plays about him, but might have a different who= value in each one.
For large collections of plays in which different encoders may be encoding different plays (e.g. Cowley, Cavendish), we use a standard system to generate who= attributes in a consistent way, as follows:
-- R for "ROLE"
-- 2 letters designating the title of the play. The abbreviations for each play’s title are determined by the encoder during document analysis.
-- 2 or 3 letters indicating the name of the character.
Thus, the who= attribute for the character Lady Happy in “The Convent of Pleasure” would have a value “RCPLH” (R=“ROLE”, CP=“Convent of Pleasure”, LH=“Lady Happy”).