A SYNOPSIS OF ROLE AND REFERENCE GRAMMAR 117
Core junctures such as these raise an interesting issue with respect to
case marking: is the domain of the case marking rules presented in section
4.6 the core or the clause? In a simple sentence it is difficult to find argu
ments for distinguishing the possible domains, but in non-subordinate core
junctures the contrast between the clause as a whole and the individual con
stituent cores is clear-cut. Evidence that the clause, not the core, must be
the domain of case marking comes from ergative languages like Newari
(Genetti 1986) and Enga (Li & Lang 1979), in which the shared actor argu
ment in a construction analogous to that in Figure 30 receives ergative case
if there is a transitive verb in the construction, even if the verb in the core
in which the argument occurs is intransitive. Thus in a sentence like X went
to kill a pig, X would receive ergative case because kill is transitive, even
though it is in the same core as go, which is intransitive. Since go and kill
are in nuclei in distinct cores, it would be impossible to account for the
ergative case on X if the ergative case assignment rule had as its domain the
core in which the transitive verb occurs. If, however, case marking has a
