Below is the introduction to the written methodology for the project:
It is difficult to summarize the contents of Ezra Pound’s 1930 poem A Draft of XXX Cantos. It is literature in the most classic sense, but it also includes passages that are informational in the most banal. It is both Pound’s original work, and a collection of fragments sourced from an amorphous global archive. It is both high modernist poem and a history of the world, by turns focused on moments of imagist perceptual clarity and catalogs of bureaucratic documents. Rather than untangle these definitional threads, my project, using the Text Encoding Initiative’s encoding language, seeks to describe the integrative forces that brought these materials together into the pages of a single volume of poetry. This task begins by identifying the elements themselves, tracing the traditions and texts Pound drew from and the figures and places from literature and history that appear in the poem. After the elements have been defined and encoded, the TEI provides an additional tagging structure through which related elements can be connected to one another. This will provide an agile tool for tracing patterns within a massive and highly complex modernist text. The TEI encoding and then the methods of presentation of that data for users seeks to capture and express the textual, media, and informational dynamics of a poem that generates meaning through the relationships between diverse elements.
The first step in the process required creating a digital text. For this, I scanned the XXX Cantos, then OCR’d and edited it using ABBYY FineReader. I imported the plain text into Oxygen XML Editor 17 and attached structural tags. These include canto divisions (), empty page break tags (), and line group (), and line () tags. The initial encoding process included tagging each person and place in the text using and tags. Initially, I did not understand how complicated this would become. Because Pound draws from multiple texts in the majority of the cantos, he often quotes or adopts alternative spellings of a figure’s name. I ended up deciding to build a separate prosopography file and create a unique xml:id for each person appearing in the text. Within the inline tag, I included the @ref attribute to point to the prosopography file for every appearance a given name. For instance, when Ovid is referred to by the Italian spelling of his name, Ovidio, in “Canto 20” the tag in the primary text appears this way: Ovidio (20.8). At present, the prosopography only contains the English spelling of the person names. In the future, I plan to include additional information for each figure, including aliases, titles, terms of address, and, if it proves valuable, to point individuals to the companion gazetteer file I have created. The gazetteer is quite similar to the prosopography, and the inline encoding of the primary text will include @type attribute to define whether it is a city, country, river, mountain, region, or neighborhood and the @ref attributes pointing to the xml:id in the gazetteer file. Each entry will also receive a regularized spelling, as often as possible, in the language of the region in which it exists.
The most complex aspect of the project is encoding the bibliographic information for each source Pound drew from and included in the text. In his critical writings, Pound was adamant that he did not want to explain why a quote, allusion, or reference appeared in his work. Furthermore, it quickly becomes clear that he made little attempt to reproduce texts in their original. The best example is “Canto 1” in which he altered the text he drew from in subtle ways. The majority “Canto 1” is comprised of a translation from the Nekyia passage from Book 11 of Homer’s Odyssey. However, Pound does not translate into English from the Greek, but from Andreas Divus’ Medieval Latin translation. To complicate the issue, Pound alters the text further by rendering it in Anglo-Saxon syntax. Representing these various layers of transmission in TEI is in not impossible, but it requires a consistent and transparent encoding structure and presentation method that makes the factors contributing to how the text appears clear to a potential user/reader. For now, details of this kind will be included in notes that are linked within the primary XXX Cantos file. This practice will extend to those areas where Pound renders sections of text in exaggerated dialects and accents.
There are several structural problems as well. My primary goal for the entire project stems from my sense that the XXX Cantos is a vast modernist poem comprised of many different elements. I was attracted to the TEI because it provides a way to encode spans of text that are addressable by different processing languages which would in turn provide multiple options for visualizing and organizing relationships within the text. The context in which an element appears, as in the Eleanor/Helen example from “Canto 2,” is essential for understanding how it relates to other sections. Creating simple indexes of names, places, and source texts is not as useful as being able to represent sortable and arrangeable spans of text. In “Canto 1” it is fairly simple to demarcate where the section that derives from The Odyssey begins and ends, but in other sections it is much more difficult. With careful reading and Terrell’s annotations, it is possible to identify the intertextual references, but many areas in which Pound filters references through dialects, or when they are mingled with other texts, require significant interpretative work to determine where to place an opening tag and then the closing tag. Some sections, such as the Malatesta Cantos (“Cantos 8-11”), draw from an array of sources recounting the same events from different perspectives. In many cases, there are varying degrees of certainty about which source Pound is referencing at any given point. As of now, I have reached provisional solutions for these challenges and the tags surrounding spans of text will undergo revision as new scholarship provides a better sense of the text’s references.
I decided to use line group element () to indicate where an intertextual reference begins and ends. Line groups tags, because they are meant for flexible structural encoding, are not confined by rigid ancestor and child hierarchies, meaning I was able to apply them pretty freely throughout the text. To keep the linking structure simple, I did not create a separate bibliography file and instead built it after the primary text. Each entry includes xml:id’s, title, author, publication, edition information as it is available for each reference that appears. I inserted @corresp attributes with the inline elements, to link the encoded span of to the corresponding entry in the bibliography.
