In classical generative phonology, the focus of attention fell on rules and rule interactions. These rules were language specific and were ordered with respect to one another into language specific rule hierarchies. With the advent of non-linear phonology, emphasis shifted to issues concerning representations. As representations became enriched, operations performed on them became more and more impoverished. The goal of these models has been to discover general, preferably universal, principles governing both the representations and the operations applying to them. The principles proposed are considered "hard" in the sense that they cannot be violated. Apparent violations are accounted for by setting up language particular parameters within or supplementing the principles. The main interest, however, lies in discovering the properties that all languages share with each other. With the appearance of Optimality Theory (OT) (Prince & Smolensky 1993), emphasis has shifted again. Although OT, just like non-linear approaches, advocates a 'rule-free' type of phonology, and it calls for universal constraints to constitute the grammar of particular languages, these constraints are not inviolable, that is, they are "soft". They can be violated, if there is a higher-ranked constraint that can be satisfied by violating a constraint sitting lower in the hierarchy. Although constraints are considered universal, their ranking is language specific. Since OT concentrates on the possible rankings of these constraints in different languages, emphasis lies primarily on how languages differ from each other. In my view, the non-linear approach and the OT approach are not incompatible with each other. On the contrary, they cover complementary empirical domains and my conviction is that they can only be really successful, if they are combined. It is this conviction that I will try to motivate in this course (on the basis of work done in my dissertation). On the one hand, the question of ranking (i.e. the question of which properties are considered more important than others in particular languages) is only interesting if we know what the properties are that can be ranked. That is, we need to have a theory of possible constraints and a theory of representations that these constraints apply to. Thus we need to have a theory about what is universal in the world's languages. The theory of Government Phonology (GP) (Kaye, Lowenstamm & Vergnaud 1985, 1990) will serve this purpose in the course. On the other hand, it has become clear over the years that constraints are not independent of each other, that there are situations when they come into conflict. Such conflicts can only be resolved by violating one or the other constraint, the choice of which is apparently language specific. This is where ranking becomes relevant, and where OT comes into play. And since ranking will be shown to be necessary, the next question is whether parameters are still necessary to express language variation, or whether they can be dispensed with in favour of only utilising ranking for this purpose. After briefly introducing the basic tenets of both GP and OT, I will motivate a particular type of synthesis of the two approaches, using an example involving the behaviour of consonant clusters in different languages. Then I will go on to show that this model is still too impoverished to account for certain types of wide-spread phenomena in languages of the world. More precisely, it seems that we need to retain a certain amount of "derivationalism" in the theory to be able to handle so-called derived environment effects, where processes seem to be aware of the "history" of the forms they apply to. I will propose to account for this phenomenon by incorporating into the model the lexical phonological notion of the Strict Cycle Condition, a particular way of handling phonology-morphology interaction (Kiparsky 1985). In the remaining part of the course, I will be concerned with different aspects of the complex, but wide-spread, phenomenon of vowel harmony, whereby all vowels in (roughly speaking) a word are required to agree with one another with respect to one of their properties. I will show that the combined model outlined above can handle successfully several phenomena that have been problematic for the above mentioned approaches when employed separately. Issues to be considered include raising harmony, disharmonic roots and suffixes, and the phenomenon of neutral vowels (the behaviour of those vowels in a system that lack a harmonic counterpart). Students who would like to take the course are expected to be familiar with basic notions of non-linear phonology (autosegmental phonology, syllable theory, and feature theory), those that can be obtained from a standard textbook, such as Kenstowicz 1994. Familiarity with GP and/or OT is an advantage, but it is not required.
Kaye, J., J. Lowenstamm & J.-R. Vergnaud (1985): The internal structure of
phonological elements: A theory of Charm and Government. Phonology Yearbook
2. 305-28
Kaye, J., J. Lowenstamm & J.-R. Vergnaud (1990): Constituent structure and
government in phonology. Phonology 7. 193-231
Kenstowicz, M. (1994): Phonology in Generative Grammar. Blackwell, Oxford
Kiparsky, P. (1985): Some consequences of Lexical Phonology. Phonology 2.
85-138
Polgárdi, K. (1998): Vowel Harmony. An Account in Terms of Government and
Optimality. Holland Academic Graphics, The Hague
Prince, A. & P. Smolensky (1993): Optimality Theory: Constraint Interaction
in Generative Grammar. ms. Rutgers University & University of Colorado at
Boulder