 Thanks so much for coming today. I'm so excited about the size of crowd we had, especially this time of the semester, when it seems like everyone is crazy busy. So it's my pleasure today to introduce Brian Croxel, who is the Assistant Research Professor of Digital Humanities at BYU's Office of Digital Humanities. Prior to joining BYU's faculty this fall, Brian was the Digital Humanities Librarian at Brown University's Center for Digital Scholarship and the Digital Humanities Strategist at Emory's Center for Digital Scholarship from 2012 to 2015. Today, Brian's topic is speaking in code, understanding, and misunderstanding the digital humanities. And please welcome me and, sorry, please join me in welcoming Brian Croxel. Thank you for being here. I feel a little discombobulated coming in. You're all here. It's sort of like when I got married, and it was awkward. It wasn't the most awkward thing about when I got married, but I'll tell you that story later if you're lucky. So thank you, Rebecca, for the kind introduction, and even bigger thank you for the invitation to speak here at the U to finally see the Marriott Library. I was just telling Rebecca on the way in. Having worked in libraries for the last seven years, I love to come and see a library, see how it's put together, how it's organized, how the library thinks about itself. I haven't actually spent much time at the U, apart from my time, a while back in the Utah Olympic spirit band. So I don't know if you can see me, but I'm the one in the ridiculous hat. There I am. And as you've heard today and seen, my title is Speaking in Code. So before we get to speaking in, let's speak about code. Oh wait, I missed the thing. Now there's the about word. So code, of course, has many definitions. The OED's first entry points to the legalistic meaning of the term, a systematic collection of statutes. So here you might think of the code of Hammurabi, which consists of 282 laws. This document from Babylon dates back to about 1754 BCE and is one of the oldest deciphered writings of any length in the world. It's come down to us on clay tablets and a massive stone stelle written in cuneiform. This code governed behavior, wages, and punishments. It made Babylonian society run. The second definition of code in the OED points to a broader interpretation of the first, quote, a collection of rules or regulations on any subject. You might think here of the code of student academic conduct at an institution like, say, the U, or perhaps the rules for boxing established by the Marques of Queensbury. These are not quite as dissimilar as you might think. The first one is a set of rules to keep your nose clean and the other one to make sure you get it bloodied in as clean a way as possible. And finally, the OED gives us one more set of definitions or code or for code, a system of military or naval signals. And here we finally get to the type of code that fascinated many of us when we were kids, the secrets that could be revealed with a decoder ring. A code is something that is clearly legible to its intended audience and completely opaque to anyone else who is listening in on or attempting to join that conversation. Codes in this sense carry critical information. And the importance of its message is always proportional to its ability to be misread or misunderstood by the majority who hear it. This is, as a new professor of digital humanities, and I keep meeting people. And this is what I do, or this is what people do to me all the time, like, what is this thing? OK, hi. OK. And finally, code is also almost always this sort of code is what J.L. Austin would call performative language. They lead directly to action. Code in this last military sense is normally understood as a noun, but of course, these days, it's equally used as a verb, especially in the context of computer programming. Code is both the thing one produces and the action. And it looks remarkably similar to the Kaniyaya form when you get down to it. In our current information age, code, noun, and verb are highly-affected concepts. We don't just use the products of those who code. We read about them and we watch movies about them. We wonder whether, and almost all of us, worry about our inability to code. We wonder whether we need to learn to code. Indeed, the only thing that Google thinks we want to learn more than coding is typing, which is, of course, sort of a prerequisite. We have, as the kids say, or as they said two years ago, a constant fear of missing out, which is a reasonable fear. Code is, after all, something that was developed to do something or communicate something without other people really understanding what you're doing. And here at last, perhaps, we get to the digital humanities. So over the last 10 years, there's been an increasing furrer over the digital humanities. Perhaps the blame of this could be laid at the feet of academic journalists. William Panapacker suggested, in the middle of the 2009 MLA convention in the Chronicle of Higher Education, that digital humanities, quote, seems like the first next big thing in a long time, and then followed it up a year later with a post titled, Digital Humanities Triumphant. By the way, I took this picture of somebody from the U after the BYU Utah game. So Panapacker's articles are just a sample of what appeared in the Chronicle inside higher ed over the last five to 10 years. Some of the excitement could also be laid at the, in addition to the academic journalism, we also had the sort of normal press. So the New York Times ran its own six-part series on the subject over the course of a year on digital humanities. In 2016, the Los Angeles Times got into the act, running a series of interviews with a dozen scholars in the field. And this year, in 2017, following the annual MLA and AHA conferences, PC Magazine of all places ran a story titled, Digital Humanities colon the most exciting field you've never heard of. So folks, you can consider yourselves hip insofar as you've actually heard of it, at least to show up here today. So all of these stories suggest in a way that digital humanities was not only the next big thing, the first next big thing in a long time, but also the first thing about the humanities that the general public itself might care about in a while. This is, of course, a specious contention on my part to give it to me. Along with all the news coverage, of course, there have been circumstances that conspired to garner the digital humanities attention on campuses. One of these was the creation of a permanent Office of Digital Humanities at the NEH, the