If you lived in Shakespeare’s England and needed to learn a foreign language, you had a few options. Using traditional methods, you could study with a tutor or journey abroad to converse with native speakers. At this time, however, it was also possible to learn foreign languages from printed dialogues in bilingual or polyglot phrasebooks (you can listen to historian John Gallagher talk about these foreign-language manuals here). Here’s an example from the widely-published Colloquia, originally composed by Noël de Berlaimont in the 1530s:
Abbie Weinberg wrote about editions of Berlaimont’s dialogues earlier this year for The Collation, paying special attention to their unique mise-en-page and diminutive size. As I’ve mentioned before on this blog, this oblong manual witnessed nearly 150 editions from Lisbon to Warsaw, and over a period of almost 300 years. As you can see in the photo above, it could include some combination of Latin, French, Dutch, German, Spanish, Italian, English, and/or Portuguese. The contents included a guide to letter-writing, a dictionary, and a short grammatical treatise, along with a lively set of dialogues.
Dialogues were also the substance of John Florio’s Italian-English conversation guides, First Fruites (1578) and Second Frvtes (1591). The first manual begins with “familiare speache,” including greetings and a bit of commentary on the theater (not all good):
Figure 2. John Florio, Florio his firste Fruites (London, 1578), STC 11096. Huntington Library, shelfmark 60820, in EEBO.
In Florio’s Second Frvtes, interlocutors would discuss the theater again, remarking on how in England one can find “neither right comedies, nor right tragedies.” Rather, there are “[r]epresentations of histories, without any decorum” (D4r).
Understandably, the dialogues of Berlaimont, Florio, and others have captured the attention of scholars, who often comment on the theatrical nature of these exchanges. In his book-length study of early modern Anglo-Italian relations, Michael Wyatt recognizes a “theatrical structure” in Florio’s bilingual dialogues (The Italian Encounter, 167). William N. West notes in a discussion of cant in Jacobean theater how “the dramatic potential” of bilingual language manuals could be brought to the stage (“Talking the Talk,” 234). More recently, Joyce Boro has stated that these multilingual dialogues have “affinities to closet drama” (Tudor Translation, 22).
These observations raise bigger questions for me, though. First, if language-learning dialogues seem so similar to Renaissance plays, then exactly why do they seem so similar? Put differently, what are the characteristics governing both genres? Second, if language-learning dialogues and drama are indeed categorically similar, which plays — in which genre, or by which author — are these dialogues most like? Finally, and perhaps most consequentially, how might this generic similarity come to influence our understanding of Renaissance drama?
To begin to answer these questions — and that “begin” is key, so if you’re looking for a definitive answer, best to stop reading now — I opted for a large-scale analysis of dramatic and language-learning dialogue texts using the linguistics analysis software DocuScope (created by David Kaufer and Suguru Ishizaki). In using DocuScope to assess dramatic genres, I’m taking an approach indebted to Jonathan Hope and Michael Witmore, who have used this software to investigate Shakespearean genre, and Ted Underwood, who reminds us how blurry our agreed-upon genres can be. Breaking language down into strings of characters, DocuScope uses more than 100 language action types, or LATs for short, to identify words and phrases common to lots and lots of texts. With this approach, I can investigate dozens of characteristics of these texts at a single blow, drilling down where appropriate in order to assess more nuanced similarities at the level of the sentence. For the dramatic texts, I used a corpus of 320 works of early English drama curated by Martin Mueller.
This is where things start to get messy, though. For the language-learning dialogues, I combined 3 EEBO-TCP texts available to me with 6 that I keyed myself, using only the English text columns, and stripping out speech prefixes to match Mueller’s corpus. (Polyglot or bilingual texts have been a relatively low priority for TCP; altogether, the keying amounted to a little over 82,000 words.) Furthermore, none of these 9 texts I’m comparing with Mueller’s corpus have yet been processed with Alistair Baron’s VARD modernization software. (The complexities of using VARD for early modern texts are documented here on the Visualizing English Print blog). Not using VARD certainly adds noise to the results — as the saying goes, “Garbage in, garbage out” — but I’m willing to take the risk in the hopes that some shared characteristics do emerge from the test. Consider this as a single and rather early stopping-point on the longer road of these questions.
Processed with DocuScope and displayed in the statistical suite JMP, the texts look like this:
Figure 3. Principal component analysis of 320 works of Renaissance drama and 9 language-learning dialogues.
Each color indicates a separate genre (as you can see, DocuScope is quite good at distinguishing comedies from tragedies). My mini-corpus of language-learning dialogues appears here in green, and occupies a diagonal path through the lower portion of the chart (keep in mind, though, that we’re looking at hundreds of dimensions at once, so “lower” or “left” have little true meaning here). Left to right according to this visualization, they are: Benvenuto Italian’s The Passenger (1612, Ita-Eng), John Florio’s Firste Fruites (1578, Ita-Eng), John Eliot’s Ortho-epia Gallica (1593, Fre-Eng), John Minsheu’s Pleasant and Delightfull Dialogues (1599, Spa-Eng), John Florio’s Second Frvtes (1591, Ita-Eng), Claudius Hollyband’s French Littelton (1576, Fre-Eng), Claudius Hollyband’s French Schoolemaister (1573, Fre-Eng), William Stepney’s Spanish Schoole-master (1591, Spa-Eng), and Noël de Berlaimont’s Colloquia (1639 ed., Polyglot).
