Browsed by
Tag: XML

A New(ish) Approach to Markup in the Undergraduate Classroom

A New(ish) Approach to Markup in the Undergraduate Classroom

By Kevin G. Smith, Ph.D. Candidate in English, Northeastern University

Note: Kevin G. Smith is a pedagogical development consultant for the WWP. His dissertation research is partially supported by a grant from the NULab for Texts, Maps, and Networks.

A few summers ago, I spent my days working in Northeastern’s Digital Scholarship Commons. As is common in that space, there were nearly daily meetings of different teams of faculty, library personnel, and graduate students working on digital projects. One of these projects was The Early Caribbean Digital Archive (ECDA). During that summer the ECDA project team was working on customizing a TEI schema to encode their texts in ways that were more in line with their decolonial archival goals. As I procrastinated on my own work, I was overhearing these amazing conversations that the EDCA team was having about the meanings and applications of certain aspects of their TEI customization. How should they tag an embedded or mediated slave narrative, for example? What to do about unnamed slaves? And how might they handle commodities? What are the ethical ramifications of encoding a slave as a commodity (or not)?

As I sat, listening to these conversations, I began to realize that it was precisely because they were encoding the texts in TEI that these conversations were happening. The act of encoding literally inscribes texts with interpretation, forcing the project team to discuss just what kinds of interpretive judgments they wanted to make. And they were important conversations: about how we represent our objects of inquiry in the humanities, about the ethics of data representation. (By no means am I the first to realize this. For a compelling example, see Julia Flanders: “The Productive Unease of 21st-century Digital Scholarship.”)

The point is that I was struck by these conversations. And I began to think about how the tension of formalization, this “productive unease,” as Flanders terms it, might be leveraged in writing classrooms. Could I somehow use the TEI to intervene in students’ writing processes, to foster these kinds of conversations about their own writing? What would that even look like?

Two years later, in the summer of 2016, I taught my first markup-based writing course at Northeastern. In the intervening years my approach shifted from using the TEI to designing a built-from-scratch XML schema for each course. Thus far, I’ve taught two courses using this method (Advanced Writing for the Technical Professions in the summer of 2016 and First-year Writing in the fall of 2016). In addition to writing their assignments in XML (using Oxygen), students in these courses engage in a semester-long, collaborative writing project: the design and implementation of an XML schema that structurally and rhetorically models a range of genres of writing.

This approach—using XML to produce texts—represents a shift from the mimetic roots of XML and its primary use in humanities research, the TEI. In the rest of this post, I want to briefly discuss this shift and its implications for the study of markup.

Teaching with Markup

 There are many wonderful examples of using the TEI and XML in classrooms. Kate Singer’s use of TEI for developing poetic vocabularies in an undergraduate class comes to mind, as does Trey Conatser’s use of XML in a first-year writing course at the Ohio State. Though, at first blush, these two markup classrooms may appear very different—one being in an upper-level literature course and the other a first-year writing course—the perceived pedagogical benefits of using markup are similar. Both pedagogues seek to foster close attention to the object of study—a poem or the student’s own writing—through what is essentially a process of annotation.

Where my approach to markup differs from these (and most traditional) classroom uses is in the thoroughly bottom-up, data driven approach to schema design (Piez, 2001). Students begin with a (basically) bare schema and—iteratively and deliberately over the course of an entire semester—design and revise the schema for a range of writing tasks using document analysis and modeling, qualitative writing research methods, and their own experiences of authorship. The result is a shift from annotation to production, from product to process.

An example may be illuminative here. A group of students decide they would like to design a schema for movie reviews. They begin the process by researching the genre—gathering examples, examining related genres, tracing the circulation and uptake of the genre, interviewing experienced writers and readers of the genre, and so on. Based on this research, the group identifies the salient structural, rhetorical, and content-based components of the genre—a movie review includes a series of paragraphs, for example; the first of these paragraphs must, according to the students, include a component called “opinion,” which has a specific definition and different types. They name these components and write a prose pseudo-schema, including documentation, attributes, dependencies, and rules for the components. The pseudo-schema is translated into an XML schema using Relax NG (by me).

An element list from an in-class schema design session with students in the First-year Writing course of 2016.

