How Quickly Nature Falls Into Revolt
When the first folio edition of William Shakespeare's works was published in 1623, "it was not clear whose idea the collected volume was or even what was the precise motivation for it" (Proudfoot, Thompson, & Kastan-1998, 8), but the inclusion of two actors that worked with Shakespeare in the publication process underscores the importance of accuracy of authorial intent in the volume. This is especially important since the actors, John Heminges and Henry Condell, state that while Shakespeare's input would have been preferred, "he by death departed from that right, [and that] we pray you do not envie his Friends, the office of their care, and paine to haue collected & publifh'd them. . .as he conceiued them" (Moston-1998, 7). Because of the direct intentions of Heminges and Condell, as well as the publishers, to stay true to what Shakespeare intended, the textual superiority of the first folio should be taken as a standard to work with, even against subsequent folios in the period (Blayney-1991, 34). This said, the delineation of Shakespeare's works into three distinct categories, Comedies, Histories, and Tragedies, by the Folio is problematic because the "organization of the volume in three sections. . .betrays the plays, which are characterized by fluidity not consistency of genre, by a continual mixing of modes" (Wells & Taylor-1987, 38). The significance of this division was noted as early as the nineteenth century when Edward Dowden singled out certain plays, which he labeled "Late Plays", as being distinct from the rest. Modern editions such as the Riverside, the Norton, and the Pelican Shakespeare have followed suit by classing Cymbeline, The Winter's Tale, and The Tempest, with the addition of Pericles, apart from the rest. This poses a quandary about the nature of authorial intent. In this paper I will examine if a "natural" grouping of genre exists in Shakespeare's plays by discussing various theories of a reader's response to a text and then introducing a method of computational analysis to Dowden's arguments based on the genre categories of the First Folio (F1).
The division into Comedy, Tragedy, and History was made with at least some intent since there were other genres present in early modern society. For example, around 1579 Sir Philip Sidney wrote both for and against the "mongrel tragic-comedy' that mingled kings and clowns" in his A Defence of Poetry (Foster-2004, 16) and "John Fletcher, who, in his preface to The Faithful Shepherdess (1609), provided the first adequate English definitions of tragicomedy" (Foster-2004, 21). This shows that there was a precedent set both in the practice as well as publication of different kinds of genres, such as tragicomedy. In addition, Polonius's speech in the second Quarto (Q2) of Hamlet, printed in 1604, emphasizes the variety of dramatic modes available as he states the players at Elsinore as "the best actors in the world, [for] either for Tragedie, Comedy, Hiftory, Paftorall, Paftorall Comicall, [or] Hiftoricall Paftorall" (Hamlet-1985, II.ii.415-416). Thus, publications prior to F1 highlight the availability of other genres to the F1 publishers. However, even though F1 was "publifhed according the True Originall Copies" (Moston-1998, 5), "Shakespeare himself does not seem to have undertaken to oversee the printing of his plays, even during his lifetime" (Wells & Taylor-1987, 2) so thinking of Shakespeare as the designator of this genre classification is illogical. This is reflected by all the subsequent printed folios using the same genres and catalogue page despite adding new plays (Brewer). In fact "the Fourth Folio of 1685 was the last Folio to be simply reprinted from its predecessor (the Third, with the seven additional plays), and by the early 1700s the plays were coming to be seen as works that required editing" (Blayney-1991, 34). This transition from reproduction to reworking in editing marks the change from actors as editors to readers such as Dowden as editors. The significance of assigning Shakespeare's works to genres remains but the reasoning behind it stems from two different sources, that of the actor and that of the reader.
In dealing with this two-sided situation, pursuing and understanding the thought-process of readers of Shakespeare is the only option available to expound upon the idea of genre categorizations. This decision has less to do with granting either priority, but is rather to avoid assumptions and polemics compared to the more straightforward analysis of a reader's response. To that end, Wolfang Iser's theory of reception explains a relationship between readers and a text.
