Missing of @gender or <gender> element for persons

[annabellasc]

We had an internal workshop at the Digital Academy Mainz in which we enriched person data from various research projects with gender information, using GenderGuesser, custom-trained LLMs, and the GND. I would like to integrate the results of this workshop into our MEI data (they are currently stored in a separate CSV). However, MEI does not provide a @gender attribute or a <gender> or <sex> element, whereas TEI offers both.

I think it is important to be able to record this information in the data in order to make female actors visible within their contexts. This would give us the opportunity to answer more research questions on this topic. One possible solution would be to add a @gender attribute to <persName>, like it's done in TEI. In TEI, a <gender> element could be modelled inside person, and its @evidence attribute could be used to specify the source of the information more precisely.

Example in TEI:

<person gender="M" role="poet">
     <persName>Ovid</persName>
</person>

Example in TEI with element gender:

<person role="poet">
     <persName>Ovid</persName>
     <gender value="M" evidence="selfConstructed">male</gender>
</person>

Proposal in MEI:

<contributor>
     <persName role="lyricist" gender="F">Henriette, Luise</persName>
</contributor>

[ahankinson]

It's important to note here (well, at least I think it's important 😁 ) that MEI does not have the <person> element and, as such, has no model of a Person per se. <persName> is largely intended to identify a person's name, but we shouldn't conflate the model for marking up a person's name with a model that represents a digital model of the person themself.

So I would be more inclined to suggest that, rather than add these attributes / elements in the <persName> that the <person> model for the TEI is explored for importing into MEI.

[gucl-mu]

Thank you, Annabella, that's really an important finding, and we should address this issue quickly and find a solution!
@ahankinson do you think a standOff like in TEI could “solve” this problem? That is, encode the persons (with gender-attribute) in the standOff and, when using the names elsewhere, only refer to the persona. Or do you see another possibility?

[annabellasc]

Yes <person is not available in MEI. That's why persName is in my opinion often modelled actually as a person entity and not as a persons name. We have the opportunity to add the attributes @roleand @auth.urito persName, which both are more about the person and not the name, right? I would be happy to be able to add <person> in MEI.

[ahankinson]

@gucl-mu There’s different levels.Lets not try to boil the ocean to make a cup of tea.

If all you need is the ability to mark up a certain string of text as representing a person name, then there is no need to bring in a full person model.

If you want to start trying to model a person’s identity, however, then using the relatively simple persName element is likely not expressive enough. Other identity qualities could be things like place of birth / death / activity, language, religious / social / political affiliations, etc.

I’m not suggesting we need to do those too, but I offer this as examples where trying to condense this all into persName would get fairly unwieldy.

[ahankinson]

@annabellasc Role is a problem because a person can have multiple roles. I’ve mentioned this before but it’s not gotten much traction — a role is always in relation to something else, and not inherent to a person. Having role in persName is a kind of “shortcut” when all you want to do is say that x person has y role, but it’s not really sufficient for Modelling the relationship of a role. (For example, how long did the person have this role?You can’t say anything about that with just a simple attribute)

Auth URI can link a person’s name to an authority that does model the person. So it’s not the model of the person, it’s the place to find the model of the person.

Likewise, a person’s dates (start / end, etc) as attributes are not sufficient to say anything about them. Where was a person born? Who was their mother? Father? Where are they buried? ) Sonthey should be seen as a shortcut and not as an example of how persName is actually a model of a person.