At a first glance, it seems many of our dialogues are just a bunch of outliers. There are probably good reasons for this, though. First, discrepancy of modernization might be one. Processing my mini-corpus of 9 language-learning dialogues with VARD could address this issue down the road, though only as well as I’d match the conventions in Mueller’s drama corpus, I think. Second, the length of texts is a factor, and an important one at that. Simply put, some of these dialogues are short, and others are very long. Hollyband’s French Littleton amounts to just under 5,000 words, while The Passenger weighs in at a whopping 103,862. Chunking the texts — that is, breaking them up into smaller units — could begin to address this issue, even if it involves its own methodological curveballs.
With those caveats on the table, it’s worth remarking that these language-learning dialogues stand out from other dialogues published in the period. According to Visualizing English Print‘s Beth Ralston, writing in March 2015 on an effort to map the ‘whole’ of early modern drama, Renaissance dialogues tend to “have lengthy monologues, which might explain why they fall mostly on the side of the tragedies.” Her diagram, published on Wine Dark Sea, appears below:
Figure 4. VEP’s principal component analysis of 704 early modern dramatic texts, with dialogues highlighted in black. Accessed 9/5/2015 at Wine Dark Sea.
This picture, however, is not what we find for the language-learning dialogues at all. Even if the complications I briefly addressed above play a role here, my corpus of 9 texts clearly falls on the side of comedies instead.
This observation leads me to ask a few more specific questions that might get us somewhere. What accounts for The Passenger‘s position among an odd mix of plays — a comedy (Nashe’s Summer’s Last Will and Testament), a tragicomedy (Heywood’s 1 Iron Age), and a tragedy (Marlowe and Nashe’s Dido, Queen of Carthage)? What about Florio’s Firste Fruites is so similar to The Fortunate Isles and their Union, a masque by Ben Jonson? And what factors might land the same author’s Second Frvtes in close relation to Thomas Dekker’s The Shoemaker’s Holiday and Jonson’s The Alchemist (both known today as fairly important “city comedies”)?
Identifying relevant LAT vectors will help us to identify the characteristics common to drama and the dialogues. It’s really messy, but have a look at the 100+ LATS used for this test (not all of them are significant):
Here, the vectors illustrate the statistical relationships most relevant for generic distinctions among our texts. The direction of the vectors indicates the relative co-occurrence of LATs, and the length of the vectors indicates the strength of their influence upon generic relationships. Focusing on the lower right-hand quadrant, we find a number of LATs commonly associated with comedies: Directives_Imperative, Elaboration_Numbers, Character_OralCues, and Interactive_Question.
If we look at Interactive_Question more closely, we find this LAT to be associated first with our dialogues, and second with a handful of Jonson’s city comedies. This LAT uses punctuation to indicate a request of specific information, which makes some sense for books designed to introduce readers to foreign vocabulary, phrases, and information. (Think of lessons in today’s language textbooks: “How much does it cost? Dov’è la stazione? Combien d’etudiants sont-ils?”) The texts in this corpus exemplifying Interactive_Question to the highest degree are Hollyband’s French Schoolemaister, Berlaimont’s Colloquia, and Stepney’s Spanish Schoole-master, each of them in our lower-righthand quadrant. We also find that Jonson’s city comedies The Alchemist, The Case is Altered, and Every Man Out of his Humour display this LAT prominently, along with Heywood’s A Maidenhead Well Lost, Dekker’s Honest Whore, and Fletcher’s The Chances. Even if the diagram presents the Hollyband, Berlaimont, and Stepney dialogues as outliers, this test suggests that they nonetheless have something in common with these particular works of drama. The next step would be to look closely at what types of interactive questions one finds in the dialogues vs. in the plays, and to assess those relationships on more qualitative terms.
With a similar approach, we can begin to investigate the relationship between Florio’s Second Frvtes and Dekker’s The Shoemaker’s Holiday, which I find a little more interesting. Looking at DocuScope’s tagged texts for the dialogue and the play, we see several coinciding LATs, among them Descript_SenseObject, Assert_First_Person, and Interactive_You_Reference. I’ll just focus on these three for now. Descript_SenseObject consists of concrete nouns, and accounts for the five senses’ perception and experience of objects. Assert_First_Person consists of self-referential language (I, me, etc.), while Interactive_You_Reference refers to a second party (you, thee, thy, etc.). These LAT definitions are based upon DocuScope’s 2008 dictionaries.
To really find out how these LATs play out, we’ll have to take a closer look at the texts. After all, as Witmore and Hope put it, “Digitally based research is not an end point: its findings need to be tested against the texts” (369). Scanning through the tagged texts of Second Frvtes and The Shoemaker’s Holiday, we can highlight and identify specific passages that feature these LATs, then compare them to see if any patterns emerge.
