You are on page 1of 1

Masterplot Clauses

One way to approach the construction of a plot in Plotto is to use what Cook called masterplot clauses. Your plot would move from an expository A clause that establishes the protagonist, to the B clause which begins and moves the conflict along, and finally to the C clause which finishes up and then resolves that conflict. These are organized in related categories and lists by Cook for ease of browsing through the clauses thematically, but any masterplot clause can be used in conjunction with another, using the A to B to C formula. Inevitably, you would then need to make decisions about more specific conflicts and characters, but this is one way to begin work in Plotto: with an idea of the large, overarching plot pieces. For instance, beginning with masterplot clauses here was my initial take on Bartleby (each clause is numbered for ease of reference for Plotto users): (A) 7. A Person of Ideals, (B) (56) Seeking to test the value of a mysterious communication and becoming involved in weird complexities, (C) (13) Comes finally to the blank wall of enigma. The next step in plot construction would be to examine conflicts (56) and (13) for potential usefulness, and to see which potential lead-up and carry-on conflicts you might want to use for your plot.

by Jonathan Armoza

Plotting in Reverse:

Topic Analysis: Plotto Masterplot Clauses of Bartleby

under supervision of Prof. Stfan Sinclair McGill University, Fall 2013

&
The Master Book of All Plots Bartleby, the Scrivener
Why combine Plotto and Topic Modeling?
As evidenced by Cooks work, theorists like Polti, folklorists like Stith, Thompson, and Propp, or in antiquity with Aristotle, there have long existed efforts to model stories. One modern approach includes programmatic topic modeling used in natural language processing. These algorithms use advanced mathematical approaches of textual collocation and frequency analysis. However, as literary scholar Lisa Rhody points out, this modeling relies on the notion that language in documents being analyzed is intended to be unambiguous as possible. This is clearly not the case in most literary texts - in fact, such texts often revel in ambiguity. Understandings gained by close readings of texts may aid in contextualizing topics derived from topic modeling algorithms. For the case of Plotto, in terms of this experiment, its almost immaterial as to whether or not Cooks model is correct or seen as reductive. By utilizing a textual modeling approach that injects understandings gained by human-based readings, data then introduced to computer-based approaches of text analysis will have then been prefigured by human understandings of that text.

What is Plotto and Why Use It?


William Wallace Cook, a prolific American dime novelist, published Plotto: The Master Book of All Plots in 1928. It was intended to help fellow writers easily construct the skeleton of plots in systematic fashion. Once what Cook called the masterplot was built, it was up to the writer to flesh out the actual story. In the three boxes on the left you can find descriptions of the primary components of Cooks plot model: masterplot clauses, conflict situations, and character symbols. Either one of the three components could be used in any order to construct a masterplot, depending on initial ideas the writer had before opening Plotto to begin work. Used by many writers over the century, hobbyists and professionals - including, almost mythically, Alfred Hitchcock, Plotto was re-published in 2011 by Tin House Books, with a foreword by literary detective Paul Collins. Looking over Plotto, I quickly realized that it was a text primed for a digital approach due to Cooks systematic structure. If a text could be modeled in a way that reflect the aims of a reader interpreting that text, could this model be made programmatic? The first step in that larger project, would be an artisanal approach - to use Plotto, but in reverse! Here, I have attempted to describe and model a text I know well: Herman Melvilles Bartleby, the Scrivener: A Story of Wall-Street.

Conflict Situations
The next way to approach the construction of a plot in Plotto is to use one or more of Cooks conflict situations. He theorized the existence of 1462 different conflict situations that could be combined to form the overall or series of conflicts involved in a masterplot. Conflicts are often complex, and Cook allows for this complexity in a few ways. They are categorized hierachically in this order from top-down: B clause number, conflict situation number, variation, and subclause. Each conflict listed also has a set of lead-up conflicts and carry-on conflicts to choose from. The conflict itself is listed as a body of text, potentially subdivided into subclauses and features symbols for character names. For example, conflict 1402, which I identified as the primary conflict for Bartleby looks like this in Plottos conflict index: (701 tr A & A-2) (705 ch A to A-2) (1021b) A, his curiosity aroused by the mysterious actions of his friend, A-2, decides to investigate the cause (704) (807) (841) (1334c) The numbers in parentheses are the lead-up and carry-on conflicts. Characters, represented by character symbols A and A-2 above are also often changed or transposed for the purpose of a suggested conflict, thus ch and tr. Symbols *, **, *** are also used to refer to conflict subclauses that can be used independently of the conflict situation within which they sit. Cook made sure that the text, the order of conflict situations, and even the characters are meant only to be suggestions to the writer while they construct their plot - the writer should feel free to shape them with their imagination as they see fit.

