1. Introduction

1In the October 2022 release of its Guidelines, the TEI introduced four major changes in the personography encoding of sex and gender:

• Removal of references to external standards for coding either sex or gender

• Revision of language relating to the encoding of states and traits supported by new literature in the Guidelines’ bibliography

• Introduction of a new <gender> element with accompanying attributes (@value and @gender and datatype (teidata.gender)

• Revision of the <sex> element and accompanying attributes with modification of the teidata.sex datatype

3The Guidelines now recognize that there is no one satisfactory global standard for the encoding of sex and gender. In the Names, Dates, People, and Places chapter, the Council’s revisions involved removing references to external standards, relying instead on discussion of states and traits and a range of examples to guide TEI personography development. The new Guidelines for <sex> and <gender> preserve backward compatibility for encoding sex, since no values are ruled out, and the changes should enhance ontological work in TEI personographies.

4The revisions in the October 2022 release come at a time of increased activity and discussion of linked-data mapping and tooling designed for TEI projects. By no longer recommending any specific external ontology, the Guidelines may better support theoretical work with ontologies developing in TEI projects. The new encoding for <gender> and <sex> is also strongly influenced by the internationalization of the TEI Guidelines and the authority of our diverse TEI community to counteract biases in computational culture—biases that have shaped the language and tooling of interoperation, that are limited by an emphasis on dominant cultures of the present, and that lack awareness of the diversity of cultural heritage. TEI projects regularly digitize texts from other times and places, and their prosopography encoding can run counter to western European or contemporary Global North assumptions, with alternative views of humanity and alternative modeling of concepts like sex, gender, and age that we associate with states and traits of existence.

2. Confronting TEI’s troubled past

5Initially the working group began with a ticket (https://​github.​com/​TEIC/​TEI/​issues/​2189) requesting the introduction of a <gender> element, knowing that the language of the Guidelines would need to be altered to support it. From Council discussions, it was known that multiple TEI projects and members of the community had called for a way to mark personal <gender>, and the group sought to address a long-standing lack of this capacity in the Guidelines. It was also known that they would need to confront a troubled history of inadequate revisions in this area.

2.1 Relying on the ISO or vCard

6In late 2007, the very first release of TEI P5 introduced the <sex> element as part of its newly formulated specifications for marking prosopography data, and to extend what was formerly available only with a @sex attribute in P4 and earlier (see example 1, below). From the beginning the Guidelines set a @value attribute on the <sex> element, for which valid values were defined by the ISO standard 5218, dating from 1977: Information interchange—Representation of human sexes:

• 0 = Not known

• 1 = Male

• 2 = Female

• 9 = Not applicable

As with other culturally-constructed traits such as sex, the way in which this concept is described in different cultural contexts may vary. The normalising attributes are provided as a means of simplifying that variety to Western European norms and should not be used where that is inappropriate. The content of the element may be used to describe the intended concept in more detail, using plain text.
(TEI Guidelines Version History)

9It was not until 2013 that the TEI would begin more thoroughly confronting the problem of relying exclusively on the ISO, through a ticket opened by Melissa Terras on Sourceforge, which also mentioned a tweet from Stephen Ramsay:

TEI uses ISO 5218:2004 to assign sexuality of persons in a document (with attributes being given as 1 for male, 2 for female, 9 for non-applicable, and 0 for unknown). This is an outmoded and problematic representation of sexuality, and in particular formally assigns women to be secondary to men.
There are other discussions online regarding how best to tackle sexuality in markup, and the problems in using ISO 5218 - see the w3c lists here: http://​lists.​w3.​org/​Archives/​Public/​public-contacts-coord/​2010JulSep/​0010.​html.
I would like to see TEI move away from enshrining women as the second sex in their markup - as Steven [sic] Ramsay tweeted:
<author/>Simone de Beauvoir</author> <sex value="2">female</sex> sigh
Can a discussion be had about how best to achieve this? Your current approach is both outmoded and offensive.
best,
Melissa
(TEI ticket #425)

10The TEI suffered embarrassment in 2013 not just for codifying females as the second sex, but also for failing to recognize the inappropriateness of the ISO for contending with the complexity of sex and gender overall, particularly in humanities research. Cultural heritage work in the TEI requires more nuance, and the TEI needs to encourage projects to define their own values.

11The TEI Council’s work through Gabriel Bodard in April 2013 untethered the TEI from the ISO, by changing the datatype of the attributes @value and @sex from the old ISO numbers to the open data.word (now teidata.word), renamed as data.sex (now teidata.sex). Declaring a specific datatype encourages a systematic encoding without imposing a problematic universal set of values (see example 1).

