Typically, I think - and I ask this as a question - college age composers tend to have tremendous bursts of inspiration, and be almost incapable of composing when not inspired. As your psychology changes in your 20s, you start thinking less of individual moments (or melodies, or motives) and more about strategies for entire pieces (like chord progressions or rhythmic structures). Then it becomes easier to just sit down and start writing, inspired or not, and at some point inspiration creeps in and lifts the piece off the ground.I tend to start with some sort of catalyst, some sort of initial inspiration, usually a melody or more abstract notion. Once that first melody is generated, then I can work away at it with a little less inspiration. I often find it difficult to return to pieces, especially if I've listened to a lot of music in the intervening time between sittings - my headspace and my relationship to the music is different. One piece of advice I've taken to heart is something Dave Douglas advised me at Banff a couple of years ago: never assume that because it's already on the page it's completed and set in stone, and that the best way to re-evaluate one's decisions is to re-copy the piece by hand. Re-writing it forces a re-thinking: do I really intend this? or is there another, a better way to achieve this effect?
In some cases - more and more frequently, actually - I try to set out objectives for myself to achieve in a piece. Sometimes it is a strictly musical challenge - writing reggae-influenced pieces without resorting to one-drop or dancehall in the rhythm section; sometimes it deals with a sound world or mood I want to achieve; sometimes I try to write a piece the entire opposite of everything I've written, like the one I'm working on now for BMI - uptempo and rocking. The success rate varies, and at certain points the music takes on a life of its own and may move away from the initial concept. I'm alright by that.
Gann wonders if it's even possible to teach composition, and the most successful composition lessons I've had dealt with process and headspace more than anything else. Usually it's one very simple piece of advice that opens a new door of perception. I very rarely write at the piano, or on any instrument, simply because I was advised to write as much as I could in my head and on paper and then move it to the piano if necessary. Later on, I found that if I write directly on Sibelius, I wind up taking the easy way out, whereas with pencil and paper the music is a lot more intentional. Michael Mossman, Don Byron, and Dave Douglas all advocate demanding certain questions of a piece before it is written, and by doing that one narrows down possibilities. I don't always start with those questions, because unlike Douglas I still do hear melodies in my head and indulge them, but once I hear those melodies I try to discover their universes.
I have had some very nuts-and-bolts composition lessons at BMI, courtesy of Jim, Mike Abene and Mike Holober, and usually it's more to do with the "lost in translation" pitfalls of orchestration than anything to do with the structure of the piece itself. Forcing myself to bring in substantial amounts of new material has made me aware of my clichés/formulas/preferences (the term varies depending on how self-critical I'm feeling), which is possibly the best composition lesson of all. I'm starting to self-identify as a composer though not solely so; I love playing too much to ever entirely leave it behind.
Lately, I've gotten into drawing inspiration from literature and film. Not by necessarily writing programmatic music, though. One piece I wrote for the trio, "Bella," was inspired by hearing Caetano Veloso's voice singing the Neruda poem (from The Captain's Verses) in my head. I "transcribed" the first stanza and worked from there. I guess one could call it a text setting, except it's not for voice. I had a similar sensation when I read Jorge Luis Borges' "Heraclito." I still need to set that.
I'm also tempted to try and utilize an organizational process I discovered by reading Murakami's Hard-Boiled Wonderland and the End of the World, a desire buttressed by watching David Lynch's Inland Empire last night. Both Murakami and Lynch set up organizational structures that initially seem fragmented, but as they progress the reader/viewer becomes aware of their properties, and some sort of unity is achieved at the end. Well, not entirely - the clues are fairly obvious in both Murakami and Lynch, but they still wind up being complete mindfucks.