The literary work has two poles, which we might call the artistic and the aesthetic: the artistic pole is the author's text and the aesthetic is the realization accomplished by the reader. In view of this polarity, it is clear that the work itself cannot be identical with the text or with the concretization, but must be situated somewhere between the two. It must inevitably be virtual in character, as it cannot be reduced to the reality of the text or to the subjectivity of the reader, and it is from this virtuality that it derives its dynamism. As the reader passes through the various perspectives offered by the text and relates the different views and patterns to one another he sets the work in motion, and so sets himself in motion too. (Iser-1978, 21)
Iser's theory proposes a dualistic relationship between the text and the reader where each has influence on the other. However, this also subjects a work of literature to a certain set of interpretational guidelines where one endpoint is authorial intent and the other is any given reader's perception of the text. In Iser's model, reader and text interact in a way that allows for a varied yet concentrated commonality between readers, since one endpoint of a given text will always be the fixed locus that is the author's actual writings. However this schema does not account for external influences, such as with what a reader has previously read, and this shortcoming arguably relegates Iser's theory to a starting point for how to understand literary reception.
Given Iser's ideas, we are able to explain how a reader and a text may be related but still lack the specificity of what exactly a reader takes note of when reading. In comparison, Peter Brooks states that a "text is seen as a texture or weaving of codes (using the etymological sense of "text") which the reader organizes and sorts" (Brooks-1992, 19). This description provides a framework to build upon since Brooks says that a text has certain patterns couched in it that are necessary for comprehension. Franco Moretti expands on this as he declares that the study of literature is about "the very small and the very large; these are the forces that shape literary history. Devices and genres; not texts. Texts are certainly the real objects of literature... but they are not the right objects of knowledge for literary history." (Moretti-2005, 76, emphasis in originalf). In other words, Moretti argues to move away from thinking about a text as a unit, instead focusing on smaller and larger systems, i.e. the composition of a text and its genre. This ties the physical typeset on the page into a codependent relationship with its more nebulous aspects, like genre and theme, as the true objects of literary goals. Brooks's codes also, through Moretti's lens, carry innate textual and thematic patterns tied to a reader's realization of literature. It is then clear that the text itself needs to be investigated as the source of the larger nebulous units such as "[The Tempest's] perfect expression to the spirit that breathes through these three ['Late'] plays" (Dowden-2003, 291). Unfortunately, closely examining the text on the page is more difficult than it appears.
The difficultly in appropriately and succinctly approaching Shakespeare's works may be best introduced by Michael Witmore as he takes stock of Shakespeare's use of words. Witmore notes that Shakespeare "used imagery" lots and lots of it "to suggest how an actor or audience member should feel about a particular moment in the story. The results of this technique are visually and aurally arresting, point us toward what literal sight cannot convey" (Purcell & Witmore-2010, 8-9). According to Witmore, Shakespeare's lexical choices elicit a second, sensual element out of the original text. The intrinsic complexity of textual features as "literature" to be read in their own right is furthered by Anne Barton who writes that Shakespeare's works "are filled with tallies, a sea of objects which continually threaten to engulf the characters" (Barton-1997, 253). Barton's comment suggests that the textual object of a "character" can come into conflict with additional features, such as Witmore's imagery, which ultimately renders the text inexorably tied together. Furthermore, with the words of the early modern era "almost any part of speech can be used as any other part of speech" (Abbott-1966, 5) which leaves a straightforward analysis of the text to be desired. To work against the convoluted nature of the text, I.A. Richards argues that "we come back to a master rule of method: It is the sense of a word in a particular use which we have to consider first, not the general meaning, the wide range of all its senses taken together, nor something we may suppose is common to them all" (Richards-1942, 111). Richards's point is that the key to interpreting the meaning is the context in which a word appears. I would like to develop that one step farther by suggesting that context also be brought into consideration with the thematic and aesthetic pole of literature as well. In using context, approaching both individual texts and groups of texts is facilitated.
