Resumptive pronouns in Baha Arabic: an experimental study

The phenomenon of resumption has been a central topic of debate for both syntacticians and psycholinguists. The debate particularly centres on whether resumption is a syntactic phenomenon or a processing one and whether this characterization differs across languages (and dependencies).
Theoretical and experimental investigations have revealed that resumption exhibits a great deal of variation across (and sometimes within) languages (Sells, 1984; Sells, 1987; Asudeh, 2004; McCloskey, 2006; Asudeh, 2012). This is further complicated by diverging interpretations of the data. Specifically, while the theoretical literature presents resumption as a wide-spread cross-linguistic phenomenon affecting longer distance dependencies and island-violating dependencies, this claim is not confirmed in experimental literature (Farby et al., 2010; Keffala, 2011; Tucker et al., 2019; Perpiñán, 2020, among others).
This dissertation examines the phenomenon of resumption in Baha Arabic, a language with a productive use of resumptive pronouns across different types of dependency structures. It aims to investigate the theoretical literature’s argument that resumption is preferred to gaps in certain syntactic configurations (i.e. islands and longer dependencies) and to elaborate on the extent to which resumption in this variety of Arabic differs from resumption in languages like English. Four experimental studies, exploiting both offline and online methods, are conducted. The findings suggest that resumption in Baha Arabic is not a uniform phenomenon, despite the argument that it mainly has a syntactic function; (i) true RPs in illi-structures constitute part of the initial derivation in binding dependencies (as morpho-syntactic features of C do not trigger movement) and (ii) intrusive RPs in wh-questions are utilized as last resort devices to fix derivation problems when movement is illicit. The availability of intrusive RPs as a syntactic last resort device is restricted to wh-questions featuring inherently D-linked fillers ‘i.e. which-fillers’.
This dissertation, furthermore, examines the extent to which the type of wh-filler phrase (which vs. what) affects the acceptability and processing of island-violating dependencies in Baha Arabic. Although no such claim had previously been made for Arabic varieties, we found that gapped island-violating dependencies with which- fillers are accepted more than gapped island-violating dependencies with what-fillers. This amelioration effect is interpreted as reflecting an extra-grammatical phenomenon.
Though not the focus of the dissertation, our findings suggest that islands are neither a purely syntactic nor a purely processing phenomenon and that a combination of both cognitive and syntactic constraints contribute to it.