Subject extraction in Jordanian Arabic

Abstract

This thesis proposes a minimalist and cartographic analysis of A-bar movement in Jordanian Arabic (JA) with a particular focus on subject extraction. Adopting the Criterial Freezing approach to A-bar movement and chain formation (Rizzi 2005, 2006, 2014, Rizzi and Shlonsky 2006, 2007), it argues that Spec,SubjP in this Arabic variety is a criterial position and, hence, subject to the effects of the so-called Subject Criterion that prevents movement from this position. In order to facilitate subject extraction from root clauses, the thesis argues that JA resorts to a set of skipping strategies ruled by the postulated D(iscourse)-linking condition of the Subject Criterion which requires Spec,SubjP to be filled with an element with the same D-linking status as the extracted subject wh-word. When the subject wh-word of a root clause is D-linked, Spec,SubjP is filled with the D-linking element ʔilli. The thesis also shows that Spec,SubjP in such cases may alternatively be filled with a deictic temporal/locative adjunct. Deictic temporal adjuncts may fill Spec,SubjP, regardless of the type of the verb used (i.e. transitive, unergative, or unaccusative), whereas deictic locative adjuncts only fill Spec,SubjP in questions with unaccusative verbs. The thesis shows that this discrepancy is due to the effects of the Phase Impenetrability Condition (PIC) (Chomsky 2000, 2001, 2005), which blocks Subj° from probing goals within the complement of v*P. The study provides evidence to the effect that locative adjuncts are adjoined to VP, whereas temporal adjuncts are adjoined to TP, something that makes them immune to the effects of the PIC. On the other hand, when the subject wh-word is not D-linked, Spec,SubjP is filled with an expletive pro. In pursuit of exploring subject extraction from embedded clauses (introduced by the complementizer ʔinn ‘that’), the thesis explores the derivations of the possible word orders used in such clauses. It also provides an account of the bound forms attached to the complementizer ʔinn, arguing that such forms are better treated as inflectional suffixes whose PF form is a consequence of the locality-ruled Agree relation between C° ʔinn and the closest c-commanded visible DP. Contra Chomsky’s (2007) feature inheritance, the present thesis assumes that C° in JA retains its uΦ-features, while T° is separately endowed with a bundle of uΦ-features, given its positive setting of the postulated T°-Φ parameter (i.e. Tº is endowed with Φ-features). iv Additionally, the thesis shows that factivity is a key factor that determines the possibility of (subject) extraction from embedded contexts or lack thereof. Unlike the clauses embedded under a nonfactive predicate such as jifkkir ‘to believe’, no extraction is possible out of clauses embedded under a factive predicate such as jiħzan ‘to regret’. A full analysis of subject extraction from nonfactive complements is provided, and the relevant observations such as the impossibility of A-bar movement of some elements within the same clause while the subject of the embedded clause is extracted are accounted for, using the feature-based approach to locality (Starke 2001). As for the ban against extraction out of factive complements, the thesis argues that such clauses are DPs, headed by a null determiner. In so doing, the thesis provides substance to the Kiparskian stand that the structural difference between factive and non-factive complements lies in subcategorization of the matrix verb. At the same time, the proposed analysis challenges several recent approaches to factive complements that have argued either for a reduced left periphery for factive complements (e.g. Haegeman 2006, de Cuba 2007) or for the presence of an operator that has the effect to block movement out of these clauses (e.g. Zubizaretta 2001, Starke 2004, Haegeman 2012).