The contextual approach to literature aids in understanding single and multiple texts but also accentuates the absence of a patterned context between a reader and the literature. John Sheriff's analogy to this problem is that:
it is as if the reader brings to the text a large stained glass window that has a pattern and many varicolored panes (representing the strategies, codes, and forms that make up his way of seeing). The text he wishes to read is another such window. He puts the windows face to face, holds them up to the light of interpretation, and sees the pattern and colors formed by the combination of the two. How can he determine what the contribution of each is to the pattern and colors that emerge? (Sheriff-1989, 22)
When Sheriff's idea of a reader-specific filter is considered alongside Iser's model, a framework for the understanding of genre emerges. In a sense, the aesthetic end Iser proposes can be viewed through this "window" but remains clustered within a limited deviation. Complementing this is the idea in which "text conventions are 'constitutive' rather than 'regulative,' i.e., they constitute rather than regulate a form or genre" (Beach-1993, 17). To rephrase, Beach combines the idea of a single text existing in parallel among others forming a group of literature. The group is then in turn regulated by the texts it encompasses because "whether an author adheres strictly to a genre or deviates from it, his intention is expressed to some degree in his basic choice of genre" (Mills-1976, 264). By thinking of a text as a constituent of a genre but only being able to read each one independently, a situation arises where "the reader is free to fill in the blanks but is at the same time constrained by the patterns supplied in the text; the text proposes, or instructs and the reader disposes or constructs" (Freund-1987, 142). Freund's statement reinforces the idea of a pattern language in a text where a reader is able to actively engage with the text but is still affected by the author's artistic end as proposed by Iser. This grants the greatest breadth of subjectivity between readers while also conforming to the predefined constraints of the author's written patterns. Unfortunately this system is not able to answer Sheriff's parting query: "how can [we] determine what the contribution of each is to the pattern?"
J.F. Burrows expands upon this by mixing quantitative data with qualitative assessments in his work on Jane Austen's novels. In his study Burrows states that some textual evaluations:
are united by the assumption, not always made so explicit, that, within the verbal universe of any novel, the very common words constitute a largely inert medium while all the real activity emanates from more visible and energetic bodies . . . [but] the neglected third, two-fifths, or half of our material has light of its own to shed on the meaning of one novel or another; on subtle relationships between narrative and dialogue, character and character; on less directed and less limited comparison between novels and between novelists; and ultimately on the very processes of reading itself. (Burrows-1987, 2)
Burrows's study reveals that statistically significant discrimination between characters, chapters, or even different books can be provided by the most frequent words in a text, such as we, us, of, and very. These findings link the importance of regular function words with "more visible and energetic bodies" of nouns, adjectives, verbs, etc. and also provide a route into considering the pattern language of literature. What this means is that the level of address, be it words in a text or groups of texts, is dependent on both its context and the sum of its parts. In support of this, Morretti states that:
quantitative research provides a type of data which is ideally independent of interpretations, as I said earlier, and that is of course also its limit: it provides data, not interpretation... Quantitative data can tell us when Britain produced one new novel per month, or week, or day, or hour for that matter, but where the significant turning points lie along the continuum - and why - is something that must be decided on a different basis. (Moretti-2005, 8)
Here Moretti is supplying the framework for which to investigate the small and large of literary knowledge. Focusing on quantitative means, while filling in the gaps with the qualitative approaches of existing literary theory, allows for a hybridization to occur where the number of certain words in a text can be viewed in correlation with groupings of genre. I aim to further the work linking the counting of specific words with the qualitative approaches of literary theorists.
In order to move beyond the calculation of commonplace words, I employed the text-tagging program Docuscope, created at Carnegie Mellon University, which "now classes over 200 million strings of English (1 to 10 words in length) into over 100 distinct categories of use or function" (Witmore-2009, 1). Its main function is to tag elements found in a given text and output the quantitative data of what it finds but, in theory, Docuscope is working towards decrypting the patterns of the aesthetic pole by imposing a uniform, prosthetic reader upon a text. Docuscope is organized in a three tier system where every one of the entries of Docuscope's dictionaries is first assigned to a single Language Attribute Type, or LAT. There are one hundred and one LATs which represent the most elaborate distinctions and provide the finest grain of similarity in analyses. Each of the LATs is then placed into one of fifty-one groups called Dimensions. Each Dimension is subsequently sited within seventeen Clusters which represent the coarsest grain of similarity of the three tiers . In short, Docuscope is able to identify many English words, like a reader with a finite vocabulary, and consider each word with equal weight in relation to the others which allows for a uniform approach to any form of literature. It also expands to include strings of words in its counting, thereby more closely resembling an actual reader and providing information about words contextually in a work. Thus this lives up to the expectations of a theory of reception while providing a direct approach to studying both small and large elements.
