Dramatic speakers

drama poem speaker
persName sp role

Encoding of dramatic speeches and speakers using sp and speaker, use of the who attribute. Use of full TEI encoding for drama to encode speech and speakers in both dramatic and non-dramatic texts. This entry discusses the differences and what encoding features are best used in each context.

Although the concept of the dramatic speaker is most familiar from dramatic texts, it also appears (particularly in early texts) as a feature of verse dialogues, prose dialogues (a familiar form for philosophical writing), and reports of trial or interrogation proceedings, as well as occasionally turning up in novels as a shorthand way of representing dialogue. The TEI provides the speaker element as part of the general provision for encoding drama, but some modification of the DTD is required to use this element in non-dramatic contexts.

Traditional dramatic texts have a typical set of textual markers (speeches, cast list, stage directions) which signal their nature as performance texts, and which appear even in cases (such as Cavendish’s dramas) where the play was arguably not written to be performed. Some of these markers, and in particular the speaker label, are also used in non-dramatic texts to signal dialogue and create a sense of performative immediacy. Although you can encode both dramatic and non-dramatic texts using the full TEI encoding for drama, the full encoding may not be necessary or useful in all cases. This entry discusses the differences and what encoding features are best used in each context.

In dramatic texts, speaker is always nested within sp. The sp element carries a who attribute which allows the speaker of the speech to be identified (using an IDREF that points to corresponding role element in the cast list); see more on the who attribute for details. As data, the speaker element and its content are redundant if the who attribute is used. What they record is the specifics of that label: the form in which the speaker name is given, the method of abbreviation, and similar details. These are important if the goal of the encoding is to present the exact content of the text in full, and they may also be important as evidence in cases where the speaker identification is uncertain and presenting the primary textual evidence is necessary. Given the documentary approach we are describing here, we recommend that speaker labels be transcribed and encoded with speaker. There is no need to use abbr to encode abbreviated speaker labels (since the full name can be supplied using the information provided by the who attribute on sp). There is also no need to use name or persName within speaker, since these names are not being used within the flow of text (in other words, they do not function discursively as names).

In non-dramatic texts such as verse or prose dialogues, it may still be useful to mark speeches and speakers. Particularly in trial proceedings, or in extended dialogues where you may want to be able to compare the styles or vocabulary of the various speakers, identifying each speech with sp and each speaker with speaker can be very helpful. If the speaker labels vary (for instance, if they are inconsistently abbreviated) and cannot be used reliably to identify the speaker for processing purposes, then you may also want to use the who attribute on sp. Since dialogues of this sort typically don’t have cast lists, you will need to create a dummy cast list somewhere in the text (we recommend using the hyperDiv , an element invented by the WWP. This cast list will contain one role element for each speaker; the who attribute on each sp will point to the id of the appropriate role element, thereby identifying the speaker.

You may choose not to use sp for non-dramatic texts (i.e. dialogues which lack the full apparatus of drama). In this case, you sacrifice the ability to identify the boundaries of individual speeches, and you gain simplicity of encoding. If you choose not to use sp for non-dramatic texts, the speaker label takes on a significant function (since it is the only signal that dialogue is taking place). If you wish to use speaker to encode it, however, you will need to modify the TEI DTD to allow speaker to appear as the first child of lg and p. The extension files supplied with this Guide include this modification.

For large collections of plays in which different encoders may be encoding different plays, we recommend using a standard system to generate id attributes for role in a consistent way to guarantee their uniqueness. An example of such a system follows:

Thus, the role ID for the character Lady Happy in The Convent of Pleasure would have a value rcplh (r for role, cp for Convent of Pleasure, and lh for Lady Happy).