Advances in Role and Reference Grammar

ATTRIBUTIVES AND IDENTIFICATIONALS 457

(71) a. M odin gorod ne byl čist.

not One City-MASC.NOM.SG NEG WaS-3sG.MASC clean-MASC.SG

"Not one city was clean"

b. *Ne bylo cisto ni odnogo goroda.

NEG WaS-3sG.NEUT clean-NEUT.SG not One City-MASC.GEN.SG

(72) a. Studenty zdes' ne duraki.

student-MAsc.NOM.PL here NEG fools

"The students here are not fools"

b. *Studentov zdes' ne duraki.

student-MAsc.GEN.PL here NEG fools

Although the situation in Russian is more variable (less lexically or syntacti­
cally fixed) for the objective/subjective distinction than it is in some other
languages, a tentative hypothesis is that only arguments intended by the
speaker to be interpreted as typical undergoers (i.e. typical theme/patient)
allow encoding with the genitive of negation. This would account for its
occurrence with direct objects, passive subjects and intransitive verbs with
a single undergoer macrorole but will not allow attributives and identifica-
tionals into this class because of their non-typical undergoer status. That is,
like n-cliticization in Italian, this construction has associated with it a strict
view of theme. Dakota and Russian would differ in that the subjective-
objective distinction is made at the lexical level in Dakota but at the contex­
tual level in Russian. Furthermore, we see once more in Russian that
attributives and identificationals as a class will fall into the default member
of the opposition.

4.4 The syntactic argument status of attributes and identification-sets

In section 2.1, it was claimed that the previous localist treatment of attribu­

tive and identificational expressions was one in which two thematic rela­

tions were assumed to be present in a syntactically intransitive structure.

An alternative schema was presented here which also assumes the same two

thematic relations but in reverse: subject = location, attribute/identifica-

tion-set=theme. In section 2.2, it is claimed (Van Valin 1991) that there

may be more core arguments in a clause than macroroles. This discrepancy

is utilized by Van Valin to account for instances of dative case-marking in

Icelandic when accusative case was expected. Briefly, a core argument not

linked to a macrorole is assigned dative case as a default. In the analysis of