National Endowment for the Humanities, created in 2008. The ODH suggested not only that the field had quite literally arrived, but that there might be money involved as well. Another change was the creation of jobs for digital humanists, both on and off the tenure track. Cluster hire suddenly appeared in departments whose hiring plan had been decimated only a year or two earlier, thanks to the financial crisis of 2008. So suddenly, digital humanities had everything a young scholar or department needed, attention, money, and jobs. But unfortunately, you're very likely not sure how to go about getting involved because digital humanities seemed very much like a code phrase. This is the enigma machine at the Museum of Computing History, which is in Palo Alto. That was the coolest thing I saw there. So the magical term digital humanities was getting things done. In this way, it worked like code. It was performative text, but it wasn't something that most people felt they had access to. So it was code about code. And then to make matters worse, Stephen Ramsey gave a talk at the 2011 MLA convention talking about, as the title put it, who's in and who's out in the digital humanities. In two sentences that were heard around the world via Twitter, Ramsey said, quote, do you have to know how to code? I'm a tenure professor of digital humanities, and I say yes, end quote. So talk about FOMO. Oh, I forgot to hit this. There it is. My script's off. There's our FOMO. So if Freud was alive, he might call us paranoid, but he's not, and it's probably because he didn't know how to code. So for all the interest in digital humanities, it seems there are some serious barriers to participating, not the least of which is simply figuring out what digital humanities is. Indeed, the unspoken debate at the center of the excellent 2012 collection edited by Matt Gold is precisely the question of what is it? What is digital humanities? That is, in many ways, the debate. The defining of digital humanities has become something like a joke among those in the field, since it's a question we're asked over and over again. That's probably why Jason Hepler built whatisdigitalhumanities.com, which pulls one definition at a time from a crowdsource database. In other words, he wrote code to expose the code words that the code writers are using that we don't understand. We could hit refresh on this thing infinitely and never get to the end. And most of the definitions, we could get to the end. There's about 900 of them in there. And most of them are pretty good. But in many ways, I think the question of what is digital humanities is sort of a red herring. It suggests that there is, in fact, an answer, that there's a signified behind the signifier. Instead of asking what digital humanities is then, I'd like to suggest that we bend the question just slightly. Instead, I'd like to talk about what digital humanities means. Now, I swear I'm not trying to pull the wool over your eyes. While these questions are closely related, they have an important difference. The first question, what is digital humanities, suggests again that there is something out there that is digital humanities and something that is not, and that we can tell the difference. The is is an equal sign. And like all equal signs, it's axiomatic. It is because it is. The second question, what does digital humanities mean, points us to something else entirely. Meaning, it turns out, is constructed. And it is constructed by people. Of course, it's not made of people. Meaning depends on context, the place where the subject is broached, and the people who are doing the talking. In other words, meaning is inevitably code. But meaning is a code that humanists are better prepared to deal with. We know how to break these codes. When we consider what digital humanities means, we are tracking a term with a history, one that varies from one person field or institution to the next. And one person's is may very well be the next person's isn't. The focusing our attention on meaning puts us in the same campus, Matthew Kirschenbaum, who in debates in DH suggests that digital humanities is a quote tactical term and one of convenience. So this convenient term has become a code for a number of different practices to get lumped together into the concept of digital humanities. It's a code word insofar that a code like SOS helps to speak quicker and more efficiently. But it's not a code that's really meant to be secretive. In other words, while digital humanities have been speaking in code, it's not a code of exclusion so much as one of convenience. That said, we digital humanists, I think, probably do need to speak in our code a little less frequently. For that reasons, I'd like to clarify what I think people tend to mean when they say digital humanities. As best as I've been able to tell over the last decade, there are five different things that digital humanities might mean. Unsurprisingly, these areas overlap. And many people mean more than one when they use the code phrase DH. But something that I think might be surprising is that when we break down what the code phrase digital humanities means is that you'll discover that you might already be engaged in a way that you hadn't considered. In other words, you probably already speak a little bit of code. There's not very many statues on this campus, it turns out, that people are willing to put on Flickr with Creative Commons licenses. So that was the best I could do. OK, so five things. The first thing that is sometimes meant by digital humanities is the humanistic examination of digital objects. How, in other words, do we use our humanist toolbox to look at an object like the personal computer, the network it operates within, or even the code it runs on? It's sort of work and take on many characteristics. You can look at Bruno Latour's philosophy and call it digital humanities. Friedrich Kittler's historicizing of different inscription media could perhaps qualify. Nick Montfort and Ian Bogost's work on the Atari 2600 in racing the beam. Art historians considering digital art and installations. We would also fit under this rubric. And critical code studies where scholars read the code of software to deconstruct the assumptions at work in building the systems that govern our world. One recent example of this is the book Ten Print, which spends 328 pages considering the ramifications of a single line of Commodore 64 code. And it's written by 10 different people at once. It's sort of a strange book. So any of this could be meant by those who use the term digital humanities. But