Picture Becomes Text, Becomes Writing: Software as Interlocutor

By and | 1 November 2012

While co-teaching a class titled TOO MANY COOKS at UnderAcademy College, our common interest in writing with the assistance of spell-check emerged. Sonny Rae Tempest’s Daily Motion Art blog post ‘A Picture Worth 11,739 Words’ provoked our idea to collaborate on a presentation for students during the course.

Allowing Microsoft Word and other programs or processing techniques to have non-trivial control over the content of a piece of writing is nothing new. In the mid-1990s, poets such as Joel Kuszai began to conceive processes which employed Word to help perform authorial tasks through the use of spell-check operations, and many others have invented methods for using the device as a compositional tool. In these mechanical poems, software serves as a type of interlocutor that sustains the writer’s experimental objective.

What is required in any type of spell-check writing is an input text, which can be anything prepared in a language a word processing program (e.g., Microsoft Word) understands. Though many approaches are possible, generally speaking an author proceeds by removing from the input text anything that is not an alphabetic character, including spaces, so as to have a block of letters to work with. From there, the block is uniformly recast into word-sized fragments – typically by use of automated processes (see A D Jameson’s HTMLGIANT blog post ‘Another way to generate text #1’ for a description of his lesson, which involves using macros for breaking text into chunks).

One can take many approaches in styling such works. In most cases, unless a strict replacement order is imposed, the end result combines objective aspects, such as Word’s analysis and consequent suggestions, with the author’s subjective choices. For this issue of Cordite Poetry Review, we derived our text, ‘Exit Ducky?’, from the issue’s cover images: a triptych of blurred faces.

After acquiring the alphanumeric code of the image by opening it in a text editor, the data was translated (via translate.google) from Chinese to English. The original coded output consists of thousands upon thousands of random characters. Encoding the binary in Chinese, a pictographic language, ensures that each character becomes its own individual word. Any other language encoding, except Japanese, will not accomplish this. We removed all non-alphabetic and non-basic punctuation characters, and then working with Word’s spell-check mechanism stylised the text. In this example, which is rare, the machine translation produced an abundance of excellent phrases. Thus, in addition to our spell-check work we also engaged in preserving and editing some machine-translated text.

This simple but time consuming process blends the creative and uncreative. The exercise obviously contains destructive qualities, but we prefer to emphasise its multi-level transformative properties. Allowing the software to dictate, at least in part, or steer the direction of this type of writing serves to provide the author with unexpected vocabulary and unforeseeable textual encounters in which compositional decisions must be made. One text, through programmatic filtering, expands into another. It is worth mention, however, that despite our use of a number of software programs (and different versions thereof) to conduct this text, the number of hours we humans spent shaping it for Cordite numbers in the range of dozens. In the end, it is by no means trivial ‘Uncreative Writing’. Certain forms of late-stage literature veer wildly from norms.

     Exit Ducky? Exit Ducky?

This entry was posted in ARTWORKS and tagged , , , . Bookmark the permalink.

Related work:

One Response to Picture Becomes Text, Becomes Writing: Software as Interlocutor

  1. Pingback: Reused, rewritten – <code>text