Localism versus Globalism in Morphology and Phonology

7. Discussion

7 Discussion The C1-LIN theory developed in part I of the book is a localist theory that makes explicit predictions about how (morpho)syntax and (morpho)phonology interact; these are developed with reference to contextual allomorphy , which constitutes the central empirical focus of the book. I chose this topic because this area provides significant insight into the relationship between syntax, morphology, and phonology, and because allomorphic interactions in language are highly restricted. The core proposal of the C1-LIN theory is that possible patterns of allomorphy in language are constrained by interacting cyclic and linear notions of locality. The predictions of this theory are defined and elaborated in numerous examples analyzed in chapters 2 and 3. The predictions of the C1-LIN theory (and of localist theories in general) contrast sharply with those made by theories with even a limited amount of global interaction between morphology and phonology . Allomorphy—and phonologically conditioned allomorphy in particular—provides an important test-case for comparing localist and globalist grammatical architectures, because precisely this phenomenon allows us to determine whether global properties of the phonology determine morphology. The argument in part II of the book identifies a number of phenomena that could in principle constitute evidence for globalism and against localism , and it shows that there is no evidence for the strong predictions of globalist theories. The argument has two components. First, as detailed in chapter 5, there appears to be little motivation for Phonological Selection—the idea that output phonology is crucial for allomorph selection—when systems of allomorphy are analyzed in detail. This point emerges from a number of case studies, including detailed analyses of case a‰xes in Djabugay and Yidił. The second—and more important— line of argument, advanced in chapter 6, shows that globalist theories predict interactions of a type that cannot be formulated in a localist theory: NL-application, where the factors determining allomorph choice are not local to the node undergoing insertion, and allomorphic vacillation, where allomorphs chosen for a particular Root change depending on the shape of outer, nonlocal morphemes. In case studies where these predictions could be manifested, as in di¤erent types of Latin verbal morphology, these e¤ects are not found. Instead, the key cases show patterns that are expected on the more restrictive localist view. The attested e¤ects could, of course, be modeled in a globalist theory; but since there are no cases in which the strong predictions of globalism are found, restraining such theories by imposing additional constraints to produce the correct results is missing the point. The fact that allomorphic interactions do not show global interaction between morphology and phonology, instead behaving as predicted by the localist theory, argues that the localist view of the grammar is correct. These results clearly have implications for how morphology and phonology interact. Although the point is less direct, they also have implications for phonology proper. Globalist theories of phonological interactions are competition-based; crucially, they involve competitions among complex objects, something that is ruled out in the localist theory of morphology and syntax developed in part I. Interface areas like allomorphy, where the relationship between morphology and phonology can be examined in detail, show the behavior predicted by localist theories. What does this mean, then, for approaches to phonology that employ globalist assumptions in order to implement competition? It is possible to consider di¤erent kinds of hybrid theories as a response to the results presented in this book. However, the type of phonological theory that fits most naturally with the no-competition theory of morphosyntax is a phonological theory with no competition among complex objects—that is, one in which the sound forms of complex expressions should be inextricably related to the generative procedure(s) responsible for constructing them in a localist and serialist fashion. The following sections outline further implications of the arguments presented in this book, concentrating first on programmatic implications of the two main parts of the book in sections 7.1 and 7.2. Section 7.3 returns to the theme broached immediately above: what it might mean to have a morphosyntax and a morphophonology that di¤er fundamentally in their organization. 188 Chapter 7 7.1 The Program I: Competition, Localism, Cyclicity As stressed several times in the preceding chapters, a main point of tension between localist and globalist architectures is their stance on competition . A grammar that generates multiple...