Accusative-genitive transitvity in the development of Russian : syntactic change and semantic motivation

Abstract The thesis considers the competition between the accusative and the genitive as means of case-marking direct objects in Old and Middle Russian. Data from Old Russian (roughly 1000 1400) and Middle Russian (roughly 1400 1700) show that case-marking of direct objects underwent a pronounced development. The development is analysed diachronically within the theory outlined in Harris & Campbell (1995) s Historical syntax in cross-linguistic perspective, which views all possible changes in syntax as instances of only three different mechanisms: reanalysis, extension and syntactic borrowing. In the thesis, changes in several different verb groups are shown mostly to proceed by way of extension. The predominant extension consists of the accusative case ousting the genitive case. However, during a shorter period the genitive extended as well, resulting in a temporary state of free variation between the accusative and the genitive in many of the verb groups, although not necessarily at the same time. In addition, one probable instance of reanalysis that had impact on subsequent extensions is pointed out. The thesis also demonstrates that various causal factors, especially the semantics of the verbs that changed their government pattern, have considerable impact on the extensions. Because of this, a semantic analysis of the factors motivating the changes is also provided. Two semantic theories are used: a cognitively founded semantic network model, Janda (1993) s A Geography of Case Semantics. The Czech Dative and the Russian Instrumental, and Dowty (1991) s theory of Proto-Roles "Thematic proto-roles and argument selection". Dowty proposes a prototypical patient (Proto-Patient) role accumulating the properties that characterise typical verbal objects. In the thesis, these properties are interpreted as a continuum with one accusative-triggering and one genitive-triggering pole. In accordance with this and applying the property continuum to the concrete changes, it is shown that accusative-marking tend to be used on more prototypical objects, whereas less prototypical objects are more prone to be genitive-marked. Moreover, the objects of verbs that show extensive variation as to which case they govern can be classified as something in the middle between a prototypical and a less prototypical object. A network model views the different meanings of a case as constituting a network of related senses. In accordance with the network concept, it is demonstrated how the different meanings of the two cases are related. Especially the genitive case is shown to have several slightly different meanings. The use of the genitive case is shown to be motivated in a number of contexts in which it has often been regarded as unmotivated. The model also demonstrates that the meanings of the accusative and the genitive case had many similarities in semantic content during the period, where they both functioned as means of direct object case-marking. The development in the different verb groups can partly be accounted for by referring to adjustments in a semantic network. The two semantic theories account for and explain different aspects of the development treated in the thesis. Whereas the Proto-Role theory is a useful device in obtaining an overall view of the object case-markings found in the verb groups, as well as providing possible explanations for the direction in which the development proceeded, the descriptions emerging from the network model help accounting also for some of the seemingly more diverging changes, particularly within the development of the genitive-marking