A split approach to concord in Arabic
by Feras Saeed
Date of Examination:2025-08-12
Date of issue:2025-12-11
Advisor:Prof. Dr. Hedde Zeijlstra
Referee:Dr. Sascha Alexeyenko
Referee:Prof. Dr. David Pesetsky
Sponsor:Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), RTG 2636 “Form-meaning mismatches”, Project number 429844083
Files in this item
Name:Thesis_revised.pdf
Size:2.09Mb
Format:PDF
Abstract
English
This dissertation develops a novel approach to the split in nominal features on postnominal modifiers. The empirical focus of this study is the noun phrase in Modern Standard Arabic and the modifiers under focus include adjectives, participles and relative clauses. The core phenomena under investigation are: (i) the split in nominal features which characterises the patterns of concord mismatches, (ii) the directionality of Agree with postnominal modifiers and cases of misplaced concord, and (iii) the variation in the word order of multiple (and different) postnominal modifiers. With respect to the first phenomenon, I investigate several instances of concord mismatches in the Arabic noun phrase which are all characterised by a split in nominal features. That is, the mismatches on the modifier always target a combination of two features, either number and gender or definiteness and case, and no other combinations of these features can be the target of concord mismatches. I examine such splits in nominal features in three contexts: (i) when the modifier is embedded in a relative clause and displays partial concord in number and gender, (ii) when the modifier follows a plural collective/nonhuman noun and displays partial concord in definiteness and case, and (iii) when the modifier occurs in between two nouns and displays split concord, agreeing with the first noun in definiteness and case and with the second noun in number and gender. The investigation of such splits directly bears on the ongoing debate in the literature concerning the location of concord in grammar and how features are distributed in the nominal system. In this respect, I propose a split approach to concord, arguing that modifiers that show feature-splits are derived from a predicate source and therefore have two domains for the valuation of their features. While number and gender are valued in a subject-predicate configuration, definiteness and case are valued in a noun-modifier configuration. Under this view, the split in nominal features depends on whether the modifier enters an Agree relation in one domain but not the other (partial concord), or if it enters Agree relations in both domains but with two different nouns (split concord). I argue that an analysis along these lines can account for the split in nominal features and derive all the attested patterns of concord mismatches in the Arabic noun phrase. It also provides a principled account for why other combinations of these features cannot be the target of concord mismatches. With respect to the second phenomenon, I investigate a context where the modifier displays misplaced concord. This is a context in which the modifier occurs in between two nouns [N1-A-N2] and is semantically modifying the second noun [N2]. In this position, the modifier cannot agree with the semantically modified noun [N2] and instead agrees in all features with the preceding noun [N1]. The investigation of this pattern of concord has implications for the current debate in the literature concerning the directionality of Agree and the locality condition under which Agree can take place. In this respect, I argue that Agree does not need to track semantic modification, and modifiers value their features once the locality condition for Agree to take place is met. Accordingly, I argue that modifiers in Arabic can only Agree upwardly with a structurally higher noun. I argue that this view of the Agree relation can derive all instances of concord in this language. With respect to the third phenomenon, I investigate variation in the word order of multiple (and different) postnominal modifiers in Arabic. While simple adjectives in this language are subject to ordering restrictions and therefore must precede the numeral, adjectives and participles that have PP complements or adverbial modifiers are not subject to ordering restrictions and therefore must follow the numeral. The investigation of this variation has implications for the current view in the literature concerning the derivation of postnominal modifiers, and whether or not all modifiers are derived from the same syntactic source. In this regard, I show that this variation poses a challenge for the view that takes the modifiers in Arabic to originate in a prenominal position, and that their postnominal position is derived when the noun undergoes leftward movement. I show that this view of the derivation of postnominal modifiers makes the wrong predictions for the attested distribution of postnominal modifiers in Arabic. Instead, I argue that all postnominal modifiers in Arabic originate in a postnominal position where they are right-adjoined to the noun. I also argue that postnominal modifiers in this language are derived from two different syntactic sources. While modifiers that are subject to ordering restrictions merge as non-clausal adnominals, modifiers that are not subject to ordering restrictions originate as predicates in reduced relative clauses. Accordingly, modifiers that originate as predicates merge above non-clausal modifiers, similar to the distribution of standard relative clauses. I argue that an analysis along these lines can derive the variation in the word order of postnominal modifiers in Arabic.
Keywords: split features; concord mismatches; adjective agreement; nominal modification; DP syntax; Arabic
