Indeed, thick formations of apparently decomposing hot potassium carbonate would form around the cathode. Sometimes they would detach from it and burst at a distance, or deflate with the emission of hot particles. I imagine that the potassium vapor formed as suggested in the PDF you linked would rather quickly form KOH.
I found interesting that heating of some sort was still going on and sufficient to cause K2CO3 to heat up that much in relatively large amounts.