The tags do come with some sensible structural rules. The intertextual references in the XXX Cantos do not always come in large, multi- or even full-line chunks. Often, they consist of a phrase or single word that comprises only part of a line. For example, in “Canto 5” Pound inserts the Italian words “ciocco,” an allusion to Dante’s Paradiso (5.17). As makes sense, line group elements do not fit within single line elements. My solution in these cases is to use the segment element (), designed to encode smaller sections of text, and to use it as I have with the line group element, including inserting @corresp to point to the bibliographic entries.
Linking Associated Elements
Identifying the people, places, and intertextual references is only the first step toward how I visualize the project. The next step is to gather a set of tags that will allow me to build links between elements to reveal the associational patterns within the XXX Cantos. In order to link the textual references to one another I will introduce the interpretation element () which I will add within each line group tag associated with the spans of text. This allows one element to participate in multiple well-defined relationships. The inline element features the analysis attribute (@ana) which can contain an infinite number of references, allowing me to link a single intertextual reference with multiple categories of association. The @ana attribute then points to a list of interpretive categories () each with its own xml:id. As an example, including the element with the span of text from “Canto 2” in which the old men discuss sending Helen back to Greece can be linked to the bureaucrats opposing Sulpicia in “Canto 25” and any other spans of text tagged with the “old men’s voices” xml:id included in the list. In addition, I will use the element with the @ana attribute to refer person and place names to categories of figures I will list in the . In this way, Odysseus, Sordello, Malatesta, and other figures can be linked through their participation in various types of associations. Odysseus with Sordello through their shared “shapeshifting” characteristics and Odysseus with Malatesta through the “hero” because they both behave bravely.
The interface I envision for this encoding of the XXX Cantos is a direct result of my reading of the text as a collection of elements related to one another in multiple ways. By encoding relationships between elements, I have created pathways that fork in multiple directions through the text, but instead of creating a hypertext version or a fundamentally different media object, I see the encoded text as a supplementary tool for reading the XXX Cantos in print. Pound was highly aware of how his poem would appear on pages of a print book and capitalized on his medium’s capabilities. In order not to interrupt contact with these calculated textual effects, I want to preserve the reader’s sense of the text in print, while providing a tool for reading it in alternative ways. I envision a user having the print text before them while using the tools my project places at their disposal. The primary interface I am proposing for the encoding will feature several menus: one each for the person and place names, the bibliography, the gazetteer, and prosopography files. Selecting a person or place name would return all the locations in the text the selection appears. A selection of a specific intertextual reference, quotation, translation, allusion, from the bibliography menu will return all the spans of text in which the selected reference appears. In addition, I will include menus that list the thematic tags I have listed in the list.
A useful precedent for this last function exists in the Algernon Charles Swinburne Project created by John A. Walsh. On the “Tools and Tactics” page of the project Walsh writes, “Through successive re-readings of the poems, I identified the following core set of concepts and images–familiar to readers of Swinburne–found throughout the volume … I then devised an encoding strategy, using standard TEI elements and attributes, to describe the set of core themes and identify instances of those themes in the text.” My goals for the this aspect of the interface are nearly identical to what Walsh describes. By tagging major themes in the XXX Cantos, multiple paths through the text open up and connections between different elements and figures are revealed. In order to ensure the project’s usefulness to the largest number of potential users, I have tried to keep my thematic tags fairly general. For example, each instance in which a key concept such as “usury” appears can now be accessed and tracked within the text by a user of the interface. The problem that remains, is that since the text is under copyright, it cannot be presented online. I have built a structure that I hope will allow me to present the data I have encoded without the copyrighted text itself.
I have encoded each line of the poem with an xml:id which includes the canto and line number: line 122 of “Canto 2” is encoded as <xml:id=”c2.122”>. A user attempting to track references to The Iliad would never be able to see the full text of any reference appearing in the XXX Cantos, but would be presented with the canto number and beginning and ending line numbers in which these references appear beginning with “Canto 2” lines 14-22 where the Trojan elders discuss banishing Helen to avoid conflict with the Greeks. This passage would also appear if the user selected “old men’s voices” from the theme menu. A user, with New Directions print copy of The Cantos would then use the interface as a supplementary tool for reading, capitalizing on the affordances of both print and digital media.
Generating these types of links is not impossible in print, but it does require a significant amount of time and effort not only to generate initially, but then also to revise or extend. Besides having the potential to sort, arrange, and mix textual patterns, the features available through the project’s interface will be extensible. If I decide the thematic tags are not as valuable to users as more robust mapping and geolocation capabilities, altering the project will be easier than if it were a book. In addition, the extensibility of the TEI encoding provides multiple options for visualization tools as well as expanding the project into the remaining 90 cantos and Pound’s other work. The motivation for my TEI encoding of the XXX Cantos originated with an interpretation of the text as a single text constructed from many interrelated elements. The text has a complex informational and thematic coherence that connects the elements together in addition to the material binding of print book in which they appear. I have no desire to alter the text as Pound designed it, but I have constructed a digital tool that encourages and facilitates alternative pathways through the poem. To this end, I have generated and applied a TEI schema that identifies, encodes, and relates the different elements in a way that takes advantage of the capabilities of digital media to enhance interactions with a complex modernist poem.