A Functional Reanalysis of Object Clitic Doubling in Persian

Nichols's (1986) typological parameterization of languages as head versus dependent marking opened a horizon in language typology and linguistic research. Dependent-marking languages are those which morphologically encode dependency between verb and its argument(s) by grammatical markings on the latter. In head-marking languages, on the other hand, the dependency is displayed via bound forms hosted by the verb. This parameterization is not categorical however, as some languages utilize both strategies to encode grammatical relations, which is referred to as double marking or locus. Persian exhibits head- and dependent-marking features at the clause level, situating itself then within the category of double-marking languages. In close connection with this, Persian allows the possibility of co-occurrence between a ra-marked direct object and a cross-referencing bound form, which is tentatively called clitic doubling. The study sought to argue that the co-occurrence of a ra-marked object and a co-referential clitic can be representative of two information-structurally distinct structures. One is referred to as clitic doubled-object construction and the other as left-dislocated object construction. Diachronically, it is assumed that the latter sets the stage for the appearance of the former, and the two constructions co-exist in modern Persian.

Theoretical Framework

Object clitics optionally appear on transitive verbs in Persian. This resurrects the classical controversy concerning the agreement versus argument status of bound forms on a par with their controlling reference phrases. Haspelmath (2013) presented four dominant views in the literature with reference to this issue: the virtual agreement view, the bound-argument view, the dual nature view, and the double-expression view. Under the virtual-agreement view, bound forms are uniformly considered purely as agreement markers even if their controllers are absent. This is normally taken in generative approaches under the rubric of “pro-drop” or “null subject”. Under the bound-argument view, bound forms are treated in any event as true arguments; no matter their controllers are present, in which case they hold an appositive/adjunct status to the verb. The presence or absence of controllers are the key solution to the agreement or argument status of bound forms in the dual-nature view. Eventually, the dual-expression view allows an argument to be expressed twice. In line with Haspelmath (2013), we regard bound forms to perform as pro-indexes, cross-indexes or gramm-indexes. Pro-indexing is defined in terms of the complementary distribution of bound forms with their corresponding RPs, such that they never occur within the same clause. This implies that a bound form can co-occur with its controlling RP within the same sentence, in which case the RP is a dislocated topic. Bound forms and their optional controlling reference phrases are authorized to be in the same clause under a cross-indexing view, leading to the consideration of reference phrases as co-nominals. Gramm-indexing is akin to what Siewierska (1999) terms as grammatical agreement, where the obligatory presence of a co-nominal to bind a gramm-index within the same clause is vital (e.g. -s in English).

This study is a theory-based investigation that tackles the role of information structure in the formation of the clitic doubled-object and left-dislocated object construction. Hence, some random sentences and question-answer pairs are presented from colloquial Persian in order to fathom out how the topic-focus articulation of an utterance characterizes the information structure in the given constructions.

It becomes clear that first, clitic doubling terminologically is a misnomer and can be representative of two distinct grammatical constructions, following Haspelmath's nomenclature. The pro-indexing construction is identified when the controlling reference phrase is outside the boundary of the clause, either within the same sentence or in the preceding discourse. The cross-indexing construction is recognizable when the controlling reference phrase/co-nominal and the clitic, indexing its feature bundles, are allowed to occur in the same clause. Second, the pro-indexing construction is topicality-motivated in the sense that, the controlling reference phrase needs to be a topic of the sentence/discourse. We maintain that pro-indexing is incompatible with the trifold focus taxonomy in Lambrecht (1994) and occurs only when the object referent constitutes part of the pragmatic presupposition. In contrast, the cross-indexing is a focus-driven operation and can be accounted for in the light of laying extra-emphasis an object with referential properties. Third, we argue that the cross-indexing construction historically evolves from the pro-indexing construction in the following fashion: a) The object referent as a dislocated nominal separate via hanging-topic constructions from the clause by an intonational pause whilst being pro-indexed by a co-referential clitic inside it; b) -Ra starts appearing on the dislocated object by analogy of its use with oblique and possessor roles, dating back to Middle Persian (Hopper & Traugott, 2003); c) The intonational pause disappears and the dislocated object incorporates into the clause, leading to the emergence of the cross-indexing construction. This explains why the pro-indexing and the cross-indexing constructions co-exist in modern Persian from a synchronic perspective.

We conclude that first, the simultaneity of the pro-indexing and cross-indexing construction confirms the reanalysis of the Persian -ra as an object marker. -Ra as a topic marker appeared initially on dislocated objects in the pro-indexing formats, which concomitantly paved the way for its companionship with clause-internal objects in cross-indexing formats. This is by no means new and cross-linguistically attested. Second, the birth of the cross-indexing construction out of the pro-indexing construction lends further credence to the emphatic nature of the former on the grounds that, the presence of a controlling reference phrase and its corresponding cross-index in the same clause is viewed as providing additional information helping to identify the referent in case the indications given by the cross-index and by the context are not sufficient (Siewierska, 1999; Creissel, 2001). Third, the co-existence of the pro-indexing and cross-indexing mediates the two opposing views in the literature. Some researchers favor object doubling (Rasekh Mahand, 2006) and some, on the contrary, defy it and support the view that those sentences which seem to contain instances of object doubling indeed represent what can be conveniently taken as object dislocation (Ghomeshi, 1997).