Methodology
The process for choosing images to add to a text’s database begins first by highlighting the text with words that may be either of interest to contemporary readers in terms of providing more context or those which are likely to result in a wide array of images for consideration, even if the meaning appears straightforward. As the sole person making these decisions for the proof-of-concept, I am fully aware of the subjective nature of this process, so I have tried my best to err on the side of highlighting anything that might be of interest to the contemporary reader of these texts. In Marlowe’s Hero and Leander, for example, “uplandish country” on line 80 of the First Sestiad does not seem as though it would yield particularly interesting results, but I found that two “uplandish” illustrations feature people, not countryside, and these give some clue as to who we might imagine the “uplandish” peasant—here imagined as “melting” in view of Leander’s beauty—to be.[1] It is easy to imagine a reader assuming the meaning of “uplandish” in context of the poem’s discussion of peasantry, but two images in the current proof-of-concept complicate and enrich that reading. The work “An Oration Against the Vnlawfull Insurrections of the Protestantes of our Time” by Peter Frarin (1566) describes its printed image depicting a rebellious cavalry force opposing “princes and magistrates” as an “uplandish band,” while the frontispiece to Robert Greene’s “A Quip for an Upstart Courtier” (1592) depicts an “uplandish” figure of a peasant with a wide-brimmed hat who is holding some sort of farming implement over his shoulder, and this figure is contrasted with a well-dressed gentleman on the other side of the frame.[2] These two images, which appear in the current form of the proof-of-concept, provoke further thought on the concept of “uplandishness” at the periphery and its relationship to refined behavior at the center, as both juxtapose the uplandish figure(s) with princes, gentlemen, and others associated with the court. These images are not simply describing the uplandish person in a vacuum, then, but inherently subjecting them to the viewer’s judgment in comparison to the other figures standing literally opposite them in visual space. One might even say that both of these images position the uplandish as foils. One finds consistency between this representation of such figures and their appearance in Marlowe’s text, where the fact that the extremity of their uplandishness is nonetheless susceptible to Leander’s beauty is a measurement of the totality of that beauty. The possible association of the upland peasant with revolt also connects to the following lines of the poem, in which the “barbarous Thracian soldier, moved with nought,” is similarly suggested to be moved by Leander. The “uplandish” peasant and “barbarous” soldier may have been more closely connected in the early modern mind than one might otherwise imagine.
Most of the highlighted words in the literary texts are nouns, such as “sparrow,” “Cupid,” or “nun.” After selecting these words, I recorded them in the database next to the line in which they appear for later reference in assembling the platform in code. Then, they were ready to be employed as search terms on EEBO. In some very rare cases, I have altered the search term when appropriate: for example, I chose to search for “eyelid” instead of “curtains of thine eye” (Shakespeare, The Tempest, 1.2.412) to find more images faster, though I attempt to avoid any such changes as much as possible and I document these altered terms. As I will discuss below, some other key changes need to be made with compound phrases when turning them into search terms.
The next step in my process has been to use the selected words and phrases from the text as search terms in EEBO’s “Advanced Search” page. Many, if not most, of the texts in EEBO have been digitized either by hand or with the aid of Optical Character Recognition (OCR) software, making them completely searchable. I have found that it is best to use the default “anywhere” search for my terms, which searches the entirety of a document for the word or phrase in question, including the title, author name, subtitle, the text block, and any notes. Utilizing the advanced search tools, I am able to specify a number of additional criteria as filters, and as mentioned above, I have always specified a time period and the need for results to be texts that feature some sort of illustration. Though established time periods are never satisfactorily coherent, my choice for the date range criterion has been the years 1550-1660, for several reasons: these years cover Elizabeth’s reign and the reigns of her successors up to the Civil War; they cover the earliest and latest materials in my proof-of-concept, such as Marlowe’s Hero and Leander and Milton’s L’Allegro; they cover a period of English textual production in which illustration practices were more or less established and illustrated texts were gaining in popularity; they likewise cover a period in which English visual culture was allowed to stabilize and develop under Elizabeth and even further under James and his patronage of Continental art; and they draw upon elements of visual culture that would not only have potentially informed the writing of the literary texts in question but their reading as they gained in popularity over ensuing years or even decades.[3] That said, however, it must be acknowledged that the choice to narrow by date is always problematic, not least for capturing everything that might matter for the task at hand, as it assumes simplistic, clean breaks to cultural phenomena where they almost never exist. Moreover, as scholars such as Jean H. Hagstrum have very rightly pointed out, one’s influences are not always contemporary, whether one is reading or writing a work (xvi). And yet, searching the entirety of EEBO also felt like a mistake; assuming, for example, that an image from 1720 would have much to offer to our understanding of Elizabethan poetry did not seem appropriate, even if the goal is not to provide proof of connection between a specific word and a specific image.
