Tigrinya Applicatives in Lexical-Functional Grammar
MetadataVis full innførsel
The primary goal of this thesis is to describe and analyze applicative constructions in Tigrinya. An applicative construction is characterized by a verb that bears an affix for an argument that either has a semantic role that is not normally entailed by the lexical meaning of the base verb or is specified as a peripheral argument. The choice of an applicative expression is motivated by semantic and discourse factors. Applicatively expressed arguments are associated with referents that possess high semantic prominence and discourse salience. The applicative phenomenon is viewed as a morphosyntactic strategy that introduces a core object function that is salient in the discourse event described by the verb. The theoretical motivation of this study is to explore the conditions that trigger object marking in Tigrinya so as to examine the semantic, functional and discourse properties of objects. The main theoretical framework used in this research is Lexical-Functional Grammar (LFG). This formalism assumes that the different linguistic information pertaining to functional, semantic and discourse structures can be modeled as interrelated parallel representations. Formal syntactic theories classify applicative constructions into symmetrical and asymmetrical types according to the grammatical properties of objects coded in double object applicative clauses. Grammatical properties such as adjacency to the verb, pronominal marking and passivization are posited as parameters of object variation, and are assumed to be characteristic of the sole object of monotransitive clauses. In double object constructions object arguments that are implicated in these structures are regarded as possessing primary object properties. A theory of object asymmetries which predicts the patterns of objects across languages based on applicative data predominantly from the Bantu languages assumes that in asymmetrical applicatives only one object argument, and most likely the applied object, displays primary object properties, whereas in symmetrical applicatives both object arguments, the base object of the verb and the applied object, display such properties (Bresnan and Moshi 1990, Alsina and Mchombo 1993). The object argument with primary object properties is assigned an OBJ grammatical function, and the argument object that lacks such traits is assigned an OBJ grammatical function. The theory of object asymmetries is formulated within LMT, the sub-theory of LFG which deals with the mapping patterns of semantic arguments to grammatical functions. In this theory it is maintained that only one OBJ function can be realized in a clause, but there can be several restricted objects. Even though there can be two primary object arguments in symmetrical applicatives, only one of them can be analyzed as OBJ because this function must be unique in the clause. Due to this restriction, there is no difference in the analysis of object functions in asymmetrical and symmetrical applicative types; in both applicative types the two objects are analyzed as OBJ and OBJ . The Tigrinya data dealt with in this research indicate that the grammatical diagnostics that are posited to distinguish between symmetrical and asymmetrical objects do not converge into a single primary object property. In some double applicative constructions, objects reflect asymmetrical properties, and in others objects symmetrical properties. In asymmetrical applicatives the primary object properties are not correlated to differentiate between the applied and the base objects. With respect to some of the grammatical tests the applied object displays the opposite properties to what is predicted by LMT. Moreover, the Tigrinya data suggest that the classification of objects as OBJ and OBJ cannot capture the similarity displayed by symmetrical objects. In Tigrinya, objects are coded in a complex interplay of word order, case marking and pronominal indexation which cannot be properly accounted for by the binary system proposed in LMT which assumes a straightforward contrast between objects. In this study we adopt the general constraint system in LFG in order to capture the different conditions on word order, case marking and pronominal indexation by which objects are distinguished. In addition, this research proposes that overt object coding cannot be regarded as a manifestation of primary objecthood in Tigrinya, since marked objects may or may not display genuine patient-like properties, but pronominally marked objects are all unified in their semantic and discourse properties.