This post responds to Matt's post from yesterday.
One of my original hopes when I started writing my dissertation was to compile an anthology of women's ekphrasis in order to answer the question, "how have women (and, frankly, men, too) looked at and responded to the visual arts with poetry?" Currently the "tradition" of ekphrasis (words representing images) has been theorized solely on the basis of works authored by male poets. In order to see if current theorizations are representative of the whole, we need to recover ekphrastic works by women. The most difficult, of course, to find are pre-Romantic poems by women.
In order to find examples of ekphrasis by women, I've had to literally stand around in libraries and skim through volumes and volumes of poetry looking for "catch phrases" or words or literary devices or, frankly, titles that seem to indicate that the poem is referencing a work of art. My "dream" would be to try to train a computer to do what I'm doing, but better.
Here's what I imagine: based on contemporary poems come up with terms, phrases, contexts, figures of speech, etc that are commonly used in ekphrasis (could you teach a computer to identify "descriptive language"?.. I suppose this would be research question #1). Then, based on the texts entered in the database somehow come up with a robust series of features that "typify" poems about visual subjects. )I suppose you'd have to go through a series of checks and balances... the computer makes suggestions... a human being determines whether or not the poem fits the criteria... tells the computer if it does or doesn't... the computer essentially "learns" from this process and modifies its results based on what is included and excluded from the generated lists.)
An activity like this would be interesting because, of course, one of the most difficult methodological questions, I think, in interart studies is finding a way to make sister art comparisons. Also, I think the process itself would spark rich and interesting questions about the "how" part of my question. "How" do poets represent images with words? The question is vexed with all kinds of interpretive presuppositions... but watching how the computer identifies "ekphrastic" language would teach us about the "how" that we can't see.
Visualization of the information in this instance might be more informative for the researcher than descriptive language... because description of description... well... seems only to be more derivitive, but perhaps illustration of description (translating back to the image through language) would be more interesting.
The end result of attempting this task, I imagine, would be two-fold. 1.) an important work of recovery (the anthology) and 2.) an attempt to answer "how" words represent images.
Posted by c_jane at February 22, 2005 1:38 PM | TrackBackThanks, CJ, a very useful example.
Can you put your finger on a particular electronic corpus (i.e., one that you know exists and that you could get access to) that you would like to mine for ekphrastic texts?
Well, I think that several of the databases at the UVa Electronic Text center would work, for example:
1. The English Poetry Database (AD 600-1900)
2. The 20th Century American Poetry Database (in order to build a up the criteria and do test runs, b/c we know more about what's out there and what kind of results we "should" get)
3. The African American Poetry Database
4. British Poetry 1780-1910
5. The American Poetry Database to 1900
I wish I knew more about the holdings at the Brown Women Writers Project, because I think that would be an excellent place to start working on recovery of women's ekphrastic poems... of course, that depends on whether or not they have volumes of poetry and what their criteria are for marking up poems.
The sad truth is that I've sat in on lectures by Julia Flanders so often... and yet I never actually learned what kinds of holdings they have. I tried to access the database, but even the table of contents required a login and password.
I think, though, that these represent a few of the most easily accessable databases. Of course, I would need to do more careful reading and analysis of the criteria for adding texts to the database (many of them say "major" authors of the period and a few others... but frankly, I'm not entirely sure what that means in practical terms).
What would be interesting is to do the same kind of mining across several different databases and see whether or not the computer "learns" different things about how poets represent images based on the database that's being mined. In other words, would a database that included "major" (male) authors produce different algorithms or what have you than a database that included only lesser-known women poets. That would be cool.
{longtime on-again, off-again lurker jumping in. Hi!}
This is an intriguing idea. My reading series (www.debaun.org) is about to do an ekphrastic event, pairing poets with visual artists and commissioning new poems for a joint reading/gallery showing. In reviewing applications with ekphrastic poetry, I've not come across anything like a standard vocabulary or language set. I'd thought this was a good thing - some of the submitted work was observational on its visual inspiration, some told a new story, some got into the mind of the painter, etc.
Am I interpreting ekphrasis too broadly? If not, how would you limit the language in any way to permit any kind of automated trawling?
I really enjoy and learn from what you write here, and hope to have the chance to read much more in the future!
David,
I wish I had exactly the right answers that I could rattle off to you about how to trawl for ekphrasis... but I don't. I think this is really the whole crux of what a project like this would be. But let me back up and answer your first question first.
