Reading’s Feelings

In a recent article in Slate magazine, I argued that the kind of reading we do with books is significantly different from that of e-books. I based my observations on the tactile differences between these two technologies. I suggested that such somatic differences changed the way words assumed meaning on the page. There has been a a great deal of new scholarship on this topic, and it seemed worthwhile to bring this to bear on current debates about the future of reading. Since we have two different words for these experiences – reading and e-reading – it seemed reasonable to identify what those differences were.

This, it turns out, made readers very angry. I mean very angry. An article on the significance of feeling for reading apparently elicited strong feelings in readers.

To test just how angry, I performed what is called sentiment analysis using new quantitative techniques on these and other reactions to discussions about reading. Sentiment analysis involves data mining language to see whether the words people use to express their ideas exhibit higher than average levels of emotionality. Using the LIWC software package developed by James W. Pennebaker, Roger J. Booth, and Martha E. Francis, I found that for certain indicators reactions to my piece were more extreme than other sets of internet comments (which as we know tend toward the extreme in general).

For example, the comments that followed a reaction to the piece that were posted on the website TechDirt exhibited higher levels of anger (and emotionality in general) than comments posted in response to a piece in Slate on Obamacare. And while the comments directly appended to my piece scored slightly lower than the Obama article, they scored fifty-percent higher on certain indeces like the use of exclamation points, as in, “What utter nonsense!”, “Get off my lawn!”, or the exclamation squared, “Give me my Kindle anyday!!” (I can just imagine the comments now: “Exclamation points! What nonsense!”)

Reading, in other words, can make people as mad as Obamacare. Think about that.

Of course it could just be that I make people as mad as Obamacare. To test this hypothesis, I checked a piece in the Los Angeles Review of Books in which Stephen Marche argued that literature wasn’t data (and therefore shouldn’t be data mined).  His comments scored even higher than mine for emotionality and fifty-percent higher than the Obamacare comments in terms of anger. No wonder Marche doesn’t want us to data mine.

To see if this pattern held more broadly, I used a new “hedonometrics” data set developed by researchers at the University of Vermont to examine levels of happiness in sample sets of tweets. For each group, I collected 100,000 tweets that contained either the keywords “reading,” “books,” or “eread,” as well as a random sample for the sake of comparison. It turns out that when people talk about reading on twitter they are significantly happier than when they talk about other random topics. Yes, happier. Much happier. And it was no different for “books” or “ereading.” I should add that my article in Slate received close to 7,000 likes on Facebook in addition to 800 or so bitter comments.

Reading, it seems, can make us both happy where to buy Keppra tablets and mad. It is a deeply bipolar category.

This shouldn’t come as a surprise. The history of reading is littered with extreme emotions, especially during times of technological change. At the end of the eighteenth century, a period that witnessed a dramatic increase in the number of books, and one that has never since abated, writers began comparing reading to a “plague,” a “madness,” and a “flood” (as in the Biblical deluge). The Enlightenment German theologian, Johann Gottfried Hoche, even called reading the new “sodom.” On the other end of the spectrum, the Romantic essayist Leigh Hunt fantasized about marrying his books. The novelist Karl Immermann, author of the famed Adventures of Baron von Münchhausen, had this to say about himself as a child standing before his parents’ bookshelves: “The sheer sight of a book would set the afflicted child in a kind of quivering curiosity. The young creature lived and breathed only in print.” And we shouldn’t forget Edgar Allen Poe, whose murderous narrator of his short-story, “Berenice,” kills his cousin and then pours her teeth out on his desk – he was born in a library.

Changes to the material landscape of reading elicit strong emotions. We may say, repeatedly, that the medium doesn’t matter, that only the words are important. But our reactions tell us something very different. The material nature of reading, our physical experience of those words, makes a difference in how we respond to them.

When readers were told that what they did with their kindles or tablets or screens wasn’t “reading,” they lashed out, much like the long history of readers before them. To say what reading is or is not is to impugn one of our most personal possessions.

I, too, have been told that what I do isn’t reading. When presenting research that used the data mining techniques like the ones I mentioned above on works of literature, there was agreement among many in the audience that that’s not reading. That kind of thing may work for Target, but not for Jane Austen (or in my case Goethe).

The point is that reading computationally is reading, too, just like reading on a Kindle is reading and so is reading books. These are just a few of the many ways that we read. Reading is a diverse ecosystem of habits and objects. It seems worthwhile to keep these differences in mind, even to the point of using different words to designate them. Reading, whether in practice or in spirit, has never been universal. Our experiences of reading are never truly the same, whoever we are and however we do it. There are as many types of reading as there are readers. That’s the beauty of reading. The more words we have to describe it, the better.