Encoding Principles and Commentary


The editions and archival materials in the Lili Elbe Digital Archive are encoded using TEI P5 guidelines (https://tei-c.org/guidelines/p5/). In addition, we have refined the TEI guidelines in our own Lili Elbe Digital Archive encoding guidelines.

Below are some specific principles we used in encoding the editions. A detailed description of each of the following principles can be found in our encoding guidelines at https://github.com/RJP43/LiliElbe_EngagedLearners/blob/master/ProjectDocs/encodingGuidelines.md.

  • We have captured people in persName elements and places in placeName elements across editions and archival materials, keeping a record of the variant names in our Prosopography. In the paratextual materials, we have tagged persNames with the historical names when we are certain the claim is factual, otherwise we have used the key ID for the fictional character. For example, we know the King of Denmark at that time was Christian X; however, we also know the King did not anul the Wegeners’ marriage so in the Introduction we have tagged him with his fictional key ID because the claim is not factual.
  • Our encoding hierarchy uses the div element to organize the basic structure of our XML and the self-closing pb element to indicate page beginnings. However, to stay consistent across editions for the page beginnings, we have moved each chapter's first page number, originally appearing at the bottom of the print page, to the beginning of the page's text before the chapter heading.
  • In instances of idiosyncratic or misspellings, and incorrect or missing punctuation, we have utilized the choice element, with a child orig element holding the original text or a self-closing orig element if the original text is missing, and a child reg element holding our regularized text. In TS-TR we have used this markup as well when distinguishing the translation (in the orig element) from the original wording (in the reg element) for instances when the translation may be especially challenging. In our HTML we have rendered the original with the regularized text on mouse over; however, if the text is missing altogether, we show the regularized text and the original is explained on mouse over.
  • When text is added, deleted, or substituted, we have used the respective add, del, and subst elements with attribute-value pairs to indicate the particular emendation.
  • We have also captured the print stylistic renderings such as bold, italics, underline, and the German emphasis (spaced letters) using the hi element and respective attribute-value pairs.
  • In instances where illustrations appear within the text, we have encoded these using the figure element, providing the caption in a child figDesc element, and alt text for the image in a child note element with a resp attribute indicating the encoder responsible for providing the alt text.
  • We use the @ symbol to indicate editor and translator notes, with the initials of the person making the comment, which is revealed upon mouse over.

Since a number of volunteers worked on various editions and documents over the course of several years, there are some inconsistencies in following the practices below that we intend to correct in the future.

  • If an appositive appears in the same sentence as the name, we’re tagging the name and not the appositive: e.g., in “the Danish painter Einar Wegener (Andreas Sparre)” we tag the names only.
    Danish painter Einar Wegener (Andreas Sparre)
    However, where no name appears, we tag the appositive.
    her great helper in Dresden
  • When identifying an unnamed person, the German and Danish gendered distinctions helped us determine the name—e.g., “her life comrade in the sunny south” is in the feminine so we tagged the person as Grete; “her truest friend in Paris” is in the masculine in G1 so we tagged the person Claude (the feminine would indicate Elena). The reference to “a friend in Berlin” (A1, ch. 3, p. 33) is feminine in G1 and thus likely refers to Baronin Schildt, but we are not sure so we do not give that phrase a key ID.
    a friend in Berlin
  • If a word or name splits across pages, we move the word to the page where the word first appears before the next page beginning (pg) element.
    My cousin, le Comte de Trempe Page 87
    My cousin, le Page 87 Comte de Trempe
  • When words like “friend” or “helper” are clear in context, we do not tag them; but when the referent is not clear, we do tag them: e.g., we tagged “French friend” (A1, ch. 4, p.3 9) because we know who that is (Charles Guyot).
    In collaboration with a French friend he wrote a book about Northern sagas
  • Generally, when tagging “the Professor” as Kreutz, we include “the.” When tagging “the German doctor” as Kreutz, we just tag “German doctor.” (Admittedly, this has varied with different encoders.)
  • In instances where the figure may be an historical person, such as the three specialists in Berlin, we tag each but without a key ID.
    A specialist inVersailles
    We give “a well-known musician” in chapter 4 a key ID even though we don’t know if this is an historical person or a fictional character, and likewise for Mrs. Teddybear in chapter 15, since these characters are referred to more than once and may eventually be identified.
    her husband, a well-known musician
    The goal is to facilitate the linking of characters across editions.

  • We did not try to make the editions consistent when the grammar differed: e.g., A1/B1 ch4, p. 36 refers to “a native of Copenhagen,” Copenhagen is tagged as a placeName, but in G1, p.26 “Kopenhagener” is an adjective describing the person not a place name so we didn’t tag it. In G1, we tag “Seineufer” (Seine banks) but in A1 “embankments” without Seine is not tagged. When a place name is used as an adjective (as in “Copenhagen slang” or “Berlin rolls”) we do not tag the place name.


