[quote="J.M.D."]
One thought I had some time ago was to morphologically specify whether (transitive) verbs are affectual or non-affectual, and fold the above construction into the class of transitive, non-affectual verbs along with verbs like "see".  So, sentences like "Tom listened to the radio" would have the same construction as "Tom saw the radio", and verbs like [i]look, talk, rely, think[/i] and others would all take direct objects without having to invoke a complex menagerie of prepositions.[/quote]

In my gjâ-zym-byn, this isn't marked on the verb itself, but it's marked by the object postposition.   There's one postposition for patients affected by the action of the verb, one for objects of attention that aren't affected but are present/within range of senses, one for topics of thinking or feeling verbs that might not necessarily be present, one for objects of quest (including a variety of verbs like "desire", "request", "search for", etc.), and so forth.

I've been gradually working through the lexicon documenting the argument structure of each verb, what postpositions or conjunctions mark its arguments (if a verb can have a clause as its argument, either subject or object, there might be restrictions on which conjunctions can introduce such an argument clause) and which ones are required and which optional.  See for example here

http://www.pobox.com/~jimhenry/gzb/nxcgtx_categ.htm#kyr

Since it's an engelang of sorts, I'm trying to make the argument structures consistent within semantic domains, e.g. all verbs of thinking or feeling have the same argument structures; but I'm finding it tricky.  Different verbs of saying/speaking seem to want different argument structures.

Some of the distinctions English makes with different verb roots gzb makes by using alternate optional argument structures for the same verb.

I'm thinking a future incarnation of my other engelang might fully specify the argument structure of each verb root in its phonological form, but I haven't worked out the details yet, and it might turn out to be too difficult.  Even though it doesn't have nearly as many morphosyntactically distinguished theta roles as gzb or Ithkuil, there are several possible ways arguments can be marked (ergative, absolutive, focus, locative, and subordinate clause), and the combinatorial set of all those being required, optional, or not allowed could be pretty large.