Begin
Conflict Situations

Method and Analysis


Plotto was designed to construct plots, and not necessarily describe them. This required a bit of inventiveness on my part. Though I kept more complex literary criticisms of Bartleby in mind while considering the plot of Bartleby, I wanted to keep as close to a literal reading of it as possible, if for the sake of this first experiment, and the literal nature of topic modeling itself. I went about using each Plotto component to this end, outlining the work via masterplot clauses and conflict situations, which, when combined provided two different divisions of the work. I encoded Bartleby with a custom Plotto-TEI schema in CATMA and separated the clauses and conflicts into separate documents, which I then ran together using a basic ten topic/optimized interval LDA topic modeling in MALLET. The results above and below show the topic weights in each clause or conflict. Though the bar graphs do not accurately display the length of each section, you can see where larger sections have a tendency to contain multiple topics of significant proportion. This is indicative of the length of the text, but also perhaps of my too-broad assumptions about a particular section, the beginning and ends of which were left to my interpretation. The labels of the topics were created in such a way that attempts to contextualize MALLETs topics: by looking to sections of text where topics were proportionally more weighted than others - in other words, given my knowledge of Bartleby, why does this grouping of words exist more notably within the clause or conflict section I have identified? Increased stratification of a section, indicates to me that my divisions need further refining. The most successful clause/conflict identification may be that which seems most boring on the graphs. Though small in sampling and thus less in topic possibilities, the 600 word conclusion which aligns clause C and conflicts 1411/1355 provides an evocative topic list I have labelled, Bartleby, humanity, and death. That said, the results do provide a fascinating confirmation of a very literal reading of the Bartleby story, where the exposition of the lawyers office sits, where Bartleby begins his preference not to, and onward.

Masterplot Clauses

Character Symbols

Problems and Future Work


As one can see in the results above, there are key points of the resultant data that need refining, and approaches to it that bear scrutiny. By prefiguring the data via Plotto modeling before it was fed into MALLET, I was quite conscious that any interpretation of outputted topics could be subject to confirmation bias since I had a hand in manipulating the data before it was analyzed. One relief to that possibility is the notion that a predetermined thematic occurring at a textual location, once modeled further by programmatic means, may showcase aspects of that location that differ from those predeterminations. On the other hand, one of the key criticisms of potential humanities usages of topic modeling is the tendency to read too much into its results - so we are not so far afield from those dangers even without my pre-intervention. As noted above, the topic distribution across Plotto components indicates further refining of the methodology is needed, and critique of that methodology should follow in suit. The larger project would then be to programmatize Plotto plot analysis, first by making the text more digitally accessible, and then attempting algorithm-oriented approaches to modeling in this way, or other related ways. I believe that more traditional interventions of digital analysis such as this hold at least two benefits: they comfort institutional fears of quantification, and they showcase the possibilities of a more humanistic computing of texts. The drama remains in the bits. And the plot thickens.

Masterplot
Bartleby
by Herman Melville

Topic Analysis: Plotto Conflict Situations of Bartleby Character Symbols


The last way to approach the construction of a plot in Plotto is examine possible combinations of characters, or, as notated within the book: character symbols. Characters are represented by letter and number sequences like A-2, BX, or GCH-B - male friend of a male protagonist, a mysterious female person, or grandchild of a female protagonist, respectively. There are over 50 character symbols. At the back of a Plotto is an entire guide consisting of listings of character combinations for plots, lists of suggestions for interactions of those combinations, which then lead back to suggested conflict situation numbers. One might begin with this step, or go to it to provide further insight on potential conflict situations. For instance, when conflict (56) did not prove fruitful for lead-up and carry-on conflicts for Bartleby, I turned to this method to determine the characters, and other potentially appropriate conflict situations. Using Plottos Index to Classification by Symbols, I found the combination of character A - male protagonist (the lawyer) - and character AX - man of mystery (Bartleby). Plotto then suggests that I take a look at several conflicts, the most appropriate of which I found to be conflict 1393, subclause two: A tries to solve the mystery of a stranger, AX ** After this, I examined conflict 1393s lead-up and carry-on conflicts 1024, 1029, 1388, 1404 and 1405 to further describe Bartleby.

<masterplot_clause>

Plotto XML

Bartleby Masterplot Clause A Bartleby Conflict Situation 822 Bartleby Masterplot Clause C

MALLET Topic Modeler

</masterplot_clause> <conflict_situation>

Latent Dirichlet Allocation

</conflict_situation>

You might also like