the scholars who work in these fields might be more likely to refer to their work with the term media studies. Some media studies scholars would welcome inclusion in digital humanities, but many others find that inclusion problematic. As a relative recent term, DH might erase the valuable work that's been happening in the field for decades. So that's the first. The second thing DH can mean, as some people, is digital scholarly communication. So there's this thing. It's called the internet. We didn't used to have it, and now we do. And it kind of changed a few things. One of the things that's changed is how scholars communicate with each other. And we still talk to each other in books and articles. We now have different ways of talking to each other and publishing middle state scholarship. Think, for example, of all the blog posts out there of recent conference talks. This was mine at the MLA in January, by colleagues as well as research blogs. Because increasing attention came to digital humanities at the same time as scholarly communication began undergoing massive shifts towards a blogosphere and publishing a middle state scholarship online. Sometimes when people say DH, they really mean what happens when the processes of scholarship goes digital. That's a melon. Oh, well, there's a melon. Which is, I was going to mention, for example, I'm just getting started at BYU. So I don't have a lot of on the ground experience there or projects to show you about from there. So I'm going to tell you about something we did with the Mellon Foundation at Brown, where we have a Mellon grant or had, I suppose. I'm not there anymore to publish large-scale works of digital scholarship over the period of four years. One of our first two projects dealt with a 17th century book of alchemy, Michael Myers, Atlanta Fugians. And it's not that Michael Meyer. So what this book is, is it's an emblem book. This is what an emblem looks like. There are 50 of these things, so this is number 26. For each emblem, there is a title in German. There is a three-voice fugue with a text in Latin. There's an epigram in German. And then on the facing page, we have the title, again, this time in Latin. So it's the same. It's a translation. We have an image, and then we have the epigram in Latin. So these are, again, translations, which, of course, suggests that this image is the same thing as this music. The following two pages are, for each emblem, are dense Latin explaining the allegory and how to do alchemical practices that are sort of mentioned in here. As I said, there's 50 of them. Some of them are very strange. And so the project is figuring out, and what I spent the two years I was at Brown, doing this figure out, how do we take this book and put it online? How are we going to digitize it? How are we going to not just show people the pages but render the text into a configurable format? It's going to involve TI encoding. But also, how do we design it to have a more modern, web-based experience? How do we add the music to it so people can finally hear this thing that's always been a multimedia work, but unless you're an extremely educated person, even in the 17th century, an individual couldn't really deal with this thing on his or her own. So we're working to create digital facsimile. We're also bolting onto the side of it, essentially an edited collection of long scholarly essays discussing the book and explaining it, what scholars have learned about it. There's short introductory essays explaining some of the concepts you might need to know, like fugues or the alchemy in general. And then building in the sort of things you might expect to have in a digital edition, search, browse, facets for exploring the emblems, cross-linking between the different things. In trying to figure out how we were going to take the different pieces of the text and put them onto a single page, we had to deal with identifying what the features were and thinking about how we might go about laying them out. This was a first attempt by our designer. This is just sort of lorem text in here. And a later iteration in which, after sitting down and talking with our scholars again, they said, you know, the image is really the thing, even though it's not the first thing when you're looking at the print book, the fugue comes first, the image is what is going to help people, most of the time, sort of bookmark it in their head, have a sense of what's there. And they wanted to feature that first. So we moved it up, redesigned the features of the page to sort of privilege that and make it more visible. So this is one example of how not just blogging, talking to each other as scholars, but putting big multi-year scholarly projects online. So the shift in digital scholars, scholarly communication means we need to think not only about how we share our work, but how we let people know about ourselves. We might need to think about building a website or at the very least getting somebody to not have my images there. Okay, have somebody build one for us. You might have to think about also how you do social media. So again, you don't have to be a digital humanist, whatever that is, to be engaged in these shifts in scholarly communication. But some of the first people to be thinking actively about such shifts were digital humanities scholars. And so for a time DH could mean scholarly communications. There's a sort of Venn diagram happening here. I think as we move on in time from 2012 or so, this definition becomes less prevalent, but it's still there. So two down, three to go. The third thing we might mean when we say digital humanities is digital pedagogy. So if digital technologies have changed how we talk to each other, they're also shifting the work that we ask our students to do. So let me take it out of the abstract here and show you a project I did in my introduction to digital humanities course a couple of years ago. In that class we tried to read Mark Z. Danielevsky's House of Leaves. Has anybody read this? Okay, we've got one. All right, well good, I've got a summary here. So published in 2000, which is sort of crazy to me, it's that long ago. House of Leaves is the story of a piece of architecture, a house that's bigger on the inside than it is on the outside. Or more properly, it's the story of the discovery by a 20-something LA club kid and aspiring tattoo artist named Johnny Truent of the manuscript of an academic treatise written by a blind man named Zampano about a documentary about a house that's bigger on the inside than it is on the outside and how that threatens the