Once the schema is drafted, each student writes an individual XML document, their own example of a movie review that responds to a unique rhetorical situation. Based on this experience, the group reconvenes to revise their schema. They might, for example, decide that the <opinion> element should be optional in the first paragraph, or decide that an additional attribute value should be added to the @type attribute, or choose to adjust the definition of the element itself. Once schema revision is complete, students revise their XML documents. And on it goes.

An example of XML markup designed for the course.

What I hope the above example illuminates is the thoroughly process-oriented approach to markup adopted in these classes. The schema is not static. It is a living document that affects and is affected by student’s experiences of composing, among other things. Neither are the student-authored XML documents static. They are repeatedly invalidated by revisions to the schema. They are subject to feedback from classmates and instructor. They must be continually revised. From a digital humanities perspective, this application of markup may seem alien. In fact, in some ways, it doesn’t even matter what the schema ends up looking like (though it can be fascinating). The object of using markup in this way is not to produce the perfect model of a genre. In fact, an understanding of genres as social actions, rather than a set of ossified textual features is central to the theoretical framework of the course. This understanding resists the idea that genres can be accurately modeled. The point of using markup is to foster productive conversations about writing, to interrupt the normal thinking and writing processes of students in productive ways. This brings us back to the conversations I overheard in the summer of 2014, eavesdropping on the ECDA when I was supposed to be writing.

An example of a markup output document for display. The XML is transformed to HTML with custom XSLT and highlighted according to XML tags.

But this approach raises new questions. How do I know if this approach is productive in the ways that I hope? What kinds of conversations are students having in these classes? How does markup function rhetorically for students when used for authorship? Does writing in XML and designing schemas for authoring contribute to students’ understanding of their writing and reading processes? Do reading and writing practices in the markup classroom transfer to other contexts? These questions just so happen to be the basis for my dissertation research, which takes as its objects of inquiry the two markup-based writing courses.

Studying (Authorial) Markup

The questions posed above present unique methodological concerns for the study of markup. A shift from product to process raises practical questions concerning how we access students’ experience using markup in this way. How can I make claims about the rhetorical and expressive capacities of authorial markup? How can I understand the role of the schema, the markup, and the platform(s) in students’ writing, reading, and thinking processes? In short, how do I study this?

Here, a slight shifting in thinking—from the digital humanities to writing studies—is helpful. While the pedagogical approach may be unconventional, my research questions are typical of writing studies research. Methods for studying student writing and experience in classroom settings are well established in the field. Although qualitative approaches to the study of markup are not typical in the digital humanities, the research questions for this project, based, as they are, on student experience, reflection, writing, and perception, necessitate the adaptation of innovative methods. To this end, I’ve employed a teacher research methodology—a systematic approach to data collection that honors the inside perspectives of teachers and students—that adapts qualitative research methods culled from ethnography, education, and writing studies research. Data for the study was gathered from direct participant observation, reflective journaling, semi-structured and directed qualitative interviews (three interviews each with nine case study students), and the collection of student writing (normal prose and XML, including version control logs for all XML files).

At this point, data collection has ended and the project is shifting to the data analysis phase. It is too soon to report results, however, early indications from student interviews point to some promising findings around student reflection and transfer, the multi-directional mediation of the schema, and students’ use of markup as a tool for generic invention and change. Here, it may be enough to assert that qualitative approaches to studying markup-based undergraduate courses may be fruitful. Indeed, digital humanities courses in general may benefit from adopting qualitative methodologies, like teacher research, to self-assess and to advocate for curricular change and institutional support.

The assignment discussed above is collected with the pilot set of teaching materials from the WWP’s pedagogical development consultants and is available here.

A (semi-)Serious Proposal to the Linguists

A (semi-)Serious Proposal to the Linguists

God, Vertue, Ladies, and Souls

A few days ago, I came across this really interesting Language Log post, which talks about capitalization in one of our Women Writers Online texts—Mary Astell’s A Serious Proposal to the Ladies (1694). In the post, Mark Liberman asks the question: “Why did authors from Astell’s time distribute initial capital letters in the apparently erratic way that they did?” Liberman looks at sentences like this one, which describes the purpose of Astell’s proposal:

It’s aim is to fix that Beauty, to make it laſting and permanent, which Nature with all the helps of Art, cannot ſecure: And to place it out of the reach of Sickneſs and Old Age, by transferring it from a corruptible Body to an immortal Mind.

