EZAFE IN THE CONTEXT OF CPS: Evidence from three Iranian languages

The present study investigates the distribution of the Ezafe (EZ) morpheme in adnominal clauses in three Iranian languages, namely Persian, Kurmanji (Northern Kurdish), and Zazaki, demonstrating that the behavior of EZ in these languages challenges the case analysis of EZ, suggesting instead a compatibility with the inversion analysis of EZ. In contrast to the prediction made by the case analysis, it is shown that EZ is required before [–N] modifiers such as CPs. First, EZ is consistently used in restrictive relative clauses (RCs) in all three languages, with Persian using an allomorph of EZ in this context, contrasting with Kurmanji and Zazaki, which use the regular form of EZ. Non-restrictive RCs present divergence: while Persian does not allow EZ in this environment, Zazaki and Kurmanji employ regular and anaphoric EZ (AEZ) forms, respectively. Following de Vries (2006), non-restrictive RCs are treated as restrictive RCs with a silent head, aligning the distribution of EZ in these languages with its distribution after a silent noun: Persian lacks EZ here, while Zazaki and Kurmanji use EZ and AEZ, respectively. Second, in Noun-Complement Clauses (NCCs), Kurmanji and Zazaki consistently use regular EZ, while Persian offers two options: the allomorph of EZ used with CPs or no EZ at all. Two possible structures are proposed for NCCs, one with and one without inversion, attributing the distribution of EZ to the structure involving inversion. This study shows that EZ is present in CP contexts and its distribution in these languages follows from the general behaviour of EZ and the syntax of N-CP structures.