Erik Simpson is Professor of English and Samuel R. and Marie-Louise Rosenthal Professor of Humanities at Grinnell College. He is the author of two books:
Hannah L. P. Brown is a Counseling Psychology Graduate Student at the University of Florida whose research combines feminist and queer theories with practical counseling questions.
Grinnell College
Olly Shortell is a graduate student of Clinical Mental Health Counseling at Prescott College.
James Jaehoon Lee, Ph.D is the Director of the Digital Scholarship Center and Associate Professor of Digital Humanities at the University of Cincinnati. His research and teaching focus on the areas of digital humanities, machine learning and data mining techniques on historical text and image archives, social network analysis, and data visualization. Much of his work has used machine learning methods applied to large text corpora and other unstructured datasets to rethink what it means to perform historical analysis. His research also investigates ways to visualize the results of machine learning algorithms in a human-interpretable way that enables non-technical audiences to glean useful information from the data. More recently, his research has branched out into fascinating collaborations applying digital humanities methods with partners in biomedical informatics, law, design, and archeology. He is the author of
This is the source
This paper uses network theory and network analysis to propose a new approach to analyzing cross-dressing in Shakespearean drama, specifically the key questions driving much of the scholarship on that topic in recent decades: What kind of disruption to the social order did cross-dressing represent in Early Modern England, and what did it mean to shift that disruption from the street to the stage? We know that the laws and customs of the age emphasized clothing that matched the outward appearances of people to their places in the social order. Network analysis allows us to analyze how characters become gendered through a non-linear, non-hierarchical lattice of social relationships. As the characters interact and social relationships change, the individual’s gendered subjectivity also transforms. We find that cross-dressing protagonists propel themselves from isolated social worlds into a complex web of relationships through cross-dressing, and that entry into sociability follows predictable patterns. By following those patterns, the characters combine roles--the broker and the heroine--that are normally separate in Shakespearean comic plots, creating a hybrid type that becomes visible through network analysis.
Using network theory to analyze cross-dressing in Shakesperean drama.
This paper uses network theory and network analysis to propose a new approach
to analyzing cross-dressing in Shakespearean drama, specifically the key
questions driving much of the scholarship on that topic in recent decades:
What kind of disruption to the social order did cross-dressing represent in
Early Modern England, and what did it mean to shift that disruption from
the street to the stage? We know that the laws and customs of the age
emphasized clothing that matched the outward appearances of people to their
places in the social order. A letter from John Chamberlain dated 1620
exhibits the Jacobean stakes of conventionally gendered dress: Yesterday the bishop of London called together all
his clergie about this towne, and told them he had expresse
commandment from the King to will them to inveigh vehemently
against the insolencie of our women, and theyre wearing of brode
brimed hats, pointed dublets, theyre haire cut short or
shorne...the truth is the world is very much out of order.
amounted to a declaration of war,
and the
results of that declaration revealed that women in
men’s clothing had assumed threatening enough proportions in the
conservative mind to be singled out in a conscientious and thorough
attempt to eliminate the style from social life
The conservatives of the time attempted, in other words, to enact the kind of restoration of propriety that Stephen Greenblatt perceives at the end of Shakespeare’s
[t]he sexes are sorted out, correctly paired, and dismissed to bliss — or will be as soon as Viola changes her clothes
properly feminine subjectivityeven while dressed as men
teleologically male
Other scholars emphasize the extent to which cross-dressing fundamentally
troubled conventional gender roles, especially in a theatrical world that
involved all-male acting companies. As Phyllis Rackin puts it, the boy actress's body was male, while the character he
portrayed was female. Thus inverting the offstage associations, stage
illusion radically subverted the gender divisions of the Elizabethan
world
category crisis
for the very
concept of gender In the place of the law of heterosexual
coherence, we see sex and gender denaturalized by means of a
performance which avows their distinctness and dramatizes the cultural
mechanism of their fabricated unity
In the context of Renaissance drama, the perceptions of cross-dressing as fundamentally conservative or disruptive, as opposed as they seem, share a good deal of common ground. Both views see the potential disruption arising from all-male casting and the ways in which female characters cross boundaries while dressed as men, for example. Much of the disagreement arises from different assessments of the power of comic resolutions to contain that disruption. When the structure of comedy involves a movement from disruption to resolution, the difference between disruptive radicalism and reactionary essentialism can amount to a question of emphasis. Can the resolution contain the disruption? How much of the conventional hierarchy does a play restore when nominally female characters assume the clothing, legal status, and dispossessions of wives in patriarchal marriage?
