Processing covert dependencies: Turkish wh-in-situ

Abstract

This study presents a reading experiment and a production experiment on Turkish wh-in-situ constructions. These constructions establish a relationship between a wh element and its scope position without displacing the element from its original argument position; thus, they are referred to as covert dependencies. The results suggest that covert dependencies are resolved actively, similar to filler-gap dependencies that have reliable cues that signal the existence of a dependency. This preference for resolving dependencies at the first grammatically available position is paralleled in production results. However, an unexpected pattern is observed in the production study, where declarative sentences are preferred over questions even though the length of the dependencies between these constructions are the same. This brings the question of whether there is a bias for statements over questions independent of dependency length optimization. A follow-up to Experiment II that includes context scenarios and controls for the frequency of different verb types can clarify the source this bias.