THEORY: Incorporating Agents vs. Patients in Verbs

This is a descriptive term for something that occurs in Basque. :) I've
been wondering what to call it for a while, since it's not incorporation
per se (the noun does not form a single word with the verb) but the bond
is clearly greater than a normal subject and object. Basically, there
are a number of phrasal verbs, often made with the verb egin "to do",
which occur with a noun that lacks the normally compulsory definiteness
/ number / case markers, and can't take adjectives. FOr example:

hitz egiten dut
word make-prog 3rd.abs-1st.erg
I make word = I speak

If this were a normal clause, this would be ungrammatical, and hitz
"word" would require the appropriate ending, in this case probably -ak.
This would give:

*hitzak egiten ditut
word-abs.plur make-prog 3rd.abs.plur-1st.erg
(I'm pretty sure this would sound strange at the very least to a native
basque speaker)

Nor can the noun in constructions like this take modifiers. For example:

lan egin dut
work make 3rd.abs-1st.erg
I made work
I worked

*lan handi egin dut
work big make 3rd.abs-1st.erg
I did a lot of work

 The bond seems to be looser than what you described though, because in
certain circumstances I have seen the stripped noun be parted from its
associated verb, although this does seem to be restricted. I'm fairly
sure that it's ungrammatical to insert other arguments or modifiers
between the noun and verb in the typical order noun verb AUX as shown
above, but certain processes like questions and negation generally
trigger the movement of arguments etc to after the AUX, and in these
circumstances I have seen the stripped noun parted from the verb,
although I'm at a loss to think of an example at this moment in time.
 Anyway, the stripped noun term is useful, and this process is extremely
productive in Basque for activities.

>Anyway, that's noun stripping.  If anyone wants further info on the
>typology of incorporation, I have an article or two in PDF that I
>could mail you.  (Including Mithun 1984, which is also available from
>jstor -- it's in Language 60 -- if you have an account.)
>
>-- Pat
>
>
>
>