Wikidata:WikiProject Performing arts/Data sources/Character roles

Character Roles in Operatic Works – Data Modelling Issues

In order to be able to represent the cast information contained in the Carnegie Hall Performance History Database, Wikidata entries for the various character roles in operatic works have to be created. As the German and English Wikipedias provide complete lists of operatic characters for many operatic works, it was decided to use this data as a starting point for the creation of the character role items instead of creating entries based on the information contained in the Carnegie Hall Performance History Database or other similar databases.

Example Items

• Die Zauberflöte (opera by Wolfgang Amadeus Mozart): The Magic Flute (Q5064)
• Carmen (opera by Georges Bizet): Carmen (Q185968)
• Aufstieg und Fall der Stadt Mahagonny (opera by Kurt Weill): Rise and Fall of the City of Mahagonny (Q760054)
• Mignon (opera by Ambroise Thomas)

Data Modelling Issues

Character role labels

Some character role labels contain just the name of the character, e.g. "Cio-Cio-san (Madama Butterfly)", "Tamino", "The aunt", "The Imperial Commissioner". The name may refer to the character's proper name or to their function or some other characteristics. Other character role labels contain a name and an attribute, e.g. "Suzuki, her maid" or "B.F. Pinkerton, Lieutenant in the United States Navy". Name and attribute are typically separated by a comma; in English Wikipedia, the attribute is in addition set in italics.

Before ingesting the data into Wikidata, the character role labels are split into a name and an attribute part:

• The name is rendered as the label of the item describing the character role. Examples: "Cio-Cio-san (Madama Butterfly)"; "Suzuki".
• The attribute (if present) is appended to the description of the item, following a semicolon. Example: "character in the opera Madama Butterfly by Giacomo Puccini; Cio-Cio san's maid" (note that the possessive pronoun has been resolved, see the note below).
• The full label in its original form (if different from the name) is rendered as an alias of the item. Example: "Suzuki, her maid" (note that the possessive pronoun has *not* been resolved).

Note that some character role labels contain possessive pronouns which refer to other characters, e.g. "Suzuki, her maid" in Madama Butterfly or "The Emperor Altoum, her father" and "The Unkown Prince (Calaf), his son" in Turandot, which poses a problem when the characters are presented in isolated form, which is the case with individual WD items. Therefore, on the Character Role Item, the original label with the possessive pronoun is entered as alias; in addition, the attribute with the possessive pronoun resolved, is appended to the description, e.g. "Cio-Cio-san's maid" (instead of "her maid"), "Princess Turandot's father" (instead of "her father"), "Timur's son" (instead of "his son").

Note that the order in which the characters are listed may vary from one language to the other (and from one production to the other).It therefore does not make much sense to indicate the order of characters on the work Item using a series ordinal (P1545) qualifier.

Collective roles

Many operatic works have collective characters; example: "Three ladies", "Three child-spirits", "Three priests", "Two armoured men", "Three slaves", as well as "Priests, women, people, slaves" (chorus) in Mozart's Magic Flute.

These characters are entered into WD in the same way as the individual characters.

A difficulty arises however when one language version gives them as collective characters, while another gives them as individual characters, e.g. Mozart's Magic Flute (German version). In this case, the more detailed version is given preference, and the other one is ignored.

Voice type or Fach?

There are two WD properties to indicate the voice type: voice type (P412) and its sub-property Fach vocal classification (P1731). While the seven basic voice types (soprano, mezzo-soprano, contralto, countertenor, tenor, baritone, bass) mainly refer to the vocal range, the German Fach system is a method for classifying opera singers according to the range, weight, and color of their voices. While the Fach system is largely aligned with the seven basic voice types, it comprises more specific types which sometimes make reference to the type of role corresponding to the type (e.g. buffo, character role, etc.).

Another difficulty arises from the fact that up until the 19th century, many opera parts were originally written for castrato voices, but are today either transposed or performed by female singers. In these cases, the original Fach indication mainly serves a historical purpose.

In order to maximize consistency, the basic voice types will be indicated for all the character roles using the voice type (P412) property. Where the database contains more specific indications, reference to the Fach system and similar specifications is made by means of the Fach vocal classification (P1731) property.

In order to maximize consistency and to avoid keeping a lot of redundant information, the basic voice types will be indicated using the voice type (P412) property; where more specific information is available (Fach or similar), it is indicated by means of the Fach vocal classification (P1731) property. Furthermore, it needs to be ensured that each instance of Fach (Q1748957) is defined as a sub-class of one of the basic voice types.

Check out: choral voice types; intermediate types; correspondence between languages. [to do]

Different voice types for different versions of the opera

Some role tables contain different voice types for different versions of the work, e.g. Mignon. In the course of the present ingest, only the data for the original version are ingested. Additional versions would need to be modelled separately in Wikidata, as the Mignon example illustrates (it is presently listed both as a "tragedy" and as an "opéra comique"). Once the different versions have been entered as separate items, different voice types can be specified for different versions of the work where necessary.