This paper seeks, if not to break that impasse, at least to shift the terms of the conversation away from the linear placement of Early Modern theatrical cross-dressing on a scale from radical disruption to the conservative reinforcement of norms. Network analysis allows us to analyze how characters become gendered through a non-linear, non-hierarchical lattice of social relationships. As the characters interact and social relationships change, the individual’s gendered subjectivity also transforms. We find that cross-dressing protagonists propel themselves from isolated social worlds into a complex web of relationships through cross-dressing, and that entry into sociability follows predictable patterns. By following those patterns, the characters combine roles — the broker and the heroine — that are normally separate in Shakespearean comic plots, creating a hybrid type that becomes visible through network analysis.
This paper seeks, if not to break the impasse, at least to shift the terms of the conversation away from the linear placement of Early Modern theatrical cross-dressing on a scale from radical disruption to the conservative reinforcement of norms. Network analysis allows us to analyze how characters become gendered through a non-linear, non-hierarchical lattice of social relationships. As the characters interact and social relationships change, the individual’s gendered subjectivity also transforms. We find that cross-dressing protagonists propel themselves from isolated social worlds into a complex web of relationships through cross-dressing, and that entry into sociability follows predictable patterns. By following those patterns, the characters combine roles — the broker and the heroine — that are normally separate in Shakespearean comic plots, creating a hybrid type that becomes visible through network analysis.
To perform our analysis, we generated network visualizations of each of Shakespeare’s plays including a crossdressing heroine using the Python library NetworkX. These visualizations, hosted at https://rnlp.net/shakespeare/, depict the relationships among characters scene-by-scene, shown as networks of nodes with edges connecting them. Across the play’s scenes, each node represents a single character and the edges linking nodes represent verbal interactions between the characters. Edge length represents the frequency of interactions between characters; i.e., characters that speak more lines to each other more frequently have shorter edges between them. Node size corresponds to the number of lines spoken by a given character in the play or scene in question. This means that the more lines a character utters, the larger the radius of their respective node is. For example, if we were interested in analyzing the visualization for Hamlet, Act 1, Scene 5 (fig. X), we see there are four nodes, one each for Hamlet, Ghost, Horatio, and Marcellus. Hamlet and the Ghost’s nodes are considerably larger than Horatio or Marcellus’s because Hamlet and the Ghost speak far more lines than either of the other two characters in the scene. Edges connect those characters that speak to one another; i.e. Hamlet is linked to Ghost, Horatio, and Marcellus, because he speaks to all of them while Ghost is linked only to Hamlet because he speaks to Hamlet and only Hamlet.
We modeled our uses of NetworkX to measure and parse speaking relationships
among Shakespeare characters after those expressed in Lee (2017). As such, rather than merely count
exchanges between characters to create a simple character map, our
visualizations include weighted relationships based on volume as well as
frequency of speech — that is, we assign greater importance to longer
utterances — ensuring a more comprehensive view of Shakespeare’s language
than was typical before 2017. Building upon former methods, we also
measured relationships using other social network analysis metrics to
analyze our results
Network visualizations have proven useful in many academic disciplines in recent years because they rearrange linear data in a two-dimensional space, foregrounding formerly latent connections as a visible structure. In literary studies and other areas of the humanities, where our data (usually language) is primarily observed one-dimensionally (through reading), it can be difficult to accurately observe trends and large scale relationships, let alone reenvision primary sources among so many other competing ideas in the academy. This is especially pertinent to scholars interested in analyzing source data in widely read and written about subdisciplines like Shakespeare Studies. It is our assertion that network visualizations are one way to engage with Shakespeare Studies and Queer Theory in an innovative way while also remaining true to the text.
