statement: Are avant-garde practices still important to you?
My main artistic concerns have involved my poetry, of late — meanings stuff for journals and books— but I continue to be influenced by experiments by artists associated with Futurism, Dada, Concrete poetry, the Oulipo and the Situationism, at least as concerned my linguistic practices that involve the visual image.
In fact, a side effect of the above project with Finnegans Wake will be a series of conceptual videos that show the program in process as it weaves through its various decodings — i.e. the viewer will “read” Finnegans Wake through watching the program consider various options as it “spell-checks” the text. I feel this is as valid a way to “read” Finnegans Wake as those proposed by Burgess, Campbell and Cage.
Most of my digital text projects are alternative “readers” — “The Dreamlife of Letters,” “Kluge” and “Suicide in an Airplane (1919)” for instance — as I still want my audience to try to parse and understand the text.
I’ve stepped back from doing digital work, largely because so much is being done now with a much higher level of programming expertise than I could ever hope to acquire, namely in the film world but also the art world, but there’s still a space for non-commercial experimenters to wrap ourselves around projects that are conceptually interesting even as they might never gain tremendous popularity.
project: Finnegans Wake, Spell-Checked
Using Python, I’ve written several programs that “spell-check” Joyce’s forbidding novel. These programs are presently rudimentary: they unpack portmanteaus, try to turn the famous 100-letter long “thunder words” into recognizable strings, try to translate foreign or specialized words into common English, and so forth. But the plan is to add several additional levels encompassing syntactic, phonetic and semantic elements of the text.
There’s a pretty substantial history of artworks that use Finnegans Wake as a source text, notably John Cage’s Writing Through Finnegans Wake (1976) and Writing for the Second Time Through Finnegans Wake (1978). There are also a number of non-traditional scholarly books that intend to “open up” Finnegans Wake, notably by Anthony Burgess (of Clockwork Orange fame) who published A Shorter Finnegans Wake (1967) and by Joseph Campbell (known for The Hero With a Thousand Faces) with his Skeletons Key to Finnegans Wake (1944).
The Initial “Spell-Check”
- “Correct” the text against English: unpack portmanteaus, try to turn the famous 100-letter long thunder words into recognizable strings, try to translate foreign or specialized words into common English, and so forth.
- “Correct” the text against other dictionaries of languages with which Joyce was known to be familiar. Available NTLK (Natural Language Toolkit) word lists include: Catalan, Czech, German, Greek, English, Spanish, Finnish, French, Hungarian, Icelandic, Italian, Latvian, Dutch, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Swedish and Tamil.
- Finnegans Wake has been translated into French, German, Greek, Japanese, Korean, Polish, Spanish, Dutch, Portuguese, and Turkish. Well-advanced translations in progress include Chinese, Italian, and Russian.
- One table in the database has an entry for every word in Finnegans Wake with a designation of where it appears in order.
- A second table will store words that could be “corrected” rather easily and a list (in array form) of every word possible correction, with low percentage words being discarded.
- A third table will store words that have been designated “portmanteaus”—words that are really two words smashed together (Lewis Carrol, who wrote Alice in Wonderland, invented the phrase). I’d like to be able to put words back together once they’ve been split and spell-checked, but also be able to have words “swapped out” when the semantic run is performed.
- Perhaps another table will store every word part — while still linking it to the portmanteau database — and as in the second table, will just have an array of high percentage options.
- I will run an analysis of sentences from Joyce’s early writings, notably Dubliners and create a set of “stereotypes” — common word orders in Joycean phrases (which can be recombined) and sentences. English is an SVO language, but it would be interesting to allow for SOV and VSO languages, such as Biblical Hebrew, Irish and Welsh with which Joyce would have been acquainted.
- Words in the sentences from the output will be parsed, tagged, and stored in the database. Most if not all of these sentences, with the exception of the very short ones, will not be parsable (asyntactic).
- Sentences from my output will be forced to fit into one of the Dubliners “stereotypes”: the “verbs” of the Dubliner sentences will only be replaced by verbs in my output, likewise the nouns by nouns, in the manner of a Mad Lib.
- If need be (most likely), some words will have to be turned into nouns, verbs, adjectives and other parts of speech to do so. Any word ending that provided the grammatical clue — in whatever language — will be turned into English. (The Latin genitive, for example, will become the English apostrophe s.)
- One of the most difficult steps will be to reproduce some of the rhythmical qualities and sound patterning of the original.
- This will rely on the CMU Pronouncing Dictionary. Daniel Howe’s “RiTa” function libraries will be helpful here.
- Ultimately, the spell-checked version of the test should sound satisfying — intonations denoting exclamations, questions, and so forth. — when read by an electronic reader (naturally, this is subjective).
- This level will involve semantic analysis of my output to determine what are the main overtones of a chunk — largely using war imagery? erotic imagery? bawdy humor? — after which the program will swap out some words that generally have a similar percentage of occurring in common English — carp, tarp and harp, for example (I’ll make a database of all of the options from the first run) — and swap out words about, say, a fish and replace it with a musical term.
- The kinds of “swap-outs” that occur on this level will have to be done, to some degree, in conjunction with those of the phonetic level. As for which takes precedence when will be a matter of trial and error.