Docuscope's own, built-in features of textual analysis offer a direct window for the user to view what it is "reading" out of a text, as seen below in the case of The Tempest:
But Docuscope itself lacks the tools to grapple with larger collections of works even though it provides the data for it so I also feed the data into a statistics software package called JMP. JMP allows for larger amounts of Docuscope's data to be visualized and examined through the use of multivariate statistics. For example, the following diagram is a JMP generated Hierarchical Cluster map, using Frequency Counts at the Cluster level from Docuscope and a Ward?s test with best guess analysis. It also incorporates a distance scale ratio in the dendrogram which means that the lines of the dendrogram are proportional to the actual statistical distance. The dendrogram itself functions as a family tree sort of visualization which shows the statistically closest connections the farthest to the left on the horizontal axis. Essentially what this does is process the information provided by the seventeen Clusters and place the results onto a two-dimensional plane. Below is Shakespeare's First Folio, according to Docuscope and JMP.
In this picture, three genres have been color-coded for the ease of viewing: Histories in orange, Comedies in blue, and Tragedies in red. The solidity in each of the colors is striking. In addition, Witmore has expanded on the qualitative corollaries that these distinct groupings represent. For example, in Richard II:
we see the formal settings of royal display, a herald offering Mowbray's formal challenge - no surprise this exemplifies history, a genre in which the nation and its kings are front and center. Yet where the passage really begins to rack up points is in its use of descriptive words, which are underlined in yellow. Chairs, helmets, blood, earth, gentle sleep, drums, quite confines...we don't think of history as the genre of objects and adjectives, but linguistically it is. Inclusive strings, in the olive colored green, are perhaps less surprising given our previous analyses. We expect kings to speak about "our council" and what "we have done." But notice that such language is quite difficult to use in comedy: even in a passage of collusion, where we would expect Mistress Page and Mistress Ford to be using first person plural pronouns, the language tends to pivot off of first person singular perspectives. The language of "we" really isn't a part of comedy. (Witmore-2010, 1) {The image Witmore references is below.}
The fact that Docuscope's results can be combined with contemporary literary scholarship is logical due to its modern creation, but the question of its suitability for older views like Dowden's remains.
However, in trying to search for what Dowden has seen in the text, it is likely that a user would impose specific judgments that might not otherwise be prominent, thereby biasing the results. To work against this I have chosen to continue using the Cluster level analysis, rather than Dimensions or LATs, since I feel that the largest degree of dissimilarity should be invested first, before movement into more restrictive spaces takes place. In addition, I would propose the least third party interference possible, the user being the third party in this case, between the text and the results. To this end, I only changed the colors of the previous F1 diagram to incorporate green for the Late Plays. In this we see that Cymbeline and Winter's Tale cluster together closely but The Tempest is far apart, which is a false positive and prompts an alternate route into discovering their commonalities.
Although the Late Plays did not cluster together on the dendrogram, that does not mean that they do not share traits in common. In fact, I would argue that not sharing a single trait in common is fundamentally impossible based on Docuscope's construction of a textual object. However this does mean that the majority of their traits are less similar to each other than they are to other plays. Using an Oneway ANOVA test, I worked around this to determine several common factors influencing the Late Plays. This particular statistical test compares the spread of variation for all of the plays on a single Cluster variable at a time which allows the viewer to pinpoint what exactly is being picked up by Docuscope and visualized in JMP. After this test, it is clear that all of the Late Plays score highly on the Clusters labeled Elaborations, Reporting, and Time Orientation but they score lower on Directing Readers, Emotion and Public Reference. To further extract what Docuscope reports as commonalities between these plays, I used Docuscope's Single Text Viewer with the play The Tempest and the high scoring items selected in color.
In the picture Time Orientation is colored orange and is what Docuscope says represents the idea of time past or future (Docuscope). The blue elements are Elaborations which are described as "actions in the stream of discourse that serve the readers' curiosity and information needs through the content they add. Texts with high proportions of elaborating feel content dense, brimming with information" (Docuscope). The most significant feature present was Reporting, colored in purple. By Reporting, a writer "is dispensing information to update a reader's mental model of a world that is both steady and changing, fixed and dynamic, routine and subject to historical change... so that readers and other audiences of the reporting can keep pace and up-to-date" (Docuscope). Docuscope's highlighting features allows for quantitative results to be mapped onto qualitative readings of a text with the added perk of the exact locality. By investigating in this way, partiality can be avoided in the findings while still preserving the original judgments.