integrity of the Navitsen family who comes to live in it. Truent reassembles the manuscript that he finds after Zampano dies, adding in his own narrative about familial loss into the footnotes, which in turn are commented on by anonymous editors. And if that is not complicated enough for you, the book does not want to be read. Just wait until you start looking at the pages. House of Leaves is quite literally a text that's too difficult to read on your own because in part it's full of codes, which means it's the perfect book to read as a team. It took the internet about four years to figure out most of what Daniel Lefsky had sort of shoved into this book. So what I did here was to say, you know what? If reading it as a class is good, and it is because I've done that before, what if we read it with more classes? And I reached out on Twitter as one does and I said, I'm teaching this book. Who out there is teaching it the same semester and who wants to do an assignment with me? And I don't know what it is yet, but we'll come up with something that all of our students can do together. So we eventually got six people working on it. Since a book is in many ways an attempt to condense all forms of media into a print volume, I think at the heart of it, it's Daniel Lefsky trying to say books still matter, despite the fact that the internet seems to be pretty cool. We wanted to sort of blow the internet, blow the different media forms that Daniel Lefsky crams into the book back out into the internet. The idea for how we might approach this came from Zach Smith's project, Pictures, and this is a title that pictures showing what happens on each page of Thomas Pynchon's novels, Gravity's Rainbow, which is exactly what you think it is. Zach Smith drew a picture for every single page of Pynchon's Gravity's Rainbow. And with that as a sort of inspiration, we decided to build a platformer for remediating every page of House of Leaves. We called it A Million Blue Pages, which is a line from the book. A Million Blue Pages collected media response of our students to individual pages of the novel. All they had to do was to post to one of our four social media platforms, Tumblr, Twitter, Vine, Rest in Peace, or Instagram, and include two hashtags, one for the project for A Million Blue Pages and the other with the page number. So our platform was tapped into the different APIs of these social media sites, crawl the services regularly, and when it found new posts that met these criteria, they got pulled into the platform. They got assigned to the pages and you could sort of see the book coming together as our students worked. We assigned our students, again, across six different institutions to work in groups and create one object for every chapter in the book. They would not only have to create an interesting object, but also write about why that page had seemed important to the overall narrative because there's, of course, in most chapters, multiple pages, which pages they pick. The only exception to that is for chapter six, which you can't see here, there's only two pages in it and people really hated that one for that reason. So working in together, we made significant progress towards making something for every page in the book and the students' work was pretty exceptional. So here's something that some of my students did. Trees matter a lot. That's a symbol in the book and so does Ash. The house in question is on Ash Tree Lane and so a couple of students took burnt match and sketched the tree. I asked afterwards if they had the original and they'd already thrown it away. Here's work another couple of students did related to that chapter six, which is so hard. They worked with, they looked at some book art stuff and so this chapter is about the pets in the house, the dog and the cat, that don't really seem to be affected by it but keep trying to run out of the house. And here was another one in which the narrative, the way in which the print is laid out in the book often echoes what's happening here. Karen helped by going out to town to purchase additional parts, including a pulley and extra rope, which is when the rope slaps down on the floor. So they cut up the page and sort of bent it to build representation of what was happening there. So a million blue pages was a complicated project but not all digital pedagogy has to be quite that complicated. Easy to use tools, make it possible to create new in-class activities or assignments. It's a project that I worked on at Brown. Professor Massimo Riva, the Italian Studies, there teaches a course annually on Boccaccio's Decameron, which is a story of tens, it's 100 stories. By told, a story a day told by 10 narrators across 10 days as they're fleeing plague-ridden providence. That's where I was. Florence, the only plagues in providence relate to Cthulhu. And so what we did, the students were putting in information about the stories and they got assigned individual narrators and they would put in stuff about the settings of the stories and a lot of other details. And we were tracking all this in a Google spreadsheet. And then from the Google spreadsheet, we can just sort of pull all that data using the Google Maps API and plot everything here. And then we used a map overlay. This is a 16th century map. It works pretty well in this region because the map maker was from Spain but the further north you get, the more it's sort of, that's where the dragons are. And so it just, it bends weird ways. And these are actually, we think are supposed to be mountain ranges. They're sort of weird tumor-looking things. But with this tool, the students who had been reading the stories for a while were able to see at a glance that certain narrators like here in dark blue, I can't remember who it is, have a sort of broader global imagination that their references or where they set stories go out further than say our white narrator. And certain days, like the second day, tend to be places you get a bit more traveling happen, which is a way to sort of see and it's a place to start asking questions from, not that it's going to tell you. So when we talk about digital humanities as digital pedagogy, we point in part to the fact that our classroom walls have become porous. We no longer need to pretend that we're the only class in the world considering the subject. We can reach out to colleagues who are teaching the same objects or history and collaborate. Our students can do the same with their work. So the fourth thing we mean when we say digital humanities is the creation of digital primary source materials or archives. So