The search parameters thus defined, I should note some choices made in my process for preparing words and phrases from the text as search terms. It has been occasionally necessary to alter words or phrases taken directly from the text in order to find a list of useful results. Though I have tried to avoid such action whenever possible, I believe reasonable scholars will allow for a change such as rewriting the poetic phrase “pine-bearing mountain” (from Marlowe’s Hero and Leander, 1.116)—which only resulted in a result of Hero and Leander itself—to “mountain pine,” which turned up the rather exciting results of an emblematic image with a subscriptio dedicated to the mountain pine bent low by the wind, as well as two botanical illustrations of the mountain pine. The focus on the pines in this alteration of the search term has allowed me to locate what early modern readers might have imagined as the mountain pines that dotted those hillsides where Marlowe would like them to imagine centaurs would “gallop amaine” in his simile (Sestiad 1.115).
Likewise, it was occasionally necessary to modify EEBO’s standard method of looking “anywhere” in the text when searching for phrases instead of single words. EEBO seems to look for words independently of one another on its default setting, and as long as both are present in a document it will appear in the list of results. The problem with this method is clear when searching for a phrase such as “corners of the earth” (Shakespeare, The Tempest, 1.2.495): EEBO presents a list of results that include “corners,” “of,” “the,” and “earth,” which is not helpful when one is attempting to locate the visualization of the specific concept being referenced in the text. The remedy in these circumstances has been to omit the prepositions or articles and utilize a proximity search. In EEBO, this kind of search is accomplished by placing the “NEAR” command between the two words, and one may additionally specify the degree to which they are near one another: the command by itself defaults to four words away from each other, but “NEAR/5” would increase that to five, “NEAR/2” would require them only to be two words away, and so on. In this way, I ended up with a much shorter list of works, but one that contained the phrase “corners of the earth”—I found, as it happens, images of Eden, a geographical abstraction, and maps as the result of this search. Some alterations are similar but ultimately simpler: for example, I changed “Circe’s wand” (Marlowe, Hero and Leander, 1.61) to “Circe NEAR wand.”
