Committee: Alec Marantz (chair), Liina Pylkkänen, Maria Gouskova, Nizar Habash (NYUAD, Computer Science Department), and Diogo Almeida (NYUAD, Psychology Department)
This dissertation sheds light on two important aspects of Arabic morphology: the status and representation of roots and that of templates (or word patterns). The main purpose of this dissertation is to investigate the role of these controversial word constituents in the representation and processing of the non-linear morphological structure of words in two varieties of Arabic: Emirati Arabic (EA) and Modern Standard Arabic (MSA).
Chapter 2 presents a linguistic investigation of the verbal morphological system of EA, with some focus on how it compares to the MSA system and in light of the main claims made by Doron (2003) in her account of Hebrew verbal templates. In this chapter, I provide arguments for separating the verbal root from templates and affixes in the analysis of Arabic and, by extension, Semitic morphology. Specifically, I argue that verbal meaning is a combination of at least two syntactic heads: Voice and little v, and a lexical head: the consonantal root. I further show that the interaction of the semantics and argument structure of the root with little v and Voice within a unified syntactic structure captures the regularities as well as the exceptions in the interpretation of the verb forms of EA.
Chapter 3 presents experiments using the subliminal speech priming technique, as developed by Kouider and Dupoux (2005) and used by Schluter (2013), and magnetoencephalography (MEG), as applied to auditory morphological processing (Ettinger, Linzen, & Marantz, 2013). Both these techniques are relatively new and are applied to EA for the first time. The subliminal speech priming technique has the advantage of tapping into the earliest stages of auditory word recognition, allowing an investigation of the relevance of the consonantal root in the processing of the EA verbal forms. The results of this study suggest that the discontinuous consonantal root in EA is an independent lexical unit, a finding consistent with root-based models of Semitic morphology. The results also point to the effectiveness of the subliminal speech priming technique as a promising way to explore unwritten dialects of Arabic such as Emirati Arabic.
In Chapter 4, I exploited a standard visual lexical decision task with concurrent MEG recordings to explore the neural correlates of morphological decomposition in MSA by examining the early stages of visual word recognition in this language. The results obtained show that words in MSA go through the same stages of word recognition, beginning with extraction of the consonantal root, in a manner completely parallel to the decomposition of words into stems and affixes in concatenative languages like English, arguing for the obligatory decomposition model of word recognition in Arabic non-linear morphology. The results of this study also hold significant implications for the long-lasting debate surrounding the status of the consonantal roots in Arabic and morphological theory in general.
The behavioural and MEG studies reported in this dissertation support the hypothesis that morphological decomposition in language processing is the same across non-linear and affixal languages, ratifying the linguistic analysis of non-concatenative morphology as essentially affixal. Additionally, the general results of this dissertation add new and original support for the claims that the Arabic consonantal root is an essential cognitive unit in representations and processing.