for many of us, this might be a familiar form of digital humanities. It's the type of work that's been happening for decades, especially in libraries. One might point for an example to the work of the Women's Writers Project. Again, started at Brown in 1988 with the process of text encoding rare texts written by women across centuries. It was a recovery project. Emory University was also engaged in a similar women's recovery project. There have been a number, we've got a project at BYU getting underway with recovery of Native American literary materials from newspapers, from tribal papers. In some ways, digital humanities' archive creation is a way to seed the ground for scholarship. This is why the NEH, Mellon, IMLS, and more have at times been willing to fund digitization in archives. Team that I worked on with in 2013 for the NEH-funded One Week One Tool Grant did something similar with APIs from Europeana, Flickr Commons, and the Digital Public Library of America. We built a tool called Serendipimatic in a week, just sort of crazy, which uses natural language processing and algorithmic randomization to find primary materials related to bibliography, syllabus, or article. So you can come, you can put in some text you've been writing, and we can find primary source materials that are maybe sort of related to what you're doing. It's this, instead of a search engine, it's a serendipity engine, so it's built to be weird and to give you sort of, hopefully, serendipitous things. It's a really hard name to spell after a while. A lot of times this sort of work of creating digitization relies on TEI, the text encoding initiative, which is a sort of approach to XML, which ends up looking like this. This is for our project at Brown with that alchemy book in which our grad student who could read the English translation, we had a 17th century English translation that we wanted to use because it was closer to the period, and he was having the correct spelling, add in deletions, because it was a handwritten manuscript and sort of mark everything that was there. And with this sort of precision of TEI, you can create a pretty good facsimile of the way a text appears. So this type of digital humanities, especially as the librarians know, is a creation of archives and responsible digitization and curation of the same is not just the groundwork for scholarship, but is a form of scholarship in its own right. People spend their lives thinking about, thinking hard about how do we record a text? How do we encode music? And we do it responsibly so that other people can sort of adhere to the same standards. So that's the fourth. And the fifth and final thing that digital humanities can mean is the digital examination of humanistic objects. So look what I did here. This is called a chiasm. Isaiah would be proud. All right, so what does it mean to use a digital tool to read history or an archeological site? In short, and this is what I tell people when they wanna know what digital humanities is, I think any field of scholarship whether you're a chemist or a sociologist or a digital humanist, your job is to look for patterns. You might look as a chemist for a pattern of precipitates following a chemistry experiment. What I imagine chemists do all these is basically take cool looking Erlenmeyer flasks and pour them into each other and sort of try and figure out what that means. It could mean a pattern of representation of racial groups and historical documents if you're a historian. We do the same thing in this fifth type of digital humanities. We just use digital tools to help us identify the patterns. These patterns can be in data sets that are simply too large for us to read as humans or they can be patterns in small data sets that are hard for humans to count or keep straight in our heads. So it turns out computers are great at solving these problems. But, and this is key, while computers, while computers can help you find those patterns, it's still up to you, the human, to decide what it means. Once you've seen the shape of the data, in other words, you still have to interpret it. And this means that no matter how much code we speak, no matter how digital we are, the digital humanities still depends as much on the humanities as it does on the digital. Because while computers are counting machines, people are meaning machines. Meaning doesn't exist outside of humans. So here's an example. My friend Rob Nelson at the University of Richmond got interested in the Civil War. He's a historian of the Civil War. And he took the four years, he was in University of Richmond, I said, the local paper, the University of Richmond, he got the four years of the Richmond Daily Dispatch during the Civil War. And he got these from ProQuest. He didn't do the work because some hardworking librarians and archivists had already done all the digitization work for him. And what he did was he wanted to, he used a computer with unsupervised machine learning to read through and find patterns of words that had co-occurrence in individual articles. So this is, and these are called, this is an approach called topic modeling. This is one set of words that is pulled out by the computer as being a topic. Negro year's reward, boy, man named jail delivery, give, left, black, paid, pay, ran, color, Richmond, subscriber, high apprehension, age, ran away, free feet, delivered. Okay, so the computer, which doesn't know English, has decided these words appear together regularly in articles and it has assigned them as a topic. And the computer gives this a number. It says this is topic 17 or topic 200 out of however many Rob asked for. If I ask you, humans, what is this a topic about? What would you say? Runaway slave ads, that's exactly what it is. And we know that because Rob paid some undergrads to go through and count all the undergrads slave ads in the four years and sort of double check that the one's computer was pulling out as being highly representative of this topic were in fact runaway slave ads. He didn't do that with everything but he did with this one. So once he's done this, once he's trained the computer and it's pulled out a topic, then you can take the entire corpus, the millions of words across four years and you can say, show me individual articles that have a high density of this topic. And so here's an example of one. Ran away from the subscriber on the sixth instant at Vienna, Virginia, a mulatto boy named Sam said boy is about 20 years of age and so on. Right, this is clearly runaway slave ad and over here the computer model says that this ad, this