Example 1. The encoding of the element <sex> before TEI P5 version 4.5.0. Line 1: TEI P5 version 1.0.0; line 2: TEI P5 version 2.4.0.

<sex value="2">female</sex>
<sex value="F">female</sex>

2.2 New Description and Remarks for <sex> and teidata.sex

13Tasked with introducing a new <gender> element and with making corresponding revisions to the <sex> element, the working group decided to step away from recommending any specific standards, in deference to projects that might locate culture-specific models for sex and gender in their own fields. Its revision simplified the remarks on the @value and @sex attributes to read: “Values for this attribute may be locally defined by a project, or they may refer to an external standard.”

14One more revision was made to the definition of <sex>, to separate it from its original association only with the <person> element:

• Originally, the description of <sex> was “specifies the sex of a person.”

• It was changed to: “specifies the sex of an organism.”

2.3 Introducing <gender> and teidata.gender

16At the same time that it introduced the revision to <sex>, the working group prepared a new specification for a <gender> element and defined it with a similar new attention to a broad scope for application. The <gender> element follows upon the recent introductions of the <persona> element in 2016 and the <persPronouns> element in 2021, each of which extends the TEI’s capacity to encode self-characterizations. Like <sex>, <gender> may be contained by the elements <person>, <personGrp>, and <persona>. The @value attribute on <gender> (and the @gender attribute for <person>, <personGrp>, and <persona>) are defined by a new teidata.gender datatype, which “defines the range of attribute values used to represent the gender of a person, persona, or character.”

18As with teidata.sex, the working group’s note on teidata.gender indicates that values “may be defined locally by a project, or they may refer to an external standard,” without specifying any standard. The note additionally cautions that “values for this datatype should not be used to encode morphological gender,” so as not to confuse the new element with <gen>, used in lexical dictionary encoding of terms. Examples for teidata.sex and teidata.gender now demonstrate categories like "NB" for nonbinary, "I" for intersex, and the use of multiple attribute values.

19Revising the language of the TEI Guidelines themselves in the Names, Dates, People, and Places chapter would prove more challenging for the working group than revising and introducing the new code specifications.

2.4 Traits, States, and Sex/Gender Systems

20The encoding of personal data about sex and gender was handled minimally in early releases of the TEI Guidelines. We do not find the words sex or gender in the Names and Dates chapter of the TEI P4 guidelines, which established a basis for encoding ranks, roles, and titles of persons. Marriage was mentioned once in the P4 chapter, as an example of an event that can be dated in time. Tellingly, the word state was restricted to the concept of a geopolitical entity, and the word trait was not used yet. In prosopography encoding, these would both soon become organizing principles for separating identifying descriptors that either change (states) or do not change (traits).

21In November 2007’s first release of TEI P5 and its thoroughly rewritten chapter 13: Names, Dates, People, and Places, we find the first discussion of state and trait, together with the Guidelines’ first discussion of biographical sex and gender. The new language and scope of P5’s first version of chapter 13 reflected the substantial efforts of a TEI Personography Task Force that met in 2006–2007, produced a paper reviewing strategies by several existing XML models of biographical markup, and developed an ODD model and proposal for TEI personography encoding (Wedervang-Jensen and Driscoll 2006, Burnard and Cummings 2007). The working group’s efforts offer a perspective on the state of XML encoding systems in the early millennium, as its report comments that four systems (EAC, GEDCOM XML, GENTECH XML, and HR-XML), unlike the TEI at the time, offered elements either for <sex> or <gender> as personal data, while the TEI only provided a @sex attribute, significantly restricting what could be marked of complicated cases (Wedervang-Jensen and Driscoll 2006). The task force prepared a preliminary ODD which introduced three distinct classes of element, described as follows:

• model.persTraitLike: “the class of elements describing generally unchanging physical or socially-constructed characteristics of a person, for example hair-colour, ethnicity, sex. . . . These characteristics of an individual are typically independent of their volition or action.”

• model.persStateLike: “the class of elements describing changeable characteristics of a person which have a definite duration, for example occupation, residence, name. . . These characteristics of an individual are typically a consequence of their own action or that of others.”

• • 7 See https://​tei-c.​org/​Vault/​Workgroups/​PERS/​persw06.​xml for the language of this early ODD p (...)

  model.persEventLike: “the class of elements describing specific events in a person’s history, for example birth, marriage, appointment. . . . These are not characteristics of an individual, but often cause an individual to gain such characteristics, or to enter a new state.”7

22Sex is represented in these statements as something either physical or cultural that typically cannot be changed, and personal gender is not represented at all. It may be that the TEI task force left the encoding of personal gender to project ODD customizations as necessary to prevent confusion with morphological gender, and to direct them to broader elements first introduced by the task force as <persState>, <persTrait>, and <persEvent>, and later modified in the Guidelines for more general prosopography applications as <state>, <trait>, and <event>.

