Akademik Veri Yönetim Sistemi
Tezin Türü: Doktora
Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Enformatik Enstitüsü, Bilişsel Bilimler Anabilim Dalı, Türkiye
Tezin Onay Tarihi: 2010
Öğrenci: DUYGU ÖZGE
Danışman: DENİZ ZEYREK BOZŞAHİN
Özet:The study combines offline techniques with online reaction-time experiments, for the first time in testing monolingual Turkish children, in order to compare the mechanisms and strategies employed by adults and children of a head-final language with rich inflection and variable word order. In addition, the study presents two off-line experiments investigating the comprehension and production strategies employed by Turkish-English bilingual children and Turkish monolingual children at the ages of 5-8. A series of experiments in this study confirmed that the subject-object asymmetry that has been reported in several other languages, as well as in Turkish, has also been observed in Turkish speaking monolingual and bilingual children in terms of their comprehension and production of Turkish relative clauses. In all of the experiments, both monolingual and bilingual children showed a better performance in subject RCs compared to object RCs. Moreover, the monolingual children presented a very similar pattern to adults in some of the experiments, which was taken to indicate that the subject-object asymmetry cannot be caused by a single factor but rather it arises as a combination of multiple factors such as ambiguity concerning the function of a lexical or morphological item, ease of local attachment to a verb (a la Gibson, 1998), deviation from the canonical word order, frequency, and perceptual factors, among others. While investigating the underlying causes of this asymmetry, the study also focused on some of the hypotheses offered to account for the strategies used in sentence processing, such as the Filler-Gap Hypotheses (Maratsos, 1974; Wanner & Maratsos, 1974; Fodor, 1978; Clifton & Frazier, 1989; Frazier, Flores d’Arcais & Giovanni, 1989; O’Grady, 1997; among others), the Parallel Function Hypothesis (Sheldon, 1974), and the Canonical Word Order Strategy (Bever, 1970), and showed that none of these hypotheses can fully account for the processing facts from Turkish relative clauses. The study argues that the present findings could be best accounted for in a constrained- based, lexicalist, and incremental framework. Two such accounts were discussed in the thesis. One is Steedman’s (1989, 2000) model of a processor with a highly lexicalized grammar, a bottom-up parsing algorithm, a mechanism that evaluates multiple sources of information in line with the parsing model of a specific language. The other is a processing model by Vasishth and Kruijff (2001) that uses a highly lexicalized grammar, a combination of a top-down and a bottom-up algorithm, and a complexity metric inspired by Gibson (1998) and Hale (2001).