| Popis: |
In this thesis I aim to clarify the implications of variation and change for linguistic theory: to examine how linguists deal with variability and to show the implications of their theoretical choices. Firstly, I discuss in detail what linguistic theories are trying to achieve in terms of description and explanation, and how they should deal with variability. Much modern linguistic theory idealizes languages as invariant states, and goes beyond this to look for language universals. It is also commonly assumed that linguistics should be a study of psychological reality. I show that linguists often adopt these idealizations without explicitly discussing the relation between invariant states and variation, universals and tendencies, the psychological and the non-psychological, even when such theories are used to account for variation and change. I show how these unexamined assumptions have led to over-complex, uninsightful models and inappropriate explanations, both for states and for changes. Following this theoretical discussion, my first case study examines Optimality Theory phonology. I argue that incorrect assumptions about synchronic universals have led to unnecessary and uninsightful elaborations of the theory. Many patterns are better characterised and explained in terms of diachronic changes, and we need to consider carefully the interaction between synchrony and diachrony. As well as qualitative variation, I also examine analyses of quantitative variation, showing that they are too powerful and have difficulties with explanation. Work on quantitative variation in syntax has similar problems with statistical significance and external explanation. My second case study considers definiteness marking in Scandinavian noun phrases. I examine how three different synchronic theories (HPSG, OT and minimalist) analyse the variation between languages, and conclude that they are either too loose or too restrictive to be particularly insightful. I then trace the diachronic development of the languages and suggest that it is more revealing to take a historical view with less abstract representations, and that frequency of usage is also significant for explaining later categorical synchronic patterns. I conclude that progress in linguistics requires a broad perspective which compares alternative models and considers how they can be motivated in the light of variation, change, and external explanation. |
