• Cantonese
• microvariation
• Optimality Theory
• phonology
• sound change
• tonology

The goal of this thesis is to investigate a hypothesis of linguistic microvariation, which old as it is in its original formulation, seems to perfectly marry with Optimality Theory, which is naturally suited to handle linguistic variation: the hypothesis that linguistic distance correlates with language geographic distance. Of little interest to generative phonology, developed to analyzed the grammar of individual speakers rather than of language communities, the problem has a direct impact on linguistic cognition if cast in the right terms. Fundamental to the developing of a theory of linguistic variation and change in generative grammar are two assumptions: OT is the first theory to my knowledge, which elegantly permits to formalize and numerically quantify the once vacuous concept of 'linguistic distance'; some language changes are determined by a change in the grammar, that is the changes observed are the result of a reorganization of the grammar that happens for reasons that can be entirely explained by observing the grammar itself. The concept, is not new and it was well known in the structuralist tradition (e.g. push-pull chain), but the forces driving the change can now be identified as elements of the grammar itself (i.e. constraints), and operating in accordance with the principles of the theory (as constraint re-ranking). The possibility to measure linguistic distance and to describe language change in formal terms by using the same set of relations and primitives offers a valid tool to investigate not only their nature, but also their relation.

The analysis requires an unusual approach to OT. Typically more oriented to language typologies rather than the description of a language, since to verify the hypothesis it was necessary to compare complete grammars - or subsets of grammars - to measure their distance with an acceptable degree of accuracy, only few related varieties were studied. Tonologies are usually simpler than segmental systems with respect to processes, structural organization and number of elements and interactions. The observation is particularly true for Cantonese, a language famous for its tonal stability, the quasi-absence of tone sandhi, and most importantly for the extremely limited interaction of its tonal system with other areas of the grammar (intonation, segments, syntax etc.). Cantonese tonology, as a very independent subset of the grammar (mini-grammar), can then be used with an acceptable approximation to verify the hypothesis.

To investigate the thesis, most of the work in the thesis had to be done to flesh out the description and the differences of very closely related grammars. After a brief introduction about the history of China and of its languages (ch. 1), some basic theoretical notions about the theory used and its application to tonology are briefly outlined (ch. 2). The fundamentals of the languages investigated is given in chapter 3, which includes an introduction to the segmental system, and a first analysis of the tonematics of the languages. Central to the developing of the mini-grammar are tonal alternations (tone changes in Cantonese). These are introduced in chapter 4, followed a phonetic analysis of characteristics crucial to their phonology and microvariation (chapter 5). Chapter 6 and 7 conclude the analysis of the mini-grammars and propose an explanation of the change the suffix responsible for the tone change. Finally, chapter 8 defines the model for the computation of the linguistic distance and relates it to the observed geographic distance.