about OBJ(theta) as a controller

[Alex Alsina]

OBJtheta is just the name that is given since 1989 to what was previously
known as OBJ2.  On this assumption, English is an example of a language in
which OBJtheta functions as a controller in functional control.  Examples
illustrating this point (such as "Tom will serve you the fish raw") are
given in Bresnan's "Control and Complementation" (1982: 323).  (The idea
is that, in this example, the NP "the fish" is the second object in this
ditransitive construction and it controls the subject of the predicative
phrase "raw".) In Joan Bresnan's 1982 theory of control, obliques
(OBLtheta) were excluded from being possible controllers in functional
control, whereas second objects (which later became OBJtheta) are
explicitly included among the possible controllers. 

The reason why one might not expect an OBJtheta to be a controller in
functional control is indicated in Helge Lodrup's reply: semantically
restricted grammatical functions should not be able to satisfy a
semantically unrestricted grammatical function.  In English (and most
languages), the controlled grammatical function is a subject, a
semantically unrestricted grammatical function since it may correspond to
any thematic role and even to no thematic role.  Thus, we don't expect an
oblique to control a subject, because an oblique is semantically
restricted, the restriction coming from the preposition, case marker,
c-structure position, etc., and so it is unable to satisfy a semantically
unrestricted function such as a subject.  If the subscripted *theta* in
OBJtheta means the same as it does in OBLtheta, namely, semantically
restricted, we would expect OBJtheta's to share with OBLtheta's the
inability to be controllers in functional control.  Given the kind of
example mentioned earlier, this doesn't seem to be the case: OBJtheta's
can be controllers, just like SUBJs and OBJs.  This clearly indicates to
me that it is a mistake to group second objects and obliques as the class
of "semantically restricted grammatical functions".  If there is some
substance to the notion of semantically restricted grammatical function,
it should not include second objects (or OBJtheta's). 

Alex Alsina