23The Guidelines’ specification of personal traits evidently posed problems for the 2011 Council, so that by December 2011’s release of TEI P5 version 2.0.0, model.persTraitLike was removed, though without comment in the release notes. The elements within it were simply moved to model.persStateLike:

• Members of model.personPart in version 1.9.0:

  model.persTraitLike [age faith langKnowledge nationality sex socecStatus trait] model.persStateLike [persName affiliation education floruit occupation residence state] model.persEventLike [birth death event] bibl

• Members of model.personPart in version 2.0.0:

  model.persStateLike [persName affiliation age education faith floruit langKnowledge nationality occupation residence sex socecStatus state trait] model.persEventLike [birth death event] bibl

24Despite the removal of the model.persTraitLike class, the language of the Guidelines’ Names, Dates, People, and Places chapter was left unaltered. The passage in 13.3.1 “Basic Principles” declared a fundamental distinction between traits and states in its guidance for prosopography markup. In that context, it introduced physical sex as a trait and mentioned gender only in passing, as determined by culture. Here is the relevant passage, which remained largely unmodified in the P5 Guidelines from 2007 until 2022:

Characteristics or traits are typically independent of an individual’s volition or action and can be either physical, such as sex or hair and eye colour, or cultural, such as ethnicity, caste, or faith. The distinction is not entirely straightforward, however: while sex is fairly obviously a physical trait, gender should rather be regarded as culturally determined, and the division of mankind into different races, proposed by early (white European) anthropologists on the basis of physical characteristics such as skin colour, hair type and skull measurements, is by many modern cultural anthropologists now considered to be more a social or mental construct than an objective biological fact. Furthermore, while some characteristics will obviously change over time, hair colour for example, none, in principle—not even sex—is immutable.
(TEI Guidelines’ Names, Dates, People, and Places, 13.3.1 “Basic Principles”)

25When the working group in 2022 fully confronted this passage, they were first surprised that these problems had not been noticed before, and wondered how it had survived so long without revision. (For example, why did no one challenge the idea that religious faith be considered an example of a trait independent of an individual’s volition? Why did no one remove the word mankind?) Upon reflection, they realized that philosophical passages like this tend not to occupy the Council or the community as much as problems in the coding specifications. Like many invisible yet ingrained assumptions in our world, we have left passages like this in authoritative places without noticing them—until we were directly motivated to search out mentions of sex and gender in the Guidelines to revise.

27In 2022, the working group revised the passage to connect gender and sex, and though it preserved the distinction of traits and states, it tried to emphasize the temporary cultural constructions that define individual characteristics:

• 9 [For example, see Renato G. Mazzolini’s (2018) historical survey of European empires and their unst (...)

Characteristics or traits are typically independent of an individual’s volition or action and can be either physical, such as hair and eye colour; or cultural, such as ethnicity and caste. The distinction is not entirely straightforward: gender and sex can be regarded as culturally determined with physical manifestations, interpreted based on the social hierarchies of a time and place. Also the division of humanity into different races, formalized in the nineteenth century to organize a hierarchy of humanity on the basis of physical characteristics such as skin colour, hair type, and skull measurements, is now recognized as a colonialist social construct.9 When such a construct defined individuals’ identities beyond their volition, the markup and documentation should reflect this as a culturally-determined characteristic.
(TEI Guidelines’ Names, Dates, People, and Places, 13.3.1 “Basic Principles”)

28The revision associates gender with sex “as culturally determined with physical manifestations.” This phrasing sought to emphasize the cultural categorization of bodily forms, rather than take for granted a universal defining system. To build on the efforts of its predecessors, the group wanted the Guidelines’ language to complicate even further the problem of marking sex and gender, and to call on TEI personography encoding to clarify its awareness of the rules of meaning-making that governed the documents we work with and the named entities they reference. To further emphasize the shifting nature of established classification systems as well as the intersection of sex, gender, and race, a new footnote was added to the sentence about racial constructions, citing Renato G. Mazzolini’s study “Colonialism and the Emergence of Racial Theories” (Mazzolini 2018). How important any cultural system’s definition of sex and gender may be to a project cannot be predicted, but we seek to provide tools for the TEI community to engage with these concepts meaningfully in all their complexity with respect to the places and times of the documents they study.

