Finite-state sign language morphophonology


Thesis Type: Doctorate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Graduate School of Informatics, Cognitive Science, Turkey

Approval Date: 2015

Student: AYÇA MÜGE SEVİNÇ

Supervisor: HÜSEYİN CEM BOZŞAHİN

Abstract:

The aim of this thesis is to investigate the computational power required for processing sign language morphophonology. This dissertation focuses on the objective of reducing autosegmental representations and rules defined by three sign language phonology models, namely, Movement-Hold Model (Liddell & Johnson, 1989), Hand-Tier Model (Sandler, 1989, 1990), and Prosodic Model (Brentari, 1998), to finite state machinery. By adopting Autosegmental Phonology framework (Goldsmith, 1976), these models are capable of dealing with both simultaneity and sequentiality observed in sign language phonology and morphology. We suggest algorithms for transforming the autosegmental representations and rules constructed within these three models into state labeled automata of One-Level Phonology (Bird & Ellison, 1994). State labeled automata are known to have regular language power. By this reduction, non-linear representations of sign languages are shown to be serializable.