>>Are you trying to say we could add "verbs" as the topmost entry of
>>this hierarchy? A bit counterintuitive, but why not, if it works.
>I'm not saying anything like that.  The Nominal Hierarchy has
>to do with NPs' capability of functioning as agents.  Verbs
>can't function as agents.  What does happen is that the
>"agentness" of the verb arguments gets marked on the verb.

No, I know the nominal animacy hierarchy has several uses; it's just that 
according to what you quoted, verbs < nouns with regards to tendency for 
erg/abs rather than nom/acc marking; and nouns would be subdivided exactly 
according to their animacy. Furthermore, the existence of split-S languages 
shows that verbs also have a subdivision of active verbs < stative verbs. 
Yes, the actual marking ends up in the nouns, but it's the verbs that govern 

>If I recall Trask correctly Basque auxiliaries are
>ergatively marked.  Main verbs usually only exist in
>infinite form.

Now this was the sort of stuff I was asking about. So would this mean verbs 
proper < auxiliaries?

I also thought of a setup similar to what you were suggesting for Sohlob. 
You added an inst/abs section above the erg/abs section, thus getting three 
divisions of the scale. Another way to get three divisions would be to 
instead add something between nom/acc and erg/abs. Frex, nom/acc agreement 
on verbs, tripartite pronouns, and erg/abs nouns.

I can't just now think of anything that could be added below nom/acc, nor if 
analogous ditransitiv alignment division would be possible, but I'll keep 
the idea in mind. I have a project where something like this would fit 
rather well.

John Vertical