Abstract:
This study addresses the processing of Genitive (GEN)-Possessive (POSS) long distance dependencies in Turkish in both Noun Phrases (NPs) and non-finite embedded clauses. The self-paced reading paradigm to investigate two factors that might play a role in determining the processing complexity of the dependency at issue: the overt presence of the GEN-marker in the sentence and the linear distance between the GEN-marked noun and the POSS-marked word, which can be a noun or a nominalized verb. To that end, we compare three models on sentence processing; namely locality accounts (Hawkins, 1990; 1994; Gibson, 2000), anti-locality accounts (Konieczny, 2000; Kamide & Mitchell, 1999) and content-addressable retrieval (McElree, 2006; Phillips, Wagers, & Lau, 2010). The results indicate NPs and non-finite clauses do not exhibit a complete parallelism with respect to the processing of the GEN-POSS dependencies. While the overt presence of the GEN-marker leads to more processing load on the POSS-marked noun in both domains, the effect of the linear distance between the GEN-marker and the POSS-marker is not uniform. The linear distance does not affect the processing complexity in NPs whereas it eases the processing of the nominalized verb in nonfinite embedded clauses. To conclude, this study shows that each dependency relation might have its own dynamics and a particular processing model might not be able to account for all types of long distance dependencies.