Once a list of results has been generated, I have developed a process over time to identify the promise of finding a useful image in each one. This task is made easier by the fact that within the list of results, EEBO will highlight one or multiple excerpts that show the use of the search term(s) in context. While searching “anywhere” in the document, this feature has saved me a great deal of work in some cases, as I was able to recognize instantly that a text authored by someone with the surname “Temple” was not the kind of document that I needed. In addition, I have had to make some choices about what usage of the word to pursue further, and it would be worthwhile explaining some of that logic here, as best I can, though naturally each case varies. Line 126 of Hero and Leander’s First Sestiad, for example, contains the word “passions.” Of course, during the early modern period, this word had multiple common meanings: it could denote the Passion of Christ, a re-enactment of that Passion, a celebration of that Passion, the sufferings of a martyr or a tale of such, any kind of suffering, bodily disorders, and strong, even overpowering emotions (OED). When looking through the list of search results, it was necessary to decide the sense in which this word was used within the text:
In this case, the passions in question clearly seem to fit the last definition in that list, indicating overpowering emotions. However, so many texts in the results list were religious in nature that it seemed an obvious choice to leave them out in favor of those related to emotion and/or management of emotions (via one’s reason, etc.). I believe this was the right choice for the project, but it means that images of Christ’s passion and the passion of various martyrs will not appear for the user at all—in this case, I am thus purposefully limiting what may be “explored” in the platform based on my reading of the denotations and connotations of words and phrases. I judged the event of Christ’s passion to be outside of the realm of what should reasonably make up the visual context for this passage, and I am fairly confident in that choice, but it is always possible that I am mistaken. In several cases, I have been surprised by the results of a search and realized that I had not grasped or otherwise known a common usage of a given word or phrase in Early Modern England. Whenever it has seemed at all relevant, for this proof-of-concept I have attempted to incorporate these findings into the database by choosing a variety of images that may all provide useful or interesting context for a passage. It is also worth noting that I am rarely, if ever, in the position of choosing from so many images in EEBO for a given term that my choice requires me to leave out images from the database that seem clearly relevant. As I have sorted through thousands of documents over the course of this project, only a handful of times have clearly relevant images been so abundant that I have left a great deal “on the cutting room floor,” as it were, but the need to select at least two and no more than five images for the proof-of-concept should also be considered one of its limitations, and one tied to my own subjective choice with limited knowledge of the period.Poor soldiers stand with fear of death dead-strooken,
So at her presence all surprised and tooken
Await the sentence of her scornful eyes:
He whom she favors lives, the other dies.
There might you see one sigh, another rage,
And some (their violent passions to assuage)
Compile sharp satires, but alas too late,
For faithful love will never turn to hate. (121-128)
Likewise, before moving on from the database work, I should explain the process of searching individual texts for images related to the search term. In most cases, this process is by no means exhaustive, and because EEBO is not specifically built for this particular kind of inquiry, the most helpful results are not always at the top of the search list. The most reliable place to find a relevant image is in frontispieces, when the search term appears in the title of a work (EEBO places these first in the results order): usually a handful of documents, at most. When a word like “barbarous” (Marlowe, Hero and Leander, 1.86) appears in the title and can easily be matched to the depiction on the frontispiece or front-facing illustration of barbarous murder, that is clearly a strong and useful connection. These kinds of connections, however, are rare for most terms, and thus more in-depth examination of the works is necessary. After browsing any works in which the search term is in the title itself, I have been presented with a list, sometimes in the thousands, of documents that are not organized in a way that benefits my task—useful images may be buried on the last page of those results. In fact, it is often when I am about to give up after searching a hundred or more documents for a relevant image that I come across one that is extremely useful.
Searching within individual documents has been possible with the use of EEBO’s full-text search feature, available for a large percentage of documents. While a preview of the search term’s appearance is available in the list of results, it is impossible to discern from that preview its proximity to any potentially useful images. My method has thus been to use the full-text search to highlight each instance of the term appearing in the text, and to move through these instances looking for connections to images. The initial criterion here is proximity: if a search term appears within a paragraph or two of an image and a link seems likely, I take a closer look to assess the relationship between the term and the image. Identifying connections is simple for terms like “angel” but less so for terms like “sigh.” As an example, Hero and Leander 1.15 describes Hero as wearing a blue “kirtle;” the word “kirtle” may be found in the anonymously authored The Pourtraiture of His Royal Highness, Oliver Late Lord Protector (1659), in which there is a description of the king’s body and effigies. These descriptions are near the end of the document—a “Kirtle robe of purple Velvet” which is “girt with a rich Embroidered belt”—while the image of the king (his effigy) in the royal garments appears on the second page. Since these are located so far apart, I spent a significant amount of time determining whether or not the image portrays the same scene described later in the document; confident enough that the royal garments are the only ones being noted here and that the image also features Cromwell’s royal garments, I decided to include this image in the proof-of-concept for the word “kirtle.” As much as possible, my assessment has erred on the side of not including images that may have no relevance to the search term. If a connection between the term and an image seems sufficient, I then add it to my database. I should mention here again that this process is not exhaustive: for the sake of completing the proof-of-concept myself in a timely manner, I usually move on from the list of results once I have three to five images that relate to the term.