Since this is a WWO text, I decided to try a bit of experimentation and see what I might be able to uncover using not just the text itself, but also the markup. For just a bit of background, the texts in WWO are encoded according to the guidelines of the Text Encoding Initiative. You do need a subscription to access the collection, but we are always happy to offer free trials, so if you don’t have institutional access or an individual subscription and are interested in reading the texts in WWO, you can find instructions for how to set up a month-long trial here. If you’re curious about the details of our markup, those are covered in our internal documentation.

The first thing I did was enlist some help from Syd Bauman and Ashley Clark, our XML developers. Syd generated a list of all the capitalized words in Astell’s Proposal, along with their immediate ancestry (i.e., the local elements around each word). We found 2,491 capitalized words in total. Reviewing the elements in this list, I could see that it was likely many words were capitalized for reasons reflected in their markup. For example, there were proper nouns (tagged with <name>, <persName>, and <placeName>), titles of other texts (tagged with <title>), and the document’s own headings (tagged with <head>). There were also some words that were simply appearing at the starts of sentences.

So, I asked Ashley and Syd to help me come up with a new list of the capitalized words in Proposal, excluding those in proper nouns, titles, headings, and at the start of sentences. That list is here (original spellings preserved). The top results are: “God” with 31 instances; “Vertue” with 31; “Ladies” with 24; and “Souls” with 21 (in case you’re wondering, the WWP does not encode “God” with <persName>; see here for more details). The rest of the top fifteen—Women, World, Good, Nature, Piety, Religious, Religion, Soul, Beauty, Education, Glory—are all the sorts of word I’d expect to see capitalized in a seventeenth-century text.

Beauty and Death

Having looked at the capitalized words in an individual file, I thought it would be worth investigating all of the occurrences of those words across our corpus. So, since “Beauty” was a commonly capitalized word for Astell (in addition to being relatively short and without too many potential spelling variations), I started with that.

I first wanted to determine if I should be concerned with weeding out the capitalized cases of “Beauty” in sentence-initial positions. A bit of exploration showed me that there weren’t many such cases, and most of these came from texts that also had instances of “beauty” capitalized in the middle of sentences. I found only a handful of clear cases where “beauty” was being capitalized just because it was at the start of the sentence, so I decided not to worry about sentence position. I did find several texts that capitalized “beauty” only some of the time—in a few cases, this seemed to indicate a distinction between personified beauty and a more general usage (e.g., contrast “Soft Beauty’s timid smile serene” with “youth and the bloom of beauty,” both from the 1824 Poetical Works of the Late Mrs. Mary Robinson); in other cases the pattern was less clear. These instances, presumably, would be one place I might start if I were investigating this phenomenon in earnest.

So, armed with the power of XPath, I set out to investigate the beauties of WWO. Here’s what I found. There are:
1577 total instances of Capital-B “Beauty” and
1863 cases of lowercase-b “beauty”
Looking across the whole corpus, that’s about 46% capitalized instances.

I repeated the search with “beautie” (to catch both “beauties” and the alternate spelling of “beautie”) and while there were fewer hits, the results were similar in terms of percent capitalized:
438 Beautie; 580 beautie (43% capitalized)

For “beautiful” I saw a different distribution:
71 Beautiful; 1619 beautiful (4% capitalized)

Since I suspected that this kind of capitalization would be more common in our earlier set of texts, I decided to narrow down the results. That just meant adding a bit of XPath before my search to look only in texts with publication dates before 1701 (198 out of 388 texts total).

