Arabic broken plurals and prosodic circumscription: A theoretical and experimental investigation

This dissertation provides a full and unified account of the Arabic broken plurals. The analysis is based on and significantly improves McCarthy and Prince's Prosodic Circumscription account of the iambic broken plural in Arabic (McCarthy and Prince, 1990). Furthermore, the results of two experiments conducted with Arabic native speakers are presented.; McCarthy and Prince's analysis successfully accounts for many iambic broken plural patterns; however, it fails to account for a substantial number of other iambic broken plural patterns. Specifically, it is argued that the iambic broken plural in Arabic displays a prosodic circumscription process that is positive, where a prosodic constituent at the left edge of the singular stem is circumscribed. This prosodic constituent is subject to the word minimality constraint that limits it to the category minimal word, which according to McCarthy and Prince in Arabic is a quantitative or moraic trochee comprising two moras.; Contrary to McCarthy and Prince's claim, I argue that all the iambic broken plural patterns in Arabic are best accounted for by a prosodic circumscription process that is in fact negative, where any prosodic constituent that can be found at the right edge of the singular stem is circumscribed, starting from the smallest possible unit and moving up the prosodic hierarchy. Furthermore, I show that the new parametrical settings I propose successfully account for the Arabic trochaic broken plural as well.; The results of an experiment with adult speakers of Arabic, as well as one with children, yield numerous broken plural forms that do not conform to the prosodic circumscription rule, as proposed by McCarthy and Prince. Indeed, these forms are similar to the actual types of broken plural patterns that are unaccounted for under McCarthy and Prince's analysis.; I argue that the results of these two experiments, as well as the broken plural patterns that actually exist in Arabic, provide strong evidence that the parametrical settings proposed for prosodic circumscription are too constrained. They severely restrict the types of iambic plural patterns that are possible and incorrectly rule out broken plural patterns that are well attested in Arabic.