article is composed 90% of top of words from the fugitive slave ads topic but also 4.26% from the groceries topic, military orders to deport, deserters and secondhand news. So there are words in here that belong to other topics that Rob has named. That's okay, we use words for lots of different reasons. But this one is pretty clear, this particular topic, the computer model it to show up like this. Once you can do that and have a sense of where these things show up in one article you can graph the entire aggregate across say the entire time of the Civil War and you can look at the rate of runaway fugitive slave ads. And then once you've got that you can ask yourself, well what's going on? Why, I mean there's up and downs but there's a couple of places where we get some peaks. What's happening here? Rob who's a historian and he's written about this so this isn't a thing I've learned on my own. He said, these two are easy to understand because these two are when the federal, the union army is near Richmond. And so people take it upon themselves to emancipate themselves. You've got an army nearby, people are distracted and so if you're an enslaved person you take off. And that explains these. However this one he says he didn't understand why. There's no union army nearby, there's a clear increase here. When he looked at some other topics though this is one that he looked at. It turns out, and it's something that I hadn't known until I read Rob's work, is that a lot of times the people that were enslaved were not always put to work by the people who owned them. There was a thriving practice of renting out your enslaved person to somebody else for generally a year at a time. And so people would put out higher and wanted ads on an annual basis and it tended to work for the calendar year. And so we've got spikes January of every year of when these for sale and wanted ads show up in the Richmond Daily Dispatch of people putting other people out for hire. But this is odd because this one's lower than all the others. This stays pretty consistent across the Civil War despite the fact that the fortunes of the South turn right around here, this stays consistent. But January of 1862, there's a dip. And when Rob put the two graphs on top of each other this spike in runaway slave ads, which is the green, timed pretty well with the drop in the for sale and wanted ads, which of course still doesn't explain it, right, correlation is not causation. But Rob hypothesizes that with the beginning of the war starting in the fall of 1861 that people had in just sort of the general excitement for people in the South, they took to emancipating themselves despite the fact there was no union army nearby and with a consequence that there was a drop in the available human labor to be farmed out at the beginning of January of 1862. So that was one result that he found in doing this work. Now, I have some slides in here. I was gonna tell you about a project I did with Hemingway, we don't have time. I don't think, I can tell you about it if you want to know more. I was gonna tell you more about topic modeling. Things I've learned. Okay, so there you have it. We've broken the code of digital humanities. People tend to mean one or more of these five things. This is, you know, if you're taking notes, this is the part you write down, I guess. The humanistic examination of digital objects, digital scholarly communication, digital pedagogy, a creation of digital archives and digitizing things and the digital examination of humanistic objects. But let's hold on just a minute. Let's turn back to Steve Ramsey. Didn't he say that you need to know how to code in order to do DH? Well, yes, he did, right? It's right there. So is everything I've been telling you some sort of mumbo jumbo designed to convince you that DH really is something that you can do was my talk just a smoke screen and attempt to throw you off the scent of the elusive digital humanists, a coded message couched in the claim of lack of code. Well, no, I don't think so because it turns out that this sentence, so these two sentences I've shown you from Ramsey's talk were largely taken out of context. If you've read the whole thing, it turns out for Ramsey, coding becomes a proxy for the notion of actually building. Building is actually what he thinks is at the heart of digital humanities and he has quote, a highly expansive definition of what it means to build something. So when we build new structures for scholarly communication, when we build new archives, when we build tools to find patterns in our data and when we ask our students to build collaboratively rather than to write to an audience of one, we are engaged in digital humanities. Does this mean that anything goes in digital humanities? No, of course not, but I'm hoping that I've given you a way to see how you could begin building yourself or to recognize that you've already begun doing so. So let me speak here at the end, not in code, but in plain text. You already possess the most basic tools you need to do digital humanities. It's not code, it's humanities training to see the meaning that the code reveals. Thank you. Question me. Tell me I'm wrong. I think this is the mic we're supposed to use, right, Angela, or Angela's proxy. Okay, thank you, Brian, that was wonderful. Do we have any questions from the audience about any of this? So I was just curious to hear, are there any tools that you're excited about learning more about? I mean, there's kind of developments happening in a lot of spaces and what has you excited right now? That's a great question. I'm excited by topic modeling. That's the thing I've been working with more over the years. So I'm still excited by the possibility of, and what I didn't show you about my Hemingway projects is my students and I took all of Hemingway and we could build a corpus. It's not hard to build your own novel set of text as long as you have a scanner and a book guillotine, which libraries tend to have. And it takes all of about four hours to get all of Hemingway digitized. And so the tension sometimes in digital humanities is this idea of big data. But the difference between what humanities people normally have is data, which is like I read one book and now I will talk about it, to even like to 10 books, like all of Hemingway, that's a pretty fun jump. And it's not hard to do that. You don't need to have ProQuest access to do that. That said, I'm also interested in something I'm doing for the first time in the class I'm teaching right now. And part of this is in the past when I've taught my digital