Here’s what I found:
872 Beauty; 415 beauty (68% capitalized)
270 Beautie; 235 beautie (53% capitalized)
36 Beautiful; 212 beautiful (16% capitalized)

For this term at least (and with all appropriate acknowledgement of the highly rudimentary nature of this search), there does seem to be a bit more capitalization in the earlier half of the collection. Next, I wanted to see what else I could do with our markup. In my review of the tags we used for capitalized words in Astell’s Proposal, I had noticed that there were quite a few occurrences of <mcr>; this is a WWP-created element for a “meaningful change in rendition.” We use it where there are changes in rendition (such as between upright and italicized text) that are neither a printer’s error nor a merely decorative shift and that we can’t encode with more specific elements (such as <emph>, <name>, &c.). It’s essentially an element that says: “we think something semantically significant is happening with rendition here, but we’re not able to say exactly what.” Liberman alluded to this sort of thing when he wrote: “[And never mind, for now, Astell’s italicization choices…]”

Thinking that there might be interesting links between capitalization and these meaningful-but-unspecified changes in rendition, I tried my “beauty” search again, but restricted my results to text inside of <mcr>.

Here’s what I found, first looking across the corpus as a whole:
102 Beauty; 16 beauty (86% capitalized)

And then just the pre-1701 texts:
83 Beauty; 5 beauty (94% capitalized)

Admittedly, the corpus is small enough that narrowing down this far means you have fairly few results. (I also tried “beautie” and “beautiful,” but there really weren’t that many once I narrowed to the contents of <mcr>; for what it’s worth, 35 out of 37 instances of “beautie” in <mcr> are capitalized.) Still, there does seem to be something potentially interesting here. Most of the time, the rendition doesn’t change with capitalization (there are, after all, 1475 instances of “Beauty” in the collection that are not in <mcr>), but when the rendition does change, there is a higher percentage of capitalization. I decided to try another keyword and see what came up. I went with “death” this time, using the same criteria that it’s short, fairly common in the corpus, and without many spelling variations (there is “deathe,” which had 5 capitalized and 138 lowercase instances overall, none in <mcr>, all from texts published before 1701). Here’s what I found:

Corpus-wide
2578 Death; 4759 death (35% capitalized)
239 Dead; 2381 dead (9% capitalized)

Pre-1701
1226 Death; 2115 death (37% capitalized)
110 Dead; 1313 dead (8% capitalized)

Contents of <mcr>
Corpus-wide
251 Death; 54 death (82% capitalized)
Pre-1701
218 Death; 34 death (87% capitalized)

These are just two specific keywords, of course; if I were pursuing this seriously, I’d want to refine the search itself and try quite a few more terms as well as other XPath variations: looking at headings and titles, checking for items in lists, perhaps comparing verse and prose, and so on.

“Friendship Cheese”

Finally, I decided to take a look at the contents of <mcr> itself, using an XQuery that Ashley Clark wrote for the WWP (affectionately nicknamed “The Counting Robot” and available here). I normalized punctuation, long s (ſ) characters, and whitespace, but preserved capitalization. I got 21,741 different strings inside of <mcr>; of those, 16,832 were unique. Many of the unique cases are not single words or short phrases, but entire sentences or clauses where the renditional shifts cannot be attributed to emphasis or quotation. The top term on the list was “God,” with 1237 results; rounding out the top-five for the corpus are: Lord, I, Love, and Author.

Of the 127 cases with 30 or more hits, all but ten are capitalized—the exceptions are: “life,” “death,” “lying,” “they,” “she,” “love,” “one,” “her,” “he,” and “royal paper.” (This last item serves as a small caveat regarding the size of our corpus: all 204 instances of “royal paper” appear in a single text, Mary Jones’s 1750 Miscellanies in Prose and Verse.) Nevertheless, I do think that these exploratory results show that there is a great deal of potential for more serious research into these features using the WWO corpus—and if anyone is interested in a project along these lines, I’d be delighted to help set that up. In fact, this is my semi-serious proposal to anyone in the research community (linguists or otherwise) who might want to take this kind of work up.

One of my favorite things about this sort of exploration is that it brings me into contact with our texts in unpredictable ways, usually emphasizing how interesting and genuinely fun our corpus is. This was no exception and I’ll end here with my personal Top Ten results from the contents of <mcr>:

  • Wretched productions! inspired by hunger and dictated by stupidity and a disposition to lying! &c &c
  • As Irish ladies pass in jaunting cars
  • Confounded Harlot!
  • Effemenate Cat
  • For Gad Madam I don’t love being baulk’d thus
  • Friendship Cheese
  • Great Cuttle’s gland
  • Hedges of the Eyebrows
  • His lisping children hail their sire’s return!
  • Julius Cesar when he was beheaded by Oliver Cromwell