29That is why the working group also slightly altered the following cautionary passage from 13.3.2.1 “Personal Characteristics.” Before October 2022, the passage read:

The meanings of concepts such as sex, nationality, or age are highly culturally-dependent, and the encoder should take particular care to be explicit about any assumptions underlying their usage of them. For example, when recording personal age in different cultures, there may be different assumptions about the point from which age is reckoned. A statement of the practice adopted in a given encoding may usefully be provided in the <editorialDecl> element discussed in 2.3.3 The Editorial Practices Declaration.
(TEI Guidelines’ Names, Dates, People, and Places, 13.3.2.1 “Personal Characteristics.”)

30The revision rearranges and adds to the first sentence and adds a footnote, as well as a new third sentence cautioning about markup of sex and gender:

• 10 [Following Gayle Rubin’s influential theory, ‘Every society also has a sex/gender system—a set of a (...)

Encoders should take particular care to be explicit about culturally determined concepts such as nationality, ethnicity, age, or sex/gender system that influence their markup.10 For example, when recording personal age in different cultures, there may be different assumptions about the point from which age is reckoned. Scholars face similar issues when marking sex and gender according to cultural systems from times and places not their own. A statement of the practice adopted in a given encoding may usefully be provided in the <editorialDecl> element discussed in 2.3.3 The Editorial Practices Declaration.
(TEI Guidelines’ Names, Dates, People, and Places, 13.3.2.1 “Personal Characteristics.”)

31This revision emphasizes the need for attending to cultural constructions. It also introduces the frequently cited phrase sex/gender system, for which we reference the well-established theory of anthropologist Gayle Rubin, whose ideas—including the provocative application of the solidus character (/)—were found particularly helpful for reflecting on the cultural intersections of sex with gender.

32The concept of the sex/gender system encourages reflection on how a culture’s views of reproduction may shape its available range of recognizable gender identities. When biology and culture are enmeshed in human life and language, we should be challenged in our editorial practices to describe the cultural bases of our prosopographic classifications. When do we follow the conventions of the time and place of our encoding, and when do we attempt to represent the ways of an unfamiliar culture? In these revisions we seek a definitive step away from the misguided universalizing assumptions of the old ISO standard that our Guidelines used to reinforce. Rather than encouraging reliance on any external standard, we urge that the encoding of sex and gender be formulated by the specialists who work on documents and study the cultures that generate them.

3. What Can We Do with the New Gender and Sex Encoding?

33The following sections propose some examples to show a variety of ways these encodings can be applied in TEI projects. They are based on English literary and performance texts familiar to the authors, and encourage contributions toward a more diverse variety of examples from non-Anglophone literary traditions.

3.1 Orlando

34Orlando: A Biography by Virginia Woolf (1928) can serve to illustrate the flexibility of the new <gender> element, whose combination with <sex> provides us with an expressive tool within the semantic spectrum to represent sex/gender identifications. In this novel, the protagonist, Orlando, undergoes a sudden change of sex. In the first chapter, Orlando is presented as a he (and the narrator further clarifies “for there could be no doubt of his sex” [15]). The change takes place in chapter 3: Orlando wakes up as a woman (126). In this chapter, the narrator goes from using the pronoun he, to briefly using their, to quickly making explicit that the pronouns to be used will be her and she:

Orlando had become a woman—there is no denying it. But in every other respect, Orlando remained precisely as he had been. The change of sex, though it altered their future, did nothing whatever to alter their identity. Their faces remained, as their portraits prove, practically the same. His memory—but in future we must, for convention’s sake, say her for his, and she for he—her memory then, . . . (127)

35The narrator leaves open the interpretation of how this change should be explained (127–28), which for us means that we can encode different interpretive hypotheses.

The change seemed to have been accomplished painlessly and completely and in such a way that Orlando herself showed no surprise at it. Many people, taking this into account, and holding that such a change of sex is against nature, have been at great pains to prove (1) that Orlando had always been a woman, (2) that Orlando is at this moment a man. Let biologists and psychologists determine. It is enough for us to state the simple fact; Orlando was a man till the age of thirty; when he became a woman and has remained so ever since. (127–28)

36Considering the information given by the narrator, the sex change could be encoded as in example 2, with the addition of a <milestone> element at the beginning of the work (identified as "ch1") and a second milestone in chapter 3 when Orlando wakes up as a woman ("ch3"):

Example 2. Encoding example that represents Orlando’s sex change.