In another sense, reading Shakespeare as a series of networks also serves to
circumvent presupposed organizing forms among characters. Because our
networks are spatially arranged by machine, these visualizations represent
the relationships among characters based only on the amount and
directionality of their language. This means that relationships in the
network are independent of qualitative taxonomies like social hierarchy,
marriage, friendship, or employment
touchstonebetween characters in the networks, he is mathematically displayed as such neither because of his position as the fool, nor because of his apt name, but simply because of how much and to how many people he speaks. From there, we might enrich our reading of Touchstone with the qualitative facts of his character, but they do not inform the network at the generative end. This revisioning of Shakespeare’s plays as two-dimensional relative spaces based on direction and amount of speech was pivotal to our analysis because it shifts focus away from conventional organizing structures that we may take for granted and onto how characters actually function in the text itself.
While this project began with a general interest in the ways in which cross-dressed heroines are relationally positioned in Shakespeare’s gender-bending plays, our research did not initially start with a hypothesis or set of assumptions that we hoped to evidence with these network diagrams. Rather, we used the network visualizations as a way to perform exploratory data analysis of Shakespeare’s comedies in a different light. Ultimately, our findings confirmed to a certain extent existing understandings of the texts, as well as some conclusions drawn by other scholars interested in this area. We viewed such confirmation as valuable because it reinforced existing qualitative interpretations with a different mode of evidence gleaned from an independent method - network analysis rather than reading. Often however, as we will demonstrate in the following sections, our findings complicate, and at times challenge previous scholarly interpretations of Shakespeare’s comedies.
Shakespeare’s comedies have been defined by inversions
of the norms of behavior,
which can temporarily provide the exhilarating sense of freedom which transgression
affords
locked up in awe and respect,
to a heightened awareness of the relation between man and nature
Criticism on the comedies has more recently moved away from the binary poles
of order and its inversion within the play’s plot, toward a more
historically nuanced understanding of cross-dressing on the stage as a
disruptive failure of gender representation. Specifically, Tracey Sedinger
has suggested that cross-dressing on stage in the early modern context
rejects a definition of sexuality in relation to embodiment or any object
Our network visualizations show that when female characters don men’s clothing, the social network constructed between characters changes. The individual characters are thrust into dynamic webs of interactions, but more generally we see the proliferation of dense and tightly clustered communities. Those communities allow the protagonists to meet a variety of characters and ultimately participate in a social world that is at once larger and more complex. Density, which measures how many interactions each character has with the other characters in the scene, increases significantly after the female protagonist cross-dresses. The simple bar graph in Figure 1 shows the typical jump in density of interactions once the protagonist cross-dresses.
Figure 2 visualizes the interactions in each play that create the tendencies summarized in Figure 1. In Figure 2, for each play, we visualize four scenes from each play, in chronological sequence. In each case, the first scene (shown at the top of the four-scene sequence) takes place before the moment of cross-dressing, and the other three afterwards. These are only selected scenes, but they illustrate the kinds of increasing density of social networks that create the overall effects captured in our analysis.
Looking at these patterns across the five plays, we see visual patterns emerging. The networks before cross-dressing are frequently structured like a barbell or a hub and spoke; the secondary characters are companions to the protagonists and do not interact with anyone but their masters and mistresses. That is, the networks cluster tightly around the major characters, who are isolated from one another. By contrast, when dressed as men, the female characters are able to join tightly bound networks in which many characters interact with each other. The high interconnectedness of these groups can also be measured by the clustering coefficient of these network, which increases in the plays after the female characters dress as men.
Clustering can be explained with the following example: if Viola knows Olivia and Olivia knows Sir Toby, then the likelihood that Viola and Sir Toby also interact is high. These complex social circles expand to include characters of different classes and professions: Julia is captured by outlaws, Rosalind barters with a shepherd, Portia dresses as a doctor of law and orchestrates a court scene, and Imogen befriends both shepherds and Roman soldiers. Thus dressing as men allows these women to escape the confines of these dialogues and participate with a larger community. Through escaping their isolation the female characters are therefore embedded in the larger network and able to become participants in the social drama of the plays.
