My primary goal for this assignment was to learn Voyant’s functionality, but I also wanted to understand how I might be able to use it in my current work. Over the past year I have supported faculty members at Hostos Community College who are building textbooks with Open Educational Resources (OER). Most of the faculty I have supported need assistance finding appropriate content or a satisfactory amount of content for their texts. They often get through 50-75% of the book on their own, and then need help identifying openly licensed content for gap areas or help identifying chapters that are sparse. I wondered if it would be helpful to faculty to create a corpus for the complete book and each chapter of their book, and then compare to their syllabus or course objectives. 

For the sake of this assignment I chose to use a textbook for EDU 105: Social Studies for Young Children, one on which I am currently working. I decided to look at how the text of the syllabus — which included course objectives and weekly topics — compared to the content of the course material the professor wrote. To do so, I created one corpus of content for the textbook — with each section a separate document — and one corpus for the syllabus.  For each I copied and pasted content from their respective websites (the books are built in LibGuides and syllabus is in Google Drive) into the “Add Texts” window on the home page. I found it very usable and straightforward. 

Here is a view of the textbook corpus with the default dashboard:

(https://voyant-tools.org/?corpus=883399c79d0ce541ce9dad5c5eb0fec4)

Here is a view of the syllabus corpus with the default dashboard:

(https://voyant-tools.org/?corpus=6c93088ad815d60fab77b9d63e02f659)

I primarily looked at the “cirrus” word cloud, term count, and links tools for my review, and asked the following questions:

• What are the most visually prominent words (based on term frequency in the content), and do they give me an idea of what the class covered?
  • Textbook: Overall I would say that it did, but I needed to use a combination of tools for the complete picture. While the “cirrus” world cloud gave me a quick visual on the most common words, I could not create two-word phrases, like social studies or early childhood. “Social” and “Studies” were separate, so it wasn’t obvious that the class was about social studies. However, when I looked at the terms tool, I could see “social” and “studies” stacked together. As the second and third most common words. I could also see the link between “social” and “studies” with the link tool. Early Childhood did require that I scroll down through the list to understand that the course was for teachers of younger children, but I understood enough of what the course was about. I would say that the links tool was the best reflection of the class content (see image below).
  • Syllabus: Regardless of the tool, the syllabus could have been appropriate for most courses. However, if this syllabus was compared to a syllabus for a course outside of the Education department, it is likely that it would be evident that it was for an education course.
    The Summary tool was actually a good complement to the visual tools, because I was able to see the most frequent words (social, studies, students) and distinctive words per document (chapter). 

• Did the scope of terms in the textbook content match the syllabus?
  • The textbook terms covered all of the identified topics and goals in the syllabus, and was more specific. I found the Terms tool to be most helpful for this comparison, but once I realized that Voyant wasn’t great with the syllabus I actually compared the visualization for the textbook to the syllabus word document outside of Voyant.
  • I hoped that comparing these two would show gaps in the content, but none of the tools I used helped with this evaluation.

• Are any chapters more content-rich than others?
  • The Reader and Summary tools was superficially helpful for this. For example, the Reader tool showed a colored bar per document, so I could see which documents (in my text these are chapters) had more words. The Summary tool counted the number of words. 
  • This would not work for pages with a number of links to websites with additional reading, nor would it work for other media, like videos and images, which this textbook included. 

I intent to continue exploring the following:

• How can I exclude what I consider to be false results? By this I mean common terms in all documents but distract from what I am reviewing? This is beyond stop words.
• Can I hide and reveal documents to compare subsets, or do I need to create a different corpus for each grouping?
• Would the visualizations change if I flattened multiple documents into one as opposed to multiple documents?
• Is it possible to see the syllabus and the textbook documents in parallel in the same screen?

Memoir and personal reflections are the texts that interest me most, so I chose to map a memoir I was reading called Haldol and Hyacinths: A Bipolar Life by Melody Moezzi. Melody Moezzi has many identities. She is a manic-depressive, Iranian-American Muslim activist, attorney, writer, and award-winning author. In her memoir she chronicles her experience of clinical and cultural bipolarity, and I wondered how this duality would project on geographical terrain. I wondered about her relationship with each of these identities, and if there was a way that mapping might display patterns in her life that were otherwise not evident in linear text. 

I started the project using ArcGIS StoryMaps, because I liked the idea of making this a scrapbook or personal journal of sorts. I used a map done in watercolor to give it a hand-drawn aesthetic, and to extend the personal. I also liked the idea of presenting different map snippets with sections of her writing, rather than my own interpretation of her writing:

However, this format quickly became problematic, because so much of her experience was of relationships between and juxtapositions: locations in the United States and Iran; locations where she was mentally well, and where she was manic; who she was when she was living American and Iranian culture. Small snippets like that displayed above were too small to show these relationships, and standing alone the bipolar quality of her identity was lost.