Interestingly, the initial results from Docuscope are often not original at all. For example Ifor Evans notes the high-scoring attributes of The Tempest as he said that Shakespeare "creates the mood of the island and, at the same time, in a most concentrated manner, reveals Prospero's previous history... Seldom had there been so much said in the plays in so few words since Hamlet described his adventures with the pirates" (Evans-1952, 184). Evans notices the packaging of information into smaller and smaller units in The Tempest, such as through the use of Elaboration, and he relates this to the Late Plays in general. This also chimes with Dowden's rational for grouping them because the:
characteristics of versification and style, and the enlarged place given to scenic spectacle, that these plays were produced much about the same time. But the ties of deepest kinship between them are spiritual. There is a certain romantic element in each. They receive contributions from every portion of Shakespeare's genius, but all are mellowed, refined, made exquisite; they avoid the extremes of broad humour and of tragic intensity; they were written with less of passionate concentration than the plays which immediately precede them, but with more of a spirit of deep or exquisite recreation. (Dowden-2003, 309)
One element that Dowden mentions consistently weaving through the Late Plays is the avoidance of extremely broad humor and tragic intensity. By noticing this, Dowden suggests a lack of Emotion, something which JMP showed the Late Plays having in common. In addition, the lack of Public Language, defined as the language of institutions and authority, is what Dowden says gives the plays a certain spirit of recreation. The harmony between Docuscope's findings and Evans's and Dowden's rational for another genre classification answers the question of Docuscope's compatibility with both contemporary and past literary scholars.
Now the question which remains is whether we can truly consider the Late Plays to be a group separate from the rest. In support of a general distinction, Alfred Harbage states that "since Pericles was not printed in the first folio, the editors were spared the problem of classification" (Harbage-1969, 1257) and thus did not have to process the Late Plays as a different genre. Evans argues against this by claiming that the Late Plays cannot constitute a separate genre because lexically "the language of these three plays cannot be treated as one, they are all different, except for the early scenes of The Winter's Tale, from the tragedies that precede them" (Evans-1952, 176). To quantitatively decide, I created a three-dimensional model with the same four color scheme for genre as before and used Principal Component Analysis to investigate.
From the model's contents, it appears that Evans and Docuscope agree that the idea of a linguistically distinct group of plays called the "Late Plays" is faulty. This is backed initially by the dendrograms, but also in a three-dimensional space with the separation of the green data points. However, I would also agree with Harbage's rational that "Pericles, Cymbeline, The Winter's Tale, and The Tempest are separately grouped... because they share certain characteristics which the grouping helps to emphasize" (Harbage-1969, 1257). With this, the idea of a fundamentally distinct genre of Late Plays is null but grouping them because of their similar traits for emphasis is a valid choice.
In using a quantitative approach and evaluating whether or not a truly separate genre of Late Plays exists according to Dowden, two answers seem to have arisen. If the smallest elements are emphasized, the conclusion would not support a separation of the Late Plays from the rest of F1, as Evans noted. But the presence of similarities in the elements, such as those which Docuscope and Harbage find, is what prompts the idea of a "palpable discrepancy between form (of the Folio) and content (of Shakespeare's actual plays)" (Wells & Taylor 38). This creates a conundrum in which the individual must decide whether or not what these plays share is truly unique from the rest of the corpus. I think that while arguments can be made either way, ultimately "Shakespeare does not supply us with a doctrine, with an interpretation, with a revelation" (Dowden-2003, 329) and so we must take the original judgments of F1 as they stand. The division of genre by Heminges and Condell is bolstered by Docuscope's findings while Dowden's view is not. Even though the similarities between the Late Plays can be seen by both readers and Docuscope, the fact that their grouping is not as significant as the others means that the original judgment of Comedy, History, and Tragedy as genres in Shakespeare's works is the most natural.