Subject-gap relative clauses are generally easier to process than
object-gap relative clauses, but the underlying causes for this Subject Gap
Preference (SGP) are difficult to disentangle. Are subject gaps better
because they are in the most prominent structural position, or because they
are associated with easily-processed unmarked case (nominative)? We present
reading-time data from Georgian to distinguish these hypotheses. Having a
split-ergative case system, Georgian totally dissociates case and
grammatical position. Nevertheless, we found strong evidence for a
structural source of the SGP. No matter case associated with the gap site,
subject relatives were more easily processed than object relatives.