English 738T, Spring 2015
Header image

Gothic Novels, Chronologically woodchipped

Posted by Allison Wyss on Saturday, April 14th, 2012 at 12:47 pm

To study how the Gothic novel might have changed over time, I divided our set of texts into four chronologically based periods.

Late 1700s: The Mysteries of Udolpho, The Monk, Caleb Williams

Early 1800s: Northanger Abbey, Frankenstein, The Vampyre

Mid 1800s: Varney the Vampyre, Wuthering Heights, Jane Eyre

Late 1800s: The Beetle, The Time Machine, Dracula, The Picture of Dorian Grey, Dr. Jekyll and Mr. Hyde, The Mystery of Edwin Drood, Carmilla

(Due to availability of texts, the last group used more than twice as many as the first. I did some test runs with a subset of the late 1800′s to see how much the large/small group would affect my results and it did not seem significant for the purposes of my analysis. I would have liked to increase size of all four groups, but I didn’t have the texts to do it.)

Splash Patterns

When I divided the gothic novels into sub-periods, I was hoping that woodchipper would find some interesting changes in the splash patterns. Based on some of Moretti’s ideas about a new genre starting out a bit shaky, then stabilizing, then branching out, I expected more variation in the earliest and latest sets than in the middle sets. I had hoped the woodchipper would prove that the early period was when the gothic was still figuring itself out (the splash patterns would be indistinct, fewer strong lines, less overlap between novels), that the middle periods were where it coalesced (strong lines, lots of overlap), and during the later period, I hoped the woodchipper would find evidence of the different subgenres diverging (strong lines, but going in different directions).

I could find some evidence to support my theory, but it was weak, and maybe I was looking too hard for it. The earliest patterns were somewhat fuzzy, but not remarkably so. The lines did get a little stronger and the overlap increased between the late 1700s and the early 1800s, but it wasn’t particularly dramatic. Then the middle 1800s definitely had the most overlapping patterns, but their lines were not noticeably sharper. I found an instance or two of nice diverging lines in the late 1800s, but also some splash patterns that were just as well aligned as earlier periods.

The following are the runs that most fit my hypothesis. Other runs don’t fit as well.

Early 1700′s: Not much definition.

Early 1800′s: Getting a little bit sharper.

Middle 1800′s: Maybe sharper yet?

Late 1800′s: Clear divergences

So I tried a more generalized chronological run, with larger groups—pre 1850 and post 1850. I didn’t think the pictures would be very good with so many novels occupying the same space at once, but I was pleasantly surprised. I could see pretty clearly that the early set overlaps much better, but has weaker lines, than the later set, which has much sharper lines and stronger divergences.

The samples I’ve chosen this time are quite typical of the set. The only problem is the splash pattern is much more difficult to see, as the titles overlap it.

Before 1850: Pretty fuzzy, but overlapping well

After 1850: Sharper but more divergent

So my larger groupings seem to illustrate the way the Gothic genre may have started out fuzzy, but then crystalized, just in time to diverge, as Moretti suggests genres behave in “Trees.” However, woodchipper was not able, in my runs, to separate the crystallization step from the diverging step.


I was also interested in the topics that the woodchipper found and had some luck finding patterns with my four chronologically based sub-periods of the Gothic novel. First, the most consistent topic, which appears at about the same rate in all periods is the one I’m calling “Good Conduct.” Some other trends emerge across time. For instance, I found a handful of topics that start out strong but fade through time. “Crime” is extremely prevalent in late 1700s, but then disappears after that. “Grief,” “Nature” and “Character” are very prevalent in both the late 1700s and early 1800s, but disappear as topics in the mid and late 1800s.  “Hope,” “Escape,” and “Goodness” are all most common in the earliest parts of the period, fade slowly, and are completely gone by the time of the late 1800’s.

Other topics move in the opposite direction, to become stronger over time. “Doubt” occurs throughout the period, but is not all that prevalent in the late 1700s, early and middle 1800s. It swells remarkable in the late 1800s. Both “Faces” and “Aristocracy” rise to prominence in middle and late 1800s, but are not seen before then. “Knowledge” first appears (just once) in mid-1800’s, but then becomes very prevalent in late 1800s. “House Interior” belongs with this group, gaining momentum in later years. It occurs once in the late 1700s, disappears in the early 1800s, but become extremely strong in the middle and late 1800s.

“Time” is the only topic in which I can see a marked ebb and flow. It first appears in the early 1800s, stays prevalent in the middle 1800’s and sticks around, but is less prevalent in the late 1800s.