Moreover, I have attempted to create a few tools that will aid me later, including a list of images that seem specific enough to be labeled for use later, but do not relate to the search term at hand, and a list of heavily illustrated works that I have come across during my searches. An example of one of these works is the Arcana arcanissima hoc est Hieroglyphica Ægyptio-Græca (1613), which, in moving through its images, I have tagged with the keywords “osiris, typhon, isis, hercules, dionysus, ibis, apis, cynocephalus, ornament.” Though these separate databases are byproducts of my search for specific terms, they could be projects in themselves.[4]
Once they’re discovered, adding these images to my database is straightforward. I first catalogue the image’s “permalink” in EEBO next to the search term in my database, with a label corresponding to the order in which I found it (1, 2, 3, and so on, for convenience’s sake). Next, I download a high-resolution image of the document page in question, renaming this image file to match its database entry. Files are stored on my local computer and backups saved to the cloud. This process has thus far ensured that I am able to revisit the location in EEBO for any of the images that I have downloaded, and it means that when one clicks on an image in the proof-of-concept, it will take them to the document where that image may be found in EEBO.[5]
The next step with each image is to crop it properly for display, and then it is ready to be uploaded to the server and given a unique filename and location there that can be connected with the database. I have had to crop each image by hand due to the fact that EEBO only allows download of an entire page image, and often pages contain more than one image or part of the image needs to be cropped out for clarity. This final step allows the image to be called by the website and to populate what I am calling the “illustration panel” when the search term linked to it appears on screen or is clicked—I will return to this functionality shortly. The database is formatted in JSON for easy manipulation by JavaScript in the website. As for the data model, each JSON entry contains the word or phrase from the text, any alterations that were necessary to make it a better search term, a list of image locations (local to the server) and a list of links to EEBO resources that correspond to these images (external to the server). My own JavaScript functions are responsible for accessing the database when a user hovers over a keyword in order to populate the illustration panel with images from that specific entry, creating them as links that will redirect the user to their source material.
Having described the task of creating and adding to the database, I should conclude this section by acknowledging that this process is by no means objective. I am aware that these choices are editorial in nature, as I suggested in the previous chapter, and I believe they should be handled carefully. As Jerome McGann highlights in The Textual Condition, print scholarship has dealt with the problem of necessary subjectivity in editing for a very long time. Because all texts are subject to change as they are put to use (McGann 180-181), even (or perhaps especially) digital ones, a project like the EMVR necessarily transforms the texts it engages. A desire to provide as much transparency as possible about the process of creating those transformations in is why the EMVR links to this Scalar book.
[1] For reference, the OED shows period use for “upland” as a reference to the countryside and/or elevated inland territory, and for “uplandish” as a type of rusticity associated with rural peasantry.
[2] Both of these images may be found in the EMVR proof-of-concept’s Hero and Leander page for the keyword in question.
[3] If I were to follow the example of Ian Lancashire and his team in the construction of Lexicons of Early Modern English (LEME), I might include a range of 1475 to 1755. I've been more conservative here for the proof-of-concept out of caution, but I do see potential reasons to expand the range in the future pending more research.
[4] These side projects also suggest that a larger database of early modern images, whether simply those in EEBO or ones culled from a wider variety of sources, would be extremely useful for the continued study of visual culture in the period. As stated previously in the dissertation, it is only recently that assumptions about English visual culture in the period have shifted and placed more value on the products of a society that was previously thought to have an “impoverished” or “underdeveloped” sense of the visual. An accessible consolidation of existing visual artifacts would go a long way toward making more research on this visual culture possible, and digitization clearly affords that possibility.
[5] Other options for on-click events are navigating to a page dedicated to the image or linking to images with similar tags, both directions to consider for future development.