CHWP B.2, publ. May 1996. © Editors of CHWP 1996. [First published in CCH Working Papers, vol. 1 (1991).]
KEYWORDS
TACT, Shakespeare, textual structures, book structures, tagging, imagery, prefixes, pronouns, exclamations, Petrarchan diction, variants
Every undergraduate student knows that references to the dramatic works of Shakespeare should be keyed to Act, Scene, and Line numbers, not to page numbers, not to running time, and not to Coles' Notes. These hierarchical divisions constitute the 'explicit structure' of the play, they are told, just as Book, Chapter, and Verse organize the text of the Bible. It is no coincidence that text retrieval software like WordCruncher, from Brigham Young University, is specifically designed to handle three, and no more than three, levels of hierarchical tags, like those of the Bible or of Shakespeare.
But Shakespeare scholars have recognized for decades that, in actual fact, plays were performed on the bare, unlocalized Elizabethan public stage without pauses for Acts or Scenes. For centuries, the same scholarly activity which overwhelms modern editions of Hamlet with more apparatus than poetry, has continued to refine and alter the Act and Scene divisions of the original quarto and folio texts, generally on the principle that a cleared stage indicates a new scene (but also on the principle that, if enough editorial predecessors have divided a scene at a certain point, so should we). Even granting that Elizabethan theatrical practice might have left the stage cleared, which is doubtful in itself, these Act and Scene divisions are far more obvious on the page, where they are now accompanied by generous white space and typographical ornament, than on the stage, where they are often unimportant or invisible. Shakespeare's original audience would have been much more affected by shifts from prose to verse, or from balcony to stage level, or from song to dialogue, than by arbitrary act or scene divisions.
On stage, when Shakespeare is performed in this flexible Elizabethan manner, what theatre scholars label 'classical scenes', or 'French scenes', are much more striking to the audience than the traditional scene breaks. Each time a new character enters or exits, the dynamic between the on-stage characters is significantly altered: for example, Hamlet's soliloquies stand apart from the scenes they precede or follow; Romeo and Juliet's love scenes end the moment the Nurse enters (whether editors change the scene number or not); and the majestic entrance of King Lear and his courtiers abruptly ends the opening small talk and bawdy jesting of Kent and Gloucester. Still more complex are concentric scenes, situations in which several distinct groups of characters are on-stage simultaneously, and the script alternates between them: in such a scene, like Love's Labour's Lost 4.3 or A Midsummer Night's Dream 5.1, the audience senses a discontinuity between social groups just as distinct as that between classical scenes.
Moreover, the fundamental dramatic experience is not one of Acts and Scenes, but of characters in action and interaction. Drama has no narrator, no single omniscient perspective; instead, there are multiple and continually conflicting perspectives, much like those we encounter in everyday life. Shakespeare never said "to thine owne selfe be true";[3] his character Polonius did -- and throughout Hamlet, Polonius is portrayed as something of a doddering old fool, spouting trite clichés and misinterpreting almost everything he sees. Neither did Shakespeare say "All the world's a stage";[4] Jaques did -- and throughout As You Like It Jaques is quite clearly a cynical, melancholy misanthrope. In many ways, Shakespeare's characters are the explicit structure of his plays; in the Globe playhouse, working manuscripts of plays were divided into individual 'parts' for each actor, not into Acts or Scenes. Finally, then, text analysis software which limits itself merely to Act, Scene, and Line divisions cannot capture the subtleties or complexities of the structure of Shakespeare's works.
Figure 1 attempts to represent graphically the distinction between simple hierarchical structures, like the traditional Act/Scene/Line divisions (on the left), and the more complex, flexible, and ultimately revealing non-hierarchical structures in which Shakespearean scholars, and probably most literary scholars, are primarily interested (on the right). Charting all this conflicting information in a comprehensible manner is not a simple task, and I have resorted to quite a bit of simplification and fabrication for illustrative purposes. In the end, this is not the neatest chart in the world -- but then that is precisely the point. TACT is the only interactive text retrieval and analysis software which allows complete freedom to use non-hierarchical divisions, so that overlapping structures (like pages and scenes), contradictory structures (like scenes or classical scenes), and fragmented structures (like the speeches of a given character, or the stints of an individual compositor, scattered throughout the play) can all be tagged and analyzed with electronic assistance and accuracy. WordCruncher still prefers simple structures like that on the left.