I then switched to use the ArcGIS app alone. This gave me some more flexibility with space and relationships, but the first thing that was obvious about mapping a personal narrative that was not my own was that a geographical map created boundaries where they may not naturally exist for the writer. For Moezzi, someone who experiences clinical and cultural bipolarity, boundaries are blurred, and shift regularly. Boundaries and names of boundary lines shift at given points in history, i.e., during the Iranian Revolution, during which time she became a refugee. Mentally, boundaries can disappear altogether in a given day, i.e.,  when she’s at once sleeping in a room in Ohio in college while also being terrorized by a glowing, green spider far away from the bed she’s physically in. 

Figuring out locations to add to the map was also more complex with a personal narrative, and in fact, she opens the book talking about whether or not the specifics of her story should be trusted at all, because of the nature of her mind and her mind on medication. I also knew that names and places were changed for privacy. Some locations were known. For example, she visits family in Tehran. She was born in Chicago, grew up in Ohio, and went to law school at Emory. However, specific streets or locations weren’t named in most cases so where I put markers on the map aren’t precise. Most significantly, a major site of her life and story takes place at Stillbrook,  a mental institution, but the name has been changed. Without a name and location, it would have been left off altogether, and her narrative in visual form would render silent a major part of her identity and activism. Who is she without her clinical bipolarity? It is not for me to rewrite her story, which I would in effect be doing. Therefore, I traced other places in the narrative to create a location in close proximity. For example, she mentioned she was at the Emory infirmary before being institutionalized, so placed a marker on a mental hospital near Emory. It is not correct, and is in fact a lie. But which inaccuracy was less harmful? I decided that it was more important to present her whole identity rather than go with precision.

Deciding on symbols for the map was an area where I pushed the boundaries of a traditional basemap available on ArcGIS, and also found it lacking. For example, I reflected her memories and experiences as an Iranian and Muslim woman with the “General Infrastructure” mosque symbol. Without context, the mosque symbol is definitely misleading! I struggled most with symbols for her mental health. I liked the fuzziness of the “Firefly” symbol, but it gave the impression that anytime she grappled with her mental health she was erratic and fuzzy. That was not the case. In fact, the various ways that her symptoms manifested almost required different symbols from episode to episode or type of symptom, i.e., insomnia, hypermania, etc. As such, I chose regular stickpins, which I wasn’t pleased with. I tried using a color coding of yellow, orange, and red to reflect severity of symptoms, but that was another point where I was making decisions about her life in visual form. Would institutionalization be red, or should moments when she was raging and suicidal prior to institutionalization be depicted as worse? Again, this was another boundaryless area, and adding color coding was something I couldn’t get comfortable with. I ran into the same problem with symbol size, which I attempted to use to signify which moments in her life had the greatest impact. Symbols introduced bias at every turn. I found it harder and harder to add any moments to the map with significance that didn’t distort her story.

As my map took shape — or rather didn’t take shape — it became clear that it was impossible to use this geographical basemap, with its preset location markers (most of which were not important to her story), to reflect the complexity of her humanity. My map would not represent the terrain of her emotional life — anguish, confusion, longing, heartbreak, and unpredictability. It was also impossible to map the silences in her story, i.e., when she was well and thriving, because a) that wasn’t the scope of her book, and b) how would I reflect that? Would all empty space be considered areas where she was well? 

Without the ability to project moments of wellness, joy, and her incredible humor onto the map, I fell into the trap of perpetuating the common story about people with disability or chronic conditions: they are their illness. I was not okay with this, and it felt irresponsible. I felt protective of her. She isn’t a fragile person, and despite several stays in psychiatric hospitals, bombarded with tranquilizers and anti-psychotics, she refused to be shamed into secrecy. Refusing to be ashamed or silenced, Moezzi became an outspoken advocate, determined to fight the stigma surrounding mental illness and reclaim her life along the way. Perhaps if mental illness was not so stigmatized, continuing with the map wouldn’t feel so off limits. After multiple attempts to add her writing to points on the map, I came to the conclusion that not every tool works for every project, and to force a project into a tool does it much harm. We must remember to bring care to our work. 

Here is my map, as it turned out:

https://arcg.is/SrTKW

(I apologize in advanced for what turned out to be a very long reflection!)