The exception to this trend is Rosalind from
(3.3.433-435)I would cure you if you would but call me Rosalind and come every day to my cote and woo me.
These networks suggest that that all five of the cross-dressing protagonists
experience a change in sociality: when they put on men’s clothing they are
no longer limited to one-on-one dialogues, but instead enter into a dense
and interwoven community that allows for the drama to unfold. King James I
recognized these effects that cross-dressing could have on the larger
society when he expressed his fear that cross-dressing would create a
world very much out of order.
However,
Shakespeare paints a different picture of its influence: rather than
creating chaos, cross-dressing transforms a sparse and isolated world into
one that multiplies social interactions.
Drawing on theory of corporate organizations, Richard Jean So and Hoyt Long
argue for the importance of brokerage functions in networks: when a network
has two subgroups divided by a gap, the broker
fills the hole in the
network and thereby enables connections among the other people who had been
separated A structural hole is a
potentially valuable context for action, brokerage is the action of
coordinating across the hole with bridges between people on opposite
sides of the hole, and network entrepreneurs, or brokers, are the
people who build the bridges
On this point, the finding of the network analysis is intuitive and consistent with textual evidence from the plays. For example, in
(II. V.95-96)Buy thou the cottage, pasture, and the flock, And thou shalt have to pay for it of us
usthat Rosalind represents. Rosalind likewise manages Silvius and Phoebe’s relationship:
Cry the man mercy, love him...so take her to thee shepherd(V.V.491). Here, Rosalind addresses one party, then the other, ultimately joining the couple together. In
I do bid thee beg my life, good lad(V.V. 490). When she complies, Imogen acts as a bridge between Lucius and Cymbeline. Likewise, in
And now am I, unhappy messenger(V.V. 116). And in
All the occurrence of my fortune since/ Hath been between this lady and this lord(V.I. 270-271). When the visual networks show that the cross-dressers are brokers, they therefore highlight a role that was already embedded in the text.
The less intuitive aspect of this brokerage function lies in the brokers’
centrality to their respective plays. The broker is more typically a
marginal figure whose importance lies in creating connections whose
importance outweighs the broker’s own. As So and Long put it, One’s eyes naturally focus upon the massive
sun
or smaller stars
that occupy our network visualizations, yet
somewhat less visible broker figures significantly populate the
literary field
In this visualization, to compare the degrees and lines across plays, the numbers are represented as a percentage of the total degrees and lines of each play. Within a given play, we can use the unadjusted number of degrees to illustrate the connectedness of the characters relative to one another, as shown in Figure 5.
All of the cross-dressing protagonists are among the three most connected characters (in Imogen’s case, tied for third) in their respective plays. In Figure 6, network visualizations of scenes illustrate the connections that constitute each protagonist’s network.
In the network of each play, the most connected characters have the smallest distances to other characters; put more technically, they minimize the sum of all distances to all other vertices. In the visualizations, therefore, the nodes of the most connected characters appear tightly clustered with the nodes of other characters.
We can take this line of thinking an important step further: not only do the cross-dressing characters combines the roles of heroine and broker, but in doing so, they also assume a dramatic centrality beyond that of their counterparts in other Shakespearean comedies. Compare the networks formed by cross-dressing heroines, for example, to the position of the main female characters in two late scenes from comedies without cross-dressing characters,
In both of these networks, non-cross-dressing female characters (Katharina, Bianca, Helena, Hermia, and Hippolyta) occupy positions at the margins of the networks. Cross-dressing female protagonists consistently gain a central position in the social networks involved in the final moves of the plays. Female characters who retain conventional dress and gender conventions, on the contrary, occupy a marginal social position during crucial scenes of narrative resolution.
