A paradigm gap in Turkish

In this thesis, I report a paradigm gap in the agreement paradigm of desiderative constructions formed with the suffix -(y)AsI in Turkish. I argue that 3PL agreeing desiderative verbs are defective by providing empirical evidence from an acceptability judgment experiment (N = 181) and corpus frequencies. To explain the existence of the gap, I first provide an analogy-based account which states that base identification procedure within the paradigm fails to reliably select a base for producing the 3PL desiderative form. That is, I argue that there is a competition between two surface forms that are both attested in speaker productions due to indeterminacy of base identification. Then, I provide an alternative account that can explain the gap in the framework of Distributed Morphology. Comparing these two accounts, I argue that the analogy-based account is better suited than a Distributed Morphology account to explain the gap in the light of the empirical findings. Besides the desiderative gap, I also discuss other gaps in the paradigm of associative plural constructions in Turkish that require different analyses. Using the tools of Distributed Morphology in this case, I argue that these gaps can be best explained with a combination of a local dislocation operation and a ban on the adjacency of two plural features. Considering the different kind of gaps reported for Turkish throughout the thesis, I suggest that our morphological models should allow for both probabilistic and deterministic generalizations.