Øyvind — That’s a great explanation of how and why we might want to include RDF ontologies within TEI. You speak of doing this in the TEI header of the file holding the encoding of, say, Megan’s novel. Such coding can also work in a separate “placeography” file storing meta-information about each place. Wherever we’d choose to store the co-referenced information about each place, we’ll want some kind of system to connect each entry on the list with the references to it in the novel, working with xml:ids. 

It would be useful for us to see more and more examples of “placeographies" that do a little bit more than store geocoordinates and alternative names. The syntax of RDF seems like it would be a great way to go, to help develop a systematic vocabulary for relationships among places for place-relationships in a poem or a novel that we might be able to use on either project. We will almost certainly come up with different sets of relationships, but perhaps there are some standard locational vocabularies we might try? 

It may not be necessary to do that: My project on networking place references doesn’t need to “map” to Megan’s. But since one critique we sometimes hear of the TEI is how little interoperability is achieved between distinct projects, despite the promise of interchangeable code, it’s something we might want to investigate. I’m not aware of a set RDF vocabulary that defines spatial relationships in the way that we’re discussing here (is there one?)—but maybe we simply produce these and compare with one another’s projects and see how they look! Ultimately, maybe what’s *interoperable* here isn’t the literal vocabulary we define, but simply the method of applying the same forms of structured syntax.

That may seem kind of obvious, but the problem (and opportunity) of interoperability has been on some of our minds lately, and your post spurred some reflection on what matters most in “cross-walking”.

Elisa Beshero-Bondar, PhD
Director, Center for the Digital Text | Associate Professor of English
University of Pittsburgh at Greensburg | Humanities Division
150 Finoli Drive
Greensburg, PA  15601  USA
E-mail: [log in to unmask]
Development site:

On Mar 24, 2017, at 12:26 PM, Øyvind Eide <[log in to unmask]> wrote:

Thanks for an interesting discussion!

Textual features such as place names and other strings referring to places are fairly easy to encode as specific strings, using ordinary inline TEI markup. When more complex spatial relationships are to be recorded it is a question how close to the text the recording have to be. Surely, it is good to have links back to the evidence from the system used to study the material, and text encoding is good to establish such links. However, I have found it useful to express relationships between place references in a different way.

This is in line with how co-reference information / place thesauri are meant to be encoded in TEI, using a place element in the header connecting through XML links all the place references in the text referring to that place.

So, even if statements such as “the place referred to by string A is east of the place referred to by string B” are based on  textual evidence, the information can still be stored disconnected from the specific textual expression establishing the relationship. A link to the paragraph (or other unit) where the statement is made can often be nice, but anything more detailed than that is generally hard to do and in my experience it is often not needed.

Thus, a combination of textual features referring to places in a broad sense (including rooms, shelves, and what have you) encoded in TEI and then relationship structures expressed partly in another formalism such as RDF can be quite useful.

All the best,


On 16 Mar 2017, at 02:48, Elisa Beshero-Bondar <[log in to unmask]> wrote:

Megan—Thank you! I’m glad my Thalaba project is helpful for you—I learned my way around Cytoscape by working on that project! I think your project on Tiempo de Silencio is fascinating: you will have many different kinds of locations and relationships among places to think about.

About the various ways to encode place in the TEI, you’ve likely been looking at Chapter 13 of the Guidelines, and the sections about encoding various kinds of geographic and geopolitical locations: . But you will want to customize your approach, and I don’t think the TEI is establishing a hierarchy of place so much as offering options for encoding. (I’m not aware, for example, of an element specifically for encoding a room.) But you can adapt the TEI to your project and find a way to encode various kinds of locations. This is an area where some pioneering efforts might be made, I think. Most of our orientation and vocabulary regarding places in digital projects is associated with mapping, but when we work with space and place in literary contexts, we frequently move into zones of mythical and abstract space, or of fictional landscapes.

For Thalaba, and other projects I’m working on, I try to keep the markup simple—<placeName> and <rs type=“place”> vs. <rs type=“metaplace”> are customizations I designed for the project (with a few other attributes in play). But I save the more complex typology of places for a separate “placeography” file, in which I store a catalogue of each distinct place, associated with information about what kind of place it is. You can dereference the markup with a prosopography file, in which each distinct place is given a distinct id, and work in the elements that help to classify it. My approach is basically to harvest widely over a pile of documents, and try to do all the classification work in one place elsewhere. Now, that’s easy for me to say, working on multiple projects at once—it’s a working method. I’ll say it can be difficult to work out how complicated a schema you want to devise, and you may just want to spend a while coding some chapters experimentally and doing a lot of lookup in the TEI Guidelines to work out your strategy.

I guess what I’m saying is that the TEI isn’t really providing a recipe to follow so much as food for thought and guidance, and definitions of the appropriate ways (plural) in which tags can be used. And I also think that encoding places as they’re referenced in literature can be especially and marvellously complex, and we stand to learn a great deal from an adventurous effort to encode them—both within the TEI and in literary scholarship!

Hope this helps a little...
Elisa Beshero-Bondar, PhD
Director, Center for the Digital Text | Associate Professor of English
University of Pittsburgh at Greensburg | Humanities Division
150 Finoli Drive
Greensburg, PA  15601  USA
E-mail: [log in to unmask]
Development site:

On Mar 14, 2017, at 11:17 AM, MEGAN MARY CYTRON DWYER <[log in to unmask]> wrote:

Hi Elisa--
Thanks so much for the link to your project! Just getting a peek at the way you are tagging--and then seeing the beautiful networks you are generating with Cytoscape--is quite helpful, not to mention inspirational.