Are you interpreting ekphrasis too broadly? The short answer is no. The definition of ekphrasis that many scholars use is "the verbal representation of visual representation." This is, of course, a problematic definition... especially if you take into consideration The New York School of poets, or those who align themselves with surrealist or abstract or abstract expressionist painters. It is also problematic, for example, when you consider that ekphrasis rightly *should* include verbal representation of arts such as pottery, quilts, landscape design, architecture, etc. These are verbal representations of forms of art persay but not verbal representations of visual representation. By not including these kinds of art forms in the ekphrastic perview, we manage to cut out poems about art in the domestic sphere... art which many women prior to the 20th century used as a "feminine" means of gaining access to subjects that were considered "too masculine" for a "lady."
Perhaps that moves too far afield from your question, though. What I think you're asking is, what are the conventions of ekphrasis? I did make a post to this effect a while back, but I'm not even going to make reference to that because I really need to redo and update it. The kinds of things you mention (e.g. "observational on its visual inspiration, some told a new story, some got into the mind of the painter, etc.") are considered ekphrastic conventions. In other words, they may narrate the image they represent, they may attempt to represent the visual with the physical presence of the language, they may speak for the artist or for the work of art (prosopopoeia), and they may even try to describe the work of art in the attempt to make it visually present for the reader (enargia). These are conventions of the genre. Another one is that by looking outward at the visual art, the artist is often compelled to make inward reflections and map the self onto the "other" of the painting.
To try to answer what I think is another part of your question: yes, this is one way to recognize a poem (or other writing) as ekphrastic. Of course, the easiest way is if you know (as you do) that the poem is ekphrastic because the poet tells you so. Either the poet at a reading says what the artwork was that was the subject of the poem, or uses the title of the artwork or artist in the title of the poem (or title of the book)or in the dust jacket to the book it indicates, these poems respond to work at such and such gallery... This, of course, is the easiest and most reliable way. This would be the first step, by the way, that I took while standing in the library flipping through volumes of poems. However, as you're likely to find, not all ekphrastic poems advertise themselves as such... and many of them are really good poems... so I had to think of another step in the process.
Everything else required closer interpretive work. I (could, and do, and have) read each poem and identify it according to its use of a convention, asking myself what conventions, if any, the poem attends to. This, of course, works, but it takes an incredible amount of time to do, and by its very methodology reinscribes the conventions as the "only" conventions of ekphrasis... So I start wondering, what happens to those poems that break the mold? How do we find them, include them, and account for them? What does this all say about our definition of the genre to begin with? The great irony, of course, being that I don't even know if they exist to begin with.
Anyway, I think the way to start answering this question is to ask more questions... For example, How do readers identify description? How do we identify speaking out for something that has no voice in a syntactical way? I suppose that's the place to start.
So, this brings me to your second question, what kinds of things do you teach a computer to trawl for in order to identify ekphrasis. Really, this is a question begins by addressing how the mechanics of language reflect the fact that they are contemplating visual media. This is not a question I can answer readily... in fact, it's one of those things that has kept me in the dissertating stage for... well... I don't even want to discuss for how long.
And, for now, that's where I'd have to stop until I could do more research. And this, of course, is what you were asking in your second question. The whole idea of "mining" for information is still very, very new to me. I'd have to learn more about the actual way the whole thing works before I could come up with a real answer to this. Unfortunately for now, I can't do that.. I have to read the set of papers in front of me and get through the prospectus for the dissertation I'm actually writing. And, this post was responding to a post by Matt Kirschenbaum asking what my wildest technological dream was regardless of whether or not I thought it was possible... What you're really asking is, is it possible?
For now my answer is, I hope so... maybe... but I need to have time to think about it... which I don't.
Oh, and by the way, David, thank you for reading!
OK... I swear... this and then I'm going to go do work...
I read all of Barbara Guest's _Selected Poems_ to try to "come up with" the poems that I thought were ekphrastic. The first time through, I missed the one that I've pretty much decided to use "The Farewell Stairway." Now, I *should* have had a clue by the epigraph "after Bella"... but not having the Italian artist in mind... I just missed it... and this is a poem in which the speaker narrates the descent of 3 women down a vortex-like stairway... I just plain miseed it. Now that I think about it, all the clues were there. The title presents a snap-shot image, the epigraph (a useful tactic, given that ekphrasis has its origins in the split between epitaph and epigraph), and the ensuing narration that visually graphs out the process of decent. But I missed it.
What I'd hope is that the computer, being a rule-monger, would at least pull this up as something to consider... because it has an epigraph that starts "after" (many epigraphs that contemplate an image indicate that the picture came first and then their poem follows... the way in which it follows is often contemplated in the poem).
Um... so that's something, I guess... Wow... I'm sooo not doing my work right now. Back to those freshman papers. In the words of a video game I heard the other night... Aei Karumba!)