Studies on derivational morphology often assume a procedural view, emphasizing the journey from morphologically simple to morphologically complex word. We present a declarative approach with the focus on the relationship between two morphologically connected words. This more static approach enables us to better locate the generalizations present in a derivational system. We test this approach on a specific body of data, namely the formation of nouns denoting ‘person’ in Russian.
After introducing Network Morphology, the declarative framework within which we base our account, and the Russian data we will be investigating, we provide as theoretical background to our proposed analysis a sketch of the role of morphology in the Structuralist tradition, and in the more important models of the early Generativists (SectionI).
Section II is entirely devoted to Network Morphology, and acts as a short survey of the recent work carried out in this framework. The focus is on the key concepts of Network Morphology, inheritance hierarchies and the idea of defaults, and on the nature of the relations that occur within a network.
Since we emphasise the relationship between words, our commitment is to a lexeme-based approach to morphology. Section III explores the implications of this, examining Aronoff’s model, and its key elements, Word Formation Rules. We construct a set of Aronovian-style WFRs that account for Russian person derivation.
Section IV constitutes the declarative account. The proposed WFRs for Russian are given a declarative interpretation. The WFRs are viewed as generalizers of derivational information, and exceptionality is characterized as the overriding of certain of the generalizations. Our account is expressed in DATR (the appendices contain the full version) which is computable, and we have therefore been in a position to demonstrate that the theory makes the correct predictions about the data it claims to account for.
Available at: http://works.bepress.com/andrew_hippisley/1/