Does Sentiment Drive Success?

Kurt Vonnegut once gave a lecture on storytelling, and explained that there were but six basic plots in literature. With this in mind, I found Matthew Jocker’s Sentiment Analysis R based program, Syuzhet. This program, using a clean, text only version of a work of literature, graphs sentiment values of a novel. The Syuzhet program uses its own lexicon of positive and negative words to create these graphs. Syuzhet has been around for quite a few years and has been updated many times to use context to ensure the most accurate graph is given for a novel. It is not infallible, but after running a few tests with the program I found it well suited for my needs, graphing a sarcastic novel correctly and highlighting the chapters with rising and falling action based only on sentiment.

Plot of Cat’s Cradle (red) and Slaughterhouse Five (blue)

This idea of the project has gone through a few evolutions as there have been quite a few set backs. Ultimately, my ideas on what literature to graph has changeg, and is still changing, based on availability of legal, public-use e-text. Originally, I wanted to graph the entire corpus of Kurt Vonnegut to prove his six plot theory within his own work. However finding any free text other than Cat’s Cradle and Slaughterhouse Five is impossible. I looked into scanning and OCR programs, but the too much time would be used to create my own e-text, edit the documents to ensure OCR recognized every character, and then format the text for analysis. After many variations, I decided to look into popular novels to compare their plot lines and success. I am still compiling this corpus as the e-text issue is impeding my progress, but once finalized my project will proceed quickly.

Ultimately, I want to utilize sentiment value technology to show trends in best selling or popular novels. Doing this would allow me to visualize and extrapolate a correlation between the two, which would open up an entirely new avenue of interpretation and understanding literature. However, this program and my project is one I plan to use for the foreseeable future as it has many outlets for research. One could look at the success of a certain sentiment plot type when examples of it were published and then map the success of the examples throughout major world events to see how history affects the success of that plot type. This type of sentiment analysis could also disprove theories that certain types of literature, romance for example, follow the same plot type every time.

Frankenstein Plotted

Taken out of context a rose by any other name is a novel. If there are only 6 formulas for the novel it would open up entire new ways of comparison across genre and historical categories. It could also help deepen our understanding of the history of novels and their success through the ages. By graphing my corpus and comparing the graphs I will not only begin to bolster the six plot argument but also compare the most successful plot types for public consumption. (511)


Booker, Christopher. The Seven Basic Plots: Why We Tell Stories. Bloomsbury, 2016.

Jockers, Matthew. “Jocker’s Blog.” Matthew L. Jockers, 16 Dec. 2017,

Moretti, Franco. “Patterns and Interpretation.” Literary Lab Pamphlet, Sept. 2015, doi:ISSN 2164-1757.

Shultes, Allison. “How One Digital Humanist Visualized the Shapes of 50,000 Novels.” Storybench, Northeastern University School of Journalism, 27 Apr. 2017,

Vonnegut, Kurt, and Daniel Simon. A Man Without a Country. Seven Stories Press, 2005.

Vonnegut, Kurt. Wampeters, Foma & Granfalloons (Opinions). The Dial Press, a Division of Random House, 2006.

Original Post

3 thoughts on “Does Sentiment Drive Success?

  1. First, it’d be interesting to learn more about this tool, Syuzhet, as I’m not familiar with the function and appreciate your analysis. Personally, I’d also be interested in this project as it’s intriguing to see a digital tool be able to analysis and categorize something that is normally labeled “emotional” (or sentimental). To be able to graph sentimental, and if proven accurate, thus graphing success of a novel, could prove to be instrumental for those with the goal of publishing their own work.
    With that said, I can see how difficult it must be to obtain digital text or translating written-to-digital text by hand to utilize this tool. Furthermore, do you have a criteria on how you’d select this text knowing you cannot use Vonnegut as of now? Naturally, you don’t want to use something outdated and old, nor should you use a recent, new novel on the chance it isn’t successful, or use a novel that has been overly-done or examined before like J.K. Rowling’s Harry Potter. Do you have something in mind now? Finally, I’d like to know more about the second set of graph-examples that use Frankenstein as I don’t see any indication of what the dotted line, red line, or faint-gray line are suppose to represent in terms of sentimental value and full (or normalized) narrative time.

  2. This is a fascinating project, though yet in its youth. I think you might consider limiting he scope of your approach so that you can clearly describe and analyze your results—for instance, “graphing a sarcastic novel correctly and highlighting the chapters with rising and falling action based only on sentiment“ is unclear to your reader, needing more detail and might use the tool to graph Frankenstein, limiting your analysis and description to that text as a first step. Vincent is correct as well to note that the graphs need more explanation—to do that thoroughly, you’ll need time and space (pages), so I strongly encourage you to work with one (or two related) texts. However, depending on your progress in developing your corpus and the way in which you frame your question, a larger pool of resources could work well—we can talk about it in conference. You will probably want more research into the novels/texts that you’re working with.

  3. Ally,
    What an interesting idea — I can’t wait to see how your project develops! Like Vincent, I am not familiar with Syuzhet, so am hesitant to offer much in terms of guidance with that. My main suggestion, however, would be to try to narrow your focus (this could be a potential thesis topic, perhaps?). One idea might be to limit your project to short stories, rather than novels? Or not trying to look at all six of Vonnegut’s basic plots, and instead identify one or two. That being said, my primary concern is that you haven’t explicitly stated how your project differs from Prof. Jockers. For instance, you could identify that you are building upon his model for the purposes of __ X___, or conversely, that you are differing with his conclusions because of __Y__ (whatever X and Y happen to be), and use your texts as evidence of that. Does that make sense and/or help? Again, I think that this idea has a great deal of potential, but the current scope might make it unwieldy for this our 571 final project.

Leave a Reply to T. Howe Cancel reply

Your email address will not be published. Required fields are marked *