Florio’s dialogues often unfold a series of related vocabulary items during verbal exchanges between two speakers. In DocuScope, the result is often a chain of Descript_SenseObject punctuated by Assert_First_Person and Interactive_You_Reference. In Florio’s first chapter, a servant named Ruspa is helping his master Torquato get dressed in the morning:
Figure 6. Tagged text of Florio’s Second Frvtes in DocuScope, with Descript_SenseObject, Assert_First_Person, and Interactive_You_Reference highlighted.
In this passage, Florio’s interlocutors deal out a series of nouns related to apparel, fashion, and hygiene, words that would appear on the facing page in Italian. DocuScope identifies 15 instances of Descript_SenseObject, and would recognize more if this text were modernized with VARD. (The human eye can identify several more, a testament to our brains’ immensely complex and sophisticated furniture). Accompanying this catalog-like list of things are first-person and second-person pronouns that cushion the vocabulary and frame it into a coherent exchange between two persons. In Florio’s lessons, there’s a need to discuss physical objects (“shirtes,” “handkerchers,” “bands,” “cuffs,” etc.) often many at a single time, as well as a tendency to attribute them to an owner as they circulate or change hands (“your,” “your,” “me,” “my,” etc.).
The Shoemaker’s Holiday features a similar kind of interplay between these three LATs, particularly in scenes between two characters that concern material goods or commodities. In these moments, and there are there are several of them, “I/me/my” and “you/your” mix together with lists of nouns. Let’s look at a particular example which I was able to identify in DocuScope’s marked-up text. In Act IV, Scene 1, a gentleman named Hammon is speaking with Jane, a shoemaker’s wife. Her husband has been sent off to war, and Hammon hopes to woo Jane while she’s alone. He speaks a scorned lover’s soliloquy at the beginning of the scene, but when he first addresses Jane, the text’s language shifts into a mode characterized by Descript_SenseObject, Assert_First_Person, and Interactive_You_Reference. Although the text processed with DocuScope has been modernized, I present the unmodernized version from EEBO-TCP here to help retain a focus on the historical context. Our relevant LATs are highlighted in red, blue, and yellow:
Sir, what ist you buy?
What ist you lacke sir? callico, or lawne,
Fine cambricke shirts, or bands, what will you buy?
That which thou wilt not sell, faith yet Ile trie:
How do you sell this handkercher?
And how these ruffes?
And how this band?
All cheape, how sell you then this hand?
My handes are not to be solde. (F3v)
In this scene, Hammon and Jane begin their interaction according a script one recognizes in several language-learning dialogues in this period (“to buy and to sell” being an important component of these manuals). With a series of Interactive_You_Reference (blue), Jane lists a number of objects that DocuScope recognizes as belonging to Descript_SenseObject (yellow): “callico,” “lawne,” “shirts,” and “bands,” and Hammon responds with similar language: “handkercher,” “ruffes,” “band,” etc. However, at a certain point, “hand” joins this list of commodities as Hammon begins to woo Jane, turning from potential customer to potential lover. Although he tries to replace the dialogue’s focus on “I/me” and “you” with a first person plural — “Good sweete, leaue worke a little while, lets play” — Jane refuses to follow him, claiming her hands as her own (“My”), and this resistance matches what Hope and Witmore observe in Twelfth Night between Cesario/Viola and Olivia (372). Altogether, the language we find here in Dekker’s play — and this is one out of several passages featuring these characteristics — resembles the instructional dialogues of Florio and other language tutors. Furthermore, the many questions in this exchange — identified in DocuScope as Interactive_Question — illustrates this scene’s similarity to the dialogues of Hollyband, Stepney, and Berlaimont, located in the lower right-hand quadrant of our PCA diagram.
Some of this might seem rather unsurprising to scholars of Renaissance drama. After all, it’s well established that comedies, and city comedies in particular, tend to focus on commodities and rapid exchanges between pairs of characters. Karen Newman’s important essay “City Talk” discusses this connection, particularly as it relates to women in Ben Jonson’s Epicene. Douglas Bruster uses the term “materialist vision” to account for this phenomenon in these plays (Drama and the Market, 38-46). So, couldn’t we identify these two passages’ similar characteristics without computers?
The answer: of course we could. The difference here, however, is that we can identify this generic similarity at the level of the sentence, in a way that can point us to fresh sites for convincing close readings. If these multilingual books for language-learners have something in common with the period’s drama after all, and we can identify it at this nuanced level, how does this knowledge affect our understanding of plays like The Shoemaker’s Holiday? Furthermore, what sense can we make out of this play’s disguised Dutchman and garbled mock-Dutch, or the pretend Spaniard in Jonson’s The Alchemist? Like Bruster, I’ve questioned the sub-genre of “city comedy” before along transnational lines, and computational approaches may offer alternative approaches to these same questions. Though we often talk about the “dramatic potential” of Florio’s dialogues, could we begin to see Renaissance plays alternatively as “staged dialogues?”
These reflections come out of some investigations I made earlier this summer as part of Early Modern Digital Agendas: Advanced Topics (find our conversations documented at #EMDA2015). I’m grateful to all participants and faculty for their suggestions, especially Michael Witmore, Jonathan Hope, and Jacob Tootalian.