She-wolves on screen

This blog post introduces the “Rome from the Ground Up” research project, which is generously funded by the Social Sciences and Humanities Research Council of Canada.

I’ve found the Romulus and Remus story fascinating since I first read it in Latin class, more years ago than I care to remember. Yet unlike so many other classical myths, it has rarely been the subject of television or film productions. This absence is strange, since the story of the twins still holds an important place in the Latin curriculum, especially before university. A few years ago, I decided to explore the existence of Romulus and Remus on YouTube, which is my personal favorite media platform. I was surprised and delighted to find hundreds of amateur (and semi-professional) accounts of the Roman foundation myth, and decided to begin analyzing them. 

The questions that I had going into the project could be divided into two parts: questions about the creators of the videos, and questions about the audience. In terms of contents, I was eager to learn how creators, especially younger creators, handled the mature elements of the Romulus story (such as sexual assault and murder), and whether this treatment could be correlated to the number of views of the video. Since my previous research focused on how ancient accounts of Romulus vary in a few important specifics, especially regarding whether he purposefully killed his brother, I was also interested to see whether one ancient author’s perspective was represented more frequently or more fully than others’. In terms of comments, I was interested in the response to the mature elements; I was also eager to see whether anyone else was watching the Romulus story across several videos, as I was (answer: yes!). Although there have been several important studies of classics and social media recently (e.g., Neville Morley on Twitter; Juliana Bastos Marques and Helen King on Wikipedia; and Donna Zuckerberg on Reddit), I am, as far as I know, the first to analyze YouTube videos, so the beginning of this research project involved learning how to approach a platform that I use for pleasure in a more scholarly way.

So far, my research team and I have watched and coded 404 individual videos for their contents, and analyzed the comments left on about ⅓ of them. By “coding”, I mean that we watched a batch of about 50 videos and came up with a list of common themes and characters; we then made a spreadsheet with these themes and noted their presence or absence in each of the videos we watched. Unsurprisingly, all of the videos have included Romulus and Remus; however, I have been impressed to see the depth of knowledge of some creators, who included more obscure characters like Acca Larentia (the wife of Faustulus, the shepherd who finds Romulus and Remus) or Picus (the king/woodpecker who watched over the twins before the she-wolf found them). Some videos also include detailed bibliographies in their descriptions.

While subsequent posts will detail the themes we’ve come across in the comments, I will here introduce the methods we’ve used to analyze them. We’ve chosen to use a mix of methods, both reading a selection of just over 8,100 in full and running the rest (67,500) through a variety of lexical tools. Some of these tools are probably familiar to you, like word clouds (a visualization of word frequency):

Some of these terms are expected, like “Romulus” or “Rome”. Others give us information about the videos’ audience, which we have then taken steps to analyze and will discuss in more detail in other posts: for example, a lot of these videos are being watched by students (“teacher”, “mr.”, “grade”), and the overwhelming response is positive (“like”, “love”, “lol”, and the emoji) – nice to see on a social media platform!

I’ve also embarked on a more in-depth analysis looking at the 150 most frequent terms and assigning them to categories and to feelings. So, using the example above, “teacher”, “mr.”, and “grade” are coded as “school”, while “like”, “love”, “lol”, and the emoji are coded as “positive”. This lets me assign a ranked score to individual terms across our roughly 67,500 comments, so that we can say (for example) commenters mention “school” 67% of the time, but “empire” only 50% of the time – these numbers are made up for the purposes of this example, but they do reflect the relative prominence of school terms in the data. These frequencies can then be used to discuss the tone of the discussions without having to read through tens of thousands of comments in depth.

In coming posts, I hope to share more of what my team and I have found from analyzing the comments of these videos!