in: Angles of Object Agreement, Andrew Nevins,Anita Peti-Stantic,Mark de Vos,Jana Willer-Gold, Editor, Oxford University Press, London , London, pp.340-362, 2022
In this chapter I propose an explanation for why oblique objects in Turkish
pattern with canonical, accusative-marked objects in relativization out of active
constructions, but with subjects in relativization out of passive constructions. I
propose that this is because the choice of the relativizing suffix in Turkish relative
clauses should be viewed as a consequence of complementizer agreement. I argue
that C surfaces as -(y)An when the valuation of its uninterpretable features is a
result of agreement with a single goal, and that it surfaces as -DIK when it undergoes
successful agreement with multiple goals. This analysis derives the atypical
behaviour of oblique objects in relativization out of a passive, as well as the more
general contrast between canonical subject relative clauses, which require -(y)An,
and canonical non-subject relative clauses, which require -DIK. In addition, the
analysis proves correct in deriving a number of exceptional configurations, where
the verb of the relative clause is affixed with -(y)An rather than with -DIK, even
though a non-subject is relativized.