Chapter 2: A framework for morphological defaults

In this chapter we introduce some foundational distinctions of Network Morphology in §2.1 and show in §2.2 how Network Morphology treats the lexeme as a locus of parallel information, where semantic, syntactic, phonological and morphological information are combined. This also means that there is potential for slippage between morphological knowledge and other linguistic levels so that it can become autonomous. In §2.3 we consider the different types of default generalization which Network Morphology makes possible. We emphasize the ability to model cross-linguistic tendencies with attribute ordering, in particular in terms of the predictions they make about neutralization and syncretism. We also show that Network Morphology has a flexible paradigm signature which can account for apparently different types of phenomena: marginal features on the one hand, and splits in the paradigm on the other. We then explain the role of normal and exceptional case defaults. The former are used for the general case that normally applies, whereas the latter are used as the last resort. Having considered these different types of default within morphology we move on to look at the default relationship with syntax, which naturally leads to a typology for morphological autonomy in §2.4, starting from a non-autonomous situation where the morphological hierarchy is isomorphic with the lexemic hierarchy, and therefore eliminable, and moving to a situation where inflectional classes create structure which is not reflected in the lexemic hierarchy. The framework, therefore, allows for different degrees of morphological autonomy and imposes constraints on possible morphological systems. We illustrate how Network Morphology treats the relationship with syntax in §2.4 and summarize in §2.5.