To answer your question, I am working on a literary cartography of Madrid, starting with the novel Tiempo de Silencio by Luis Martín-Santos, which--in addition to its other charms--functions as a radical reworking of Madrid's literary (and artistic) spaces via a complex web of overlapping geographic and metaliterary references and allusions, along with a strong dose of realism regarding demographic shifts in post-war Madrid. On a deeper level, I'm interested in the mechanisms that authors use to create such elaborate mental mapping and geographic awareness in the mind of the reader.

The philological approach to literature that I was taught here in Spain is very much like markup to begin with and what you wrote about Cytoscape resonates with me: "A great reason to try plotting a network from XML structure might just be because you have a hunch that you might learn something interesting based on relationships you’re beginning to see in your markup."

Syd, Lou, and Thomas--
Thanks for the practical advice on tackling places and the notion of naming. I'm looking for a way to balance detailed tagging of what is actually there in the text, while also capturing what is stirred up in the mind of the reader and you have given me plenty of food for thought. This particular text plays with ellipsis to such an extent that naming is quite relevant, even/especially in its absence. There's no doubt that I need to capture all of that information.

Question: is there any established hierarchy of places from macro to micro (e.g. planet, continent, country, region, city, street, building, room)?


2017-03-14 15:32 GMT+01:00 Thomas Elliott <[log in to unmask]>:
To agree with Lou by disagreeing on one point:

I wouldn't describe the distinction that TEI makes between "place name" and "place" (and between personal name and person) as "old-fashioned". Rather, it's an essential, scholarly/analytical distinction, and one that makes TEI exceptionally well-suited for working with textual sources that mention -- and (in combination) structured information about -- past places, imagined places, contingent places, etc. "Place" is a cognitive construct. "Place name" (in text encoding) is a sequence of characters in a writing system that are understood to refer to a place. There can be other textual references to a place: descriptions, euphemisms, puns, allusions, ethnica, etc.


Tom Elliott, Ph.D.
Associate Director for Digital Programs and Senior Research Scholar
Institute for the Study of the Ancient World (NYU)

On Tue, Mar 14, 2017 at 6:21 AM, Lou Burnard <[log in to unmask]> wrote:
Interesting set of questions!

I'll just address this one for now:

"the physical places I want to tag ... manifest in the text in a variety of ways"

The TEI is a bit old fashioned perhaps in making a strong ontological distinction between tagging about a named entity (a place) and its manifestation (a placename). That's because it is, after all, the TEXT encoding initiative, not the THING encoding initiative. So the name of a place as it appears in a document is an intimate part of the document, which you might well want to tag as signalling something or other. Whereas all the things you might want to say about the thing being signalled are probably not present in the document at all, or not in the same place, or not in the same way. So we have both <placeName> for the former and <place> for the latter. You might lump together all the names used for a place as child elements of a <place> element, along with some indication of the way in which the names are applied, and also along with geographical information relating to the place as a real world object. That isn't a part of your text in the same way as the references to it and you would typically not put it inside your transcription but inside the metadata for describing it. You might simply flag all the references as they occur within the document and indicate that each one is "about" the same thing by means of a @ref or @key attribute on the reference. You might do both!

On 14/03/17 01:41, Megan Cytron wrote:
First off, apologies if this isn't the correct forum for this question.

I'm just embarking on a doctoral dissertation in which I hope to do a spatial analysis of a rather complex 20th-century novel. I'm a programmer (in another life) and have quite a bit of experience with GIS and XML and was planning to use it to tag the text I'm analyzing so I can generate statistics, charts, maps and other data-driven visualizations of the novel's inner workings.

After (re)discovering TEI (I worked with it way back in the 90s in a completely different way), it seems that I could adapt it to my use case, but I'm a bit uncertain of how best to capture all of the variables I want to look at. I have found many examples of TEI encoded texts, but none that go too deep into this particular brand of literary criticism. For example, these are some of the variables that interest me with a few examples:

--mappable places  (physical locations in the plot, literary overlapping, allusions, trajectories, overlapping with author's biography)
--changes in position (forward/backward/turns, uphill/downhill, stops,  fast/slow, inside/outside)
--narration and voice (1st person, 3rd person, subjective/objective/omniscient, stream of consciousness)
--semantic fields (colors, sickness, science, feelings, physical states, atmospheric conditions)
--rhetorical devices (anaphora, metonymy, paranomasia, juxtaposition)
--intertextuality (allusion, quotation, parody)
--time alterations (analepsis/prolepsis/flash sideways/foreshadowing, anachronism)
--verbs (type, aspect, mode, tense)

I'm fully prepared to invent my own tagging scheme, if it makes sense to do so, but if something already exists and has been implemented with success, I'd much prefer to replicate what I can and contribute another example (or many!).

Is anyone else using TEI in this way? Are there any resources (aside from the obvious TEI Guidelines) and especially examples that would be useful for someone just starting out? Or groups where folks are talking about projects like this and sharing experiences?

In particular, I'm finding the "place" and "placename" tags to be somewhat confusing for what I want to do, because the physical places I want to tag (with lat/long coordinates and some fencing) manifest in the text in a variety of ways (proper names, metonymy, ellipsis, ironic allusions, via demonyms applied to people or objects, etc.).

Any help or general "pointing in the right direction" that you may be able to offer would be greatly appreciated!