Rob Haden <[log in to unmask]> writes:
> I was wondering what your methods are for deriving nouns from verbs.  More
> specifically, how to derive instances of verbal activities from verbal
> roots/stems -- i.e., "(a) kick" from "(to) kick".

Tyl Sjok can also produce the patient and agent regularly.  It uses a
particle to do this.  Very easy structure: agent before predicate,
patient after predicate:

   jo   = I
   ljot = to ingest, (the event of) ingestion
   i    = <participant particle>

   jo ljot.     -- I eat/drink.
   ljot jo.     -- I am eaten.

   i ljot       -- the eater
   ljot i       -- the 'eatee' (haha: E.T. :-)), ie., food

(There is no morphology, so these processes all work on clause level.)

One of my current conlang sketches (project name 'S7') does it
completely differently, but also regularly.  It is head marking and
thus marks the valence at the head.  It will be able to use these
valence markers to also produce agent and patient participant 'entity
states' (no nouns in the language).  But I haven't found out how yet.