The XP position in (4) may
be filled with pro, which is fully referential as commonly assumed. NC
structures are thus inherently imbued with referentiality. Even without an overt
lexical complement, e.g. as elements of the verbal complex, NCs thus suffice to
(5a), neither referential subject nor object NC markers are present, which
yields an infinitival interpretation.
(5b) contains an NC marker which introduces a second person subject. Since there
is no object NC marking, ‘see’ seems to have an intransitive feel here, as
the translations suggests. In (5c), the NC marking for both subject and object
are present, so ‘see’ is interpreted as transitive.
subject and object NC markers are fully referential, they need to be linked to a
previously established discourse referent. Introduction of a new discourse
referent or shifts from one discourse referent to another are typically executed
via full overt NPs (cf. Taylor 1994: 98 and Wilkendorf 1994: 17; both
investigate discourse structure in N¨maandE,
NC markers, on the other
hand, are used by themselves to express continuity in reference (Taylor: 98,
Wilkendorf: 20). Thus, NC markers gain their referentiality by virtue of being
discourse anaphors (Canonici 1995: 31).
has independent pronouns, so-called ‘absolute pronouns’, but they are only
employed to put additional emphasis on the discourse referent:
the subject is marked with an absolute pronoun, the verb can be visibly focussed
(probably via movement to initial focus position), which yields a contrastive
intepretation emphasizing the action:
literature generally agrees that the reason for the emphatic character of
absolute pronouns lies in the independent referentiality of the noun class
marking on the verb.
 To say that there is no AgrS may actually be too strong. As will become clear later, one may argue that ku- is, in fact, an expletive-like placeholder in AgrS. In any case, the example at hand does not contain typical referential agreement.