humanities courses I was in an English department. And so I felt like I could be as Englishy as I want. And now I'm in, BYU has a, in the College of Humanities, there's an Office of Digital Humanities. And we have a minor, but it's not in the English department and I'm not an English professor. And so I wanted to, as I redesigned the course this year, I wanted to have some more stuff on visual culture. And so we're doing a section that starts with GIFs and looks at GIFs and the way in which reaction GIFs work. But then we're gonna move on to doing, this is work by Kevin Ferguson, recent article in Digital Humanities Quarterly in which he takes all of the frames of this article, particularly these about Disney films and he averages them to do a sort of distant reading of visual objects. And so this is every frame averaged of Sleeping Beauty. And he's got a lot of stuff in here about how he does it. He uses a tool called Image Plot to do it. This one is every image in Alice in Wonderland. The color here is not great, at least not at my angle, but you can actually kind of see yellow and the blue of Alice. He writes down here that Frozen, one of the things that was interested in Frozen is the fact that there are sort of two blobs showing up and that a lot of the scenes in Frozen are composed of two people standing next to each other. So I haven't done this yet. I'm not sure, so my students, I are gonna do this this semester. So I'm excited to try and stretch myself a bit past text and to start thinking about visual images. Yale's Digital Humanities Group has been doing some interesting work on this as well with their collection of Vogue and things like that, so. Sure. Oh, we've got a couple hands. We'll start here and then go there next. And the microphone is working, even though it sounds like it's not projecting. It's captured on audio. Thanks for the talk. Sure. Topic modeling. So my meaning machine in my brain tells me I could have come up with that more simply, but I could be wrong because maybe I misunderstood. So if I just came in with a question, well, I should look for a fugitive slave ads, wouldn't that have short circuited or made much more quick the search that he did? And so I'm wondering, is there a way of talking about DH and maybe it already does and I just misunderstood that helps us ask better questions? Because it feels like that was a complex way of getting to something that I probably could have searched myself relatively quickly. Is that not true? Well, the paper wasn't broken down in such a way. It hadn't been tagged by humans. It was a scan thing, but it wasn't tagged in such a way to say this thing is this. And so by having the computer read it, he was able to sort of pull out the ones that were the slave ads. So it did save time. But this is always a tension that of course you could sit down and do it yourself. So would the argument be that the word fugitive would have been dispersed enough across a variety of different articles that it wouldn't have given me those ads or I mean, I feel like I could think of five words that not 20 but five, that would have done the same thing as the 20 words that the topic modeling gave me. Well, and that's part of, I have a whole hour talk I do on topic modeling because it's so backwards and non-intuitive and how it works. But part of the power is that it can grab words that you don't think are related to the topic and it'll show those up. But I think part of what you can get at with topic modeling as an approach is that you can ask questions of a whole corpus and that yeah, if you think of one particular thing to ask, it's there but in the case of, I mean here I will actually show you my Hemingway one. So I took all the Hemingway stuff and here are two of the topics that I pulled out. Would one could good see old man time like Med said, make much never no thought welcome always first, said don't went it's light. So these aren't very interesting. You look at those and you think gosh, I guess Hemingway wrote in English which is kind of what these are. But when I took these topics and these are, these were in the model I ran, these were the two most prominent topics on a particular pass through. This one is extremely high percentage wise in some of his late fiction. And it's relatively low in the early novels. But what was interesting is a Garden of Eden which was a posthumously published novel and he never finished it but was written in the 1950s. So he started writing it after across the river and into the trees lined up a lot more with this topic or it was low in this topic. And then this topic four does sort of the opposite. The early fiction is high in topic four. Late fiction is low but Garden of Eden again is high. And so this I think or what I'm trying to find more out about it is does this point of sort of stylistic change in Garden of Eden was Hemingway consciously trying to write more like these novels or because it's posthumously published and it was a 1500 page manuscript that Scribner's editor Tom Jenks took and cut down into a 200 page novel cutting out a whole major plot line was that a case of Tom Jenks looking to these early novels and saying like yeah I don't really like that Hemingway. I want this Hemingway to be more like this or I mean that's one way of doing it. Another thing I realized is that and there are more words than just these in the topics but this includes some words that are in present tense it's I'm whereas these are pretty much all in past or conditional tense and when we did some more tests on it what we realized is that this is this topic comes out of dialogue. These are people speaking because in novels you narrate in the past tense but you speak in the present tense and so we found another way to sort of pull out dialogue. So I didn't have this question until I did this and maybe I was looking for a question to ask once I'd spent the time on it. But I think that's part of the when you use a computer you end up seeing things from a different angle. Sure. Great talk, loved it. I'm more on the computer programming side of things than I am on the humanitarian side of things. So it's really interesting to me to see how a lot of recent advancements in machine learning and artificial intelligence in sort of a general sense affect the field and recently there's been a lot of development in machine learning for iOS and Android where very recently they're now both very capable of making use of machine learning. How do you see that these recent advancements affecting the field? That's a great question. I don't know that I have an answer for