Narratives of trans lives challenge conventions of textual scholarship and practices of digital encoding. With Lili Elbe’s narrative, the first decision we had to make in encoding this text was whether or not to tag “Andreas Sparre” and “Lili Elbe” as the same person, in keeping with current practices in transgender scholarship. We chose instead to adhere to the narrative logic and tag them as separate persons. For while contemporary practice would have us refer to the subject as Lili Elbe from her birth, that was not how Lili understood her life’s trajectory. She insisted that she was born in the women’s clinic (Staatliche Frauenklinik) in Dresden where she had three of her four surgeries. The narrative presents Andreas as a thoroughly masculine man, and Lili as a hyper feminine woman. They were two beings in one body. Andreas had to die for Lili to live.

Because this narrative is a fictionalized account of Lili Elbe’s life, we also decided to distinguish the historical persons (e.g., Einar Wegener) from their characters in the story (Andreas Sparre), giving them separate IDs. We did not want to conflate a character, such as Professor Kreutz, with the actual doctor, Kurt Warnekros. Similarly with places. Most historical places are given their actual names, but Beaugency, where the Wegeners vacationed, is named Balgencie in the narrative and encoded as such. And Hirschfeld’s Institute for Sexual Science in Berlin is named Institute for Psychiatry in the American first edition, thus we have retained that name in our encoding of the editions even though the name varies in the Danish and German editions. Although we know the name of the cemetery where Lili Elbe is buried, Trinitatis Cemetery (Trinitatisfriedhof), we do not give it an ID since it is unnamed in the narrative. The Prosopography provides all names used for any one person or place.

Yet these distinctions between fictional and historical names are upheld inconsistently in the text. The name “Lili Elbe” appears in the narrative but also in articles from the time, and as Lili’s signature on some letters in the archives. Lili Elbe is at once historical and fictional. We have given Lili Elbe one key ID. The Copenhagen art dealer, Ole Georg Haslund, an historical figure, is named Haslund in the narrative, and so we have given that character “haslund” as a key ID. Likewise, Eugen Steinach is named as an historical person in the Introduction but also in the narrative and is identified by his last name. Captions to the illustrations use both historical and fictional names and we have encoded them as such, distinguishing the character from the historical person. In the case of Claude, who is given no last name in the illustration caption, we have chosen to use the historical name as the key ID since it is a photograph of the actual person. Elsewhere in the narrative he is identified by his fictional name.

We have not encoded gender distinctions in this edition for this project has brought home to us the inadequacies of current encoding practices when it comes to capturing gender. (These difficulties and possible responses to them are discussed in detail in Pamela L. Caughie, Emily Datskou, and Rebecca Parker’s “Storm Clouds on the Horizon: Feminist Ontologies and the Problem of Gender,” Feminist Modernist Studies vol. 1, n. 3 (2018): 230-242. https://doi.org/10.1080/24692921.2018.1505819.) Moreover, for our project, using TEI P5 guidelines (https://tei-c.org/guidelines/p5/), gender ontologies are complicated by differences across languages. That Danish and German are gendered languages and English is not makes any effort to mark gender consistently especially complex. And concepts of gender change across time as well. In the modernist era, “gender” was beginning to be used synonymously with “sex” in English-language writings, whereas in German and Danish, “gender” was solely a linguistic concept, not an identitarian one. And that is consistent with past iterations of TEI guidelines still in use that categorize gender in terms of sex attribute values (e.g., M and F) and provide a “gen” element used specifically in relation to a linguistic representation of gendered words.

Transgender compounds this problem. Specifying precisely when the transition from Andreas to Lili begins and ends in this narrative is handled differently in the different editions. After the first operation, the Danish edition explicitly uses hende (her) for the patient while the German and English-language editions leave the gender indeterminate. The narrative specifies that already with her first surgery, an orchiectomy, Lili exhibits female characteristics, such as her “splendid soprano voice” (A1, p. 128) and feminine handwriting. But it is the second surgery, the implantation of an ovary, that is necessary to confirm that she is more than a castrated man, as Lili refers to the being who entered the Women’s Clinic (A1, p. 284). And yet Lili insists on a final surgery, a vaginoplasty, because she believed it would allow her to become a mother, accepting reproduction as the hallmark of a woman. No current encoding practices could possibly capture these nuances of gender attribution.

We also faced a challenge in how to note a character in chapter 9 who is described in terms of his disability: he is a “cripple” due to a hunchback. While this character is a newspaper boy and has red hair—other markers of identity we might have chosen for the key ID—we used the dated and problematic term “theCripple” for his ID because that is how he is designated in the narrative and it is precisely his disability that marks him as significant to the story.

TEI Headers: Specific information regarding the encoding of each text can be found in the teiHeader of the source XML and in the Document Metadata section of each text's HTML reading view. We have also recorded in each document a series of respStmts that indicate who on the project team is responsible for what in each text. The teiHeader also holds specific information regarding the original archive location of each text. While we have much of the information included for the launch, additional information will be added in the coming months.


Lili Elbe Digital Archive Encoding Guidelines