458 LINDA SCHWARTZ
attributives and identificationals presented here, the non-subject core argu
ments of the construction seem not to have syntactic argument status at all
— they do not take oblique case or a preposition. It was proposed that
these be analyzed as being incorporated into the predicate in the languages
under investigation, accounting for their lack of argument marking and
behavior and also for their agreement properties. Moreover, the attribute
and identification-set arguments would seem to be natural candidates for
incorporation, due to their non-referential interpretation. In Dakota, since
there is no copula, attributives and identificationals take on full verbal func
tion, admitting agreement morphology, while in languages where a copula
is used in these constructions, agreement marking (if present) is distributed
over the predicate complex.
We might, however, expect that not all languages would incorporate
the non-subject argument into the predicate in such constructions, and this
expectation would seem to be borne out in a language like Hausa. In
Hausa, attributive and identificational expressions take several distinct
forms, as illustrated in the examples in (73)-(75).
(73) a. Audu malami ne.
Audu teacher-MASc STABILIZER-MASC
"Audu is a teacher."
b. Motar nan sabuwa ce.
car this new STABLIZIER-FEM
"This car is new."
(74) a. Littafi yana da nauyi.
book he-is with heaviness
"The book is heavy."
b. Audu yana da karfi.
Audu he-is with strength
"Audu is strong."
(75) Bala yana a gajiye.
Bala he-is at tiredness
"Bala is tired."
In the examples in (73), the second element of the expression, categorically
a noun, follows the subject, and a "stabilizer" appears in final position,
inflected for number and gender (the so-called stabilizer does not have any
properties which are exclusively verbal). These structures show no evidence