check your alert.log see if the instance was initiated at all. I said SGA because when I faced this problem in NT was because large pool was full but this was because I was using MTS and my large pool size were too small.
When I start the db from svrmgrl I didnt get the error message you got, instead it simply says oracle not available and when i tried to shut it down it just hangs even with abort option however these shutdown and process were still recorded in alert.log I fixed it by shutting the db from services and start from either svrmgrl or services.

I cant think of anything since our situation might be different