The theme that repeated itself among our resources this week was the relationship between DH and story. I would further unpack story to include truth and agenda. From an information literacy perspective I often think about the following questions (not in order of importance):

“What agenda is influencing story and truth?”
“Whose story is considered the truth?” 
“How do we recognize the truth?”

On the most universal level stories and truth are curated, and there is great power in curation. Throughout history we’ve witnessed colonizing powers curate the stories we’ve come to believe as true, without mention of agenda. For example, David Scott refers to the question of agenda in On the Question of Caribbean Studies when he asks, “What is the point (political, conceptual, disciplinary, moral) of mobilizing this particular image, rather than some other, of the Caribbean in these particular discourses?” He goes on to reflect, “The ‘dependent’ character of the existing literature ‘reflects the fact that hitherto most of the researches in this area have been conducted by visiting social scientists from the United States or Britain, and have been guided by theories and themes of interest developed in studies of societies and cultures outside the British Caribbean.’” 

Even with the best intentions scholars and librarians categorize and archive materials with their own language, which may curate the community out of its own story. In Introduction: The Digital Black Atlantic Kelly Baker Josephs and Roopika Risam include two essays about practices of neglect that ultimately shape the story of the Caribbean. In “Dividing the World for Library Collection Development” the authors “wrestle with the need to free Caribbean sources from sedimented library practices of categorization while keeping them detectable within the new worlds of access offered by digital technologies.” In “A Paper Archive Sojourner’s Notes to Black Digital Humanities,” the author, Nadine Chambers, “queries the digital sedimentation of human error and neglect in archives of the black Atlantic.” Even unintentionally we sow the seeds of incorrect stories, and those stories have a lasting impact on the world outside of the academy.

The Digital Humanities grant us the opportunity to use data and narrative to question the stories we take for granted, or know little about, and to create new stories in multimodal formats. This is a great plus for access. Story has the power to mobilize, and I am excited that the “definition” and scope of Digital Humanities in 2019 includes resistance. In  their chapter “A DH that Matters” in Debates in the Digital Humanities 2019, Matt and Lauren state:

“We are convinced that digital humanists can contribute significantly to a larger technically and historically informed resistance. By enabling communication across communities and networks, by creating platforms that amplify the voices of those most in need of being heard, by pursuing projects that perform the work of recovery and resistance, and by undertaking research that intervenes in the areas of data surveillance and privacy.”

Torn Apart / Separados is a powerful, participatory DH project that can be used for resistance and mobilization. It gives urgency to the stories of the humanitarian crisis of asylum seekers to a much wider audience than academia. It allows the public to ask new questions about curated information, and to tell comprehensive, true stories that we definitely do not get in the news media if consumed on an article by article basis. Even though this project also has a social and political agenda, I really appreciate the efforts the team has taken to make their work transparent, open, and arguably inclusive. The project team makes a point to note that the visualizations and data are just one part of a larger story, and the site as a whole contextualizes reflections on the data from the team itself, as well as historians, activists, artists, and writers who provide their own readings of the visual stories. That said, even this team, like the authors included in Introduction: The Digital Black Atlantic acknowledge that with best intention data is “imprecise, impure, and as much a tool for incarceration and control as it is revealing the truth.” They make it possible to investigate their work, because the site includes a fantastic bibliography with links to a full Zotero resource, an open syllabus on immigration, and nearly all of their data in an open repository in GitHub. They include data sources, and their visualization tools in the credits portion of the site. 

Not only is the data and visualization component transparent, but I also appreciate all of the background information on the contributors. For example, author bios link to personal websites and Twitter accounts, as well as other publications from said authors to see the landscape of their portfolio / with whom they affiliate. For example, I love that Gaiutra Bahadur writes about memory maps and story maps to give a humanistic story about immigration in addition to geodata.

I also thought about social aspect of DH, and decided to look at the hashtags about this project on Twitter to see how the public is engaging with it. A quick review illustrated that the story about the project has been shared by WIRED, DH contributors across the globe, NPR Latino USA, Library Journal, and researchers in many disciplines. One wonderful use of Twitter from a contributor, Manan Ahmed links a Trump tweet including a non-truth to a section of #TornApart v2. 

There is so much potential for this project and others to influence open pedagogy, and the stories we learn in classrooms. I intend to share this with the OER community, which is always looking for new and relevant teaching materials. I wonder how else this has been circulated in the education community?

On a final note, this project put action in perspective for me as an individual. I often feel powerless around issues like immigration, as I am not near the border. But seeing V2 of this project illustrated that there is a very local way to engage. In fact, my local representative received $19M ICE dollars since 2014. This will definitely be a visual story that I will use when she’s up for reelection!