The chances of actually making use of our notes, as I’ve argued earlier, hinges heavily upon their immanent visibility – being able to see which passages have been marked and annotated within a text without having to cross-reference separate documents. Shifting to digital text not only achieves the enhanced searchability and modularity we saw in the shift to digital notes, it also allows us to streamline the two by layering annotation and citation within the same document.

Once I got accustomed to reading on a screen, I began taking notes in Adobe Acrobat, which enabled me to write paragraphs worth of notes on a specific page or within a specific highlighted passage and to save these annotations directly to the PDF file itself (rather than some proprietary layer of metadata like many of the off-brand PDF annotators). The marginalia, in other words, was no longer “marginal.” The most ephemeral traces of every reading could be inscribed in the moment and with the greatest possible detail within the text itself.

This shift to intratextual annotation was largely solicited by my coursework, which was heavily invested in the close reading of individual texts. While this method served me quite well for a few years, I eventually rediscovered the antitheses of intra- and extratextual annotation when it came time for qualifying exams. Intratextual annotation sacrifices the ability to categorize citations from multiple sources necessary for more comparative projects and classes. However detailed and graphically intuitive my PDF annotations might have been, they still relied on the document being opened to the page in which this content was embedded. Skimming through the pages of Absalom! Absalom! I could see acute instances of catachrestic, figural excess in Faulkner, but I could not see how these resembled or differed from the works of other American and Continental Modernist authors or how Faulkner fit into a larger historical picture. I could expound Freudian motifs in any given work, but I could not easily juxtapose these with the passages to which they were referring or other works making similar references. Some form of compromise was in order if I was to have any hope of writing coherently about the hundred or so texts on my lists. As devoted as I had been to the askesis of close reading, I realized that my workflow was not sufficient for this more comprehensive task (essentially, the task of thematization against which I had bridled for so long). Faced with the reality of academic evaluation and advancement I needed to thematize…or fail!

This was the birth of what I would later call the citation tree. After I had read and annotated the various PDFs on my list, I would copy-and-paste the citations from the PDFs into a categorical, outline scheme structured around the individual author and the specific work. Unlike the intratextual citations I had been doing in Acrobat, where all of the thematic associations could be listed together within the embedded annotation, the citation tree required that one of these themes be declared primary and all of the citations associated with that theme, subordinate.  I considered the idea of pasting the same citations under multiple categories, but found that this caused more trouble than it was worth (in much the same way that the shortcuts I used to bridge categories in the folder tree strained its overall functionality). The difficulty of manually maintaining and syncing multiple instances of citations made it easier to limit each instance to one thematic heading even though it compromised the more democratic representation of themes embedded in the PDF notes (where a more nuanced sense of the connections between thematic elements were grouped by the citation itself rather than the primary theme).

I was surprised to find that the flexibility and general clarity gained by making these difficult thematic choices overall outweighed (or at least mitigated) the perils of thematization. With a larger sampling of citations from a specific work – about 30 per text – the categories grew more nuanced than they had been in my previous class notes and essays. A larger number of thematic subdivisions applied to a larger sampling of citations allowed for a more concrete negotiation of the conceptual territories involved in each. The citation tree not only forced me to prioritize themes that would easily translate into exam essays, it also required me to justify the inclusion of passages that I felt were important despite their lack of exam relevance.

Transferring notes from inside the texts to an external outline was greatly facilitated by digital text, but it also revealed a number of problems. I must confess the citation tree never really bloomed. One of the simplest reasons for this was that, after a certain point the mass of text became too large to properly handle in a Word document. The number of possible branches (i.e. subcategories) was limited to the number of indentations that could be accommodated by the standard width of a page. PDF pages that appeared to be a perfect facsimile of the printed book might be riddled with errors in generated by the ClearScan OCR I used to convert the page scans into searchable, selectable text. Eventually, the spelling and grammar checking ability of Word gave out all together and the entire document grew too unwieldy even for me, let alone my exam committee. This all became quite clear when one of my advisers attempted to actually print a draft of the citation tree that would probably have required an actual trees worth of paper to complete. He informed me later that his son was the one to intervene and cancel the job, explaining to him that this was not the kind of document intended for print.

Copying these citations into the citation tree also meant effacing the contextual relevance of the intratextual annotations that tended to emphasize more discrete phrases. The tree made it possible to survey and recall far more material than before, but at the expense of the kind of emphasis that might be inferred from within the context of the page. In order to preserve some of this context I had to expand the citation range to the paragraph level or beyond. While I tried to preserve the emphasis of the previously highlighted phrases by putting them in boldface, this was a step backwards in several respects.

In short, neither mode of annotation could successfully model the complexity of themes without diminishing context. Faced with this material resistance, we must reconsider how to effectively communicate context to an audience removed from the graphic supplement of the page. This discrepancy between the broader citations made for dissemination (teaching, conferencing, discussion) and the more acute citations needed for rhetorical and linguistic analysis is one of the core infrastructural problems. If these are already present in my own relatively small project, how can we hope to accommodate networks of users working on multiple projects with varying styles of citation and annotation? What kind of infrastructure might model the concentricity (and eccentricity) of citations and annotations in a way that would enable many readers to interface with one another in a collective textual environment?