This thesis takes the adaptation of Arabic loanwords into Turkish as a case to reflect on and contribute to the ongoing debate of loanword phonology of the Perceptual approach (Boersma, 2009; Peperkamp & Dupoux, 2003; Peperkamp et al., 2008; Silverman, 1992), Phonological approach (LaCharité & Paradis, 2005; Paradis, 1995; Paradis & LaCharité, 1997, 2001, 2008; Peperkamp et al., 2008; Silverman, 1992) and a medial hybrid model of both phonetics and phonology (Kenstowicz & Suchato, 2006; Shinohara, 2004; Smith, 2006; Chang, 2008 and Dolus, 2013). The thesis includes two types of data: corpus-based and experimental. The corpus of the Arabic loanwords into Turkish comprises 1118 words from which vowel mappings and residual effects of gutturals on neighbouring vowels were identified. Based on the concept of uniformitarianism (Murray, 2013) present-day sound changes must have been governed by the same principles or laws which operated in the past. Thus, one of the goals of this work is to model the grammar of Osmanlica speakers in the perception of modern day Turkish speakers of the residual effects of vowels neighbouring gutturals.
In these effects the Arabic vowels /a/ and /u/ are adapted as /a/ and /u/ in Turkish vowels neighbouring guttural sounds (emphatics, uvulars and pharyngeals); however, the vowel /i/ is borrowed as the Turkish high back unrounded vowel only surrounding emphatics and the uvular q and as /i/ elsewhere. It was concluded that the corpus data patterns can be best accounted for by using a hybrid model of phonetics, phonology (of both source and native language) and with the effects of orthography. In addition, the role of bilinguals as the active borrowers in the adaptation process is especially corroborated.