The topic of “Existence” is very prevalent in the early 1800’s and occurs only there. I am sure, however, that this is due to one novel’s obsession with the topic (Frankenstein). My samples aren’t large enough to stop one novel from throwing the results and so I’m skeptical of other one-period topics, such as “Hearing,” and “Sky”(only middle 1800s), “House exteriors,” “Mind,” and “Sleep” (only late 1800s).

So then, what can I make of the way certain topics build over time and others fade? The topics that start strong then fade are “Crime,” “Nature,” “Character,” “Hope,” “Escape,” and “Goodness.” The topics that build up later in the period are “Doubt,” “Faces,” “Aristocracy,” “Knowledge,” and “House Interior.”

An easy observation is the way “Hope” gives way to “Doubt” as the two can be seen as opposites. Perhaps in some ways the gothic becomes less optimistic through the years.

These results also suggest to me that the early Gothic was more concerned with the extremes of human behavior. Think of the powerful extremes of “Nature.” Then think about the way a person’s “character” might move between “Crime” and “Goodness.” In contrast, the later topics appear more ambiguous to me. Maybe the later novelists are making more nuanced arguments. There may be a longing for certainty or “Knowledge,” but “Doubt” suggests there no real way to get it. Then I think about the way you might use  “Faces” and “House Interiors” to find or impart knowledge—it would be an interpretive act. It makes me think of a less imperious, more sophisticated search for truth—perhaps even understanding it as less certain and more interpretive.

This idea appeals to me–that of gothic novels moving, over time, from certain judgment of good and evil toward a more nuanced and complicated approach to knowledge and understanding. The woodchipper has not proved this movement, of course. But it suggests a direction for further study.

You can follow any responses to this entry through the RSS 2.0 You can leave a response, or trackback.

10 Responses

  • Charity Hancock says:

    Hey Allie – Did you get your category labels from the 100 topics Travis sent, or did you ascribe them yourself? For instance, what did you call that “Felt, Made, Conduct, Received, Heart” category?

  • Great, thorough post–this was fun to read. I appreciated how thoughtful you were about what you expected to see and what you actually saw–you did a great job of covering your thought process from conceiving the experiment to running it to considering its theoretical implications.
    I’d be interested in hearing more about your definition of the Gothic–Caleb Williams feels like an outlier in the group, but from the order in which the layers are displayed in your image, I couldn’t tell how similar it was to the other two texts from that period. I also wonder whether you tested these against any control groups of non-Gothic novels from the same time period (and what you found if you did).

    • Allison Wyss says:

      Thank you.

      I want to show you what that image looks like with the layers in another order, but I’m not sure how to put an image in a reply. Hmm… Well, Caleb does match up pretty well to the other two in that instance. (The order of those layers, by the way, is a rather frustrating aspect of woodchipper. To change it, you have to go back to the start, empty your list and add them in a different order.)

      Honestly, I didn’t put a lot of thought into Caleb being an outlier. We labeled it Gothic at the start of the project and it fit the time frame I wanted. But I did run it, just now, against some of the texts we’ve been calling “most gothic” and it fits pretty well. And in my earlier runs of many texts at once, to establish how this whole thing was going to work, Caleb did not stand out in any way that I noticed.

      I haven’t done very much in the way of testing the Gothic novels against a control group because we’ve had some delays in getting our control texts loaded. We probably started with too small a number, and then had some bad luck with the texts.

  • Neil says:

    Very interesting hypothesis, Allie. Larger data samples would be crucial to make the case. Also, being able to run a large dataset of randomly selected British novels through Woodchipper to see what topics emerged would also be useful in seeing just how generically specific the topics you’re finding actually are.

  • Clifford Hichar says:

    Your study is really quite fascinating. I wonder if, based on prevalence of certain words and themes during certain periods you could then draw conclusions about what was happening in the world outside of literature as well? We’ve seen in so many instances how writers turned to fiction to express their concerns with society (I think of Caleb Williams and the conversation it enters into about the French Revolution and concerns of the period, also Dracula and the fears that book expressed about the aristocracy, immigrants, women, etc). It would be fascinating if in some way you could use the woodchipper to look not just at trends in literature, but see how literature enters into the larger concerns of a people at a particular moment of history. Perhaps even see what concerns radiate not just in one genre, but across multiple genres at a particular period? All of what you have found is quite exciting.

  • Pingback: Data Analysis Group Post - Technoromanticism

  • Pingback: lustro piotrków

Leave a Reply