Figure 2 lists the categories of structural information I have tagged in preparing electronic texts of Shakespeare's tragedies for use with TACT, and offers sample codes from the opening of Hamlet Q2. These texts are the original, old-spelling quarto and folio texts, complete with typographical errors, duplications, redundancies, and omissions -- all the bibliographical gems which editors work so hard to eradicate in modern editions.[5] Because my thesis research examines these early texts in some detail,[6] I have tagged an abundance of bibliographical structures, such as Compositor Stints, Signatures, Formes, and Type Fonts. In preparing your own dramatic texts, whether of Beaumont and Fletcher, Gilbert and Sullivan, or indeed Shakespeare himself, you might well consider some of these items unnecessary, and others that I may have overlooked crucially important, to your purposes. TACT imposes very few limits on the amount or kind of structural information you can identify.[7]
Figure 3 offers a facsimile of the first page of the opening scene of Hamlet in the authoritative second quarto edition. Contrast it with figure 4, which shows the tagged text file corresponding to that single page of the quarto. Clearly, the file in figure 4 is intended to be read by MakBas, not by human beings. The initial series of codes is important to identify features of the entire text (like Author, Title, Publisher, etc.) in anticipation of the day when someone might want to combine all three texts of Hamlet, or texts of all the tragedies, into a single TACT database. Other initial tags are important to set default values; otherwise TACT will report only question marks (???) as values for title page material found before the first tags.
One necessary evil, readily apparent from figure 4, is the redundancy of some of the textual markup. For example, italicized text is surrounded by curly braces, {thus}, so that in TACT's on-screen displays the typeface change will be immediately apparent to the user. Codes in angle-brackets, however, like "<FONT Italic>" and "<FONT Roman>", are also necessary to permit TACT to conduct searches based on this typographical information. Most complicated of all are the speech prefixes: each speech prefix itself is presented as if spoken by an imaginary character, the stage director, abbreviated DIR, who must be identified as both a speech prefix and a normalized speech prefix (see below). So after the codes "<SP DIR>" and "<NSP DIR>" comes the actual speech prefix itself, for example "{Bar.}". Then a tag follows to indicate that the speech itself is spoken by the character indicated in the speech prefix, "<SP {Bar.}>", and finally a Normalized Speech Prefix, or the modern-spelling unabbreviated character name, appears, as for example "<NSP Barnardo>", so that all speeches by a given character can be identified, regardless of errors, inconsistencies, or abbreviations in the original texts, and searched as a whole. Margins 200 characters wide were necessary to fit all these tags in, and often 200 characters were insufficient, requiring a variety of compromises and more complicated work-arounds. Speech prefix lines were the primary culprits, too: compare the relative simplicity of lines 12 and 13, for instance, which constitute the middle of Barnardo's first multiple-line speech.[8]
That, in a nutshell, describes some of the many layers of explicit structures in Shakespeare's plays, and the ways in which an electronic text can be tagged to make use of all this structural information in a TACT database. Once this time-consuming process is complete, however, the real fun begins. Using a TACT database of Hamlet Q2, for example, can revolutionize our perception of the play, enhancing our attention to detail and subtle verbal patterns, allowing free play of the scholarly imagination to experiment, explore, and enjoy the play in fresh new ways. This is the way I find TACT most useful, as a sort of electronic 'doodle pad' for browsing Shakespeare's texts, not merely to answer specific questions, but also to discover new questions to ask.
I am not a computer scientist, and what I do with Shakespeare cannot really be considered 'literary computing'. In my work, the computer remains a largely transparent tool, like printed concordances, dictionaries, or my word processor; frequency distributions generated by TACT may surface in a few footnotes, and relevant quotations uncovered in the KWIC [Variable Context] display may appear in my finished argument, but statistics do not dominate my papers and I try not to let the limitations of hardware or software dictate my approach to the plays. What I value about TACT is that it does not take us away from the plays themselves, but returns us to the text with a new immediacy and accuracy, reorganizing our perception of the plays and heightening our sensitivity to inherent verbal patterns. TACT has the capacity to vastly improve all forms of Shakespearean criticism, by emphasizing the primary importance of the words themselves, and by immersing us in the text of the plays.
Usually, a reader starts Hamlet at Act 1, Scene 1, and reads through the text from beginning to end (or, in the case of most undergraduates, from the beginning until they find a videotape of Olivier's version). In figure 5, I have tried to represent this graphically as 'horizontal' reading, a linear process from left to right through the scenes as printed. This should always be the primary way to understand Shakespeare, of course -- in performance, there is no alternative. Using text retrieval software like TACT, however, it becomes possible to read a play vertically as well (see also fig. 5), synchronically, by cross-section. Browsing the TACT database of Hamlet, selecting a variety of words, phrases, or categories, and using the KWIC [Variable Context] or Index [KWIC] displays, the various acts and scenes line up around the verbal points selected. Doubtless a reader with a remarkably good memory could read vertically like this while reading horizontally, but for most of us it would require numerous re-readings of the play, with a variety of coloured highlighter pens in hand. And, inevitably, we would miss useful evidence.
[Continue]
[1] Editorial note. The reader should be aware of a few changes that have been made to the names of TACT components since this article was written. The term TACT is now used for the overall system of text retrieval programs, the two main components of which are MakeBase (formerly MakBas), which converts a text file into a textual database, and UseBase (formerly called TACT), which allows one to search the database. The old Index and KWIC displays are now called respectively KWIC and Variable context. Categories are now known as Groups. Cf. J. Bradley, "TACT Design", CHWP, B.1; cf. notice on availability of TACT.