<person>
  <persName>Orlando</persName>
  <sex from="#ch1" to="#ch3" value="M"/>
  <sex from="#ch3" value="F"/>
</person>

37But what about Orlando’s gender? We could decide that Orlando’s nonplussed attitude is due to a nonbinary identification of their gender, which we could describe with the term fluid (example 3):

Example 3. Encoding example of a possible interpretation of Orlando’s gender.

<gender value="fluid"/>

38We could also encode the interpretation of the two groups of people, described by the narrator, who wanted to prove that either Orlando was always a woman, or that Orlando was still a man after the sex change. According to the first group, for example, Orlando’s description should be as follows (example 4):

Example 4. Encoding example of the interpretation that Orlando has always been a woman.

<person>
  <persName>Orlando</persName>
  <sex from="#ch1" to="#ch3" value="M"/>
  <sex from="#ch3" value="F"/>
  <gender value="W"/>
</person>

39We could consider that the sex change entailed a gender change (example 5):

Example 5. Encoding example of the interpretation that both a sex and a gender change take place.

<person>
  <persName>Orlando</persName>
  <sex from="#ch1" to="#ch3" value="M"/>
  <sex from="#ch3" value="F"/>
  <gender from="#ch1" to="#ch3" value="M"/>
  <gender from="#ch3" value="W"/>
</person>

40These examples represent only a simplified, small set of encoding possibilities, since Orlando’s transformation could be annotated in other ways, and other combinations of values for both <sex> and <gender> could be relevant. As an alternative, we could add other milestones and a further <gender> element, pointing to the moment of transition expressed in the narration through the use of pronouns, as evidenced by a nonbinary identification in chapter 3 (“Their faces remained, as their portraits prove, practically the same”). This case could be complemented with the use of the element <persPronouns> (example 6).

Example 6. Encoding example that combines the use of <sex>, <gender>, and <persPronouns>, the latter used as evidence for the values of <gender>.

<person>
  <persName>Orlando</persName>
  <sex from="#ch1" to="#ch3" value="M"/>
  <sex from="#ch3" value="F"/>
  <gender from="#ch1" to="#ch3L2" value="M"/>
  <persPronouns evidence="narration" from="#ch1" to="#ch3L2" value="he">he/him/his</persPronouns>
  <gender from="#ch3L2" to="#ch3L4" value="NB"/>
  <persPronouns evidence="narration" from="#ch3L2" to="#ch3L4" value="they">their</persPronouns>
  <gender from="#ch3L4" value="W"/>
  <persPronouns evidence="narration" from="#ch3L4" value="she">she/her/hers</persPronouns>
</person>

3.2 Performance Texts

Example 7. Encoding example of a <castList> that makes explicit the supposed sex of the actor and the gender of the role to be performed.

<front>
  <lb/><castList>
    <lb/><head>The Actors Names.</head>
    <lb/><castItem>
      <role gender="m">Boſola</role>,
      <actor sex="m">I. Lowin</actor>
    </castItem>
    <!-- other castItem elements -->
    <lb/><castItem>
      <role gender="f">The Dutcheſſe</role>,
      <actor sex="m">R. Sharpe</actor>
    </castItem>
    <!-- other castItem elements -->
  </castList>
</front>

43For a gendered character who performs a different gender, such as the eponymous silent woman in The Persons of the Play in the 1616 Jonson Folio (figure 2), we might add the @gender attribute to the <roleDesc> element as follows (example 8):

Example 8. Encoding example of a gendered character that performs a different gender.

<castList>
  <castItem>
    <roleName>EPICOENE.</roleName>
    <roleDesc gender="m">A yong Gent.</roleDesc>
    <roleDesc gender="w">suppos'd the silent Woman.</roleDesc>
  </castItem>
</castList>

4. Conclusion

44Members of the TEI Technical Council devote much of their time to patching and improving the schema specifications of the Guidelines and to maintaining the programming that coordinates their publication and release. The prose of the Guidelines precedes and determines the elements, attribute classes, and examples we provide, but rarely do we contend extensively (or comfortably) with the Guidelines’ conceptual framework. Perhaps our best long-range guidance enhances awareness of possibilities beyond our inevitably limited examples. Council members might well adapt the Hippocratic stance, “first, do no harm,” for our work in maintaining the Guidelines as a living document and functioning schema. Just as we try to preserve backward compatibility when introducing changes, we should also try to lift restraints to encourage our community’s diverse varieties of research. Our working group’s discovery of outdated language in the Guidelines prodded us to reflect on the variety of digital humanities projects that encounter sex and gender, and the reflection gave rise to new discussions and possibilities.

SVN keywords: $Id: jtei-beshero-bondar-237-source.xml 1304 2025-04-28 15:44:40Z pietro.liuzzo $