Considering a cross-dressed heroine in terms of networked sociability helps us see that character’s gender in terms of relation as well as identity: the cross-dressing not only allows a character initially gendered female to act like a man, so to speak, but it also allows her to connect like a man. The social worlds of all of these plays are male-dominated, in that characters of all genders speak primarily in response to men, but we find, in the data as well as intuitively, that male characters respond to men more than female characters do, reflecting the plays’ tendency to incorporate same-gender and well as mixed-gender conversations. The cross-dressed heroines, however, respond to male characters even more than other male characters do. As Figure 8 illustrates, when wearing men’s clothing, the heroines inhabit the social relations of men (or even statistical super-men), even to the point of leaving behind the conversational connections to other women that have previously defined their relationships.
These visualizations share a common assumption: that gender in the plays arises initially from the categorizations that the text offers us and, subsequently but crucially, from the nature of the characters’ interactions. In this latter aspect, gender is fundamentally relational. In a more conventional linear model of gender, masculinity and femininity constitute two poles, and the focus of analysis is an individual character using cross-dressing to traverse the distance between those poles. Rather than investigating the gender identity of isolated characters — no matter how complex those identities may be — our approach instead describes the social relations within dynamic systems that produce the performed effects of gender in the plays.
The grounding of our approach in textual rather than contextual data creates
our focus on the carnivalesque effects of cross-dressing within the action
of the plays. We do not mean, however, to minimize the importance of other
approaches: for instance, Rackin’s helpful insistence on the gendering of
the actors in all-male performances of the plays implies another kind of
network visualization, showing every interaction as a male actor (in both
senses) speaking exclusively to other male actors, with our conception of
mixed-gender networks operating in ironic tension with that all-male
theatrical world. And that is not to mention yet another layer of
complication: that the male
actors of those companies may also have
experienced their gender as relational or non-binary, thus implying in
their irrecoverable subjectivities
In closing, we step back from the landscape of nodes and edges to consider the implications of our analysis for reading passages of these plays in light of a networked conception of gender. Jean Howard has made the case that the cross-dressing heroines fail to disrupt gender relations fundamentally because the cross-dressing women are reabsorbed into the conventional social world by the end of of the plays. Howard argues that in the conclusion of
real threats are removed and both difference and gender hierarchy reinscribe
shows a woman manipulating [gender] representations in her own interest,Howard still maintains that the play ultimately has the
primary effect...of confirming the gender system
is never perfectly achieved
Howard’s reading persuasively accounts for the conservative movement of these comic plots as well as the disruptive and incompletely contained undercurrents that cross-dressing creates. We perceive another reading that moves away from the conservative/radical opposition at the heart of that reading. In
(I. II. 43-45)O, that I served that lady, And might not be delivered to the world Till I had made mine own occasion.
(II.III.40-41)O Time, thou must untangle this, not I. It is too hard a knot for me t’ untie.
knotof society, even though she may shape it through her decisions.
Viola thereafter voices a new understanding of her agency that is contingent
on her social circumstances:
(V.I.263-265)make her own
occasion.
Even her most fundamental statement of named
identity, I am Viola,
becomes a dependent
clause that is shaped by the space she inhabits, namely, the place
as well as the time
and fortune
of her circumstances.
Furthermore, all of those contingencies build on the fundamental irony of
acting, that the pleasure of watching the play involves understanding that,
in voice and body, Viola is not Viola at all.
We propose that network analysis allows new and complementary understandings
of the kinds of identity and contingency, being and relationship, that
enable performed gender to make, and be made by, its own occasions. Our
visualization of these Shakespearean networks illustrates cross-dressing
not as an individual character’s movement a simple spectrum of more or less
disruptive gender performance, but rather as an act of social
reconfiguration made more visible by visualization methods such as social
network analysis. By cross-dressing, the heroine moves to the center of a
complex social world and brokers its transformations during the part of the
play in which she dresses as a man. Even after she changes back into a
woman’s clothing, in a seemingly dismaying regression back to the governing
social norms that James I’s commandant
sought to preserve, the
play’s resolution comes about subtly in response to her more fundamental
remaking of its networked communities.
If Sight and Shape be True: The Epistemology of Crossdressing on the London Stage,