- Gather your texts first off and put them all into one giant file.This may be a bit scarey particularly if you have worked hard on them but don't worry, keep your original files in case this all goes pear-shaped.
- Now use normal word processor techniques to subvert your content:
- Use outline view and make a poem of your titles, switch them or use headings to bring lines together in surprising ways;
- Make a list of your bugbear words, mine are things like still, just, quiet and transparent and they crop up in many a poem; so set up a search and replace and see what results you get;
- Set up some random coloured text styles; define your poems in this way and then group the text based on colour. Who saw that coming?
These are workshop exercises so keep an open mind; they may not make poems but you may notice some themes or openings in your work that you didn't before.
But this is where things can go hardcore with some text mining techniques: these are actually the kind of methods used to analyse masses of text for example at spy listening posts to pick up if we citizens are talking about naughty or illegal stuff, by search engines to increase the stickiness of links or management consultants pick out the gist of a massive database.
- So this time get your big file at the ready and pick out some mood words, love, hate, cherish or some colours and see how often you use them using Find.
There are more technical ways of doing this by setting up a database (sentiment analysis) but we are not doing that now. I discovered that I hardly used colours at all in my poetry a few years ago and this resulted in a profound change in my writing.
- If possible search on terms that you think define your writing or better still you fear might be omitted. How many poets particularly women poets are writing about their work for example? Is that off limits? Or can't we admit that we do something other than write and quite often enjoy it?
- Now it is the slash and burn; this is a big ask but go for it. Text mining uses extraction techniques to glean key points, themes and clusters from text. It is possible to do this by popping your work into XML form for example to enable semnantic weight and indexing. But us lazy poet geeks want instant satisfactions so here goes.
- Cut every fifth line! Yes, just go for it. Or third. Or the sixth and seventh every time. Wow! I know. Butchery of your beautiful hard-earned work but this gives you a sense of freedom. It is the old cut up approach of the Dadaists but performed on your own work. What is the result? Odd conjunctions of thought, a better understanding of your vocubulary and its tropes, new insights or glimpses of what could be. (And if the thought scares you, remember to ensure your old files of those poems still exist separately before you make the big working file for these techniques.)
- One final idea from here, although to be honest, there is plenty more you could do. I like this one a lot though. Strip out all your punctuation using Search and Replace; if you are really hard core make everything lower case using the Font function too.
At this stage, I might convert all my text into ballad form and sing it. Whatever. The result will be a different way of looking at your work. The form you choose to redefine it is infinite. A sonnet. A haiku. A ghazal. Tweet poetry.
Here is a link to a dryish definition of text mining from wikipedia and there are some software links if you have money to spend on this http://en.wikipedia.org/wiki/Text_mining.
If you want to experience free form without the hassle of the big file approach, go and get your old newspapers like William Burroughs and all, snip them up and reassemble.
This is fun too if you only want to play: http://www.nightgarden.com/shannon.htm.