Modern readers: What can social media tell us about contemporary literature?
“In the past, when someone read a book they might have left nothing but a couple of lines in the margins and that’s all we know about what they thought of it,” explains Dr Anouk Lang, a lecturer in Digital Humanities at the University of Edinburgh, “For the longest time, if you were doing research into reading it was incredibly difficult to find resources. Historically you had to go through someone’s library and, if you were lucky, you might find margin notes or maybe some comments in a letter.”
It’s impossible to say what a writer such as James Joyce would have made of the modern phenomena of social media. Perhaps he would have disdained the relentless torrent of tweets. Then again, perhaps an author famous for piecing together fragments of conversation into rich narrative strands might have found Twitter an incredible resource for inspiration.
What we do know is that, given humanity’s freshly discovered propensity for sharing their thoughts online, scholars of the future will have a vastly greater trove of information to plunder when they attempt to contextualise the writers of today and understand the impact that they had upon publication.
“Scholars have done amazing work looking at historical material on people's relationship with books,” Lang says, “But I work on the contemporary period and there’s a vast amount of online data that’s available.”
In her current research project, Developing Methods for Analysing and Evaluating Literary Engagement in Digital Contexts, Lang is looking at ways to capture people's reactions to reading from social media websites like Twitter and analyse them in a scholarly manner. Her work is being funded by the Arts and Humanities Research Council (AHRC), and involves taking tools that may have been developed for different purposes and applying them specifically to the field of reading studies.
“I'm looking at tools from the digital humanities, and I'm trying to use what other folks have achieved and bring those to the study of reading,” she says, “There are some superb scholars in this field, but they don't always have the technical knowledge to manage the software that can help them analyse large datasets. There's a tremendous amount of data out there in terms of reader habits that businesses like Amazon are using, and researchers should be taking advantage of that too.”
In the course of her research, Lang has focussed her study on two online communities. One is LibraryThing, an internet group with nearly two million members specifically designed for discussion of books, and the microblogging site Twitter, where people share views about anything and everything in 140 characters or less.
Gathering these datasets, however, means using tools such as the open source Twitter Archiving Google Spreadsheet, a script by Martin Hawksey which automates searches of the social platform and saves Tweets to an online spreadsheet. Lang says that she's building on the work of others in related fields.
“The social scientists are the people who have been on top of this for ages,” Lang says, giving the example of the ESRC-funded project Collaborative Online Social Media Observatory (COSMOS). “They’ve been very good at gathering the data and finding the best tools to analyse it. It's somewhat newer to us in the humanities.”
Dr Edmund King of The Open University is a Research Associate attached to the Reading Experience Database, an archive of materials documenting readers' reactions to literature that stretches from 1450 to 1945.
“Books, or newspapers, or magazines, or broadsheets don't read themselves,” King explains, “Readers have their own agency, their own matrices of pre-existing ideas and prejudices that shape the consumption of new texts. We can't really understand the significance of a piece of writing without trying to get some idea of its impact on individual readers.”
King believes that studying reading through social media will potentially reveal a more representative range of reactions than historically, when typically only the rich kept diaries or made margin notes.
Just as important as the core research, then, is Lang’s documenting of the tools she’s using to analyse social media and laying the groundwork for non-technical researchers to follow.
“I gathered data over eight months,” says Lang, “And now that phase of the project is over I’ve been analysing them and finding out what kind of software works best to look at them. The reason these kind of tools are important is that if you were to sit down and read something like ten thousand tweets as a human reader, it would take you a long time, and you’d probably find it difficult to notice many of the more subtle patterns in it.”
In order to find these kinds of patterns in the data, Lang is investigating tools which are already commonplace in the world of commerce. Big business takes big data seriously, and Twitter’s “firehose” is routinely subjected to keyword and sentiment analysis by marketing departments keen to understand how a brand is perceived online and react to negative comments quickly. For example, mobile network operator T-Mobile responds to 86% of queries and complaints via social media accounts, whether or not they are directly addressed to the firm.
“You can slice Twitter data in a lot of different ways,” Lang explains, “You can look at it chronologically or geographically, for example, or put it through software used by corpus linguists to look for emotionally charged words or patterns of keywords. Topic modelling is also useful, as it let us group comments by the topics people are talking about.”
This map shows tweets with geolocation metadata that contain the terms 'alice' and 'munro', gathered around the time that Munro won the Nobel Prize for Literature in 2013. The dataset is small (694 tweets) but some interesting patterns can be seen. As well as a predictable concentration of tweets in the populous cities of eastern Canada, the presence of tweets in Latin American and western European countries suggests that Munro has a significant readership in translation in those countries as well.
While Lang is the first to admit that there are problems with using data from services like Twitter and LibraryThing, for instance the fact that their users are unlikely to be representative of the population as a whole, she’s already finding interesting types of behaviour that differentiate the two. For example, she says, looking at the way the two groups commented on Joseph Boyden’s novel The Orenda, Twitter users were more likely to talk about celebrities who were connected with the book. LibraryThing discussions, meanwhile, were more involved with the content and themes.
Lang continues to prepare more observations for publication, but already the future history of reading is in jeopardy. The fact that Twitter only allows researchers to gather a limited number of tweets through regular searches before charging for access already hampers her work and leads to a sense of urgency around collecting this kind of data for future reference. Dr King agrees, and he also warns of the danger that “bit rot” — the degradation of data stored in digital formats — poses for scholars who work with born-digital materials.
One thing that Lang is sure of, however, is that there is relevance buried in all those tweets and they are worth saving. “Many of the tweets from the Arab Spring are already lost,” she says. “In a hundred years, scholars studying reading in the early twenty-first century will be grateful that we saved a little corner of the internet for them.”
Article by Adam Oxford