it. I don't know, what do you think? I mean that's obnoxious, but I haven't thought of that at all. There are a handful of University College London or King's College, I can always get the two mixed up because they both have digital humanities centers. A couple of years ago they released a tool called TextStall that is, so you can do text analysis on your iPhone or your Android and BYU has a tool called word cruncher that they've been working on Android and iOS versions of, which again are sort of just basic text analysis, word counting, but as far as neural networks, I mean this was the thing that Yale just did, they just released, or is it on their website? Nope, oh heck, where is it? Go figure. They just, they essentially took, this is terrible that I can't find it. It's more interesting if you can see the pictures. They took, here we go. They took a set of, okay. So they took a collection of photographs from the Miserve-Kunhardt collection at the Beinecke of Victorian photographs and similarity based on approximate nearest neighbors from 2048 dimensions and the penultimate tensor of an inception, you know, this just sounds like they're making things up, but essentially using neural networks to find nearest neighbors to find people that sort of look like each other, but then they from that went to, oops, Peter Leonard, the director of the lab, spent a weekend training a generative adversarial network on 25,000 19th century portraits, these people never existed, so he started using a neural net to come up with portraits of people that didn't exist. So, I mean the question about iOS and Android, the fact that we all have supercomputers that talk to space in our pockets mean that we have the ability to, you know, to do things we couldn't before. I'm not really sure what this platform at the moment does that, you know, a computer doesn't for it, so I don't have a good answer to that, but I think this is some of the direction that people are starting to move and thinking about how can neural nets change the way we think about how we think of our humanities things. Thanks, I think we have time for one more question. Does anyone have a question burning? So, I'm not too familiar with digital humanities as a field, even though I guess I do some stuff that would fall under some of your categories, but one thing I've noticed in your talk as you've listed the five things is, obviously there are correlates in the sciences of the way they use digital things to do scholarship in all of those five ways, but one thing that's done a lot increasingly in the sciences that you don't mention humanities is using digital tools to get citizen science done, to say get people involved in doing research or just to get an army of volunteers to get stuff done that you couldn't do yourself. Are there equivalent projects in the digital humanities where it's citizen humanities? That's a great question. There are, University College London has, this is just one example, University College London has the papers of Jeremy Bentham. Not only do they have his papers, they've actually got his corpse. He's embalmed, they bring him out once a year to the faculty meetings. Like this is in the charter of the university. So Bentham is there with them. So they've had, this project's been going on for, gosh, it's probably 2010, 2011, and so they've got the manuscripts and you can go in and you can transcribe them. You can get a page and type it up and then they check them against each other. So there are some of those things. I was working with the project at Brown again with a faculty member in history to build a database of indigenous slavery in America. So one of the things that I, again didn't know a whole lot about is the fact that indigenous slavery, while not on the same scale as enslavement of African, people of African descent, there was indigenous slavery as well. And so he's interested in sort of capturing this record and creating, getting stories from people about ancestors that were enslaved, finding documents in local archives. And he was really passionate about making it so that the public can contribute. And the challenge a lot of times with that, and I told him was that a good crowd sourcing project, and you see this with things like at Zooniverse, if you're tagging safari photos, I like to do that one, is that a team has already secured all of the materials. And then you can sort of, once you've got the materials, you can get a group of people who are passionate about your project. At Emory we had a project transcribing that they had in the theological library collections they had some letters from a cardinal. And so they were able to get the sort of local Catholic community interested in transcribing these on an online basis. It's hard if you don't already have the collection materials, how do you get the people together? And so that's where we go to that fourth definition. A lot of times it's up to an individual researcher working with a library or another group to get materials, to digitize them, which is not a quick process. And then with some thoughtful marketing and reaching out to people for whom the audience is there, you can get people to do that. Unfortunately it's not nearly as easy as people think it's going to be. There is not yet a sort of correlate to fold it, the sort of protein folding thing that people were doing on the PS3. But there are other projects where people can come and transcribe and that's the main thing that people are doing at the moment because it's what you need a human to do. There's the things that machines simply can't do. Machines will probably never be able to do, well not for 100 years, hand writing from the 15th century, it just doesn't work. So I'm here for a while so if you still have questions I'd love to have them thank you again so much. Before we thank Brian, there's two things I'd like to say real quick. One, I just wanted to thank Lisa Swanstrom and David Rowe who were really instrumental in setting this talk up today and making sure we had refreshments which please enjoy after this. And then the second thing I wanted to say is that the Digital Matters Lab is new here at the University of Utah. We have new social media accounts so if you enjoyed this talk, if you want to know more about digital humanities please do follow us on, I think it's you of you digital matters on Facebook and you digital matters on Twitter. And I think we have an Instagram account too but I don't have that one memorized. There you go, digitalmatters.utah.edu. Now please do go ahead and thank Brian for his great talk today.