Character Roles in Operatic Works – Mapping of the Dataset Compiled from Wikipedia

Mapping Table

The mapping table below contains the data mapping between the Character Role Database, compiled from Wikipedia entries, and Wikidata.

Class (SPA / Wikidata)	Property (in datafile)	Property Value (in datafile)	Property (SPA / Wikidata)	Refers to class (SPA) / possible values	Remarks
Work (operatic work) opera (Q1344)	Work_en	n/a (String)	Len	String	To be ingested only if the label is missing in WD.
	Work_de	n/a (String)	Lde	String	To be ingested only if the label is missing in WD.
	Work_WD	n/a (WD Q number)	Q	WD Q number	to be used for matching purposes
	CharacterRole_en	n/a (String)	characters (P674)	Character Role operatic character (Q50386450)	indicate the series ordinal (P1545) as a qualifier when linking to the Character Role
	CharacterRole_de	n/a (String)
Character Role (operatic character) operatic character (Q50386450)	CharacterRole_Len (substring of role label preceding the first comma)	n/a (String)	Len	String	To be ingested only if the label is missing in WD.
	CharacterRole_Lde (substring of role label preceding the first comma)	n/a (String)	Lde	String	To be ingested only if the label is missing in WD.
	CharacterRole_Aen (full role label as given in the source)	n/a (String)	Aen	String	To be ingested only if Aen is different from Len.
	CharacterRole_Ade (full role label as given in the source)	n/a (String)	Ade	String	To be ingested only if Ade is different from Ade.
	CharacterRole_Den	n/a (String)	Den	"character in the opera XY by ABC; attribute of the role"	The name of the opera and the composer need to be pulled from WD. If the designation of the role contains an attribute (such as "Suzuki, her maid" or "B.F. Pinkerton, Lieutenant in the United States Navy"), the attribute is provided after a semi-colon. In cases where the original attribute contains a possessive pronoun, the latter needs to be resolved by hand before the data ingest (e.g. "character in the opera Madama Butterfly by Giacomo Puccini; Cio-Cio san's maid").
	CharacterRole_Dde	n/a (String)	Dde	"Figur in der Oper XY von ABC; attribute of the role"
	VoiceType	(various voice type and Fach indications)	voice type (P412) or Fach vocal classification (P1731)	See the table in the Typology section for a list of voice types and Fächer.
		buffo	Fach vocal classification (P1731)	buffo (Q53837302)
		child's voice	Fach vocal classification (P1731)	child's voice (Q6164901)
		silent role			variants: silent; silent actor; does not sing; non-singing; non singing; tacet; mute; mute role; non-speaking
		microphoned male singer
		child, non singing
		sprechgesang		sprechgesang (Sprechgesang (Q1500983))
		spoken role			variants: speaking role; speaking part; spoken part; spoken; actor
		offstage spoken role for male voice
		spoken and prerecorded
		narrator
		speaker
		dancer			variants: male dancer
		mime
		(contains "breeches role" or equivalent)			variants: en travesti, trouser role
	Work_WD	n/a (WD Q number)	present in work (P1441)	Work (operatic work) opera (Q1344)
Performance (world premiere)	(implicit)		instance of (P31)	opera performance (Q47467927)
				world premiere (Q2500107)	Use in combination with the qualifier of (P642).
	Work_WD	n/a (WD Q number)	based on (P144)	Work (operatic work) opera (Q1344)
			of (P642)		To be used as qualifier in combination with <instance of (P31)> world premiere (Q2500107).
	PremiereCast_en	n/a (String)		Person / Agent	Does not exist in the German version. Groups of humans, such as choirs, ballet ensembles, etc. need to be identified in the dataset.
	PremiereCast_WD	n/a (WD Q number)		Person / Agent
	(implicit)		Len	(title of the opera in the original language)	The title of the opera in the original language needs to be pulled from WD. To do: The handling of original work titles in non-Latin script needs to be clarified.
	(implicit)		Lde	(title of the opera in the original language)
	(implicit)		Den	"world premiere of the opera XY by ABC"	The name of the opera and the composer need to be pulled from WD.
Person / Agent human (Q5) group of humans (Q16334295)	PremiereCast_en	n/a (String)			Does not exist in the German version. To be ingested only if it is missing in WD. Groups of humans, such as choirs, ballet ensembles, etc. need to be identified in the dataset. Missing items need to be created ahead of the data ingest.

Notes Regarding the Interpretation of the Source Data

• Corresponding entries in different language versions do not contain the same number of character roles: Note that in some cases, several roles may have been combined into a collective role. In this case, the more detailed version is retained.
• What to do if we have a database entry in English and in German? – Shall we attempt to match the Character Role items? [to be clarified]

Issues / Potential Errors in the Source Data

• Some role tables in Wikipedia (example) contain information about the creative team. This information needs to be skipped when harvesting the data from Wikipedia.