accidentals and substantives

xml/xsl: having your textual cake and eating the image too

Tomorrow Matt and I drive to Charlottesville to attend Blake Camp, the annual pilgrimage to the IATH roundtable in the recesses of Alderman library at UVa to strategize the future of the William Blake Archive. I'm already bracing myself for the perennial heated discussion of our textual transcriptions. Like many other thematic research collections on the web, the WBA delivers both high-quality digital facsimiles and carefully prepared ascii transcriptions of its primary source materials. The images enable the study of iconic codes that are unceremoniously flushed out of a low-res letterpress or ascii reproduction, while the transcription offers increased legibility (Blake's serried calligraphy can strain the eyes) and, perhaps more importantly, finely structured search queries and other character-based manipulation. These two demands--one pictorial, the other textual--are met by separate documents rather than unified in one. But the desire for unification is ever-present. The debate over the transcriptions nearly always revolves around issues of iconicity: how far should we go in trying to render the iconic codes of Blake's poetry and prose? Can we capture his vertical and horizontal spacing, his mirror-writing, his multi-directional text? One might suppose that because the transcriptions are accompanied by images, the former might not feel the onus of the latter. But in contemporary editorial theory, the word is never just a word, it is always first and foremost an image. So we strain the limits of current browser technology to deliver granular representations of course ascii marked up in sgml. Phenomenologically we want an image, but ontologically we want a text. Or, to put it another way, we want a machine-readable text and a human-readable image to coalesce in one and the same object.

The next phase of the project involves freeing ourselves from a proprietary software package called Dynaweb and converting all our sgml files to xml/xsl. Theoretically, the implementation of xsl stylesheets should finally allow us to have our cake and eat it too. I've seen stylesheet-rendered ascii that can float over other text, drop the ascenders and descenders of letters, and lay claim to the screen in a breathtaking variety of spatial configurations.

Theoretically words converged on images long ago. Technologically they have yet to do so, but certainly xhtml/css and xml/xsl style sheets are a step in that direction.

[For excellent thumbnail histories of words and images "joined at the hip" yet "enmeshed in histories of separation" (as Morris Eaves has astutely put it), see Eaves, "Graphicality," in Reimagining Textuality, ed. Neil Fraistat and Elizabeth Bergmann Loizeaux (Madison: University of WP, 2002) 99-122) and Matthew Kirschenbaum, "The Word as Image in an Age of Digital Reproduction: Lessons Learned from Humanities Computing," in Eloquent Images (Cambridge: MIT P) forthcoming.]