Animacy effect on Turkish object relative clause production among young adult native speakers of Turkish

This study investigates animacy influence on Turkish RC formation and its consistency in different tasks/contexts. Meanwhile, the interference of semantic (e.g. frequency) and syntactic variables (e.g. structural difficulty) in RC formation is also analysed. Therefore, a Picture Description and Metalinguistic Awareness Task were applied to two participant groups composed of young adult Turkish native speakers (nG1=46, nG2=58). After combining the data of groups, animacy effect on RC passivization in each task was tested by Chi-square test and their RC outcomes were compared by Kappa to check the consistency of the animacy influence. Finally, RC and statements in METU Turkish Corpus were analysed and outcomes of 105 RCs and 657 statements data were compared with participant RC use outcomes by using Two-Proportions Analysis to check a possible frequency or structural simplicity effect on RC formation. In the end, a significant effect of animacy was observed in both tasks; however, the comparison outcomes attested no agreement between the tasks in terms of RC preferences. Thus, two factors were assumed to have interfered in RC production process in either task: visual salience and task difficulty. Corpus RCs also demonstrated a significant animacy influence, which was initially associated with the fine-grain version of Tuning Hypothesis; however, the two-proportions test results indicated some other factors that could have influenced the RC preferences in corpus like genre or some other semantic (e.g. concreteness) or discourse (e.g. topicalization) factors. Meanwhile, corpus statements did not demonstrate any reactions to animacy, yet they were concluded to share more similarities with Object RCs than passive RCs in terms of argument structure and word order, which were presumed to be the reasons for high rate of active ORC preference in participant data and for the statistical similarities between participant RC and corpus statement preferences as shown by two-proportions test.